Magnetic Resonance Angiography: Principles and Applications

Magnetic Resonance Angiography

wwwwwwwwwwww

James C. Carr, MD Timothy J. Carroll, PhD Editors

Magnetic Resonance Angiography Principles and Applications

Editors: James C. Carr, MD Director of Cardiovascular Imaging Associate Professor of Radiology and Medicine Northwestern University Feinberg School of Medicine Chicago, IL USA [email protected]

Timothy J. Carroll, PhD Associate Professor of Biomedical Engineering Director of MRI Research Department of Radiology Northwestern University Chicago, IL USA [email protected]

ISBN 978-1-4419-1685-3 e-ISBN 978-1-4419-1686-0 DOI 10.1007/978-1-4419-1686-0 Springer New York Dordrecht Heidelberg London Library of Congress Control Number: 2011940427 © Springer Science+Business Media, LLC 2012 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. While the advice and information in this book are believed to be true and accurate at the date of going to press, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

“I would like to dedicate this to my wife, Jean and my son, Nate, who taught me to love books.” Timothy J. Carroll “I would like to dedicate this book to my wife, Maria, and children, Cristian and Sebastian, without whose constant support, dedication and inspiration, this work would not be possible.” James C. Carr

wwwwwwwwwwww

Preface

It is with tremendous enthusiasm and excitement that we introduce the book Magnetic Resonance Angiography: Principles and Applications to those who are interested in and involved with the ever-expanding field of magnetic resonance angiography (MRA). Vascular disease remains to this day the central cause for many serious clinical conditions, such as stroke, myocardial infarction, and peripheral arterial disease, each of which can precipitate further downstream complications, which can ultimately lead to serious morbidity and death in many individuals. While huge progress has been made to develop novel therapies for vascular pathologies, such as minimally invasive endovascular stents and stem cell therapies, much of these approaches still depend on an accurate depiction of the appearance and extent of abnormalities within the blood vessels. The original gold standard for diagnosing vascular disease was catheter-based digital subtraction angiography, which of course is invasive for the patient and can result in significant complications, albeit rare with current technology. The concept of inserting a needle directly into an artery so that iodinated contrast can be injected rapidly to opacify blood vessels under X-ray visualization will seem barbaric to readers of this textbook. In fact, it was not long ago that the accepted standard for diagnosing peripheral vascular disease was translumbar aortography, where the abdominal aorta was directly accessed percutaneously for diagnostic angiographic purposes. While such diagnostic techniques seem prehistoric and dated in the modern era, it is also a testament to how much progress has been made in the area of noninvasive vascular diagnosis that we have this attitude. Given the invasive alternatives, much effort has been spent on developing alternative noninvasive diagnostic techniques for assessing vascular disease. One of these tools, Doppler ultrasound, is used in routine clinical practice today to assess conditions, such as carotid artery stenosis and peripheral vascular disease. Ultrasound has the advantage of being noninvasive, cheap, and easily portable; however, it is most useful for superficial vessels, not having the penetration to image deeper structures, such as the intracranial vasculature or pulmonary circulation. Computed tomography (CT) is one of the most common diagnostic tools used today in medicine and can also be employed to image the vasculature. CT has achieved particular success with the recent development of multidetector scanners that allow rapid acquisition speeds at spatial resolutions that approximate digital subtraction angiography. While CT is quick and easy to operate, two major drawbacks include exposure to ionizing radiation and the requirement of injected potentially nephrotoxic-iodinated contrast for vessel opacification. Finally, magnetic resonance imaging (MRI), which is also used in all areas of medicine today, is ideally suited for imaging the vasculature. It is a noninvasive imaging tool too; however, in contrast to CT and catheter-based angiography, it does not use ionizing radiation and gadolinium, which is used as the contrast agent, is relatively nontoxic compared to iodinated agents. While MRI also has some disadvantages, such as not being able to scan patients with devices, such as pacemakers, it is the only modality that can combine anatomic depiction with functional assessment in the same study.

vii

viii

MRI emerged as a vascular imaging tool about 30 years ago with the discovery of time-offlight (TOF) imaging, where it became apparent that, as blood flowed through an external magnetic field, it emitted a signal that could be used for imaging purposes. TOF was the original noncontrast MRA technique; however, scan times were long and images were plagued by flow artifacts that could result in misdiagnosis. The technique has benefited from acceleration strategies developed over the years and is still used routinely in the intracranial circulation. The field of MRA really took off in the 1990s with the development of contrast-enhanced MRA. Contrast-enhanced MRA relied on an intravenous injection of a gadolinium-based extracellular contrast agent which, when imaged with a T1-weighted gradient echo pulse sequence, produced bright images of the blood vessels with suppression of the background tissues. The principal impact of contrast-enhanced MRA was that it produced images that were similar to conventional angiograms in very short periods of time. Much of the development in the field of MRA over the following decade focused on improving acquisition speed and spatial resolution for contrast-enhanced MRA. More recently, a condition known as nephrogenic systemic fibrosis (NSF) was described and its cause was linked to patients with renal dysfunction who were exposed to high doses of gadolinium. A lot more is known about NSF now and the condition has nearly been eliminated with better screening of patients and using lower doses of gadolinium. One less talked about consequence was that the field of MRA diverted its attention away from contrast-enhanced MRA to the development of noncontrast MRA techniques. The result was the development of a myriad of new techniques over the last few years that has become confusing to practicing clinicians and scientists in the field. Magnetic Resonance Angiography: Principles and Applications is designed to bring together into a single textbook all of the different MRA techniques, both contrast-enhanced and noncontrast, current contrast agents and implications for NSF, and strategies for applying these techniques in different clinical situations. The book is targeted to physicists, physicians (particularly those specializing in imaging), MRI technologists, residents, fellows, and students, both doctoral and postdoctoral. The book does not claim to have all of the answers and reflects a snapshot of current thinking in the field. Already there are new developments on the horizon. However, we hope to be able to provide a concrete basis for comprehending current MRA techniques and the clinical protocols in which they are applied so that new techniques can be more easily understood. With this objective in mind, we have divided the textbook into two parts. Part I (Chaps. 1–16) is focused on MRA techniques and part II (Chaps. 17–29) is focused on clinical applications. Part I begins with a review of MRI physics as it pertains to MRA. There are chapters devoted to all of the main MRA techniques, including contrast-enhanced MRA, time-of-flight, and phase contrast. There are several chapters devoted to newer noncontrast MRA techniques. A couple of specific areas, such as time-resolved angiography and coronary MRA are addressed independently. We have also attempted to describe in more detail-specific topics, such as highfield MRA, susceptibility-weighted imaging, acceleration strategies, such as parallel imaging, vessel wall imaging, targeted contrast agents, and low-dose contrast-enhanced MRA. Part II encompasses all of the clinical applications for MRA. Each chapter is divided into an initial “techniques” part, which describes the MRA techniques and protocols for that disease and vascular territory, and an “applications” part, which describes the pathology and imaging findings relevant to the disease state being discussed. There may be some repetition of techniques previously described in part I, although not with the same degree of detail. It is hoped that the techniques and protocols discussed will provide a foundation for the reader to develop his/her own protocols. We have deliberately avoided providing “canned” protocols, which may be somewhat restrictive given the numerous MRA techniques currently available. The “applications” part is designed to provide a comprehensive description of different pathologies together with MRA imaging findings. This should be particularly useful to physicians in practice, residents, and fellows in training. We have devoted specific chapters to NSF and its implications for contrast-enhanced MRA, MRI contrast agents, and newer topics, such as interventional MRI.

Preface

Preface

ix

Finally, we would like to thank all of the contributors to this book, without whom this text would not be possible. Each author is a highly respected expert in the field of MRA and this book would not have been feasible without their contribution and dedication. We would like to thank all our colleagues in the MRI community whose inspiration, support, and friendship have been invaluable. We would specifically like to thank our respective mentors, Dr. Paul Finn, MD, and Dr. Charles Mistretta, PhD, whose past and ongoing support has provided guidance and encouragement for many years. We would like to thank Springer for bringing this book to fruition, specifically Jennifer Donnelly and Frances Louie, whose patience and effort have been immeasurable. Chicago, IL Chicago, IL

James C. Carr, MD Timothy J. Carroll, PhD

wwwwwwwwwwww

Contents

Part I MRA Techniques 1

Basic Principles of MRI and MR Angiography .................................................. Frank R. Korosec

3

2

Time-of-Flight Angiography ................................................................................. Seong-Eun Kim and Dennis L. Parker

39

3

Phase-Contrast MRI and Flow Quantification ................................................... Bernd Jung and Michael Markl

51

4

Technical Aspect of Contrast-Enhanced MRA ................................................... Honglei Zhang, Wei Zhang, and Martin R. Prince

65

5

Time-Resolved, Contrast-Enhanced MR Angiography Using Cartesian Methods ...................................................................................... Stephen J. Riederer, Clifton R. Haider, Casey P. Johnson, and Petrice M. Mostardi

75

6

Flow-Dependent Noncontrast MR Angiography ................................................ Mitsue Miyazaki, Satoshi Sugiura, Yoshimori Kassai, Hitoshi Kanazawa, Robert Edelman, and Ioannis Koktzoglu

89

7

Low-Dose Contrast-Enhanced MR Angiography ............................................... Kambiz Nael, Roya Saleh, Gerhard Laub, and J. Paul Finn

107

8

Vessel Wall Imaging Techniques ........................................................................... Rui Li, Niranjan Balu, and Chun Yuan

113

9

Noncontrast Coronary Artery Imaging ............................................................... Allison Hays, Robert G. Weiss, and Matthias Stuber

129

10

Contrast-Enhanced MR Angiography of the Coronary Arteries ...................... Qi Yang and Debiao Li

141

11

MR Angiography and High Field Strength: 3.0 T and Higher.......................... Harald H. Quick and Mark E. Ladd

149

12

Susceptibility Weighted Imaging and MR Angiography.................................... Samuel Barnes and E. Mark Haacke

157

13

Non-Cartesian MR Angiography ......................................................................... Walter Block and Oliver Wieben

169

xi

xii

Contents

14

Parallel Imaging in Angiography ......................................................................... Nicole Seiberlich and Mark Griswold

185

15

Targeted Agents for Wall Imaging ....................................................................... Emily A. Waters and Thomas J. Meade

199

Part II

Clinical Applications

16

Intracranial Arterial and Venous Disease ........................................................... Dariusch R. Hadizadeh, Horst Urbach, and Winfried A. Willinek

213

17

Carotid and Vertebral Circulation: Clinical Applications ................................. Sugoto Mukherjee and Max Wintermark

225

18

Thoracic Aorta ....................................................................................................... Emily Ward and James C. Carr

239

19

Pulmonary MRA .................................................................................................... James F.M. Meaney and Peter Beddy

253

20

Abdominal Aorta and Mesenteric Vessels ........................................................... Klaus D. Hagspiel and Patrick T. Norton

269

21

Renal Vascular Diseases ........................................................................................ Tim Leiner and Henrik Michaely

283

22

MRA: Upper Extremity and Hand Vessels .......................................................... Ruth P. Lim and Vivian S. Lee

297

23

Lower Extremity Peripheral Arterial Disease..................................................... Jeremy D. Collins and Timothy Scanlon

319

24

Noninvasive Imaging for Coronary Artery Disease ............................................ Reza Nezafat, Susie N. Hong, Peng Hu, Mehdi Hedjazi Moghari, and Warren J. Manning

337

25

Venous Imaging: Techniques, Protocols, and Clinical Applications ...................................................................................... Amir H. Davarpanah, Philip Hodnett, Jeremy D. Collins, James C. Carr, and Tim Scanlon

351

26

Pediatric MR Angiography: Principles and Applications .................................. Bharathi D. Jagadeesan and David N. Loy

365

27

Contrast Agents for MR Angiography................................................................. Christoph U. Herborn

381

28

CE-MRA in the Age of Nephrogenic Systemic Fibrosis ..................................... Aditya Bharatha, Sean P. Symons, and Walter Kucharczyk

387

29

Emerging Interventional MR Applications ......................................................... Clifford R. Weiss, Aravindan Kolandaivelu, Jeff Bulte, and Aravind Arepally

395

Index ................................................................................................................................

403

Contributors

Aravind Arepally, MD, FSIR Division of Radiology, Piedmont Healthcare, Atlanta, GA, USA Niranjan Balu, PhD Department of Radiology, University of Washington, Seattle, WA, USA Samuel Barnes, MS Research Assistant, Department of Radiology, Loma Linda University Medical Center, Loma Linda, CA, USA Peter Beddy Consultant Radiologist, St. James’s Hospital, Dublin, Ireland Aditya Bharatha, MD, FRCP(C) Neuroradiology Fellow, Department of Medical Imaging, University Health Network, Toronto, ON, Canada Walter Block, PhD Associate Professor, Departments of Biomedical Engineering, Medical Physics, and Radiology, Wisconsin Institute for Medical Research, University of Wisconsin-Madison, Madison, WI, USA Jeff Bulte, PhD Professor and Director, Russell H. Morgan Department of Radiology and Radiological Science, The Johns Hopkins University School of Medicine, Baltimore, MD, USA Jeremy D. Collins, MD Assistant Professor of Radiology, Department of Radiology, Northwestern Memorial Hospital and Northwestern University Feinberg School of Medicine, Chicago, IL, USA Amir H. Davarpanah, MD Department of Radiology, Yale School of Medicine, New Haven, Connecticut Robert R. Edelman, William B. Graham Chairman, Northshore University Health System, Evanston IL, USA J. Paul Finn, MD Professor of Radiology, Medicine and Biomedical Physics, Department of Radiology, Ronald Reagan UCLA Medical Center, Los Angeles, CA, USA Mark Griswold, PhD Associate Professor, Department of Radiology, University Hospitals of Cleveland/Case Western Reserve University, Cleveland, OH, USA E. Mark Haacke, PhD Director, MR Research Facility, Department of Radiology, Harper Hospital/Wayne State University, Detroit, MI, USA Dariusch R. Hadizadeh, MD Department of Neuroradiology, University of Bonn, Bonn, Germany

xiii

xiv

Klaus D. Hagspiel, MD Professor of Radiology, Cardiology and Pediatrics, Director, Division of Noninvasive Cardiovascular Imaging, Department of Radiology, University of Virginia Health System, Charlottesville, VA, USA Clifton R. Haider, PhD Department of Radiology, Mayo Clinic, Rochester, MN, USA Allison Hays, MD Assistant Professor of Medicine, Division of Cardiology, Department of Medicine, Johns Hopkins Hospital, Baltimore, MD, USA Christoph U. Herborn, MD, MBA Associate Professor of Radiology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany Philip Hodnett, MD Department of Cardiovascular Imaging, Northwestern University Feinberg School of Medicine, Chicago, IL, USA Department of Radiology, New York University, NY Susie N. Hong, MD Advanced Cardiac Imaging Fellow, Cardiovascular Division, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA Peng Hu, PhD Assistant Professor, Department of Radiology, Ronald Reagan Medical Center, Los Angeles, CA, USA Nobuyashu Ichinose, MS Senior Specialist, MRI Systems Development Department, Toshiba Medical Systems Corporation, Otawara, Japan Bharathi D. Jagadeesan, MD Endovascular Surgical Neuroradiology Fellow, Washington University School of Medicine, Barnes-Jewish Hospital, Mallinckrodt Institute of Radiology, Saint Louis, MO, USA Casey P. Johnson Department of Radiology, Mayo Clinic, Rochester, MN, USA Bernd Jung, PhD Department of Radiology, Medical Physics, University Hospital Freiburg, Freiburg, Germany Hitoshi Kanazawa, MS Senior Manager, MR Engineering, MRI Systems Development Department, Toshiba Medical Systems Corporation, Otawara, Japan Yoshimori Kassai, MS Group Manager, MRI Systems Development Department, Toshiba Medical Systems Corporation, Otawara, Japan Seong-Eun Kim, PhD Research Associate, Department of Radiology, Utah Center for Advanced Imaging Research, Salt Lake City, UT, USA Ioannis Koktzoglou, PhD Assistant Professor of Radiology, The University of Chicago, Northshore University Health System, Evanston IL, USA Aravindan Kolandaivelu, MD, BS Assistant Professor of Medicine, Division of Cardiology, Cardiac Arrythmia Service, Johns Hopkins Hospital, Baltimore, MD, USA Frank R. Korosec, PhD Professor of Radiology and Medical Physics, Department of Radiology, University of Wisconsin Hospital and Clinics, Madison, WI, USA Walter Kucharczyk, MD, FRCP(C), FIASTUM Professor, Departments of Medical Imaging and Surgery; Director, MRI and Spectroscopy, Joint Department of Medical Imaging, Senior Scientist; Hans Fischer Senior Fellow, Institute of Advanced Studies, University of Toronto, Toronto, ON, Canada Toronto General Research Institute, Toronto, ON, Canada Technical University of Munich, Munich, Germany Department of Medical Imaging, University Health Network, Toronto, ON, Canada

Contributors

Contributors

xv

Mark E. Ladd, PhD Director, Erhan L. Hahn Institute for MRI, University Duisburg-Essen, Essen, Germany Gerhard Laub, PhD Director, MR R&D West, MR Division, Siemens Healthcare USA, Pleasanton, CA, USA Vivian S. Lee, MD, PhD, MBA Vice Dean of Science, Professor of Radiology, Physiology and Neurosciences, Department of Radiology, New York University Langone Medical Center, New York, NY, USA Tim Leiner, MD, PhD Associate Professor of Radiology, Department of Radiology, Utrecht University Medical Center, Utrecht, The Netherlands Debiao Li, PhD Director, Cedars-Sinai Medical Center, Biomedical Imaging Research Institute, Los Angeles, CA, USA Rui Li, PhD Senior Fellow, Radiology Department, University of Washington, Seattle, WA, USA Ruth P. Lim, MBBS, MMed, FRANZCR Assistant Professor, Department of Radiology, New York University Langone Medical Center, New York, NY, USA David N. Loy, MD, PhD Instructor, Washington University School of Medicine, Barnes-Jewish Hospital, Mallinckrodt Institute of Radiology, Saint Louis, MO, USA Warren J. Manning, MD Cardiovascular Division, Department of Medicine, Department of Radiology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA Michael Markl, PhD Department of Radiology, Medical Physics, University Hospital Freiburg, Freiburg, Germany Director of Cardiovascular MR Research, Associate Professor of Radiology and Biomedical Engineering, Northwestern University Feinberg School of Medicine, Chicago, IL, USA Thomas J. Meade, PhD Eileen M. Foell Professor of Chemistry, Biochemistry and Molecular and Cell Biology, Neurobiology and Physiology, and Radiology, Department of Chemistry, Northwestern University, Evanston, IL, USA James F.M. Meaney, MB, FRCR, FFR Professor, Department of Radiology, Trinity College Dublin, St. James’s Hospital, Dublin, Ireland Henrik Michaely, MD Associate Professor of Radiology, Section Chief of Vascular and Abdominal Imaging, Institute of Clinical Radiology and Nuclear Medicine, University Medical Center Mannheim, Mannheim, Germany Mitsue Miyazaki, PhD Senior Fellow, MRI Department, Toshiba Medical Research Institute, Vernon Hills, IL, USA Mehdi Hedjazi Moghari, PhD Research Fellow, Department of Medicine, Cardiovascular Division, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA Petrice M. Mostardi Department of Radiology, Mayo Clinic, Rochester, MN, USA Sugoto Mukherjee, MD Assistant Professor, Section of Neuroradiology, Department of Radiology, University of Virginia Health System, Charlottesville, VA, USA Kambiz Nael, MD Radiology Resident, Department of Radiological Sciences, Ronald Reagan UCLA Medical Center, Los Angeles, CA, USA

xvi

Reza Nezafat, PhD Assistant Professor of Medicine, Director of Translational Cardiovascular Imaging Program, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA Patrick T. Norton, MD Assistant Professor of Radiology, Cardiology and Pediatrics, Division of Noninvasive Cardiovascular Imaging, Department of Radiology, University of Virginia Health System, Charlottesville, VA, USA Dennis L. Parker, PhD Director, Department of Radiology, Utah Center for Advanced Imaging Research, Salt Lake City, UT, USA Martin R. Prince, MD, PhD Professor of Radiology, Department of Radiology, Weill Medical College of Cornell University, New York, NY, USA Harald H. Quick, PhD Director of MR Imaging, Institute of Medical Physics, University Erlangen-Nürnberg, Erlangen, Germany Stephen J. Riederer, PhD Professor, Department of Radiology, Mayo Clinic, Rochester, MN, USA Roya Saleh, MD Researcher, Department of Radiology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA Timothy Scanlon, MD, MRCPI, FFR RCSI Department of Cardiovascular Imaging, Radiology, Northwestern University Feinberg School of Medicine, Chicago, IL, USA Consultant Radiologist, Limerick Regional Hospital, Ireland Nicole Seiberlich, PhD Research Associate, Department of Radiology, University Hospitals of Cleveland/Case Western Reserve University, Cleveland, OH, USA Matthias Stuber, PhD Professor and Director, Department of Radiology, Center for Biomedical Research, University Hospital Lausanne, Lausanne, Switzerland Satoshi Sugiura Chief Specialist, MRI Systems Development Department, Toshiba Medical Systems Corporation, Otawara, Japan Sean P. Symons, MD, MPH, FRCP(C) Neuroradiologist, Department of Medical Imaging, Sunnybrook Health Sciences Centre, Toronto, ON, Canada Horst Urbach, MD Professor, Department of Radiology/Neuroradiology, University of Bonn, Bonn, Germany Emily Ward, MB, BCh, BAO Department of Radiology, Northwestern Memorial Hospital, Chicago, IL, USA Emily A. Waters, PhD Department of Chemistry, Northwestern University, Evanston, IL, USA Clifford R. Weiss, MD Assistant Professor of Radiology and Surgery, Division of Vascular and Interventional Radiology, Department of Radiology, Johns Hopkins Hospital, Baltimore, MD, USA Robert G. Weiss, MD Professor of Medicine and Radiology, Department of Medicine, Johns Hopkins Hospital, Baltimore, MD, USA Oliver Wieben, PhD Assistant Professor, Departments of Medical Physics and Radiology, University of Wisconsin-Madison, Wisconsin Institute for Medical Research, Madison, WI, USA Winfried A. Willinek, MD Associate Professor, Department of Radiology, University of Bonn, Bonn, Germany

Contributors

Contributors xvii

xvii

Max Wintermark, MD, MAS Associate Professor of Radiology, Neurology, Neurological Surgery and Biomedical Engineering, Department of Radiology, University of Virginia, Charlottesville, VA, USA Qi Yang, MD, PhD Attending Radiologist, Department of Radiology, Xuanwu Hospital, Capital Medical University, Beijing, China Chun Yuan, PhD Professor, Department of Radiology, University of Washington, Seattle, WA, USA Honglei Zhang, MD Assistant Professor, Department of Radiology, Weill Medical College of Cornell University, New York, NY, USA Wei Zhang, MD Research Fellow, Department of Radiology, Weill Medical College of Cornell University, New York, NY, USA

wwwwwwwwwwww

Part I MRA Techniques

wwwwwwwwwwww

1

Basic Principles of MRI and MR Angiography Frank R. Korosec

Introduction Magnetic resonance imaging (MRI) is a very versatile and useful imaging modality. It is capable of providing a wealth of diagnostic information, including information regarding blood flow. Magnetic resonance methods that provide images of the arteries are referred to as magnetic resonance angiography (MRA) methods. A variety of MRA methods exist. These methods can be categorized as noncontrast-enhanced and contrast-enhanced methods. The noncontrast-enhanced methods rely on the motion of blood (phase-contrast MRA, TOF MRA) or the magnetic properties of blood (steady-state free precession MRA) to differentiate signals from blood and stationary tissues, whereas contrast-enhanced methods require the intravenous injection of a contrast material to differentiate signals from blood and stationary tissues. In addition to providing information regarding the morphology of the blood vessels, some MRA methods can provide information regarding blood velocity or volume flow rates. If the MRA acquisitions are cardiac-gated, they can provide information regarding changes in blood velocity or volume flow rate throughout the cardiac cycle. Most MRA methods permit large volumes to be imaged, and the image sets may be retrospectively processed to provide observation of vessels from any perspective. Because MRI provides high-contrast images of soft tissues, MRI exams may be performed together with MRA exams to yield information regarding blood flow and target organ status in the same exam. In addition to providing information regarding vascular and tissue anatomy and blood flow, MRI can be used to obtain information regarding diffusion and perfusion (and a host of other qualities), making it a very

F.R. Korosec, PhD () Department of Radiology, University of Wisconsin Hospital and Clinics, E3/311, 600 Highland Avenue, Madison, WI 53792-3252, USA e-mail: [email protected]

effective modality for obtaining a comprehensive assessment of vascular disease. There are a great number of MRA methods, and each derives vascular signal by employing features and imaging parameters that take advantage of differences in physical properties between blood and stationary tissues. In order to most effectively utilize the MRA sequences and reap their greatest benefits, it is essential to have a good understanding of the principles of MRI; knowledge of the strengths, limitations, and capabilities of the different MRA methods; and comprehension of how the imaging parameters, and features of each of the MRA methods influence image quality. The physical principles of MRI and MRA will be briefly described in this chapter. Details of the MRA methods are described in more detail in later chapters.

MRI Overview MRI offers a number of benefits over some of the other diagnostic imaging modalities. First, MRI derives signals using a magnetic field and radiofrequency energy, not ionizing radiation or radioactive materials. Second, MRI is able to demonstrate, with striking contrast, differences in signal intensities among different soft tissues. A host of MR imaging parameters can be modified to exploit a variety of tissue-specific properties to manipulate image contrast. Third, MRI is capable of providing tomographic images of any plane without requiring movement of the patient or any of the equipment. The images may be acquired as a series of two-dimensional (2D) slices, or as a true three-dimensional (3D) volume. MRI may be used to image nuclei of atoms containing an odd number of protons and/or neutrons. The nucleus of the hydrogen atom, 1H, on the water molecule satisfies this criterion as it is composed of a single proton. Because of its abundance in the body, the hydrogen nucleus on the water molecule is routinely imaged in MRI. The protons on the fat molecule (as well as the protons on some other molecules) also appear in MR images.

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_1, © Springer Science+Business Media, LLC 2012

3

4

MRI systems use a strong magnetic field, B0, which is uniform throughout the imaging volume in the bore of the system, and is stable over time. The magnetic field strength is measured in tesla (T) or gauss (G), where 1 T = 104 G. The most commonly used clinical MRI scanners are whole-body imaging systems, with bore diameters of 60 cm or 70 cm,

F.R. Korosec

Fig. 1.1 A state-of-the-art 3 T MRI scanner

and magnetic field strengths of 1.5 T or 3 T. A state-of-the-art 3 T MRI system with a bore diameter of 60 cm is shown in Fig. 1.1. These systems use superconducting magnets, and require a reservoir surrounding the magnet to be filled with liquid helium periodically. The direction of the magnetic field is along the long axis (z-axis) of the bore. Resistive and permanent magnets also are used in MRI systems, but the field strengths of these systems are lower. The magnetic fields of some of these systems are aligned vertically. The hydrogen nucleus possesses a magnetic dipole moment and, from a classical physics perspective, will interact with a magnetic field as if it were a tiny bar magnet. The nuclear magnetic dipole moment of a hydrogen nucleus can either align or antialign with the strong magnetic field of the MRI scanner. In an ensemble of hydrogen nuclei, the majority of the nuclei will align with the magnetic field because it requires less energy to align than to antialign. The net sum of the nuclear magnetic dipole moments from an ensemble of nuclei will yield a bulk magnetization that is aligned with the applied magnetic field. This concept is demonstrated in Fig. 1.2. It is this bulk magnetization that is considered in discussions regarding MRI. In MRI, signal is generated when the bulk magnetization, M, is “tipped” out of alignment with the applied magnetic

Fig. 1.2 (a) Shown on a single xyz coordinate system for comparison with one another are vector representations of magnetic dipole moments aligned with (on the surface of the upward facing cone), and antialigned with (on the surface of the downward facing cone) the main magnetic field, B0. (b) Each vector has a longitudinal component, Mz, and a transverse component, Mxy. (c) Summing the transverse components yields a net sum of zero magnetization (since the orientation in the transverse plane is random, if enough dipole moments are considered, for each one

pointing in a given direction, statistically, there is likely one pointing in the opposite direction, resulting in total cancelation). (d) Summing the longitudinal components yields a net excess of magnetization aligned with the main magnetic field (since it requires less energy for the magnetic dipole moments to align with the magnetic field). (e) The result of summing the contributions from an ensemble of dipole moments is a net, or bulk, magnetization that aligns with the orientation of the main magnetic field, B0

1

Basic Principles of MRI and MR Angiography

5

Fig. 1.3 The bulk magnetization, M, rotates, or precesses, when it is in the transverse plane. The receiver coil detects an oscillating signal from the component of magnetization that points toward it (My in this case). The rate of precession and, therefore, the frequency of the detected signal, are proportional to the strength of the magnetic field, B0. In this diagram, the magnetic field is aligned along the z-axis

field, B0. The mechanism for tipping the magnetization is described below. The maximum signal is generated when the aligned (longitudinal) magnetization is tipped 90° so that it is perpendicular (transverse) to the direction of the applied magnetic field. The longitudinal magnetization is referred to as Mz, and the transverse magnetization is referred to as Mxy. The transverse component of magnetization rotates, or precesses, around the main magnetic field as shown in Fig. 1.3. The precessing transverse component of the magnetization can be detected using a receiver coil oriented perpendicular to the direction of the main magnetic field. The precessing magnetization induces a current in the receiver coil, proportional to the rate of change of the transverse component of the magnetization. Thus, it is the precessing transverse component of the magnetization that produces the MR signal. Because the bulk magnetization spins around the main magnetic field, and because the bulk magnetization is composed of nuclear magnetic dipole moments, the nuclear magnetic dipole moments are often referred to as “spins” in the context of describing the nuclear magnetic resonance phenomenon. The precessional rate of the bulk magnetization is proportional to the strength of the applied magnetic field, B0, and is given by the Larmor equation: ω 0 = g B0 ,

(1.1)

where ω0 is the precessional frequency, g is a constant referred to as the gyromagnetic ratio (its value depends on the nucleus under consideration, 42.58 MHz/T for hydrogen), and B0 is the strength of the main magnetic field. At a field

Fig. 1.4 When a magnetic field, B1, is applied along the x¢-axis in a frame of reference that rotates at the Larmor frequency, the bulk magnetization, M, will precess around B1 in the y¢z¢ plane. This method is used to tip the magnetization away from the longitudinal axis so that a component, Mxy, appears in the transverse plane

strength of 1.5 T, the precessional frequency of the bulk magnetization associated with the hydrogen atom is 63.87 million revolutions per second (63.87 MHz). The precessional frequency is commonly referred to as the Larmor frequency. When dealing with precessing magnetization, discussions are often made easier by introducing the rotating frame of reference. The rotating frame of reference is a frame of reference, such as a coordinate system, that rotates with the object that is being investigated. In this case, the object that is being investigated is the precessing bulk magnetization. Thus, if the xyz reference frame is made to rotate about the z-axis such that the x- and y-axes rotate at the precessional rate of the magnetization, it will appear as if the magnetization is not moving with respect to the rotating x- and y-axes, because the bulk magnetization will always stay in the same position relative to the x- and y-axes. Therefore, in the rotating frame, the precessional frequency of the bulk magnetization is zero. In the rotating frame, it is as if the main magnetic field, B0, is having no effect on the magnetization, so it can be ignored. With the introduction of the rotating frame, and the absence of B0 in this frame, it can be understood that the bulk magnetization can be tipped away from the longitudinal axis simply by applying a magnetic field perpendicular to the longitudinal axis. When this is done, the bulk magnetization will precess around this field. So, for example, if a magnetic field is applied along the x-axis in the rotating frame, the magnetization will precess in the yz plane as shown in Fig. 1.4. The magnetic field that is applied perpendicular to the bulk magnetization to tip it away from the longitudinal axis is referred to as the B1 magnetic field. Note that, in order to

6

Fig. 1.5 As viewed from the stationary frame of reference, the magnetization, M, precesses around the z-axis (at the Larmor frequency) as it is tipped away from this axis, tracing out a trajectory similar to the one shown here. The magnetic field, B1, is applied at the Larmor frequency so that it remains perpendicular to the precessing magnetization, M (ensuring that it is continuously acting to tip the magnetization away from the z-axis)

tip the magnetization away from the longitudinal axis, the B1 field must rotate at the Larmor frequency. If B1 does not precess at the Larmor frequency, the B0 field can no longer be ignored, and the magnetization will precess about some effective field determined by the combination of the B0 and B1 fields. Because the B1 magnetization precesses at 63.87 MHz, and this is in the radiofrequency range of the electromagnetic spectrum, the B1 pulse is often referred to as a radiofrequency (RF) pulse. The process of tipping the magnetization is often referred to as RF excitation. The strength of the RF excitation pulse is on the order of 50 mT, and the duration is typically on the order of just a few milliseconds. If the magnetization were observed from a stationary frame of reference as it were being tipped away from the longitudinal axis, it would be seen to precess about the longitudinal axis as it was being tipped, as shown in Fig. 1.5. The bulk magnetization can be brought from the z-axis into the transverse plane by applying a B1 pulse of the appropriate strength and duration. A B1 pulse that has such an effect is referred to as a 90° RF pulse, since this is the angle traversed by the tipping magnetization. As will be described later, it often is advantageous to tip the magnetization less than 90°. In this case, only a component of the magnetization will be brought into the transverse plane, and the signal will be proportional to sin(q), where q is the tip angle. MRI systems with higher field strengths produce more magnetization (more dipole moments align with a magnetic field of higher strength), which leads to generation of higher signals. Scanners with magnetic field strengths of 3 T are becoming more prevalent, and MRA methods performed on these systems yield excellent results [1–5].

F.R. Korosec

A host of RF coils are available for MRI and MRA applications at various magnetic field strengths. Coils that are cylindrical in shape may be used for transmission of the B1 pulse, and reception of the MR signal. These coils are typically referred to as transmit/receive RF coils. Coils that are not cylindrical in shape generally do not provide a uniform tip angle spatially, so these coils are used only for reception of the MR signal. These coils are referred to as receive only coils. When these coils are used, a cylindrical body coil that is built into the bore of the MRI system is usually used to transmit a spatially uniform B1 pulse. Some coils are composed of multiple smaller coils, or elements, each of which sends its signal to a separate receiver (amplifier, analogue-to-digital converter, etc.). Smaller elements are sensitive to noise from smaller regions of anatomy, and the noise detected by each element is uncorrelated. When the signals and noise from all elements are combined, the result is an image with improved signal-to-noise ratio relative to an image acquired with a single coil that is sensitive to noise in the entire volume covered by the multiple coil elements. These coils are referred to as phased array coils. Vendors currently are providing MRI systems with 32 or more receivers (channels) so phased array coils with 32 elements may be used. For some phased array coils, the number of elements exceeds the number of receivers in the MRI system. With these coils, signals from multiple elements may be combined and sent to a single receiver. Alternatively, the operator may specify that only certain elements be activated during a scan. For example, in a head/neck/spine coil, the operator may chose to activate only the elements required to image the head. In a subsequent scan, the operator may chose to activate only the elements required to image the neck. Some RF coils available for MR imaging are shown in Fig. 1.6.

MRI Contrast Mechanisms In MRI, varying degrees of contrast between different tissues can be achieved by modifying the MR imaging parameters. In this section, three magnetic properties of tissues, namely T2 relaxation, T1 relaxation, and proton density, are described. Later in this chapter, the imaging parameters that are modified to manipulate contrast based on these properties are discussed.

T2 (or Spin–Spin) Relaxation Time It already was shown that the bulk magnetization from a sample can be tipped into the transverse plane using a radiofrequency pulse. It also was shown that the transverse magnetization precesses, and as it does, it induces a signal in the

1

Basic Principles of MRI and MR Angiography

7

Fig. 1.6 Shown here are some of the coils used for MRI and MRA applications, including (a) an 8-channel head coil, (b), an 8-channel head and neck coil, (c) a 4-channel phased array peripheral vascular coil, and (d), an 8-channel phased array torso coil

receiver coil. What was not mentioned is that this transverse magnetization diminishes very rapidly. The rate at which the bulk transverse magnetization decreases is characterized by a constant referred as the T2 decay constant or the T2 relaxation time. Because it refers to relaxation of the transverse magnetization, it is often referred to as the transverse relaxation time. Also, because the effect is caused by spins interacting with neighboring spins, the decay constant is often referred to as the spin–spin relaxation time. The mechanism responsible for this diminution of transverse magnetization is described below. Recall that spins precess at a rate determined by the magnetic field that they experience (the Larmor equation). Recall also that, at 1.5 T, the spins from hydrogen precess at a rate of 63.87 MHz. In tissues, the magnetic field that each nucleus experiences is not exactly equal to the external magnetic field. This is because the magnetic field experienced by a spin is affected by the magnetic fields of the spins in its microscopic neighborhood. The microscopic neighborhood around a spin can cause the spin to experience a field slightly larger, or slightly smaller, than the applied external magnetic field. In tissues, the spins are constantly in motion, so the microscopic neighborhoods of spins are continually changing. The net result is that spins that are tipped into the transverse plane precess at slightly different frequencies (even if they are subjected to a uniform applied external magnetic field), and the precessional frequencies of the spins change over time. This is true even if the spins are in a macroscopically

homogeneous tissue, because of the presence of microscopic inhomogeneities. Due to the differences in the precessional frequencies of spins in different microscopic neighborhoods, the spins throughout the tissue eventually get out of synchronization with one another. That is, the spins that precess faster get ahead of those that precess slower. It is this dispersion, or dephasing, of spins that causes the net transverse magnetization to diminish over time. The behavior of the magnetic dipole moments during T2 relaxation is shown in Fig. 1.7. This figure shows that, in the rotating frame, spins that precess faster than 63.87 MHz, appear to precess clockwise (as viewed from the +z-axis), and spins that precess slower than 63.87 MHz appear to precess counterclockwise. As the spins get more out of phase with one another, components along the −x-axis cancel components along the +x-axis, and eventually components along the −y-axis cancel components along the +y-axis. The time it takes for the transverse magnetization to decrease to 37% of the initial magnetization is referred to as the T2 relaxation time. The T2 relaxation time is determined by the mobility of the spins in the tissue. Thus, different tissues with different spin mobility have different T2 relaxation times. The longer the T2 of a tissue, the longer its transverse magnetization persists. Differences in T2 times of different tissues can be exploited in MR imaging to achieve contrast among different tissues. Spin dephasing is exacerbated by inhomogeneities in the main magnetic field, B0. This is because the precessional rate

8

F.R. Korosec

Fig. 1.7 The rate of precession of spins is affected by neighboring nuclei. In the rotating frame of reference, spins that precess faster than 63.87 MHz appear to precess clockwise (as viewed from the +z-axis) relative to the reference frame, and spins that precess slower than 63.87 MHz appear to precess counterclockwise relative to the reference frame. This dephasing of the spins causes the magnitude of the transverse magnetization, Mxy, to decrease over time. The plot shows the

magnitude of the transverse magnetization as a function of time for two tissues. The time that it takes for the transverse magnetization to diminish to 37% of its initial magnitude is characterized by the T2 time constant. Different tissues have different compositions of nuclear neighbors, and therefore have different T2 time constants. This characteristic is exploited in MRI to obtain different signals from tissues with different T2 time constants

is dictated by the field strength (the Larmor equation). If the field varies spatially, then spins in different locations will have different precessional rates. This will lead to spin dephasing and loss of transverse magnetization in a manner similar to that responsible for T2 decay. Other factors contribute to spin dephasing as well (susceptibility differences, velocity distributions, acceleration distributions, etc.) The decay constant that characterizes how long it takes for the transverse magnetization to diminish to 37% of its initial strength when spin–spin interactions (T2) and all other effects are considered is referred to as T2* (T2 star). The formula for calculating T2* is

Recall that the net longitudinal magnetization is zero after the application of a 90° radiofrequency pulse. This is because after the application of the 90° RF pulse, the number of spins antialigned with the magnetic field is equal to the number of spins aligned with the magnetic field. In other words, the energy imparted to the spin system by the RF pulse causes half of spins to line up in a direction opposite to the magnetic field. After the application of the 90° RF pulse, antialigned spins give up energy to the lattice of the tissue, causing them to once again align with the B0 field. As spins continue to give up energy, aligned spins increasingly out number antialigned spins. The time constant that characterizes how long it takes for the longitudinal magnetization to return to 63% of its initial thermal equilibrium value (the value that it had before it was subjected to the RF pulse) is referred to as the T1 relaxation time. Because this process results from interactions of the spins with the structure of the tissue, or the tissue lattice, the T1 time constant is often referred to as the spin–lattice relaxation time; and because it characterizes the time it takes for the longitudinal magnetization to regrow, it often is referred to as the longitudinal relaxation time. The behavior of the magnetic dipole moments during T1 relaxation is shown in Fig. 1.8. As is the case with T2 relaxation, the T1 relaxation time is determined by the mobility of the spins in the tissue. Different tissues have different T1 relaxation times. The longer the T1 of a tissue, the longer it takes for its longitudinal magnetization to regrow. Differences in T1 times of different tissues can be exploited in MR imaging to achieve varying degrees of contrast among different tissues.

1 1 1 = + , * T2 T2 T2¢

(1.2)

where T2¢ is a time constant that characterizes the loss of transverse magnetization from all factors other than tissue specific, spin–spin interactions. Whereas dephasing due to T2 is advantageous because it permits tissues to be differentiated from one another, dephasing caused by other effects tends to unnecessarily diminish the transverse magnetization. Thus, it is desirable to minimize effects of signal loss from sources other than spin–spin interactions.

T1 (or Spin–Lattice) Relaxation Time At the same time that the transverse magnetization is diminishing, the longitudinal magnetization is increasing.

1

Basic Principles of MRI and MR Angiography

9

Fig. 1.8 After the magnetization is tipped into the transverse plane by the B1 pulse, the spins begin to collide with macromolecules that compose the lattice of the tissue, they give up energy, and they flip from the antialigned state to the aligned state. As more spins flip to the aligned state, the longitudinal magnetization, Mz, grows back to its initial thermal equilibrium value, M0. The plot shows the magnitude of the

longitudinal magnetization as a function of time for two tissues. The time that it takes for the longitudinal magnetization to regrow to 63% of its initial magnitude is characterized by the T1 time constant. Different tissues have different T1 time constants. This characteristic is exploited in MRI to obtain different signals from tissues with different T1 time constants

Table 1.1 Typical T1- and T2-relaxation times for selected tissues in a 1.0 T magnetic field

Table 1.2 Relative T1- and T2-relaxation times of different states of matter

White Matter Gray Matter Fat CSF

T1 (ms)

T2 (ms)

400 500 180 2,000

90 100 90 300

Solid Semisolid Liquid

T1

T2

Long Intermediate Long

Short Intermediate Long

CSF cerebrospinal fluid

Facts Regarding T1 and T2 Relaxation Rates As mentioned above, different tissues have different T1 and T2 values. This permits different tissues to be displayed with different signal intensities in MR images. The signal intensities of different tissues can be altered by manipulating the imaging parameters. Typical relaxation times for selected tissues (in a 1.0 T magnetic field) are listed in Table 1.1. It was stated above that both T2 and T1 relaxation times are dependent on the mobility of spins. The spins in liquids are fairly mobile. Therefore, liquids have long T2 values (see T2 of CSF in Table 1.1). At 1.5 T, the T2 of arterial (oxygenated) blood is about 250 ms, and the T2 of venous blood is about 220 ms. The spins in solids are relatively immobile, so, solids have short T2 values. In fact, the T2 values of solids are so short, that signals from solids do not appear in MRI images; the signals disappear before they can be detected. The T2 values of tissues are intermediate between the T2 values of liquids and solids.

T1 relaxation also is dependent on spin mobility, but in a fairly complicated manner. At 1.5 T, the T1 of arterial and venous blood is about 1,200 ms. The relative T1 and T2 times of solids, semisolids, and liquids are summarized in Table 1.2. These characteristics are schematically plotted in Fig. 1.9. For blood and most tissues, the T1 values increase with field strength (e.g., the T1 of arterial and venous blood is about 1,600 ms at 3 T), whereas the T2 values decrease slightly. The T1 and T2 values of blood and tissues may be decreased by intravenously injecting a gadolinium-based contrast material. Although T2 and T1 relaxation are described separately, they occur simultaneously. That is, as the spins are dephasing in the transverse plane, spins are also going from being antialigned with the B0 field to being aligned with it. Spin flips influence T1 and T2 (as spins flip, they lose coherence in the transverse plane). T2 is additionally influenced by spin– spin interactions. Thus, for most tissues imaged using MRI, T2 < T1. This means that the net magnetization in the transverse plane disappears before the longitudinal magnetization fully regrows. It is only for liquids that T2 may approach T1.

10

F.R. Korosec

Fig. 1.9 Schematic representation of (a) relative T1 relaxation times and (b) relative T2 relaxation times, of various states of matter

Bloch Equation The behavior of the net magnetization of protons due to excitation and relaxation can be summarized by the Bloch equation: M x i + M y j ( M z - M o )k dM , = M´g Bdt T2 T1

(1.3)

where M is the net magnetization, B represents the applied magnetic fields (including B0, B1, and gradient fields) and i, j, and k are unit vectors in the x, y, and z directions, respectively. Note that the behavior depends on both T1 and T2, which results in contrast between tissues with different T1 and T2 values. Solving this governing equation for M, with time-varying magnetic field gradients, yields the signal equations for different imaging sequences. In addition to being influenced by T2, T1, and proton density, the signal in MRI is influenced by diffusion, perfusion, magnetic susceptibility, chemical shift, temperature, magnetic field strength, motion (including that from flowing blood), presence of a contrast material, and many other factors. MRA techniques achieve contrast by harnessing the effects of motion (phase-contrast and TOF MRA), or the effects of injecting a gadolinium-based T1-shortening contrast material (contrast-enhanced MRA), or the inherently long T2 of blood (steady-state free precession MRA). The physical principles of the more common MRA techniques are introduced below.

Imaging Parameters and Their Effects on Image Contrast Signal intensity in MR depends on a number of factors, and there are a host of parameters in MR imaging sequences that can be modified to accentuate the influence of these factors. The signal intensity is proportional to the magnitude of the transverse magnetization at the time that the receiver is turned on. The rate at which transverse magnetization diminishes is characterized by the time constant T2. Different tissues have different T2 values. Tissues with short T2 values lose transverse magnetization quickly, whereas tissues with long T2 values lose it more slowly. Thus, allowing time to

elapse between tipping the magnetization transverse and sampling it allows achievement of signal differences from tissues with different T2 values. The time from when the magnetization is tipped transverse until the signal is detected is controlled by the MR imaging parameter TE (echo time). Because imaging sequences with longer echo times achieve signal differences based on variations in T2 decay times, they are referred to as T2-weighted imaging sequences, and the images produced with these sequences are referred to as T2-weighed images. The magnitude of the transverse magnetization is influenced by the magnitude of the longitudinal magnetization, since a component of longitudinal magnetization becomes transverse after the application of the RF pulse. After the longitudinal magnetization is tipped transverse, the longitudinal component begins to regrow. The rate of regrowth is characterized by the time constant called T1. Different tissues have different T1 regrowth times. In MR imaging, the longitudinal magnetization must be tipped many times in order to encode enough information to map the signals to the proper locations in the image (as described below). If only a short time elapses between applying the RF excitation pulses, the longitudinal magnetization will not fully regrow. The longitudinal magnetization from short T1 tissues will regrow more than the longitudinal magnetization from long T1 tissues. Thus, allowing only a short amount of time to elapse between applying RF excitation pulses will lead to different amounts of longitudinal magnetization being tipped transverse, which will result in different signal intensities from different tissues based on variations in T1 regrowth times. The time between applying RF excitation pulses is controlled by the MR imaging parameter TR (repetition time). Because imaging sequences with shorter repetition times achieve signal differences based on variations in T1 regrowth times, they are referred to as T1-weighed imaging sequences, and the images produced with these sequences are referred to as T1-weighed images. In order to minimize T2-weighting in these scans, the shortest possible echo time is used. In order to minimize T1-weighting in T2-weighted scans, a longer TR is used. When a long TR and a short TE are used, the signal intensities in the images are not dependent on T1 or T2 differences. In this case, the signal intensities depend on the density of hydrogen nuclei (on the water molecule) in each tissue. Since the hydrogen nucleus consists of a single proton, these imaging sequences are referred to as proton-density-weighted (also r-weighted) imaging sequences, and the images produced with these sequences are referred to as proton-densityweighted images. When is it desirable to achieve signal intensities in images that are strictly dependent on T1 values, T2 values, or proton densities of various tissues, a class of imaging sequences called spin-echo sequences is used. Typical TR and TE values used to achieve T1-, T2-, and

1

Basic Principles of MRI and MR Angiography

11

Table 1.3 TR and TE imaging parameters used to achieve different weighting in magnetic resonance images of the brain using a spin-echo imaging sequence at 1.5 T T1-weighting T2-weighting r-weighting

TR Short (400 ms) Long (3,000 ms) Short (3,000 ms)

TE Short (20 ms) Long (100 ms) Long (20 ms)

density-weighed images of the brain using a spin-echo imaging sequence are shown in Table 1.3. Spin-echo imaging sequences typically have scan times and properties that make them inappropriate for many MRA applications. Most MRA methods employ sequences called gradient-echo imaging sequences. In gradient-echo sequences, short TR and short TE values are used. This results in the signal for each tissue being dependent on combined effects of the T1, T2, and proton density of the tissue. There are several gradient-echo sequences used for MRA. Each yields signals that have different dependencies on the physical properties of the blood and tissues being imaged. For these sequences, additional imaging features may be employed to enhance signal differences between blood and stationary tissues. Also, imaging parameters, including tip angle, may be adjusted to enhance signal differences.

MR Image Formation To produce images in MRI, it is necessary to determine the origins of the signals so that they can be mapped to the appropriate positions in the image. This mapping of signal to position is accomplished by applying magnetic field gradients of predetermined amplitude and duration. To say that a magnetic field has a gradient means that it has a different strength at different locations in space. In MR imaging, the gradients are applied so that the magnetic field strength changes linearly with position. Magnetic field gradients can be applied along each of the three axes in the MR scanner. A magnetic field gradient has units of G/cm (or mT/m), and is represented as Gx, Gy, or Gz, depending on the axis along which it is applied. The gradient magnetic field adds to, or subtracts from, the main magnetic field B0. Thus, if a magnetic field gradient is applied along the x-axis, the total magnetic field at any position along the x-axis is given by B0 + Gx x, where x is the position along the x-axis. The value of x is 0 at the center of the magnet (isocenter of the gradient), and has greater positive values with increasing distance from the center in one direction, and greater negative values with increasing distance from the center in the other direction. If a 1 gauss/cm magnetic field gradient is applied in a 1.5 T scanner, the total magnetic field 10 cm from the isocenter of the gradient in one direction is 15,010 G

Fig. 1.10 The net magnetic field produced by applying a linear magnetic field gradient in the x direction. The lengths of the vectors represent the strength of the magnetic field. The gradient strength is denoted as Gx, and it is superimposed on the main magnetic field, B0. Note that when a gradient is applied along the x direction, the field varies linearly in the x direction, but is constant along the y and z directions

(15,000 G + 1 G/cm × 10 cm) and the total magnetic field 10 cm from the isocenter of the gradient in the other direction is 14,990 G (15,000 G – 1 G/cm × 10 cm), where the minus sign in the latter case is due to the negative direction with respect to the isocenter of the gradient. The vector representation of the magnetic field strength as a function of position resulting from the application of a linear magnetic field gradient is shown in Fig. 1.10. As was described previously, in the rotating frame of reference, the main magnetic field, B0, can be ignored. So, with this formalism, the total magnetic field that must be considered in the rotating frame becomes Gx × x for a magnetic field gradient applied along the x-axis, Gy × y for a magnetic field gradient applied along the y-axis, and Gz × z for a magnetic field gradient applied along the z-axis. For the remainder of this chapter, all discussions will pertain to the rotating frame of reference. Vector diagrams representing application of magnetic field gradients along the x-, y-, and z-axes in the rotating frame of reference are shown in Fig. 1.11. The strength of the magnetic field gradient can be plotted as a function of time. Such a plot is referred to as a pulse sequence timing diagram. Figure 1.12 demonstrates the characteristics of a pulse sequence timing diagram. In pulse sequence timing diagrams, the horizontal dimension represents time, and the vertical dimension represents the strength of the gradient at each point in time. The timing diagram shown in the top row of Fig. 1.12 represents a gradient that is applied along the x-axis (Gx) at a strength of 0.75 G/cm, for 4.0 ms. The gradient is turned on 1.0 ms after the clock starts, and it is turned off 5.0 ms after the clock starts. If the gradient were applied at a strength of 0.5 G/cm, and the timing parameters remained the same, the pulse sequence

12

Fig. 1.11 Vector representations of magnetic field gradients applied along the x-, y-, and z-axes. These representations ignore the contribution from B0. The length of the vector at each position represents the strength of the magnetic field at that position (less the contribution from B0). Note that all of the vectors point in the direction of the main magnetic field, meaning that they all add to, or subtract from, the main field, B0. In MRI, no magnetic fields are applied perpendicular to the main magnetic field (except the B1 pulses). In these diagrams, some of the vectors have

Fig. 1.12 (Top left) A timing diagram showing the strength of a magnetic field gradient (vertical axis) as a function of time (horizontal axis). Shown on the top right is a vector representation of the magnetic field gradient. The length of the vectors at each position (along the x dimension in this case) represents the strength of the magnetic field at that position. The bottom row shows a timing diagram and vector representation for a magnetic field gradient having a lower strength than the one shown in the top row

timing diagram would look like the one shown in the bottom row of Fig. 1.12. Typically, pulse sequence timing diagrams are used to give a rough indication of gradient strength and timing, and

F.R. Korosec

been intentionally omitted to reduce clutter. To complete Gx, the vectors at each x position should be reproduced along the y dimension (as shown in Fig. 1.10) and then each of those should also be reproduced along the z dimension. To complete Gy, the vectors at each y position should be reproduced along the x dimension, and then each of these should be reproduced along the z dimension. For Gz, the vectors at each z position are reproduced along the x dimension, but to complete the diagram, each of these vectors should be reproduced along the y dimension

so they do not include markings indicating actual times or gradient strengths. The remaining pulse sequence timing diagrams shown in this chapter will not include such markings. Also, the timing diagrams shown in Fig. 1.12 are drawn as if the gradients can be turned on and off instantaneously, which is not realistic. In reality, it takes time to ramp up the gradients to full strength, and an equivalent amount of time to ramp them down to zero strength. Magnetic field gradients are characterized by their maximum strength, and the time it takes to ramp them up to maximum strength. On state-of-the-art MRI scanners with high-performance gradients, the maximum gradient strength is about 5 G/cm (50 mT/m), and the amount of time it takes to ramp up the gradients on these systems is about 0.25 ms. Often times the gradient strength is divided by the ramp time to characterize the performance of the gradients in a single number, referred to as the gradient slew rate. For the numbers above, the gradient slew rate would be 200 mT/m/ ms. Stronger and faster gradients facilitate shorter TR and TE times, which is beneficial for MRA methods (as described later in this chapter). For the remaining pulse sequence timing diagrams shown in this chapter, the gradient ramp times will be represented by a sloped line before and after each gradient application. The remainder of this section describes how magnetic field gradients are used to select the magnetization that is going to be tipped into the transverse plane and then encode the signal from this magnetization so that it can be mapped to the proper position in the image.

1

Basic Principles of MRI and MR Angiography

13

Slice-Selection

Fig. 1.14 A spatially selective B1 pulse (right side of figure) is constructed by combining B1 fields that have different frequencies (left side of figure). The range of frequencies used to construct the B1 pulse is referred to as the bandwidth of the B1 pulse. The bandwidth of this pulse is −w3 to +w3, or ±w3. The spatially selective pulse shown here can be used in conjunction with a magnetic field gradient to generate transverse magnetization from spins having Larmor frequencies in the range −w3 to +w3

In order to form an image of a single slice, signal must be generated from only spins that are within that slice. This is accomplished by tipping only the magnetization within the slice of interest into the transverse plane so that it may be detected by the receiver coil. This can be achieved by applying a magnetic field gradient, and at the same time applying a specially tailored RF pulse. With the application of a magnetic field gradient along the z-axis, the Larmor equation (in the rotating frame) becomes w = g Gzz, where g is the gyromagntic ratio for protons, Gz is the strength of the magnetic field gradient, and z is the position along the z-axis. This means that, when the magnetic field gradient is being applied, spins at different positions along the z-axis have different Larmor frequencies. This is demonstrated in Fig. 1.13. By viewing Fig. 1.13, and by recalling that spins are only affected by an RF pulse that is transmitted at a frequency equal to the Larmor frequency of the spins, it can be realized that spins in a particular slice can be selectively tipped into the transverse plane by applying a magnetic field gradient, and at the same time applying an RF pulse that is made up of B1 fields that precess at the same frequencies as the spins that are within the slice that is to be imaged. An RF pulse made up of B1 fields having a range of frequencies will affect only spins that have a Larmor frequency that is within this range. Spins that have a Larmor frequency outside this range will be unaffected by the RF pulse. An RF pulse that is made up of B1 fields precessing within a range of frequencies is shown in Fig. 1.14. The range of

frequencies of the B1 fields making up the RF pulse is referred to as the bandwidth of the RF pulse. The RF pulse will affect only magnetization that has Larmor frequencies that are within the bandwidth of the RF pulse. Such a pulse is often referred to as a spatially selective, or slice-selective, RF pulse. The gradient that is applied to create a distribution in the Larmor frequencies of the spins so that magnetization from the spins within a specific slice can be selectively tipped is referred to as the slice-selection gradient. Changing the strength of the slice-selection gradient causes a change in the thickness of the slice that is selected (if the bandwidth of the RF pulse remains unchanged). Increasing the gradient strength causes spins in a thinner slice to be affected by the RF pulse, and decreasing the gradient strength causes spins in a thicker slice to be affected by the RF pulse. This is because increasing the gradient strength causes magnetization with Larmor frequencies that are within the bandwidth of the RF pulse to span a smaller distance, and decreasing the gradient strength causes magnetization with Larmor frequencies that are within the bandwidth of the RF pulse to span a larger distance. This phenomenon is demonstrated in Fig. 1.15. In Fig. 1.15, the bandwidth of the RF pulse includes frequencies from −625 Hz to +625 Hz (written as ±625 Hz), so it affects only spins that have Larmor frequencies within this range. If only a single slice is to be imaged, it is desirable to place that slice at the isocenter of the gradient. This is because the isocenter of the gradient is also the center of the magnet

Fig. 1.13 When a magnetic field gradient is applied along the z-axis, spins at different locations in the z dimension have different Larmor frequencies. The frequency as a function of position can be determined using the Larmor equation, w = g Gzz, where Gz denotes the strength of the magnetic field gradient, and z denotes the location along the z-axis

14

Fig. 1.15 The amplitude of the slice-selection gradient determines the thickness of the slice that is imaged. (a) A strong gradient causes the magnetic field strength to change rapidly with position, so when a strong gradient is applied, spins with Larmor frequencies in the range affected by the B1 pulse are clustered close together. (b) A weak gradient causes the magnetic field strength to change slowly with position, so when a weak gradient is applied, spins with Larmor frequencies in the range affected by the B1 pulse are spread out over a large distance

and the main magnetic field is most uniform at this point. So, if a single slice is to be imaged, the patient table is moved so that the anatomy that is to be imaged moves to the isocenter of the magnet. All of the discussion above pertained to imaging a slice at isocenter. If two slices are to be imaged, they cannot both be placed at the isocenter of the gradient so there is a mechanism to image slices off of isocenter. Imaging slices off of isocenter is accomplished by changing the frequencies of all of the B1 fields that make up the RF pulse. The bandwidth remains the same, but the frequencies of all the B1 fields that make up the RF pulse are offset to correspond to the position of the slice. A pulse sequence timing diagram demonstrating a sliceselective RF pulse and a slice-selection gradient is shown in Fig. 1.16. Here it is demonstrated that the gradient and RF pulse are applied simultaneously. The duration of the RF pulse is a function of its bandwidth. The duration of the gradient is equal to the duration of the RF pulse; the gradient must be on during the entire application of the RF pulse in order to maintain the distribution of spin frequencies. No other gradients may be on during application of the sliceselection gradient. The strength of the gradient is a function of the slice thickness, as explained above, and the strength of the RF pulse is a function of the tip angle. The greater the amplitude of the RF pulse, the farther the spins tip in the time that the RF pulse is applied. The negative lobe of the slice-selection gradient is used to refocus the spins. The spins along the slice-selection direction get out of phase with each other because the application of the slice-selection gradient causes the spins along this direction to precess at different frequencies while they are being tipped into the

F.R. Korosec

Fig. 1.16 A pulse sequence timing diagram showing a slice-selective RF (or B1) pulse, a slice-selection gradient, a phase-encoding gradient, a frequency-encoding (or readout) gradient, and the signal detected during activation of the receiver

transverse plane. Reversing the gradient for the appropriate amount of time causes the spins to come back into phase with each other.

Frequency-Encoding Now that a slice has been selected by tipping only the spins within that slice into the transverse plane, the signals from these spins must be encoded so that when they are detected, it can be determined from where they came, so that they can be mapped to the appropriate positions in the image. Differentiating the signal along one dimension of an object can be accomplished by applying a magnetic field gradient along that direction. Just as a slice-selection gradient applied along the z dimension causes spins at different locations along the z dimension to have different Larmor frequencies, a spatial-encoding gradient applied along the x dimension causes spins at different locations along the x dimension to have different Larmor frequencies. Thus, when the signals from the spins in the image slice are detected during the application of a magnetic field gradient, the frequencies of the signals can be used to map them to the proper location along this one dimension in the image. Because this gradient encodes position by giving spins at different positions different frequencies, it is referred to as the frequency-encoding gradient. It also often is referred to as the readout gradient because it is applied at the same time that the receiver coil is “reading out” the signal. An example of the effect that a frequency-encoding gradient has on the spins in an image slice is shown in Fig. 1.17. To understand more clearly how the frequency-encoding gradient allows the position of the spins to be determined,

1

Basic Principles of MRI and MR Angiography

consider the case of an object placed in the MR scanner as shown in Fig. 1.18. While the frequency-encoding gradient is being applied, the spins in the object precess at frequencies

15

Fig. 1.17 Application of a frequency-encoding gradient along the x dimension causes spins at different locations in the x dimension to have different Larmor frequencies. The signals can be mapped to the proper position in the image based on their precessional frequencies

that are dictated by where the spins are located along the magnetic field gradient. If the receiver coil is turned on while the frequency-encoding gradient is being applied, the detected signal will be a composite of all of the signals at all of the precessional frequencies of the spins in the object. When the signal is sufficiently sampled, the frequencyencoding gradient and the receiver coil are turned off and the signal is amplified, digitized, and sent to the signal processor. The signal processor performs a fast Fourier transform (FFT) on the sampled signal to determine the frequencies of all the signals composing the detected composite signal, and the relative number of spins precessing at each of the frequencies (signal intensity at each frequency). The output of the fast Fourier transform can be mapped to the columns of pixels (picture elements) in an image as shown in Fig. 1.18. So, the frequency-encoding gradient allows the relative amount of signal in each column of the image to be determined. To form an image, the spatial distribution of the signals in each column needs to be determined. This is the topic of the next section. Before this is addressed, a few aspects of frequency-encoding are discussed. The strength of the frequency-encoding gradient is chosen such that spins at one edge of the field-of-view (FOV) precess at a predetermined maximum frequency (added to the 63.87 MHz resulting from B0), and spins at the opposite edge of the FOV precess at the same predetermined maximum frequency but in the opposite direction (subtracted

Fig. 1.18 (Upper left) The sample, denoted by black squares in this example, extends over three of the six columns of the image. (Lower left) The detected signal contains contributions from all of the spins in the object, all precessing at frequencies determined by their location along the applied gradient. In this example, because the object extends over three of the six columns, the detected signal is composed of

signals having three different precessional frequencies, −w1, w1, and w2. (Lower right) Applying a fast Fourier transform to the detected signal reveals the frequencies that are contained in the signal, and the relative number of spins at each frequency. (Upper right) This information is used to map the signals from the spins to the appropriate column in the image

16

from the 63.87 MHz resulting from B0). The receiver is set by the operator to accept only frequencies within this predetermined range. This range of frequencies is referred to as the receiver bandwidth. A typical receiver bandwidth is ±16 kHz, which means that spins at one edge of the FOV precess at +16 kHz (faster than 63.87 MHz), and spins at the opposite edge of the FOV precess at −16 kHz (slower than 63.87 MHz). By keeping the receiver bandwidth the same and changing the strength of the gradient, the FOV in the frequencyencoding direction can be changed. Increasing the gradient strength decreases the FOV by causing spins that have precessional frequencies within the receiver bandwidth to be contained in a smaller region, and decreasing the gradient strength increases the FOV by causing spins that have precessional frequencies within the receiver bandwidth to extend over a larger region. The field of view is not restricted to the center of the magnet. Off-center FOVs are achieved by changing the frequencies that are accepted by the receiver. The receiver bandwidth is kept the same, but each of the frequencies that is accepted is changed by some amount depending on how far off of isocenter the FOV is shifted. A pulse sequence timing diagram, showing the timing of the frequency-encoding gradient and receiver activation relative to the timing of the slice-selection process is shown in Fig. 1.16. No other gradients may be applied during application of the frequencyencoding gradient.

Phase-Encoding The slice-selection gradient applied simultaneously with a tailored RF pulse allows the spins in a particular slice to be tipped into the transverse plane to generate signals, and the application of the frequency-encoding gradient permits determination of where the signals originate from along one dimension. To form an image, all that remains to be determined is from where along the second dimension in the image plane the signals originate. Encoding along the second image dimension is achieved by applying another gradient, this time along the y-axis. This gradient, because it gives signals at different positions and different phases, is referred to as the phase-encoding gradient. The phase of the magnetization is simply a measure of how far it precesses in the transverse plane during a given period of time. Phase is an angle and it is typically measured in degrees. If the magnetization in the transverse plane precesses one full revolution, it is said to have accumulated a phase of 360°. The phase-encoding gradient is applied after the magnetization is tipped into the transverse plane, and before it is read out; that is, it is applied after the slice-selection gradient is

F.R. Korosec

Fig. 1.19 (a) After the slice-selection process, the spins in the selected slice lie in the transverse plane, and are all oriented in the same direction (in this case, they all lie along the x-axis). (b) When a gradient, Gy, is applied along the y dimension, it causes spins at different locations along this dimension to precess at different frequencies. By the time the gradient is turned off, spins at different locations along the y dimension have accumulated different amounts of phase. This phase-encoding process is used to encode signals along this dimension with a unique phase so that when they are detected, they can be mapped to the proper position in the phase-encoding dimension in the image

applied and before the frequency-encoding gradient is applied. While the phase-encoding gradient is being applied, the spins at different locations along the y-axis precess at different rates. Thus, by the time the phase-encoding gradient is turned off, the spins at different locations along the y-axis will have precessed different amounts and thus will have accumulated different amounts of phase. This is shown schematically in Fig. 1.19. After the phase-encoding gradient has been turned off, the frequency-encoding gradient is turned on, and the signal from the selected slice is detected. In order to resolve all points along the y dimension, additional information is needed. The way this information is obtained is by applying additional phase-encoding gradients. After the application of each phase-encoding gradient, a frequency-encoding gradient is applied, and the signal is sampled again with the new phase-encoding value. The application of each phase-encoding gradient requires a separate TR. In each TR, a phase-encoding gradient of slightly greater amplitude is applied. This causes the spins to accumulate a slightly larger phase in each TR. The amplitude of the phaseencoding gradient is increased with each TR because, for any single application of the phase-encoding gradient, components of the magnetization from spins at different locations point in opposite directions causing the signals from some spins to partially cancel the signals from other spins. In order to appropriately determine the signal in each pixel along the phase-encoding dimension, the total number of phaseencoding gradients that must be applied is equal to the total number of rows of pixels that are to be resolved in the phaseencoding dimension (it is also true that the frequencyencoded signal needs to be sampled one time for each column

1

Basic Principles of MRI and MR Angiography

17

of pixels that is to be resolved in the frequency-encoding direction). An important difference between phase- and frequency-encoding is that increasing the number of rows of pixels in the phase-encoding direction results in an increase in scan time, whereas increasing the number of columns of pixels in the frequency-encoding direction does not affect the scan time (except in fast gradient-echo acquisitions, where increasing the number of frequency-encoding values necessitates an increase in the TR of the imaging sequence). It should be noted that for consistency of terminology in this chapter, frequency-encoding is being associated with rows of pixels in the image and phase-encoding is being associated with columns of pixels in the image, but the phaseand frequency-encoding dimensions can easily be swapped in practice. If it is desirable to have greater resolution along a particular dimension, the scanner operator would align the frequency-encoding along this dimension since aligning the phase-encoding along this dimension would necessitate a longer scan time. Just as the FOV in the frequency-encoding dimension can be adjusted by modifying the strength of the frequencyencoding gradient, the field of view in the phase-encoding dimension can be adjusted by modifying the strength of the phase-encoding gradients. Also, the center of the field of view can be offset in the phase-encoding dimension, just as it can be offset in the frequency-encoding dimension. Figure 1.16 shows the timing of the phase-encoding gradient with respect to the slice-selection gradient, the frequency-encoding gradient, the RF pulse, and the activation of the receiver. The phase-encoding gradient is shown with many amplitudes, representing the need to apply one phaseencoding amplitude for each row of pixels that is to be resolved in the phase-encoding dimension. It should be remembered that each phase-encoding value is applied in a separate TR. All other waveforms remain the same for all TR intervals. The scan time necessary to acquire the data to produce an image of a single slice in MRI is given by the following equation: Scan Time = TR ´ N y ´ Ave,

(1.4)

where TR is the repetition time of the imaging sequence, Ny is the number of phase-encoding values applied (one per TR), and Ave is the number of times each phase-encoding value is sampled for purposes of averaging the data. For certain imaging sequences used in MRI, the TR is chosen to achieve the desired image contrast (e.g., T1-weighting, T2-weighting, density-weighting). If the TR is much longer than the time necessary to acquire a single phase-encoded line of data from a given slice, then the data acquisition can be interleaved so that data from multiple slices may be obtained in a TR interval (i.e., excite slice 1, sample phase-encoding 1; excite

slice 2, sample phase-encoding 1; excite slice 3, sample phase-encoding 1; – continue until the TR interval has elapsed, then – excite slice 1, sample phase-encoding 2; excite slice 2, sample phase-encoding 2; excite slice 3, sample phaseencoding 2; – and repeat until all phase-encoding values have been sampled.) For this interleaved mode of acquisition, the scan time for imaging multiple slices is the same as that required for imaging a single slice, as long as the data for a given phase-encoding value for all slices can be acquired before the TR time elapses. If data are acquired from individual slices sequentially, such that all data for a given slice are acquired prior to collecting data for the next slice, then the scan time is increased by a factor equal to the number of slices imaged. This is true for the types of sequences most commonly used for MR angiography (i.e. fast gradient-echo sequences). The above methods, where individual slices are excited one at a time (in an interleaved fashion, or individually), and magnetization from only one slice is in the transverse plane at any given time, are referred to as 2D multislice imaging methods. If data are collected from contiguous slices, data are collected from a 3D volume, but the acquisition methods are still referred to as two-dimensional, multislice imaging methods. MR imaging may be performed using three-dimensional acquisition methods. With 3D acquisition methods, a component of the magnetization from a volume, or slab, containing all the slices to be imaged is tipped into the transverse plane in every TR, and the information is mapped to individual slices using a phase-encoding method. For 3D acquisitions, the pulse sequence timing diagram shown in Fig. 1.16 would be modified to include a phase-encoding gradient on the z-axis (sometimes referred to as slice-encoding, or slabencoding gradient). The phase-encoding gradient on the z-axis can be applied anytime after the application of the slice-selection gradient and before the application of the frequency-encoding gradient. With 3D acquisitions, data for every slice are sampled in every TR throughout the acquisition, so interleaved acquisition is not possible. For 3D acquisition methods, the scan time equation above must be multiplied by the number of slices imaged.

MR Data (k-Space) A representation of the matrix of data points acquired in an MRI scan is shown in Fig. 1.20. In MRI, analog data are sampled and converted to digital numbers. Each number (sampled point) is represented as a dot in this figure. Each dot along a row represents a different frequency-encoded sample (the number of columns of dots is equal to the number of columns of pixels that may be represented in the frequency-encoding dimension of the image). Each dot along a

18

column represents a different phase-encoded sample (the number of rows of dots is equal to the number of rows of pixels that may be represented in the phase-encoding dimension of the image). More frequency-encoding values (or more phase-encoding values) corresponds to more rows (or columns) of pixels across a FOV, leading to improved spatial resolution in an image. Actual data acquired during an MRI scan are shown in Fig. 1.21. In this figure, the dots have been replaced with a representation of the actual digital samples (the numeric values). Large numbers are represented with high-intensity signals, and small numbers are represented with low-intensity signals. In MRI, these numbers that are sampled correspond to Fourier weighting coefficients. A full discussion on Fourier

Fig. 1.20 In MRI, data are acquired at discrete points in what is referred to as k-space. The dots in this figure represent the locations of the acquired data points. The resolution in MR images is determined by how far from the origin data are sampled ( ± k x ,max, ± k y ,max ). The size of the unaliased field-of-view (FOV) in MR images is determined by the spacing between sampled points ( Dk x , Dk y )

Fig. 1.21 (a) MRI data. The signal intensity represents the magnitude of the Fourier weighting coefficients. The characteristics of the frequencyencoding gradient (in combination with the receiver bandwidth) determine which points are sampled along each row, and the characteristics of the phase-encoding gradient determine which points are sampled along each column. (b) An axial image of a human brain obtained after applying a two-dimensional fast Fourier transform (2D FFT) to the acquired data

F.R. Korosec

theory is beyond the scope of this chapter. However, a few concepts are introduced. According to Fourier theory, images can be formed by appropriately weighting and combing line-pair patterns (more specifically, cosine and sine function patterns). Fourier coefficients are the weighting values that must be applied to the line-pair patterns to produce the desired image. For twodimensional images, the Fourier coefficients can be arranged in a 2D matrix, and this matrix of coefficients may be displayed in what is referred to as k-space. The location of each point in k-space is associated with a line-pair pattern in image space. Locations along the kx-axis correspond to vertical line-pair patterns, locations along the ky-axis correspond to horizontal line-pair patterns, and locations along the diagonal axes correspond to diagonal line-pair patters. Locations near the center of k-space correspond to few line-pairs per image, whereas locations farther from the center of k-space correspond to more line-pairs per image. To fully represent an object in an image, an infinite number of line-pair patterns (and associated weighting coefficients) would be required. In MRI, only a subset of the Fourier coefficients are sampled, due to scan time considerations. The limited number of Fourier coefficients is confined between –kx,max and + kx,max in the frequency-encoding dimension, and –ky,max and +ky,max in the phase-encoding dimension. These Fourier coefficients, therefore, define the maximum spatial resolution in the image (highest density line-pair patterns in the x and y dimensions). The Fourier coefficients beyond these values are not sampled. The spacing between sampled Fourier coefficients defines the size of the FOV in MRI. The larger the spacing between the sampled k-space points, the smaller the FOV. This is demonstrated in Fig. 1.22. Another thing to note about k-space is that, in MRI, the imaging gradients are used to determine which k-space

1

Basic Principles of MRI and MR Angiography

19

Fig. 1.22 As the distance between k-space samples increases (in this case by acquiring fewer phase-encoded lines of data), the size of the field-of-view (FOV) decreases. When there is anatomy outside of the FOV, it “wraps around” and overlaps onto the anatomy within the FOV. This mismapping of the signal that originates from outside the FOV is referred to as spatial aliasing

points are sampled. The area (amplitude × duration) of the frequency-encoding gradient determines which point along the kx direction is sampled during data acquisition, and the area of the phase-encoding gradient determines which point along the ky direction is sampled. Negative gradient areas move through k-space along the negative kx and ky axes, and positive gradient areas move through k-space along the positive kx and ky axes. Once the FOV in the phase-encoding dimension is selected, Dky is determined, and the changes in the area of the phase-encoding gradient from one TR to the next can be calculated. Likewise, once the FOV in the frequency-encoding dimension is selected, Dk x is determined, and the changes in the area of the frequency-encoding gradient from one sample to the next can be calculated (in conjunction with the sampling interval of the receiver, which is dictated by the receiver bandwidth). Once the number of phase-encoding values ( N y ) and frequency-encoding values ( N x ) is determined, the number of k-space points can be determined, the kmax points can be determined Ny ö Nx æ çè ± k x ,max = ± 2 ´ D k x and ± ky ,max = ± 2 ´ Dky ÷ , and the ø duration of the frequency-encoding gradient, and maximum area of the phase-encoding gradient can be determined. By strategically applying combinations of magnetic field gradients, k-space may be sampled using a variety of trajectories. Instead of traversing k-space on a rectilinear grid [6] (typically referred to as Cartesian sampling), k-space can be sampled along 2D radial lines [7], where each radial line passes through the center of the k-space, and the radial lines are arranged with equal angles between them (like spokes on a wheel). Alternatively, k-space may be sampled along spiral trajectories [8], starting from the center and spiraling

outward. Other sampling trajectories exist as well, such as 3D radial [9], stack of spirals [10], 3D cones [11], as well as numerous others. The density of phase-encoding lines and readout samples required are determined by the Nyquist sampling criterion and can be described by the following equations, Dt £

2π , g Gx FOVx

(1.5)

2π , g TPE FOVy

(1.6)

DGy £

where Dt is the readout sampling interval (determined by the receiver bandwidth), Gx is the frequency-encoding gradient amplitude, FOVx and FOVy are field of view sizes in the x and y dimensions, DGy is the phase-encoding gradient amplitude step size, TPE is the duration of each phase-encoding gradient, and it is assumed that the imaging gradients can be turned on and off instantaneously (i.e., ramp time equals zero). If the requirements defined by these equations are not met, the reconstructed images will have aliasing artifacts, where copies of images will be superimposed as shown in Fig. 1.22. Finally, it has been described that points near the center of k-space represent low spatial frequency information (few line-pairs per image), and points at the periphery of k-space represent high spatial frequency information (many linepairs per image). This characteristic is often generalized by saying that the center of k-space contains the contrast information for the image, and the periphery of k-space contains the detail information for the image (information regarding

20

F.R. Korosec

Fig. 1.23 The images in the bottom row were produced by performing a fast Fourier transform (FFT) on the k-space data shown in the top row. Each image is shown directly below the k-space data used to produce it. These images demonstrate that the data at the center of k-space is responsible for representing the contrast in the image, whereas the data at the periphery of k-space is responsible for representing the detail in the image

the edges of objects). This concept is demonstrated in Fig. 1.23, which shows the image content derived from the center of k-space in one image, and the content derived from the periphery of k-space in a separate image.

Signal-to-Noise Ratio One parameter commonly used to characterize image quality is the signal-to-noise ratio (SNR). This is the ratio of the amplitude of the signal (desired information) to the amplitude of the noise (undesired information), and is expressed as Signal/Noise. Generally speaking, a higher SNR indicates better the image quality. In MRI, the SNR depends on a large number of factors, including the field strength of the MRI scanner, the imaging coil used, the temperature of the object being imaged, the imaging sequence selected (i.e., spin-echo, gradient-echo, fast spin-echo, etc.), the imaging parameters that are used to manipulate image contrast (i.e., TR, TE, TI, tip angle, etc.), and a host of other parameters. Once the exam is setup, and the imaging parameters are selected to produce the appropriate contrast in the image, there are still a number of parameters that may be changed for a variety of reasons (modify anatomic coverage, resolution, scan time, etc.). These parameters affect the SNR according to the following equation [12]: SNR μ voxel volume ´ acquisition time.

(1.7)

The variable acquisition time is the time that the receiver is turned on and data are being acquired, and a voxel (volume element) is the volume of the object represented by a pixel in the image. The proportionality symbol appears because the SNR is dependent on the scan setup, as well as the scan parameters responsible for image contrast as described above. The SNR is dependent on the voxel volume since larger voxel volumes result in more spins contributing to the signal from each voxel. The SNR also is dependent on the time spent acquiring data. When more data are acquired, the signal increases coherently, whereas the noise (because it is random) increases incoherently (sometimes adding constructively and sometimes destructively). The result is that the ratio of the signal and noise increase as the square root of the amount of data acquired. The voxel volume can be calculated as FOVx / N x ´ FOVy / N y ´ sl thick, where FOVx and FOVy are the fields of view in the frequency-encoding and phase-encoding directions, respectively, N x and N y are the number of frequencyencoding and phase-encoding values, respectively, and sl thick is the slice thickness. The amount of time spent acquiring data is equal to the time spent acquiring each frequency-encoding value (Dtx), multiplied by the number of frequency-encoding points sampled Nx (Dtx × Nx = duration of readout), multiplied by the number of phase-encoding values Ny (since the each frequency-encoding value is sampled at each phase-encoding value), multiplied by the number of times every data point is resampled for purposes of averaging,

1

Basic Principles of MRI and MR Angiography

21

Ave. The time spent acquiring each frequency-encoding point is equal to the inverse of the sampling rate (sampling rate = receiver bandwidth = RBW, so Dtx = 1/RBW). So, the total time spent acquiring data is equal to ((Nx × Ny × Ave)/ RBW). The equation for SNR becomes: SNR 2D μ

N x ´ N y ´ Ave (1.8) FOVx FOVy ´ ´ sl thick ´ , Nx Ny RBW

where, FOVx and FOVy are the fields of view in the frequency-encoding and phase-encoding directions, respectively, N x and N y are the number of frequency-encoding and phase-encoding values, respectively, slthick is the slice thickness, Ave is the number of signal averages, and RBW is the receiver bandwidth. For 3D acquisitions, data for every slice are sampled in every TR throughout the acquisition. This increases the SNR of every slice by the square root of the number of slices imaged. So, for 3D acquisitions, the SNR equation is as follows:

allow for high scan time reduction factors. Parallel imaging can be used to obtain higher spatial resolution or coverage in the same scan time, or to obtain the same resolution and coverage in less scan time, or various combinations of the above. Reducing the scan time is particularly useful when scans must be completed during a breath-hold interval or when temporal resolution is desired.

Magnetic Resonance Angiography The most widely used MRA techniques can be categorized as contrast-enhanced, TOF, phase-contrast, or steady-state MRA techniques. In the remainder of this chapter, a brief overview of the physical principles of these techniques is provided, examples of clinical applications are shown, and the benefits and limitations of each of the specific methods are briefly discussed.

Contrast-Enhanced MRA SNR 3D = SNR 2D ´ N z ,

(1.9)

where N z is the number of phase-encoding values in the slab thickness direction (the number of slices encoded).

Parallel Imaging When multielement receiver coils are used, scan times may be reduced using a class of methods referred to as parallel imaging methods. SMASH [13] (simultaneous acquisition of spatial harmonics), SENSE [14, 15] (sensitivity encoding), and GRAPPA [16] (generalized autocalibrating partially parallel acquisitions) are examples of parallel imaging methods. These methods rely on the sensitivity of the different coil elements to determine information about the spatial location of the detected signals, thereby reducing the number of phaseencoding values necessary for image formation. With these methods, the scan time reduction factor, R, is dependent on the number of elements contained in a coil. Coils with more elements permit greater R values. If coil elements are distributed in more than one dimension, then the amount of data in both phase-encoding dimensions in 3D acquisitions can be reduced. Reducing the amount of data acquired results in a reduced SNR [see (1.8)]. Parallel imaging methods suffer from an additional reduction in SNR-based on coil geometry. This additional reduction in SNR is characterized by what is called a g-factor. With current coil configurations and SNR limitations, the scan time reduction factors afforded by parallel imaging methods are on the order of 4–6. Parallel imaging methods are particularly effective when used with 3D CE-MRA where the SNR is sufficiently high to

Contrast-enhanced MRA (CE-MRA) methods have gained widespread acceptance due to their ease of use, and their ability to quickly, reliably, and robustly produce high-quality diagnostic images of large vascular territories. In many clinical practices, CE-MRA has replaced X-ray DSA as the method of choice for imaging certain vasculature, including the carotid and vertebral arteries, the aorta and renal arteries, and the vessels of the lower extremities. Contrast-enhanced MRA techniques [17] achieve signal differences between blood and stationary tissues by manipulating the magnitude of the magnetization, such that the magnitude of the magnetization from moving blood is larger than the magnitude of the magnetization from stationary tissues. Manipulating the magnetization to produce signal differences in CE-MRA techniques is achieved not only by employing the appropriate imaging sequence parameters, but also by injecting a contrast material intravenously to selectively shorten the T1 of the blood. By implementing a T1weighted imaging method, appropriately synchronized to acquire data during the first pass of the contrast material through the arteries of interest, images can be acquired that show arteries with striking contrast relative to surrounding stationary tissues and veins. A 3D CE-MRA image of the arteries in the lower extremities is shown in Fig. 1.24.

Contrast Material Currently, the contrast materials most widely used for CE-MRA are gadolinium-based. The gadolinium atom is highly paramagnetic. It has seven unpaired electrons in its

22

F.R. Korosec

Fig. 1.24 A coronal MIP image of the vessels of the thighs of a healthy volunteer acquired using a coronal 3D contrast-enhanced MRA acquisition method applied after intravenous injection of a gadolinium-based contrast material

outer shell that interact with the hydrogen nuclei on the water molecules. This interaction results in an increased regrowth rate of longitudinal magnetization from these hydrogen nuclei (reduced T1), so they appear bright in images acquired with T1-weighted MR acquisition methods. The T1-shortening induced by the gadolinium-based contrast material is described by the following equation: 1 1 = + R1 ´ [Gd], T1 T10

(1.10)

where T1 is the shortened longitudinal relaxation time of blood in the presence of gadolinium, T10 is the longitudinal relaxation time of blood in the absence of gadolinium, R1 is the relaxivity of the gadolinium-based contrast material, and [Gd] is the concentration of the gadolinium-based contrast material. Since gadolinium atoms are highly toxic, they must be chelated before they can be injected into the blood vessels. Currently there are several different chelates widely used in CE-MRA, which include gadobenate dimeglumine (Multihance), gadopentetate dimeglumine (Magnevist), gadodiamide (Omniscan), gadoversetamide (OptiMARK), and gadoteridol (ProHance). These materials are permeable through the blood vessel walls, resulting in enhancement of signals from tissues over time, as the contrast material makes its way through the arteries, into the veins, and eventually into the stationary tissues. These chelates are referred to as extracellular contrast materials. Contrast materials that are not permeable through vessel walls, and remain in the blood pool for more than an hour without leaking into the surrounding stationary tissues, have been developed (e.g., gadofosveset trisodium: Ablavar) [18, 19]. These so-called intravascular, or blood pool, contrast materials

can be used to increase the imaging time in order to achieve greater coverage and higher spatial resolution. The drawback of increased acquisition time is that it results in venous enhancement, which leads to difficulty in evaluating the arteries. Methods are being developed to separate arterial signal from venous signal, but are not yet available clinically. An additional benefit of some of the intravascular contrast materials is that they have a higher relaxivity. In other words, they provide greater vascular signal during the first pass of the contrast material by causing a more dramatic decrease in the T1 of blood (per unit volume of contrast material) compared to the currently available extravascular contrast materials. Exposure to gadolinium-based contrast materials has been associated with the development of nephrogenic systemic fibrosis (NSF) [20] in patients with compromised renal function who also have other confounding health issues. NSF is an irreversible, often fatal, disease. Screening measures are now in place to identify patients at risk of contracting this disease. Since awareness has been heightened, and screening measures have been implemented to identify high-risk patients, the incidence of this disease has been nearly eliminated. The gadolinium-based contrast material is typically injected intravenously through an 18–22-gauge angiocatheter. The angiocather is typically inserted into a large, easily accessible vein, such as one of the veins in the antecubital fossa. However, other injections sites may be selected as well, including the forearm, wrist, or the back of the hand. Injection into the right arm provides the most direct route to the heart, increasing the likelihood of distributing a tight, highly concentrated bolus of contrast material to the vessels of interest. The size of the angiocatheter is dependent on the size of the vessel and the desired injection rate. Injection rates typically range from 0.5 to 4.0 ml/s for CE-MRA. Injection volumes range from 0.1 to 0.3 mmol of contrast material per kg of patient body weight (mmol/kg), with typical values in the range of 20–40 ml of contrast material. Higher volumes usually provide images with higher SNR in which the small vessel detail is better delineated. The intent is to reduce the T1 of blood to as low as 50 ms during the first pass of the contrast material. The injection of the contrast material is immediately followed by a flush of normal saline. The injection may be performed manually, or using an MR-compatible, computer-controlled injector as shown in Fig. 1.25. With a computer-controlled injector, the volume and injection rate are precisely programmed to ensure reproducible results. With manual injection, the clinician controls the injection and can better monitor the status of the patient.

Imaging Sequence CE-MRA typically is performed using a three-dimensional, RF-spoiled, fast gradient-echo imaging sequence [21]. The

1

Basic Principles of MRI and MR Angiography

23

Fig. 1.25 Typical patient setup for a 3D CE-MRA acquisition of the renal arteries. An MRI-compatible computer-controlled injector is used to administer the gadolinium-based contrast material into a vein in the antecubital fossa. One syringe is used to inject the contrast material, and the second syringe is used to inject a saline flush. The anterior and posterior sections of an 8-channel torso coil (square items with Velcro straps) are placed on top of and underneath, respectively, the patient to image the region of interest

pulse sequence timing diagram looks like the one shown in Fig. 1.16, with the addition of a phase-encoding gradient on the slice-selection axis. Also RF spoiling is used to eliminate, or spoil, the transverse magnetization so it does not contribute to signal in subsequent TR intervals. With RF spoiling, the RF pulse is applied such that the longitudinal magnetization is tipped into a different location in the transverse plane (relative to the y-axis) for each TR. Eliminating the coherence of the transverse magnetization so it does not persist from one TR to the next makes RF-spoiled sequences less T2*-weighted, and therefore more strictly T1-weighted (as long as TE Ⰶ T2*). Imaging occurs after the longitudinal magnetization has reached its steady-state equilibrium value (a state where the signal remains constant from one TR to the next since the amount of longitudinal magnetization tipped away from the longitudinal axis by the RF pulse equals the amount of magnetization that regrows during each TR interval). RF spoiled gradient-echo imaging sequences are given different names by different MR vendors, including SPGR (spoiled gradient recalled echo), FLASH (fast low angle shot), and T1 FFE (T1 fast field echo). The steady-state equilibrium magnetization for an RF-spoiled gradient-echo sequence (normalized for protondensity and assuming negligible T2* decay and perfect spoiling) is given by the following equation [22]: M xy (q ,TE ) =

M 0 sin (q )(1 - E1 ) - TE/T2* e , 1 - E1 cos (q )

(1.11)

Fig. 1.26 Calculated steady-state signal as a function of tip angle for tissues with various T1 values using a spoiled gradient-echo imaging sequence with a TR of 5 ms. Note that the signal initially increases with increasing tip angle, but then decreases. See text for details. Note also that the signal increases with decreasing T1 value. For a given tip angle, the relative signal difference (contrast) between tissues can be determined. For 3D CE-MRA, it is desirable to maximize the signal different between blood and stationary tissues

where E1 is defined to be exp( - TR/T1 ) , M 0 is the thermal equilibrium magnetization value, q is the tip angle of the imaging sequence, TR is the repetition time of the imaging sequence, TE is the echo time of the imaging sequence, T2* is the time constant characterizing the decay rate of the transverse magnetization of the tissue under consideration, and T1 is the time constant characterizing the regrowth rate of the longitudinal magnetization of the tissue under consideration. The signal as a function of tip angle and T1 for an RF-spoiled gradient-echo sequence with a TR of 5 ms and TE  T2* is plotted in Fig. 1.26. It can be seen that the signal increases as the T1 value decreases. In CE-MRA, the injection of a gadolinium-based contrast material shortens the T1 of blood such that it gives rise to a large signal. All other tissues with longer T1 values give rise to lower signals. The graph also shows that the signal initially increases with tip angle as more magnetization is tipped into the transverse plane. However, as the tip angle is further increased, a point is reached where the tip angle is sufficiently large that it depletes the steady-state longitudinal magnetization that remains after tipping (and it is the longitudinal magnetization that contributes to the transverse magnetization in subsequent TR intervals), so the resulting signal gets smaller with increasing tip angles. The tip angle at which the largest signal is obtained is referred to at the Ernst angle. Rather than striving to achieve the largest signal from blood in CE-MRA, it is more desirable to maximize the image contrast (greatest signal difference between blood and other tissues).

24

Thus, tip angles greater than the Ernst angle are typically used when prescribing CE-MRA sequences. Prescribing an RF-spoiled gradient-echo sequence with a short TR and a large tip angle will yield signal differences based on differences in T1 relaxation times. As long as the scan is synchronized so that data are acquired when the contrast material is in the arteries of interest, the resulting T1weighted images will show the short-T1 arterial blood with a high signal, and the long-T1 veins and stationary tissues with diminished signal. In order to acquire the majority of the data for the contrast-enhanced angiogram during the first pass of the contrast material (when the concentration of the contrast material in the arteries is high, and the concentration in the veins and stationary tissues is low) it is necessary to use a short scan time. When imaging vessels in the abdomen, where it is necessary to acquire all the data during a breath-hold interval, it also is necessary to use a short scan time. To achieve short scan times in CE-MRA, a short TR is used, and a reduced data set is sampled. The TR may be reduced by using a specialized fast gradient-echo sequence, a high receiver bandwidth, the minimum echo time, a fractional-echo readout, a reduced number of frequency-encoding samples, and highperformance magnetic field gradients. A reduction in the data set may be achieved by employing a fractional FOV, a reduced number of phase-encoding values, and a reduced number of slices. Reducing the number of frequency-encoding values, phase-encoding values, or slices, for a specified volume of coverage, results in reduced spatial resolution in the MRA images. Designing CE-MRA imaging protocols requires compromises, and the best protocols are those that achieve the appropriate balance between scan time, anatomic coverage, and spatial resolution. The apparent spatial resolution may be increased by using zero filling [23, 24] to increase the number of pixels and/or slices in the reconstructed data set.

Scan Synchronization As has been described, in order to ensure the best possible image quality with CE-MRA, it is essential to properly synchronize data acquisition with the arrival and passage of the contrast material. If acquisition is completed before the arrival of the contrast material, the blood will not generate enough signal to appear in the angiogram. If the contrast material arrives during the scan – after data acquisition has begun, but before it is completed – the SNR of the arteries may be low, or artifacts may be present in the images [25]. The artifacts may manifest as a less intense area in the center of the vessels, ringing or replication of the vessel edges, or demonstration of only the edges of the vessels. The appearance of the artifacts depends on the size of the vessels

F.R. Korosec

affected, at what time during the acquisition the contrast material arrives in the vessels being imaged, the order in which the data are acquired, and the rate at which the concentration of the contrast material changes during the scan. If the data are acquired too late, the arterial signal will be diminished, and the veins and stationary tissues will be enhanced. Several methods have been developed to ensure proper timing of the acquisition relative to the passage of the contrast material. In one method, a timing scan is performed prior to acquiring the MR angiogram. For this timing scan, a small bolus of contrast material (1–2 ml) is injected, and then two-dimensional images are rapidly and repeatedly acquired [26]. From these images the arrival time of the contrast material can be determined and used to calculate when to start the acquisition of the three-dimensional angiogram after injecting the full bolus. In other methods, the signal in a volume [27] or an image [28] is monitored and acquisition of the angiogram begins when it has been determined that the contrast material has arrived in the vessels of interest. Finally, time-resolved methods have been developed that repeatedly acquire two-dimensional [29, 30] or three-dimensional [31–34] angiograms continuously during the passage of the contrast material, obviating the need to prospectively determine when the contrast has arrived. In addition to demonstrating the peak arterial frame, time-resolved methods provide information regarding the temporal characteristics of the passage of the bolus of contrast material. Images acquired using a commercially available time-resolved, CE-MRA technique known as TRICKs (Time-Resolved Imaging of Contrast Kinetics) are shown in Fig. 1.27.

Data Acquisition Order A current challenge of CE-MRA is obtaining high spatial resolution in the short amount of time available between arterial and venous enhancement. To address this issue, a method has been developed in which the data acquisition order is modified to acquire the low spatial frequency data early, during enhancement of the arteries, and the high spatial frequency data later, after enhancement of the veins and stationary tissues. This data acquisition order is referred to as elliptical centric phase-encoding order [35, 36]. It allows the acquisition time to be extended to increase the spatial resolution of the images without incurring substantial interference from enhancing veins and stationary tissues. In images, low spatial frequencies demonstrate the bulk of the contrast information, whereas high spatial frequencies demonstrate the detail information (see Fig. 1.23). Acquiring the low spatial frequency information (the contrast information) early in the scan, when only the arterial blood is enhanced, minimizes the amount of signal from veins and stationary tissues in the angiogram.

1

Basic Principles of MRI and MR Angiography

25

Fig. 1.27 Coronal MIP images of 8 of 13 time frames showing the vessels in the lower extremities of a patient acquired using the timeresolved 3D CE-MRA method called TRICKs. Note that the vessels in the

patient’s right leg fill much sooner and much more rapidly than the vessels in the patient’s left leg. The anatomical (nonvascular) MR images showed that this patient had a torn Achilles tendon in their right leg

Fig. 1.28 With 3D CE-MRA, it is common to subtract a “mask” image set, acquired prior to the injection of the contrast material, from an image set acquired during injection of the contrast material. This is an effective means of eliminating signals from stationary tissues that are common to both data sets, allowing the vessels to be much better

visualized. In this data set, the signal from the patient’s bladder is bright due to a prior injection of contrast material used to image vessels in a different anatomic location. Subtraction effectively eliminates this signal as well. Subtraction was performed on the source images. Shown here are the MIP images

If the contrast material reaches the veins and stationary tissues late in the scan, when the high spatial frequencies (the detail information) are being acquired, only the edges of these structures will appear in the images. This edge information from the veins and stationary tissues usually does not significantly interfere with observation of the arteries.

angiogram as shown in Fig. 1.28. The subtraction must be performed on the source slices prior to creation of the MIP images in order to fully retain the content of the three-dimensional data set. Mask subtraction in CE-MRA is analogous to subtraction in X-ray DSA in that the signals from stationary tissues (which remain constant between acquisitions) are eliminated by the subtraction, whereas the signals from vessels (which increased after the injection of the contrast material) persist after the subtraction. Disadvantages of subtraction are that it reduces the signal-to-noise ratio of the angiogram by a factor of 2 (because noise from two images, but signal from only one image, contribute to the angiogram), and it renders the angiograms susceptible to misregistration artifacts that result from patient motion.

Mask Subtraction The signal difference between vessels and stationary tissues can be increased in contrast-enhanced MRA methods by acquiring a precontrast mask image set and subtracting it from the postcontrast image set to produce a subtraction

26

F.R. Korosec

Multistation Exams Imaging more than one vascular territory, or station, during a single exam session presents challenges for CE-MRA. Two approaches are currently in use for performing multistation CE-MRA. In one approach, a separate injection of contrast material is administered just prior to scanning each territory [37]. There are benefits in this multi-injection approach, but the image quality can be adversely affected in two ways. First, the volume of contrast material available for each injection is only a fraction of the maximum allowed, because the volume must be split into multiple injections. The reduced volume leads to lower SNR in the angiograms. Second, residual contrast material remaining from the early injections enhances arteries in the mask acquisitions for the later injections. Because the arteries are slightly enhanced in the mask images, subtraction of the mask images leads to a signal reduction in the arteries in the subtracted angiogram. If the mask images from all stations are acquired before the first injection, they cannot be used to eliminate venous enhancement in the later images caused by residual contrast material remaining from the earlier injections. One benefit of a multi-injection method over a single-injection method (described below) is that more time can be spent imaging each station to achieve higher spatial resolution and extended coverage. This is possible because the scan times do not need to be limited in order to chase the bolus of contrast material. In a second approach utilized for multistation CE-MRA, multiple territories are imaged after administration of a single injection of contrast material [38–40]. After the administration of the contrast material is initiated in this single-injection approach, an image set is rapidly acquired from the first station, then the table is moved to bring the second station into the sensitive region of the MR scanner, and a scan is performed to image the second station. This table movement and data acquisition cycle is continued until all of the stations are imaged. With this method, it is necessary to scan quickly in order to image all the stations when the arterial signal is enhanced, and before the veins and stationary tissues enhance. Scanning quickly limits the spatial resolution and coverage that can be achieved at each station. Often the bolus of contrast material advances very quickly, and enhancement of veins and stationary tissues is inevitable in the later-acquired stations. When single-injection, multistation exams are performed, the mask images for all stations are acquired prior to administration of the contrast material, to prevent the contrast material from enhancing the vessels in any of the mask images. Finally, the rate of the injection must be reduced in

order to extend the duration of the contrast material administration to ensure that contrast material is present during imaging of all the stations.

Time-of-Flight MRA Just like CE-MRA techniques, TOF techniques [41, 42] derive contrast between flowing blood and stationary tissues by manipulating the magnitude of the magnetization, such that the magnetization is large for moving blood and small for stationary tissues. However, unlike CE-MRA techniques, TOF MRA techniques do not require the injection of a contrast material. Instead, TOF MRA techniques rely on the fact that blood is in motion, and stationary tissues are not. By appropriately adjusting the imaging parameters and scan prescription in TOF MRA techniques, a large signal may be obtained from moving blood, while a diminished signal is obtained from stationary tissues.

Imaging Sequence TOF methods can be implemented using two-dimensional or three-dimensional acquisition. For both acquisition methods, a spoiled, fast gradient-echo sequence is used. For twodimensional acquisition, thin slices are imaged, and the pulse sequence timing diagram is similar to the one shown in Fig. 1.16. For three-dimensional acquisition, thin slabs are imaged and the slabs are encoded into slices using a phase-encoding method. The pulse sequence timing diagram for three-dimensional TOF MRA is similar to the one shown in Fig. 1.16, with the addition of a phase-encoding gradient on the slice-selection axis. TOF sequences often employ additional gradients on the slice-selection and frequency-encoding axes to refocus unwanted phase accumulations accrued by spins that are in motion during the application of these imaging gradients. These additional gradients typically are referred to as flow-compensation gradients, velocity-compensation gradients, or first-momentnulling gradients [43, 44], and are discussed further in later chapters. They serve to reduce signal loss caused by intravoxel dephasing of spins traveling at different velocities. When a spoiled, fast gradient-echo sequence is used, the signal from tissues decreases with exposure to an increasing number of RF pulses. The longitudinal magnetization as a function of the number of RF pulses, n, is given by the following equation [45]:

(1 - (cosq E ) )+ M (cosq E ) M (n,q ) = M (1 - E ) n

z

1

0

1

1 - cos q E1

0

1

n

n ³ 1,

(1.12)

1

Basic Principles of MRI and MR Angiography

Fig. 1.29 Calculated signal for a spoiled gradient-echo imaging sequence with a flip angle of 30° and a TR of 5 ms shown as a function of the number of RF pulses experienced for tissues with various T1 values. Note that the signal decreases with increasing number of RF pulses experienced. Note also that the equilibrium signal is larger for tissues with shorter T1 values

where M z- (n,q ) is the longitudinal magnetization just prior to the nth RF pulse, E1def exp( -TR /T1 ) , M 0 is the thermal equilibrium magnetization, and q is the tip angle. given by (1.11). The steady-state equilibrium signal is often referred to as the saturation (or saturated) signal. The signal achieved using a spoiled gradient-echo sequence with a tip angle of 30°, a TR of 5 ms, and TE  T2* is plotted in the graph shown in Fig. 1.29. The signal is shown as a function of number of RF pulses experienced, and the signals are plotted for several T1 values. The graph shows that, independent of the T1 of the tissue, the signals decrease with exposure to an increasing number of RF pulses, until a steady-state equilibrium value is reached. In TOF MRA, the goal is to subject the flowing blood to a very few RF pulses, and to subject stationary tissues to a large number of RF pulses, thereby achieving a signal difference between blood and stationary tissues. The graph also shows that tissues with short T1 values have high steady-state equilibrium signal levels. Thus, tissues with short T1 values, like fat, show up bright in TOF MRA methods. Ensuring that flowing blood experiences only a few RF pulses can be accomplished by imaging slices, or thin slabs, oriented perpendicular to the primary direction of flow. When this is done, the moving blood enters the slice fully magnetized, experiences only a few excitation pulses, and then flows out of the slice before it becomes saturated. This ensures that the signal from blood will be relatively large. This phenomenon is referred to as inflow enhancement, or flow-related enhancement. The stationary tissues, however, remain in the slice, or slab, throughout image acquisition, and so they give rise to a diminished signal because the

27

magnetization from them is saturated due to the constant exposure to the excitation pulses. Data collection typically does not begin until the signals from stationary tissues are brought to their steady-state equilibrium level. The number of excitation pulses experienced by moving blood as it traverses the imaging slice (or slab) depends on a number of factors including the thickness of the slice (or slab), the velocity of the blood, the orientation of the vessel relative to the image slice, and the TR of the imaging sequence. In general, thinner slices (or slabs), faster-flowing blood, vessels oriented perpendicular to the slice (or slab), and longer TRs lead to increased vascular signal. A longer TR, however, also leads to increased signal from stationary tissues (less saturation). Therefore, an intermediate TR typically is selected with TOF MRA methods. Increasing the tip angle leads to diminished signal from stationary tissues, but can also lead to increased saturation of blood that experiences multiple excitation pulses. As a result, an intermediate tip angle typically is selected with TOF MRA methods. These factors and others must be carefully considered when prescribing a TOF MRA sequence.

Two-Dimensional TOF MRA For two-dimensional acquisition, data are acquired from multiple slices stacked contiguously along the long axes of the vessels of interest. Because the image quality is best if the slices remain perpendicular to the direction of flow, 2D TOF MRA methods are best suited for imaging vessels that are straight, such as the carotid arteries, or the vessels in the lower extremities. The data from the slices can be retrospectively reprojected or reformatted to demonstrate long segments of the vessels. With 2D TOF MRA, the spoiled, fast gradient-echo method is prescribed so that thin slices (1–3 mm), oriented perpendicular to the long axis of the vessels of interest are imaged. Prescribing the slices in this manner increases the likelihood that the blood will experience only a very few radiofrequency excitation pulses as it flows through the image slice. When thin slices are imaged, a moderately large tip angle (60°) can be used to suppress the signal from the stationary tissues without substantially suppressing the signal from blood that quickly moves through the image plane. Even when thin slices are imaged, the moderately large tip angle can cause saturation of the signal from slowly moving blood, such as that in the carotid bulb. The degree of saturation can be reduced by decreasing the tip angle and/or increasing the TR of the imaging sequence; however, it must be realized that this also will increase the amount of signal from stationary tissues. Increasing the TR also will lead to a longer scan time. When imaging vessels that contain pulsatile blood flow, it is necessary to synchronize the acquisition of data to the cardiac cycle. Cardiac leads are placed on the patient, and the

28

F.R. Korosec

Fig. 1.30 (a) To produce these cardiac-gated 2D TOF MRA coronal images of the lower legs, (b) a stack of contiguous axial slices was imaged. The data set was acquired in 4 scans (different shaded slices in (b)); the table was moved between scans so that the imaged slices were always in the most uniform magnetic field to yield the highest image quality. The images comprise a three-dimensional data set from which maximum intensity projections (MIPs) at any obliquity can be produced retrospectively

ECG signal is fed to the MRI scanner. Data are then acquired only during systole in order to take advantage of the maximum inflow of fresh, fully magnetized blood, thereby reducing the likelihood of saturating the blood. Acquiring segments of data always at the same point in each cardiac cycle also reduces ghosting artifacts that otherwise would occur due to the changing velocity of blood throughout the cardiac cycle. Images of the arteries in the lower extremities acquired using an ECG-gated, 2D TOF MRA method is shown in Fig. 1.30. With 2D TOF MRA, a spatial saturation pulse [46–48] can be applied parallel to the image slice at the beginning of each TR to reduce or eliminate unwanted signal from blood flowing into the image slice from a particular direction. For example, when imaging the vessels in the neck, applying a saturation pulse superior to an axial image slice is an effective means of eliminating signal from blood in the jugular veins, as shown in Fig. 1.31a. The signal from the blood in the jugular veins is diminished or eliminated as the blood flows through the region affected by the saturation pulse. The venous blood, therefore, has little signal to give when it flows into the image slice. If a saturation pulse was not used, and the blood in the jugular veins was left unsaturated, these vessels would interfere with observation of the carotid arteries as shown in Fig. 1.31b. When imaging the neck, if the saturation pulse is prescribed inferior to the image slice, the signal from the carotid arteries can be eliminated to produce an image of the jugular veins as shown in Fig. 1.31c. In general, inferior saturation pulses are used to suppress the signal from arteries above the heart and veins below the heart, whereas superior saturation pulses are used to suppress the signal from veins above the heart, and arteries below the heart.

Three-Dimensional TOF MRA For three-dimensional TOF acquisition [49, 50], a slab, oriented perpendicular to the long axis of the vessels of interest, is imaged and the slab is encoded into thin slices using a phase-encoding method. Because a slab is imaged, a small tip angle (30°) must be used so the signal from blood that remains in the slab does not become too saturated. The small tip angle necessary to preserve signal from blood also leads to an undesirable preservation of signal from stationary tissues. Therefore, when 3D TOF MRA methods are applied, other mechanisms must be implemented in order to improve contrast between blood and stationary tissues. Some of these mechanisms include magnetization transfer [51–53], fat-andwater out-of-phase imaging, and/or ramped tip angle [54] excitation. These methods are described in later chapters. Three-dimensional TOF MRA methods offer smaller voxels, shorter echo times, and inherently higher signal-tonoise ratios than 2D TOF MRA methods. These are features common to all three-dimensional acquisition methods as described previously. The use of phase-encoding, rather than slice-selection, leads to the ability to image thinner slices (better resolution in reformatted images, smaller voxels), and to use a shorter TE. The combination of the small voxels and the short echo time leads to a reduction in the amount of signal loss caused by intravoxel dephasing. Threedimensional TOF MRA images of the cerebral vessels are shown in Fig. 1.32. Even when using magnetization transfer, fat-and-water out-of-phase imaging, and ramped excitation pulses, contrast between blood and stationary tissues can be small in

1

Basic Principles of MRI and MR Angiography

29

Fig. 1.31 (a) In 2D TOF MRA of the neck, a spatial saturation pulse applied superior to, and very near, each axial imaging slice will eliminate signal from blood in the veins to yield an arteriogram. (b) When no spatial saturation pulse is applied, both arteries and veins appear in the image, making interpretation difficult. (c) Applying a spatial saturation pulse inferior to, and very near, each axial imaging slice will eliminate signal from blood in the carotid arteries to yield a venogram

Fig. 1.32 (a) Axial, (b) coronal, and (c) sagittal MIP images produced from axial images of the intracranial vessels of a healthy adult acquired using a 3D TOF MRA technique

3D TOF MRA when thick slabs are used to achieve extended coverage. To achieve greater coverage with reduced saturation effects, 3D TOF MRA data can be acquired from multiple thin slabs arranged perpendicular to the vessels of interest using a method referred to as multiple overlapping thin slab angiography (MOTSA) [42, 55]. The multislab method combines the thin-slice benefits of two-dimensional acquisition (reduced saturation of blood) with the benefits of three-dimensional acquisition (including an inherently high signal-to-noise ratio, small voxels, and a short echo time) in an effort to provide high quality images as shown in Fig. 1.33. With 3D TOF MRA, a spatial saturation pulse can be applied just outside the slab at the beginning of each TR to eliminate signal from blood that will flow into the imaging slab as described previously for 2D TOF MRA. However, with 3D TOF MRA, the saturation pulse is not as effective.

This is because, as the blood that was initially saturated traverses the slab, the longitudinal magnetization from this blood begins to regrow, and the once-saturated blood eventually gives rise to signal. In general, the farther the oncesaturated blood penetrates into the slab, the less effective the saturation of the signal due to the increasing amount of time that elapses between saturation and signal detection.

Phase-Contrast MRA Phase-contrast (PC) MRA [56, 57] methods differ from CE-MRA and TOF MRA methods in that PC MRA methods provide a direct quantitative measure of the velocity of the flowing blood. Like TOF MRA methods, PC MRA methods can be acquired using 2D or 3D acquisitions. The 2D acquisitions, since they are rapid, can be cardiac-gated to provide

30

F.R. Korosec

Fig. 1.33 (a) Coronal MIP of axial images of the carotid arteries acquired using a multiple overlapping thin slab angiography (MOTSA) technique with a ramped tip angle, and a TE such that fat and water are out of phase with each other. (b) Magnified and cropped oblique MIP of the left carotid bifurcation

Fig. 1.35 A timing diagram for a 2D phase-contrast MRA pulse sequence. The timing diagram is similar to the one shown in Fig. 1.16 with the addition of a bipolar flow-encoding gradient on the frequencyencoding axis (shown as a dotted grey line to distinguish it from the other magnetic field gradients) Fig. 1.34 The transverse magnetization, Mxy, precesses when a magnetic field is applied. The amount that the magnetization rotates is denoted by the phase angle, f

dynamic information regarding blood flow throughout the cardiac cycle. Phase-contrast MRA techniques derive contrast between flowing blood and stationary tissues by manipulating the phase of the magnetization, such that the phase of the magnetization is zero for stationary tissues and non-zero for moving tissues. Phase is a measure of how far the magnetization precesses, or rotates, from the time it is tipped into the transverse plane until the time it is detected. The phase of the precessing transverse magnetization is shown in Fig. 1.34.

Imaging Sequence The pulse sequence used for phase-contrast acquisitions is a modified, spoiled, fast gradient-echo sequence. The most

significant modification to the gradient-echo pulse sequence timing diagram is the addition of a bipolar flow-encoding gradient as shown for 2D acquisition in Fig. 1.35. For 3D acquisition, a phase-encoding gradient would be added to the slice-selection axis. As described below, in order to encode flow in all directions, a data set must be acquired with the bipolar flow-encoding gradient applied on each of the three gradient axes, and to determine and eliminate unwanted phase accumulations from sources other than the bipolar flowencoding gradient, a reference data set also must be acquired. The effect of applying the bipolar flow-encoding gradient in PC MRA is that spins will accumulate a phase in proportion to their velocity. For example, stationary spins will accumulate no net phase shift. This is because the first lobe of the gradient (positive polarity) will impart a positive phase to the stationary spins, and the second lobe of the gradient (negative polarity) will impart an equivalent negative phase to the stationary spins, resulting in no net phase accumulation for stationary spins at the end of the application of the flowencoding gradient. So the signal associated with voxels containing stationary spins will be zero.

1

Basic Principles of MRI and MR Angiography

31

Fig. 1.36 (a) Axial, (b) coronal, and (c) sagittal MIP images of the intracranial vessels of a patient acquired using an experimental 3D PC MRA technique that employs an very rapid undersampled 3D radial

acquisition method (Images courtesy of Kevin Johnson, University of Wisconsin-Madison)

Moving spins, on the other hand, will be left with a net phase accumulation at the end of the application of the bipolar flow-encoding gradient. Since the spins are in motion, they will experience different field strengths during application of the positive and negative lobes of the bipolar flowencoding gradient, and therefore the phase imparted by the positive lobe will not equal the phase imparted by the negative lobe. The resulting phase is proportional to the velocity of the spins since faster moving spins travel farther between (and during) application of the positive and negative lobes of the bipolar flow-encoding gradient. This leads to a bigger difference in the gradient strength experienced by the spins during application of the positive and negative lobes of the gradient, which translates to a greater difference in the phase imparted to the spins by the positive and negative lobes of the bipolar flow-encoding gradient. Note that phase will only be imparted to spins that move along the axis containing the bipolar flow-encoding gradient. The phase imparted to spins by the application of a bipolar flow-encoding gradient on the frequency-encoding axis is given by the following equation:

data set is subtracted from each of the three flow-encoded data sets to eliminate phase accumulations resulting from sources other than the bipolar flow-encoding gradient (magnetic susceptibility, eddy currents, measurement imperfections, etc.). Alternatively, the bipolar flow-encoding gradients may be strategically applied on more than one axis simultaneously, to improve the signal-to-noise ratio of the subtracted images [58, 59]. These alternative methods still require the acquisition of four scans to acquire velocity information in all three Cartesian directions. The scan time for encoding flow in all three Cartesian directions using two-dimensional acquisitions is 4 × TR × Ny × Ave, where TR is the repetition time of the imaging sequence, Ny is the number of phase-encoding values acquired, and “Ave” is the number of signal averages. If flow is encoded in only a single direction, the factor of four is reduced to a factor of two. With two-dimensional acquisitions, a single thick slab is typically imaged. If multiple slabs are imaged, the scan time is determined by multiplying the above equation by the number of slabs imaged. Phase-contrast methods may also be implemented using three-dimensional acquisition [60, 61] as shown in Fig. 1.36. Benefits of three-dimensional acquisition include an inherently high signal-to-noise ratio (due to the effective averaging of signal that is acquired throughout the entire scan) and small voxels (due to the thin slices made possible by the encoding method employed). Also, three-dimensional image sets can be retrospectively reprojected or reformatted to permit observation of the vessels from any orientation. When phase-difference processing is used, images reformatted perpendicular to a vessel can be used to determine volume flow rate through that vessel. Three-dimensional acquisition is more time consuming than two-dimensional acquisition and thus is used less frequently. Due to the long scan times associated with three-dimensional acquisitions, they are not commonly cardiac-gated.

f v = ±g Gbp vFEt 2 ,

(1.13)

where g is the gyromagnetic ratio of the spins, Gbp is the amplitude of the bipolar flow-encoding gradient, vFE is the velocity of the spins along the frequency-encoding axis, t is the duration of each lobe of the bipolar flow-encoding gradient, and the ± indicates that the sign of the phase depends on the sign of the velocity, and whether the first lobe of the bipolar flow-encoding gradient is positive or negative. This equation ignores the ramp times of the bipolar flow-encoding gradient. In order to encode flow in all directions, the bipolar flowencoding gradient must be applied on each of the three gradient axes in separate TR intervals. In addition, a fourth nonflowencoded acquisition must be acquired. The nonflow-encoded

32

F.R. Korosec

Fig. 1.37 Vector diagrams showing the values that are measured for each pixel when implementing (a) phase-difference, (b) complexdifference, and (c) magnitude processing of phase-contrast data. In phase-difference processing, the phase angle, f, through which the magnetization rotates during each of the two acquisitions is

measured. In complex-difference processing, the distance between the ends of the magnetization vectors from the two acquisitions is measured. In magnitude processing, the magnitude of the magnetization vectors from the two acquisitions is measured, and the average is calculated

The scan time required for encoding flow in all directions using three-dimensional acquisitions is 4 × TR × Ave × Ny × Nz, where Nz is the number of slice-encoding values acquired and all other abbreviations are as described previously. If flow is encoded in only a single direction, the factor of four is reduced to a factor of two. With phase-contrast acquisition methods, the subtraction provides high contrast between vessels and stationary tissues, permitting large fields-of-view to be imaged with minimal detrimental effects from saturation, as long as a relatively small tip angle (20°–30°) is used.

by each pixel. Additionally, spins moving in one direction are assigned a bright (white) signal, whereas spins moving in the opposite direction are assigned a dark (black) signal as shown in Fig. 1.38. So, a cursor or region of interest (ROI) can be placed on the phase-difference images, and a direct measure of the velocity (mm/s) of the blood can be determined. In addition, if the cross-section of a vessel is demonstrated in the image, information regarding the volume flow rate of the blood (ml/min) in that vessel can be derived [62]. Furthermore, if the acquisition is cardiac-gated [63, 64], and a series of images is produced demonstrating the cross-section of a vessel throughout the cardiac cycle, then the acquired series of images can be evaluated to determine volume flow rate in that vessel throughout the cardiac cycle. The physiologic effect of vascular pathology can be evaluated by comparing the volume flow rate on the contralateral side with that on the ipsilateral side, or by comparing the volume flow rate before and after a physical or drug-induced challenge. The accuracy of volume flow rate measurements obtained using phase-contrast MRA is dependent on imaging parameters and placement of the ROI [62]. Because phase-difference images show direction of flow in the form of black or white signal, these images may be used to effectively differentiate arteries from veins when they are aligned antiparallel to each other. More importantly, phase-difference images can be used to detect retrograde flow in cases such as in subclavian steal syndrome, where a stenosis or occlusion of the subclavian artery proximal to the vertebral artery causes reversed flow in the latter. In complex-difference processing, vector subtraction is performed to determine the signal in each pixel as shown in Fig. 1.37b. Complex-difference images are shown in Fig. 1.38. With complex-difference processing, the signal in the images is dependent on the motion of the blood (as it is in phase-difference images), but the dependence is not linear (as it is in phase-difference images). Furthermore, the direction of blood flow is not represented in complex-difference

Processing Methods The data acquired with phase-contrast techniques can be processed in three different ways to produce phase-difference, complex-difference, and magnitude images. In phase-difference processing, phase subtraction is performed to determine the signal in each pixel as shown in Fig. 1.37a. For phase-difference processing, the phase of the magnetization is measured at the point in time when the receiver is activated. Since the phase of the magnetization is influenced by factors in addition to the bipolar flow-encoding gradient, typically two measurements are made, with only the bipolar flow-encoding gradient changed between the measurements, so the phase imparted by other factors remains the same for the two measurements. When the data for the two measurements are subtracted, the phase accumulation imparted by the other factors is eliminated since it is the same for the two measurements. The phase imparted by the bipolar flow-encoding gradients is opposite in the two measurements, so subtraction yields an enhancement of the phase caused by application of the bipolar flow-encoding gradient. In phase-difference images, the signal value of each pixel is linearly proportional to the velocity of the spins represented

1

Basic Principles of MRI and MR Angiography

33

Fig. 1.38 Sagittal images of the midline intracranial vessels demonstrating that data acquired using phase-contrast MRA can be reconstructed to produce phase-difference images showing flow in the (a) right/left, (b) anterior/posterior, and (c) superior/inferior directions. In these images, the intensity of the signal is proportional to the velocity of the blood. The intensity in these images can be squared, summed, and the square root can be calculated to produce a “speed” image as

shown in (g). Data acquired using phase-contrast MRA can also be reconstructed to produce complex-difference images showing flow in the (d) right/left, (e) anterior/posterior, and (f) superior/inferior directions. The intensity in these images can be squared, summed, and the square root can be calculated to produce a “speed” image as shown in (h). Finally, the phase-contrast data can be reconstructed to produce a magnitude image as shown in (i)

images, so flow in opposite directions is not represented as black and white as it is in phase-difference images. Therefore, complex-difference images are not used to determine quantitative information regarding blood velocity or volume flow rate, but are used instead for demonstrating the morphology of the vessels. In magnitude processing, the length of the transverse magnetization vector is measured to determine the signal in each pixel as shown in Fig. 1.37c. This is the processing method traditionally used to produce images in MRI. With phase-contrast methods, the phase of the magnetization is manipulated, but the strength of the magnetization, or its magnitude, is unaltered. Thus, displaying the magnitude of the magnetization permits simultaneous demonstration of vessels and stationary tissues as shown in Fig. 1.38.

methods, only velocities within a certain range will be accurately represented in the images. The range of velocities that will be accurately represented is determined by the velocityencoding (Venc) parameter selected when the imaging sequence is prescribed. Blood traveling at velocities higher than the Venc value will be misrepresented in the image, so this value must be chosen carefully. This misrepresentation of velocities outside of the Venc value is referred to as velocity aliasing and appears differently in phase-difference and complex-difference images. In phase-difference images, aliased signal is easy to identify by an abrupt change from very dark to very bright signal as seen in the image shown in Fig. 1.40, or vice versa (see transitions in Fig. 1.39 for phase-difference processing). In complex-difference images, velocity aliasing manifests as a decreasing signal for blood traveling at velocities beyond the Venc value (see Fig. 1.39 for complex-difference processing), and blood traveling at some velocities above the Venc value gives rise to no signal at all. Thus, with complex-difference processing, selecting a Venc that is too low can yield very misleading results that are not easy to identify. With both phasedifference and complex-difference processing, choosing a Venc value that is too low should be avoided in order to

Velocity Encoding and Velocity Aliasing In phase-contrast images, the signals are proportional to the velocities of the imaged spins, as shown in the graphs in Fig. 1.39. The proportionalities are different for phase-difference and complex-difference processing. For both processing

34

Fig. 1.39 Graphs demonstrating the signal as a function of velocity for (top) phase-difference (PD) and (bottom) complex-difference (CD) processing. For phase-difference processing, the signal is linearly proportional to the velocity of the spins – for spins with velocities within the range defined by the velocity-encoding (Venc) value specified by

F.R. Korosec

operator. For complex-difference processing, the signal is sinusoidally proportional to the speed (the signal is the same for positive and negative velocities) of the spins – for spins with velocities within the range defined by the velocity-encoding (Venc) value specified by the operator

veins in an arteriovenous malformation (AVM) can be displayed in separate images that are sensitive to different velocity ranges as shown in Fig. 1.41.

Balanced Steady-State Free Precession MRA

Fig. 1.40 With phase-contrast MRA methods, a velocity-encoding, Venc, value must be selected. Velocities above this value will be misrepresented, or aliased. In phase-difference-processed images, such as the one shown here, velocity aliasing is easy to identify (arrows) as an abrupt change from very dark to very bright signal, or from very bright to very dark signal (not shown)

prevent velocity aliasing. Choosing a Venc value that is too high also should be avoided as it leads to magnetization from all blood accumulating only small phase shifts, which results in low SNR, and a small signal range. To avoid velocity aliasing, some a priori information regarding the anticipated pathology may be helpful in determining the appropriate velocity-encoding value. Alternatively, different velocity-encoding values may be used in different scans to highlight different vessels. For example, arteries and

MRA also can be performed using balanced steady-state free precession (bSSFP) imaging sequences [65]. In bSSFP imaging, both the longitudinal and transverse components of magnetization are maintained in a steady-state condition. In order to ensure that the transverse component of magnetization is not spoiled, all of the imaging gradients must be completely balanced so that the net phase accumulation imparted to the spins from one TR to the next is zero. A pulse sequence timing diagram for a bSSFP sequence is shown in Fig. 1.42. Note that, between RF pulses, the area under the positive gradient waveforms equals the area under the negative gradient waveforms, ensuring that the net phase imparted to any spin by the imaging gradients during this interval is zero. Depending on MRI vendor, this imaging sequence is called FIESTA (fast imaging employing stead-state acquisition), trueFISP (true fast imaging with steady-state precession), or bFFE (balanced fast field echo). Since the transverse magnetization is maintained from one TR to the next, those tissues with long T2 values give rise to large signals in bSSFP methods. Tissues with short T1 values also give rise to large signals. The signal intensity is proportional to T2/T1. The signal intensity achieved with signal

1

Basic Principles of MRI and MR Angiography

35

Fig. 1.41 Phase-contrast MRA images can be acquired with different velocity sensitivities (using different Venc values) to highlight vessels containing blood flowing in different velocity ranges. These sagittal complex-difference-processed phase-contrast images of an arteriovenous malformation (AVM) were acquired with velocity-encoding

Fig. 1.42 A timing diagram for a balanced steady-state free precession (bSSFP) gradient-echo pulse sequence. Note that the gradient waveforms on all three axes are balanced so that the positive area equals the negative area to ensure that no net phase is imparted to any spins during the TR interval. This allows the transverse magnetization to achieve a steady-state value

steady-state free precession methods (when steady-state has been achieved) is given by the following equation [66]: M xy+ = M 0 ´

(1 - E1 ) ´ sin(q ) , 1 - ( E1 - E2 ) ´ cos(q ) - E1 E2

(1.14)

where M xy+ is the transverse magnetization immediately following the RF pulse, E1,2 def exp( -TR /T1,2 ) , M 0 is the thermal equilibrium magnetization, and q is the tip angle. If TE = TR/2, then the transverse magnetization that is detected is given by the following equation [67]: M xy = M xy+ ´ E2 .

(1.15)

A cardiac image acquired with a bSSFP method is shown in Fig. 1.43. Owing to its relatively long T2, blood appears

values of (a) 20 cm/s and (b) 80 cm/s. Note that the slower flow in the nidus of the AVM is better demonstrated in the image shown in (a), whereas the faster flow in the feeding arteries is better demonstrated in the image shown in (b)

Fig. 1.43 One of 16 images of a patient’s heart acquired during a single breath-hold interval using a cardiac-gated, balanced steady-state free precession (bSSFP) method. With these methods, multiple images of the same anatomic slice are typically acquired. Each image shows the heart in a different phase of the cardiac cycle. When all of the images are displayed in rapid succession (as a movie loop), the cardiac dynamics are revealed

bright in this image, providing high contrast between it and the surrounding myocardium. Other tissues with large T2/T1 ratios also appear bright in images acquired with bSSFP methods, limiting the application of these methods for MRA applications. Fat and venous blood have large T2/T1 ratios. Fat suppression may be achieved to varying degrees of success by using water-selective excitation pulse, or repeated spectral fat-saturation pulses. Similarly, venous suppression may be achieved to varying degrees of success by using spatial saturation or inversion pulses. Balanced SSFP methods suffer from signal loss in regions of magnetic field nonuniformities, since, under these circumstances, the requirement for zero net phase accumulation

36

between TRs is not satisfied. Thus, shimming the magnetic field is very important when using bSSFP methods. In order to reduce phase accumulations caused by remaining magnetic field nonuniformities, short TRs are employed with bSSFP methods. Balanced SSFP methods have proven useful for cardiac imaging applications [68], but owing to the remaining challenges faced by these methods, they currently are not widely used for MRA applications.

Summary Under appropriate conditions, and when properly implemented, all MR angiographic methods can yield high-quality diagnostic images. Three-dimensional MRA methods offer small voxels, short echo times, inherently high signal-tonoise ratios, and the ability to retrospectively reformat image volumes to show tortuous vessels from any viewing angle. Three-dimensional CE-MRA methods can provide high quality vascular images with less sensitivity to artifacts (caused by intravoxel dephasing, signal saturation, cardiac pulsatility, and patient motion) and shorter scan times than noncontrast-enhanced MRA methods. Intravenous injection of a T1-shortening contrast material provides high SNR angiograms with high contrast between vessels and stationary tissues. Relatively high spatial resolution and large volume coverage may be achieved in reasonable scan times. In 3D CE-MRA images, vascular pathologies are well delineated. Owing to its many attributes, 3D CE-MRA is widely used for imaging vessels throughout the body, and in some situations is replacing X-ray DSA as the imaging method-of-choice. Three-dimensional TOF MRA methods are often used for imaging the intracranial arteries, as well as the extracranial carotid and vertebral arteries. Two-dimensional TOF methods are sensitive to signal loss from intravoxel dephasing and often are used as a highly sensitive and specific method for screening for stenoses of the carotid or lower extremity arteries. Two-dimensional TOF also may be used with an inferior spatial saturation pulse to image the intracranial veins. Twodimensional TOF MRA may be cardiac-gated to provide images in regions affected by cardiac pulsatility. Phase-contrast methods offer qualitative and quantitative information regarding blood velocity or volume flow rate. Two-dimensional phase-contrast methods offer short scan times, and often are used to provide multiple images, each sensitive to a different range of velocities, or are cardiacgated to provide velocity or flow information throughout the cardiac cycle. Three-dimensional phase-contrast images require long scan times, but the volume of information may be retrospectively analyzed to yield average velocity or flow information through any vessel, no matter what its orientation. The use of a bipolar flow-encoding gradient makes

F.R. Korosec

these sequences particularly susceptible to signal loss caused by intravoxel dephasing. bSSFP methods are widely used for cardiac applications, but owing to several remaining challenges, are not yet routinely used for MRA applications.

References 1. Bernstein MA, Huston J 3rd, Lin C, Gibbs GF, Felmlee JP. Highresolution intracranial and cervical MRA at 3.0T: technical considerations and initial experience. Magn Reson Med. 2001;46:955–962. 2. Thomas SD, Al-Kwifi O, Emery DJ, Wilman AH. Application of magnetization transfer at 3.0 T in three-dimensional time-of-flight magnetic resonance angiography of the intracranial arteries. J Magn Reson Imaging. 2002;15:479–483. 3. Willinek WA, Born M, Simon B, Tschampa HJ, Krautmacher C, Gieseke J, Urbach H, Textor HJ, Schild HH. Time-of-flight MR angiography: comparison of 3.0-T imaging and 1.5-T imaging-initial experience. Radiology. 2003;229:913–920. 4. Leiner T, de Vries M, Hoogeveen R, Vasbinder GB, Lemaire E, van Engelshoven JM. Contrast-enhanced peripheral MR angiography at 3.0 Tesla: initial experience with a whole-body scanner in healthy volunteers. J Magn Reson Imaging. 2003;17:609–614. 5. Gibbs GF, Huston J 3rd, Bernstein MA, Riederer SJ, Brown RD Jr. Improved image quality of intracranial aneurysms: 3.0-T versus 1.5-T time-of-flight MR angiography. Am J Neuroradiol. 2004;25:84–87. 6. Edelstein WA, Hutchison JM, Johnson G, Redpath T. Spin warp NMR imaging and applications to human whole-body imaging. Phys Med Biol. 1980;25:751–756. 7. Glover GH, Pauly JM. Projection reconstruction techniques for reduction of motion effects in MRI. Magn Reson Med. 1992;28: 275–289. 8. Meyer CH, Hu BS, Nishimura DG, Macovski A. Fast spiral coronary artery imaging. Magn Reson Med. 1992;28:202–213. 9. Barger AV, Block WF, Toropov Y, Grist TM, Mistretta CA.Timeresolved contrast-enhanced imaging with isotropic resolution and broad coverage using an undersampled 3D projection trajectory. Magn Reson Med. 2002;48:297–305. 10. Thedens DR, Irarrazaval P, Sachs TS, Meyer CH, Nishimura DG. Fast magnetic resonance coronary angiography with a three-dimensional stack of spirals trajectory. Magn Reson Med. 1999;41:1170–1179. 11. Gurney PT, Hargreaves BA, Nishimura DG. Design and analysis of a practical 3D cones trajectory. Magn Reson Med. 2006;55: 575–582. 12. Macovski A. Noise in MRI. Magn Reson Med. 1996;36:494–497. 13. Sodickson DK, Manning WJ. Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med. 1997;38:591–603. 14. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999;42:952–962. 15. Weiger M, Pruessmann KP, Kassner A, Roditi G, Lawton T, Reid A, Boesiger P. Contrast-enhanced 3D MRA using SENSE. J Magn Reson Imaging. 2000;12:671–77. 16. Griswold MA, Jakob PM, Heidemann RM, Nittka M, Jellus V, Wang J, Kiefer B, Haase A. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med. 2002;47:1202–1210. 17. Prince MR, Yucel EK, Kaufman JA, et al.: Dynamic gadoliniumenhanced three-dimensional abdominal MR arteriography. J Magn Reson Imaging. 1993;3:877–881. 18. Kroft LJ, de Roos A. Blood pool contrast agents for cardiovascular MR imaging. J Magn Reson Imaging. 1999;10:395–403. 19. Perreault P, Edelman MA, Baum RA, Yucel EK, Weisskoff RM, Shamsi K, Mohler ER 3rd. MR angiography with gadofosveset

1

20.

21. 22.

23.

24.

25.

26.

27.

28.

29.

30.

31. 32.

33.

34.

35.

36.

37.

Basic Principles of MRI and MR Angiography trisodium for peripheral vascular disease: phase II trial. Radiology. 2003;229:811–820. Thomsen HS, Morcos SK, Dawson P. Is there a causal relation between the administration of gadolinium based contrast media and the development of nephrogenic systemic fibrosis (NSF)? Clin Radiol. 2006;61:905–906. Zur Y, Wood ML, Neuringer LJ. Spoiling of transverse magnetization in steady-state sequences. Magn Reson Med. 1991;21:251–263. Haacke E, Brown R, Thompson M, Venkatesan R. Magnetic Resonance Imaging: Physical Principles and Sequence Design. New York: John Wiley and Sons, Inc.; 1999:455. Du YP. Parker DL. Davis WL, Cao G. Reduction of partial-volume artifacts with zero-filled interpolation in three-dimensional MR angiography. J Magn Reson Imaging. 1994;4:733–741. Bernstein MA, Fain SB, Riederer SJ. Effect of windowing and zero-filled reconstruction of MRI data on spatial resolution and acquisition strategy. J Magn Reson Imaging. 2001;14:270–280. Maki J, Prince M, Londy F, Chenevert T. The effects of time varying intravascular signal intensity and k-space acquisition order on three-dimensional MR angiography image quality. J Magn Reson Imaging. 1996;6:642–651. Earls J, Rofsky N, DeCorato D, et al. Breath-hold single-dose gadolinium-enhanced three-dimensional MR aortography: usefulness of a timing examination and MR power injector. Radiology. 1996;201:705–710. Foo T, Saranathan M, Prince M, Chenevert T. Automated detection of bolus arrival and initiation of data acquisition in fast, threedimensional, gadolinium enhanced MR angiography. Radiology. 1997;203:275–280. Huston J, Fain S, Riederer S, et al. Carotid arteries: Maximizing arterial to venous contrast in fluoroscopically triggered contrastenhanced MR angiography with elliptical centric view ordering. Radiology. 1999;211:265–273. Wang Y, Johnston D, Breen J, et al. Dynamic MR digital subtraction angiography using contrast enhancement, fast data acquisition, and complex subtraction. Magn Reson Med. 1996;36:551–556. Hennig J, Scheffler K, Laubenberger J, Strecker R. Time-resolved projection angiography after bolus injection of contrast agent. Magn Reson Med. 1997;37:341–345. Korosec F, Frayne R, Grist T, Mistretta C. Time-resolved contrastenhanced 3D MR angiography. Magn Reson Med. 1996;36:345–351. Willig DS, Turski PA, Frayne R, et al. Contrast-enhanced 3D MR DSA of the carotid artery bifurcation: Preliminary study of comparison with unenhanced 2D and 3D time-of-flight MR angiography. Radiology. 1998;208:447–451. van Vaals JJ, Brummer ME, Dixon WT, Tuithof HH, Engels H, Nelson RC, Gerety BM, Chezmar JL, den Boer JA. “Keyhole” method for accelerating imaging of contrast agent uptake. J Magn Reson Imaging. 1993;3:671–675. Willinek WA, Hadizadeh DR, von Falkenhausen M, Urbach H, Hoogeveen R, Schild HH, Gieseke J. 4D time-resolved MR angiography with keyhole (4D-TRAK): more than 60 times accelerated MRA using a combination of CENTRA, keyhole, and SENSE at 3.0T. J Magn Reson Imaging. 2008;27:1455–1460. Wilman A, Riederer S, Huston J, et al. Arterial phase carotid and vertebral artery imaging in 3D contrast-enhanced MR angiography by combining fluoroscopic triggering with an elliptical centric acquisition order. Magn Reson Med. 1998;40:24–35. Huston J, Fain SB, Wald JT, Luetmer PH, et al. Carotid artery: elliptic centric contrast-enhanced MR angiography compared with conventional angiography. Radiology. 2001;218:138–143. Earls JP, Patel NH, Smith PA, DeSena S, Meissner MH. Gadoliniumenhanced three-dimensional MR angiography of the aorta and peripheral arteries: evaluation of a multistation examination using two gadopentetate dimeglumine infusions. AJR Am J Roentgenol. 1998;171:599–604.

37 38. Ho KY, Leiner T, de Haan MW, Kessels AG, Kitslaar PJ, van Engelshoven JM. Peripheral vascular tree stenoses: evaluation with moving-bed infusion-tracking MR angiography. Radiology. 1998; 206:683–692. 39. Earls JP, DeSena S, Bluemke DA. Gadolinium-enhanced threedimensional MR angiography of the entire aorta and iliac arteries with dynamic manual table translation. Radiology. 1998;209: 844–849. 40. Meaney JF, Ridgway JP, Chakraverty S, et al. Stepping-table gadolinium-enhanced digital subtraction MR angiography of the aorta and lower extremity arteries: preliminary experience. Radiology. 1999;211:59–67. 41. Keller PJ, Drayer BP, Fram EK, et al. MR angiography with twodimensional acquisition and three-dimensional display: work in progress. Radiology. 1989;173:527–532. 42. Parker DL, Yuan C, Blatter DD. MR angiography by multiple thin slab 3D acquisition. Magn Reson Med. 1991;17:434–451. 43. Pattany PM, Phillips JJ, Chiu LC, Lipcamon JD, et al. Motion artifact suppression technique (MAST) for MR imaging. J Comput Assist Tomogr. 1987;11:369–377. 44. Haacke E, Lenz G. Improving MR image quality in the presence of motion by using rephasing gradients. AJR Am J Roentgenol. 1987;148:1251–1258. 45. Haacke E, Brown R, Thompson M, Venkatesan R. Magnetic Resonance Imaging: Physical Principles and Sequence Design. New York: John Wiley and Sons, Inc.; 1999:461. 46. Felmlee JP, Ehman RL. Spatial presaturation: a method for suppressing flow artifacts and improving depiction of vascular anatomy in MR imaging. Radiology. 1987;164:559–564. 47. Edelman RR, Atkinson DJ, Silver MS, et al. FRODO pulse sequences: a new means of eliminating motion flow, and wraparound artifacts. Radiology. 1988;166:231–236. 48. Ehman RL, Felmlee JP. Flow artifact reduction in MRI: a review of the roles of gradient moment nulling and spatial presaturation. Magn Reson Med. 1990;14:293–307. 49. Ruggieri PM, Laub GA, Marsaryk TJ, et al. Intracranial circulation: pulse-sequence considerations in three-dimensional (volume) MR angiography. Radiology. 1989;171:785–791. 50. Schmalbrock P, Yuan C, Chakeres DW, et al. Volume MR angiography: methods to achieve very short echo times. Radiology. 1990;175:861–865. 51. Wolff SD and Balaban RS. Magnetization transfer contrast (MTC) and tissue water proton relaxation in vivo. Magn Reson Med. 1989;10:135–144. 52. Pike GB, Hu BS, Glover GH, Enzmann DR. Magnetization transfer time-of-flight magnetic resonance angiography. Magn Reson Med. 1992;25:372–379. 53. Edelman RR, Ahn SS, Chien D, et al. Improved time-of-flight MR angiography of the brain with magnetization transfer contrast. Radiology. 1992;184:395–399. 54. Nagele T, Klose U, Grodd W, Petersen D, et al. The effects of linearly increasing flip angles on 3D inflow MR angiography. Magn Reson Med. 1994;31:561–566. 55. Blatter DD, Parker DL, Robinson R. Cerebral MR angiography with multiple overlapping thin slab acquisition. Radiology. 1991; 179:805–811. 56. Moran PR. A flow zeugmatographic interlace for NMR imaging in humans. Magn Reson Imaging. 1982;1:197–203. 57. Dumoulin CL and Hart HR. Magnetic resonance angiography. Radiology. 1986;161:717–720. 58. Pelc NJ, Bernstein MA, Shimakawa A, Glover GH. Encoding strategies for three-direction phase-contrast MR imaging of flow. Magn Reson Imaging. 1991;1:405–413. 59. Hausman R, Lewin JS, Laub G. Phase-contrast MR angiography with reduced acquisition time: new concepts in sequence design. J Magn Reson. 1991;1:415–422.

38 60. Dumoulin C, Souza S, Walker M, Wagle W. Three-dimensional phase contrast angiography. Magn Reson Med. 1989;9:139–149. 61. Pernicone JR, Siebert JE, Potchen EJ, et al. Three-dimensional phase-contrast MR angiography in the head and neck: preliminary report. AJR Am J Roentgenol. 1990;155:167–176. 62. Korosec F and Turski P. Velocity and volume flow rate measurements using phase contrast magnetic resonance imaging (review article). Int J Neuroradiology. 1997;3:293–318. 63. Nayler GL, Firmin DN, Longmore DB. Blood flow imaging by cine magnetic resonance. J Comput Assist Tomogr. 1986;10:715–722. 64. Pelc NJ, Herfkens RJ, Shimakawa A, Enzmann DR. Phase contrast cine magnetic resonance imaging. Magn Reson Q. 1991;7:229–254.

F.R. Korosec 65. Scheffler K, Lehnhardt S. Principles and applications of balanced SSFP techniques (review article). Eur Radiol. 2003;13: 2409–2418. 66. Haacke E, Brown R, Thompson M, Venkatesan R. Magnetic Resonance Imaging: Physical Principles and Sequence Design. New York: John Wiley and Sons, Inc.; 1999:480. 67. Scheffler K, Hennig J. Is true FISP a gradient-echo or a spin-echo sequence? Magn Reson Med. 2003;49:395–397. 68. Carr JC, Simonetti O, Bundy J, Li D, Pereles S, and Finn JP. Cine MR angiography of the heart with segmented true fast imaging with steady-state precession. Radiology. 2001;219: 828–834.

2

Time-of-Flight Angiography Seong-Eun Kim and Dennis L. Parker

Introduction Time-of-flight (TOF) angiography is a widely used means of producing angiographic images without the need for the injection of contrast agent. In this chapter, we discuss the MR imaging (MRI) physics which governs the creation of TOF angiograms. We also describe the MRI pulse sequence used for TOF imaging, sources of artifact, and compensatory mechanisms to reduce the deleterious effect of these artifacts.

MR Signal in Time of Flight The flow sensitivity of MRI methods is based on TOF effects, where the amplitude of the signal from flowing blood changes as it moves into the imaged volume, and phase effects, where motion of the blood during applied gradients results in a phase change due to motion. Both flow phenomena can be used to differentiate flowing spins from stationary spins by evaluating either the magnitude or the phase in the acquired MRI data. TOF effects influence the signal intensities of moving blood in MR image in nonangiographic applications. For example, flow voids resulting from the motion through a slice- and flow-related enhancement (FRE) have commonly been seen in both spin echo and gradient echo imaging, respectively. TOF image contrast is based on establishing a difference in the longitudinal magnetization of moving spins relative to stationary spins. The TOF effect was first reported in a nonimaging application by Suryan [1] and in the imaging of blood vessels by Hinshaw et al. [2]. The magnetization of a bolus of flowing blood is typically modified at one location and detected a short time later at another location. Since time elapses between the modification and detection S.-E. Kim, PhD () • D.L. Parker, PhD Department of Radiology, Utah Center for Advanced Imaging Research, Salt Lake City, UT, USA e-mail: [email protected]

of the flowing magnetization, this effect is referred to as the “TOF” effect. TOF techniques in MRA can be divided into those which yield a high signal from flowing spins and low signal from the background tissue (white blood) and those which yield strong signal from the background tissue and little signal from the flowing spins (black blood). White blood (sometimes called bright blood) techniques are the most commonly used for angiography and are the focus of the remainder of this chapter. In white blood TOF, static tissue is suppressed by using a spoiled gradient echo (SPGRE) sequence with relatively short TR. The spoiling suppresses (i.e., saturates) the signal from static tissue, and the TR is adjusted to be long enough so that a sufficient amount of blood can flow into the imaging plane creating contrast between flowing and static tissue. TOF images can be acquired as two-dimensional (2D) or three-dimensional (3D) images. The acquisition of a 3D volume receives signal simultaneously from the entire volume of interest while a 2D acquisition receives signal sequentially from a series of image slices, one slice at a time. In both cases, images are stored as a 3D image dataset. For physician review, a projection angiogram from the 3D dataset is created by projecting the reconstructed image values through the 3D image volume, most commonly using the maximum intensity projection (MIP) [3]. 3D techniques in general have the advantage of higher signal-to-noise ratio (SNR) and higher spatial resolution than 2D techniques at the expense of lower blood vessel signal because the blood remains in the slab for a significant fraction of the imaging time. 2D TOF images generally have higher contrast between blood and background tissue than 3D techniques but have lower spatial resolution due to slice thickness larger than 2 mm. To reduce blood signal saturation while maintaining the SNR of 3D acquisition, Parker et al. [4] developed the multiple overlapping thin 3D slab acquisition (MOTSA) technique. MOTSA acquisitions have the high SNR typical of 3D acquisition with improved vessel contrast common to 2D acquisitions.

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_2, © Springer Science+Business Media, LLC 2012

39

40

The goal of TOF MRA is to provide an accurate depiction of the blood within the vascular lumen (artery or vein) noninvasively and various TOF techniques have been successful in meeting this challenge. The best results are achieved when appropriate pulse sequence parameters are matched to each clinical application. In some cases, alterations to the standard SPGR pulse sequence, such as the use of magnetization transfer (MT) saturation, flow compensation, MOTSA [5], or TONE [6], can improve diagnostic accuracy. In general, TOF MRA techniques are clinically useful in the evaluation of small aneurysms, atherosclerosis, vasospasm, and inflammatory vasculitis.

S.-E. Kim and D.L. Parker

blood with plug flow, uniform velocity throughout the radius of the vessel, with speed v in a blood vessel with flow perpendicular to a slice to be imaged, as shown in Fig. 2.1. During the time, TR, between RF pulses, the fluid moves a distance, dz. Thus, in the time, TR, a length, dz, of nonexcited blood moves into the imaged slice. If dz is greater than slice thickness, z, the entire vessel segment within the slice is replaced by fresh inflowing blood shown as region I in Fig. 2.1a. If dz is less than z, then there will be sections of thickness dz that see one, then two, and three RF pulses as shown as region I, II, and II, respectively, in Fig. 2.1b. If the velocity of blood is exactly z/TR, the full slice thickness, z = vTR, will be completely replaced with fresh inflowing blood. The critical speed (2.1) is defined as:

Quantification of the Time-of-Flight Effect Vc ≡

z . TR

(2.1)

The TOF effect has been reviewed by Axel et al. [7], Gullberg et al. [8], and Nishimura [9] who modeled the signal on the basis of the Bloch equations for a variety of pulse sequence schemes for both plug and laminar flow. If the spins in stationary tissue experience a large number of RF pulses, the longitudinal magnetization of the stationary spins approaches a steady-state equilibrium value that is independent of position within the slice. However, when flowing spins, such as in blood, are flowing into and out of the slice, they may be subjected to fewer RF pulses resulting in a different steadystate magnetization. In general, the TOF effect leads to a diminished blood signal in spin echo imaging. However, in gradient echo images with short TR, the TOF effect results in inflow signal enhancement, which increases the signal from flowing blood relative to static tissue. As a simplistic example, consider

When the speed of blood is faster than Vc, the blood in the vessel that lies in the selected slice is completely refreshed by blood containing unsaturated spins. The fresh blood results in a higher signal relative to the stationary tissue signal since stationary tissue experiences many more RF pulses and is much more saturated. This effect is called wash-in, inflow enhancement or FRE. If v < Vc, partial saturation of the blood will begin to take place for distances into the slice greater than vTR. Fresh (fully magnetized) blood flowing into the imaging slice restores some of the signal intensity lost to partial saturation. In the simple example given above, the flow direction was perpendicular to the slice plane. Inflow enhancement of flow signal decreases for blood flow that transverses the blood vessel obliquely, and increases as

Fig. 2.1 TOF effects in the presence of plug flow with speed v in direction z. (a) When the velocity is higher than the critical velocity as given in (2.1), spins in the blood region I experience only one RF pulse and then exit the imaging slice before the next RF application. Spins in region II are not affected by the first excitation but by the next RF pulse.

(b) When the blood is moving slower than the critical velocity, the saturation effect can be understood by dividing the slice into multiple segments. Blood located in each segment experiences a different number of RF pulses, RF(n), resulting in different saturation effects depending on n

2

Time-of-Flight Angiography

41

the blood’s speed increases above Vc. Inflow enhancement also increases as slice thickness is reduced since Vc is proportional to the slice thickness, z. If the perpendicular component of velocity exceeds the critical velocity, there is complete inflow replacement in the imaging slice during each TR. A further increase in velocity results in no further increase in blood signal enhancement and might even begin to decrease the signal due to intravoxel phase dispersion. To extend this mathematical model to 3D TOF and MOTSA sequences, the slice thickness (z) can be replaced by the slab thickness, Nzdz. If the blood vessel is straight and perpendicular to the imaging slab, the critical velocity (2.2) is given by Vc =

N z dz TR

(2.2)

and for velocities greater than the critical velocity, the blood inside the slab is completely replaced by fresh blood flowing from the outside of the slab. For standard SPGR sequences, the TOF effect can be explained as the difference in signal saturation between flowing and static spins. Spins flowing into a slice may be less saturated than static spins resulting in angiographic contrast differences. The longitudinal magnetization of a stationary spin (i.e., after a series of RF pulses) with equilibrium magnetization, Mo, and longitudinal relaxation time, T1, subjected to the SPGR pulse sequence is given by [10]: M z ss =

M 0 (1 − e − TR/T1 ) , 1 − e − TR/T1 cos q

(2.3)

where q is the flip angle and TR is the repetition time of the sequence, T1 is the longitudinal relaxation time of blood which is almost same as T1 of tissue, and M0 is the equilibrium longitudinal magnetization. Before reaching the steady state, the longitudinal pulse (2.4) is given by: M z (n − ) = M z ss + (e − TR / T1 cos q )n −1 ( M 0 − M z ss ) n ≥ 1. (2.4) The transverse magnetization after the nth RF pulse (2.5) is given by M + (n) = M z (n − )sin qe − TE/T2 .

(2.5)

In cases when TR is much shorter than T1, as the number of RF pulses, n, increases, the contribution of the second term in (2.4) gets very small. Under these approximations, the transverse magnetization after the nth RF pulse (2.6) can be simplified as: M+ =

M 0 (1 − e − TR/T1 ) sin q e − TE/T2 . 1 − e − TR/T1 cosq

(2.6)

As the TR/T1 ratio decreases, the transverse magnetization given in (2.6) monotonically decreases. In other words, saturation of the MRI signal increases. The maximum signal occurs when the flip angle is the Ernst angle (qE = arcos(exp(−TR/T1)), which implies that saturation is dominant over the creation of transverse magnetization for q > qE. Inflowing fresh or fully magnetized blood entering into the imaging slice restores some of the signal intensity lost to partial saturation. Inflow enhancement increases with the blood velocity and as the imaging slice becomes perpendicular to the velocity direction. It also increases as slice thickness is reduced. As mentioned previously, if the perpendicular component of velocity exceeds the critical velocity, there is complete inflow replacement in the imaging slice during each TR. A further increase in velocity yields no further increase in the signal enhancement in blood and might even begin to decrease the signal due to intravoxel dephasing effects which are discussed in the next section. When the blood experiences a single RF excitation pulse in an SPRE sequence, the amount of inflow enhancement in the spoiled gradient sequence is given by the difference (2.7): M + (n) − M + = M0 sin q (e − TR/T1 cosq )n −1 ⎛ (1 − e − TR/T1 ) ⎞ − TE/T2* 1 − . ⎜⎝ (1 − e − TR/T1 cosq ) ⎟⎠ e

(2.7)

By using (2.7), the inflow enhancement effect can be simulated and quantified. Figure 2.2 shows computer simulations based on (2.7). In these simulations, we studied how the normalized signal of the flow enhancement effect (FRE) changes as the flip angle (q) increases for two cases: (a) TR/T1 = 0.02 and (b) when TR/T1 = 0.1. The normalized signal was obtained by dividing (2.6) by M0e−TE/T2*. The simulation demonstrates that FRE increases as n decreases when the TR and slice thickness are fixed. For a fixed TR and slice thickness, increasing n means faster flow velocity. This trend continues until v = Vmax, at which point the flow experiences only one RF pulse and the flow enhancement effect reaches a maximum. If n = 1, the maximum transverse magnetization (2.8) is given by: *

M + (n = 1) = M0 sin q e − TE/T2 .

(2.8)

Figure 2.2 shows that if there is no partial saturation (n = 1), the higher RF flip angle (90°) results in a maximum FRE. Comparing the curves, we see that the ratio of TR/T1 decreases and the flow enhancement intensity increases. When velocity v is smaller than the critical velocity, the flow enhancement effect can be modeled by subdivision of imaging slices into multiple compartments and summation of the geometric series shown in Fig. 2.1.

42

S.-E. Kim and D.L. Parker

Fig. 2.2 Plots of the flow enhancement effect (FRE) simulated from (2.8). The label n is the number of excitation RF pulses experienced by spins located in the imaging slices

In general, arterial velocity is an interesting function of the branching nature of the vascular bed. From the aorta to the capillaries, as the vascular branches, the vessel crosssection becomes progressively smaller. However, the number of branches increases in such a manner that the total cross-section increases as the branching increases. Thus, from geometry alone, arterial velocity decreases with distance from the heart. In general, the velocity remains sufficiently high for successful inflow enhancement until diameters smaller than 0.2 mm are reached. For vessels that can be seen with TOF MRA, the velocities become relatively constant (less pulsatile) for the more distal segments.

Phase Dispersion and Flow Compensation Signal intensity in MRI TOF imaging depends upon the imaging pulse sequence and the geometry and nature (velocity profile, pulsatility, etc.) of flow. Because of viscosity, flowing blood experiences frictional forces from the surrounding blood and vessel wall in addition to the force due to the pressure drop along the vessel. These forces generally result in a form of laminar flow. Laminar flow is characterized by a parabolic velocity profile, where the velocities in the center of the vessel are greater than those at the vessel wall. The magnetic field gradients used for spatial signal encoding in MRI impart a velocity-dependent phase at the time of signal acquisition. Because the velocity of blood varies considerably over dimensions that are much smaller than an image voxel, laminar flow can result in a range of phases for the signal generating spins within a voxel. If not properly compensated, this phase dispersion results in a loss in signal from flowing spins. High-velocity fluid motion through static vessels and arterial branches, such as the carotid bifurcation, produces complex flow patterns, including flow vortices (recirculation) and unsteady, nonrepetitive pulsatile, or turbulent flow.

Flow vortices and unsteady, turbulent flow increase phase dispersion of spin coherence due to the multiple directions of motion, acceleration, and higher order motions. Even simple pulsatile flow can result in magnitude and phase signal variations that are a complex function of time. These temporal changes can cause spatial misregistration of pulsatile flow in images. Venous flow is much less pulsatile than arterial flow. Flow velocities in the human body under normal conditions range from a few mm/s up to 180 cm/s. The phase dispersion among spins having the same constant velocities can be recovered with addition of the gradient waveform lobes known as “flow compensation” or “first-order gradient moment nulling.” If the spins are flowing through a gradient m(G), the time-dependent phase of flowing spins (f) at the location r (2.9) is given by f (t ) = ∫ g G (t ) × r (t )dt a f (t ) = r0 × ∫ g G (t )dt + v × ∫ g G (t )tdt + ∫ g G (t )t 2 dt +  2 a f (t ) = r0 × g × m 0 + v × g × m1 + × g × m2 +  (2.9) 2

where the expansion is obtained from a Taylor series expansion of r(t), r0 is the initial location of the spins, v is the velocity, a is an acceleration of the flow, and mj is the jth moment of the gradient. Equation (2.9) demonstrates that the behavior of the phase accumulated by the moving spins depends on the initial position, velocity, and gradient strength. If the velocity of each flowing spin is not constant, the phase accumulated after gradient application is not the same for each spin and varies with the velocity. This phase dispersion, if uncorrected, results in signal loss. To illustrate the effects of velocity on signal phase, an example of RF and gradient waveforms for a conventional SPGR sequence is shown in Fig. 2.3a. Slice selection (SS) is performed by the RF pulse in conjunction with the slice

2

Time-of-Flight Angiography

43

Fig. 2.3 3D SPGR pulse sequence for TOF acquisition without flow compensation (a) and with flow compensation in the slice selection and readout direction indicated by black arrows (b). The abbreviations

correspond to data acquisition (ADC), RF pulse (RF), readout (RO) gradient, phase-encoding (PE) gradient, and slice selection (SS) gradient

selection z gradient. The amplitude of the slice selection gradient is determined by the desired slice or slab thickness and the bandwidth of the RF pulse. A refocusing lobe is placed after the end of the RF pulse to compensate for the phase dispersion that occurs after the effective tipping of the magnetization into the transverse plane. The readout (RO) gradient, Gx, that is applied immediately after the RF pulse includes a dephasing lobe prior to signal acquisition so that the received signal is completely in phase at the time, TE, from the RF excitation. The amplitude of the readout gradient is determined by the desired resolution and the readout sampling bandwidth. Finally, a small gradient pulse is applied to the y gradient to create a variation in phase in the y direction during signal readout. This pulse sequence is repeated and the y gradient is stepped through values from a negative maximum to positive maximum. The SPGR works well for stationary tissues. The zeroth moment, m0, of both the slice selection and readout gradients is zero at the echo time, TE. The maximum signal occurs for the pulse sequence step, where the y phase-encoding gradient passes through the value, 0. At the time, TE, for this signal measurement, defined as the center of k-space, the stationary spins are completely in phase across the imaging volume. For the signal acquired when the y phase-encoding gradient has zero amplitude, all moments of the y gradient are zero. However, m1 and the higher gradient moments for the slice selection and readout gradients are not zero at time, TE. Thus, moving spins experience an additional phase shift due to motion during these gradients. Velocity dispersion due to laminar flow results in phase dispersion within a voxel and a net signal loss from flowing spins. Velocity-dependent phase dispersion can be corrected by a process known as velocity compensation or first-order gradient moment nulling (Fig. 2.3b). This is accomplished by adding additional area, positive and negative, to the gradient pulses used for to the slice selection and readout.

The magnitude of the lobes can then be adjusted to ensure that the moments of the waveform, when integrated over time, do not contribute to the velocity-dependent phase dispersion. In other words, flow compensation gradients “null” or zero the effects of phase dispersion. The amplitudes of the two compensation gradient lobes for each gradient are adjusted to null m0 and m1 at the times t = 0 and t = TE, respectively. In this manner, the phase of stationary and uniformly flowing spins is the same, independent of the velocity of flow velocity. An additional artifact can occur because of the difference in timing between the phase-encoding and readout gradients. If the time of the phase-encoding gradient is TP, there will be a position shift of v*(TE − TP) between the time the “y” coordinate is encoded and the “x” coordinate. Because the image is assumed to be recorded at time, TE, the blood may appear shifted from its actual position at that moment. When the flow is in the readout or phase-encoding direction, the shifted fluid appears to remain within the blood vessel. However, when the flow is diagonal within the readout and phase-encoding plane, the element of fluid appears to shift away from the vessel center with a distance of shift that is proportional to the flow velocity. If the flow is pulsatile, this shift will appear to change in size during the pulse sequence, resulting in blurring and ghosting artifacts. Such artifacts are often seen in the phase-encoding direction originating from the diagonal vessels in the circle of Willis. This last artifact can be eliminated by recognizing that the shift is due to a nonzero, velocity-dependent first moment of the phaseencoding gradient. By adding another lobe to the phaseencoding gradient, it is possible to step m0 through the values needed for imaging while at the same time nulling m1 at the time, TE. In 3D TOF, the rephrasing z gradient is usually used as a slice selection phase-encoding gradient. If first-order phase-encoding flow compensation is desired, both extra lobes on the z gradient can be stepped to achieve the desired m0 while maintaining a zero m1 at the time, TE.

44

Due to the additional lobes on the gradient waveforms, the minimum TE is increased. Spins experiencing higher orders of motion, such as constant acceleration, are not compensated by this technique. Acceleration terms, corresponding to the m2 moment, occur when flow is pulsatile or changes direction and in this case, some artifacts or signal loss can occur. Usually, this acceleration-caused signal loss is generally quite small and higher order flow compensation is usually not performed. However, it is possible to compensate for phase dispersion induced by constant acceleration by adding an additional lobe to the flow-compensating gradients. This is called second-order gradient motion rephasing or second-order gradient moment nulling, which compensates for constant acceleration. Changes in the rate of acceleration are described as jerk. Jerk can be compensated by third-order gradient moment nulling or third-order gradient motion rephasing with the addition of yet another lobe to the flow-compensating gradients. Signal loss due to these higher order terms is usually much smaller than the loss that would occur from the longer TE required to implement them such that most flow compensation techniques currently in use only compensate for first-order effects from constant velocity flow. If the TE of the pulse sequence can be made very short, all gradient moments will be small and there is no need for flow compensation. Thus, the phase dispersion in an ultrashort echo time sequence, such as a 3D radial acquisition technique, which requires no slice selection gradients and no readout prephasing, is minimal and there is no need for additional flow compensation [11].

S.-E. Kim and D.L. Parker

Flow-related enhancement can be reduced by the saturation of signal from the flowing spins, which depends on the number of pulses experienced by the blood, repetition rate, and blood T1 recovery time, and by intarvoxel phase dispersion.

Saturation effects are important in the setting of slow or in-plane flow, and can be minimized by using thinner slices and relatively long TR. Thus, 2D TOF [12] has the advantage in the setting of slow flow and is often the technique of choice for venous imaging. 2D TOF is also used in evaluation of cervical carotid stenosis to detect slow flow distal to a high-grade stenosis [13]. In 2D TOF imaging, a gradient echo sequence, usually 2D SPGR, is used to sequentially acquire a set of adjacent thin slices, generally 1–3-mm thick. TR, in a range of 20–30 ms, is used with a flip angle of 50–70°. If the velocity of flow is close to the critical velocity of 3–15 cm/s, as given in (2.1), the fluid in each slice experiences only a few RF pulses. The simulation shown in Fig. 2.2 demonstrates that the flow enhancement effect is maximized for flip angle in a range of 50–70°. The short TR and higher flip angle result in enhancement of the contrast between blood and background tissue, as there is insufficient time for longitudinal (T1) recovery of the static tissue magnetization and blood flowing into the slice is exposed to only one or two RF pulses for typical arterial velocities of 10–100 cm/s. The imaging plane is generally selected to be perpendicular to the flow direction, such as the axial plane for the carotid arteries. If arterial flow is principally along one axis and venous returns in the opposite direction (such as in the neck, with flow to the brain via the carotid arteries and return via the jugular veins), slice-selective saturation pulses can be used to eliminate the signal from flow in one direction. For example, to eliminate the signal from the jugular veins, a saturation RF pulse is applied superior to the axial slice (Fig. 2.4). To maintain saturation during the acquisition of all slices, this saturation pulse moves together with the axial slice as each subsequent slice is acquired. Blood flowing in the craniocaudal direction in the jugular vein is, thus, saturated before it flows into the imaging slice and produces no signal. Because the saturation pulse distinguishes arteries from veins only by the direction of blood flow, retrograde flow can result in the unwanted saturation of the desired vessel.

Fig. 2.4 Coronal MIP images of a 2D TOF acquisition of a healthy volunteer using the same acquisition parameters as Fig. 2.3. Coronal MIP image with no spatial saturation pulse (a) and with saturation pulse

superior to the imaging slices to suppress the venous signal (b) (images courtesy of Dr. Tae-Sub Chung, Gangnam Severance Hospital, Seoul, Korea)

2D TOF

2

Time-of-Flight Angiography

45

Fig. 2.5 2D TOF acquisition of a patient volunteer with atherosclerosis with TR = 28 ms, TE = 6.7 ms, flip angle = 50° , FOV = 16 cm, matrix = 256 × 256, slice thickness = 3 mm with 20% gap between slices.

(a) Coronal MIP; (b) axial source images; (c) 2D black blood T1w images

Transverse (axial) images acquired using 2D TOF in a patient with several severe stenoses just distal to the left and right bifurcation are shown in Fig. 2.5. Black blood 2D T1weighted images (Fig. 2.5c) confirm that atherosclerotic plaque caused the stenosis seen in 2D TOF. Figure 2.5a is obtained as the MIP through the “stack” of acquired image (Fig. 2.5b). Note that MIP images are not true angiograms but rather a projection produced from the source images. Any tissue with a short T1 relaxation time such as fat, depending on the exact scan technique used, may be hyperintense in the source images and thus on MIP images they may be potentially confused for flow. Further, the MIP image can be artifactual because only the brightest point along a projection line appears in the MIP image. It is possible that pathology or other important image details are masked behind brighter structures in the MIP image. Thus, when interpreting scans, it is important to review both the MIP and source images. The latter provide a means of assessing surrounding tissues and anatomy and often make identification of artifacts due to motion, signal loss, and the presence of fat easier. Pulsation and vessel motion induced by pulsation can cause artifacts in 2D TOF imaging. Flow compensation in the slice selection and frequency direction is helpful to reduce the inconsistent phase generated by pulsation [14]. Cardiac triggering can further reduce the artifact induced by pulsation. Fractional echo readout and a tailored excitation RF pulse are used to reduce the TE, since a shorter TE reduces the artifact induced by pulsation. Another drawback of 2D TOF is that complex flow, such as that seen distal to a stenosis, can have high-order motions (acceleration, jerk, etc.) and is not easily compensated for and can result in signal loss. Complex flow patterns specifically lead to signal loss in the region of a stenotic lesion and, thus, overestimation of the degree of stenosis. Although the slice thickness of

2D TOF sequences is usually less than 2 mm, 3D acquisitions can be used to produce images with thinner slices, which, together with the use of shorter TE, reduce the artifactual signal loss [15].

3D TOF Having intrinsically higher resolution and shorter echo times than 2D TOF, 3D TOF sequences suffer less from intravoxel phase dispersion. In 3D TOF imaging, a single slab of 3D slices is acquired. Like 2D TOF imaging, the slab is oriented to be perpendicular to the direction of flowing blood to ensure good inflow enhancement. SNR and the contrast between in-flowing blood and tissue in 3D TOF depend on the slab thickness and number of slices per slab. The transverse plane is often selected for a carotid or an intracranial 3D TOF acquisition to maximize the inflow enhancement effect; however, an oblique slab orientation is sometimes used, depending on the vessel geometry, to allow the desired imaging volume to be covered with fewer slices, which results in a shorter scan time. In 3D TOF acquisition, the blood inside the imaging slab generally experiences multiple RF pulses, so a smaller flip angle compared to that in 2D TOF is used to maximize the inflow enhancement effect. Further, spatial saturation is not usually applied in 3D TOF. Although spatial saturation could be applied in 3D TOF acquisition, it would increase the minimum TR and the venous signal in 3D TOF is already suppressed, except on a few entry slices, by the slow venous flow and thicker slab. Short TRs (~20 ms) in 3D TOF also result in reduced contrast between flowing and stationary tissue, including muscle and fat, which generally have a short T1 relaxation time compared with blood. Magnetization

46

S.-E. Kim and D.L. Parker

Fig. 2.6 (a) Schematic representation of TONE pulse. The flip angle of the TONE pulse increases as a function of distance in order to equalize the signal from flowing blood (indicated by arrow). (b) Coronal and sagittal MIP of 3D TOF acquisition of a volunteer at

1.5 T with and without TONE technique. The MIP image constructed with TONE shows the uniform vessel signal around edge (images courtesy of Dr. Tae-Sub Chung, Gangnam Severance Hospital, Seoul, Korea)

transfer is often used in 3D TOF to suppress the signal from the stationary tissue and to increase further the contrast between the flow and tissue. To minimize artifacts induced by turbulent or pulsatile flow, the echo time in 3D TOF is made as short as possible. Fractional echo readout and a short duration or asymmetric RF excitation pulse are used to achieve the shortest possible echo time. For situations where the signal from lipid can be a problem, an echo time can be selected, where the signal from lipid and water is out of phase, yielding better contrast between the two at the expense of a slight increase in problems due to the longer echo time. This out-of-phase echo time varies with the magnet field strength. When imaging with a thick 3D slab, the number of pulses experienced by the blood increases with distance into the slab. RF excitation that is uniform across the 3D slab results in greater inflow enhancement where the blood enters the slab, whereas distal slices in which blood has experienced

more RF pulses experience less flow enhancement signal. To compensate for the resulting variation in signal saturation, it has been found useful to make the RF excitation tip angle be a function of distance into the slab. For example, ramped RF excitation, where the tip angle is small at the slab entrance and increases with distance into the slab, is commonly used for 3D TOF to make the flowing blood signal across the imaging slab as uniform as possible [6, 16]. Ramped RF pulses, with spatially varying flip angle profiles parallel to the direction of flow, are often called tilted optimized nonsaturating excitation (TONE) pulses (Fig. 2.6). A variety of different flow compensation strategies are used in 3D TOF. Generally, flow compensation on the readout and slab selection gradient is used. In this case, the individual phase-encoding steps in the slice and phase-encoding gradient are not compensated. As described above, phaseencoding flow compensation can be used to compensate each step of both phase-encoding gradient waveforms. Although,

2

Time-of-Flight Angiography

47

Fig. 2.8 Axial MIP of 3D TOF acquisition of a healthy volunteer at 3.0 T (a) with linear view order and (b) with centric view order. Notice that the ghost artifact caused by pulsatile flow that propagates in the primary phase-encoding direction is reduced when centric view order is used

Fig. 2.7 Images obtained from four types of acquisition, from top to bottom. a–b, linear view ordering with (a) slice selection/frequencyencoding flow compensation and (b) three-directional flow compensation (3DFC). c–d, centric view ordering with (c) slice selection/ frequency-encoding flow compensation and (d) 3DFC. The cross-hatch artifacts arising from blood flow pulsations in the source images and other vessels near the circle of Willis are clearly evident in the two images in the top row. The artifacts are reduced by both 3DFC (second row of axial images) and centric view ordering acquisition order (third row), and the greatest reduction is observed when 3DFC is combined with centric view ordering acquisition (bottom row). More importantly, from the MIP images (right column), we note the change in apparent lumen diameter of the M1 segment (arrow) when using 3DFC. Without 3DFC, the obliquely oriented vessels have bright rims that are consistent with distortion or signal “pileup” (from Parker et al. [17], with permission)

compensation [20–22]. Ultrashort TE pulse sequences are also useful for scanning the brain in the presence of a metal clip or stent. In a 3D TOF acquisition, the order of k-space acquisition can be rearranged to reduce the prominence of artifacts induced by pulsatile flow (Fig. 2.8). A linear view order acquisition generally causes ghosts that are dominant along the phase-encoding direction [23]. An elliptical centric view order acquisition, which acquires the centric k-space first, spreads the ghosts evenly in the phase- and slice-encoding directions.

such phase-encoding flow compensation generally results in an increased echo time, it has the advantage of eliminating the misregistration artifact if blood is flowing obliquely to the frequency-encoding and one or both of the phase-encoding gradients (Fig. 2.7) [17]. To reduce the echo time in highresolution 3D TOF acquisition with three-direction flow compensation, a variable TE technique has been introduced [18, 19]. A variable TE technique in 3D TOF with threedirection flow compensation can minimize the TE at the center of k space. In this technique, k space of the 3D TOF is divided into several segments with different TE. Flowcompensation gradient lobes are calculated for each segment and the echo time at each segment is minimized. This results in a shorter echo time at the center of k space and a reduced flow-related signal void due to long echo times.

3D TOF with magnetization transfer has been employed for intracranial MRA to suppress the signal intensity of background brain tissue [24]. An MT pulse is a spectrally selective RF pulse that reduces the signal from tissue that has higher amounts of large molecules. Water molecules that are in contact with macromolecules generally move slowly and have a broad NMR resonance. Off-resonant excitation can saturate the magnetization on these water molecules. When these water molecules exchange with free water, the net magnetization in the tissue is reduced. Because blood has a lower concentration of macromolecules, off-resonance MT pulses can be used to selectively saturate the magnetization in station tissues with minimal effect on blood magnetization. In 3D TOF, MT is used to suppress the signal intensity of the brain parenchyma while leaving the signal from blood unaffected, thus improving smaller vessel visibility. Figure 2.9 shows a comparison of intracranial 3D TOF images acquired without and with the inclusion of MT pulses. An MT RF pulse generally requires a long pulse duration and higher RF power. Thus, MT pulse application causes increased TR and further increases the already long scan time of high-resolution 3D TOF acquisition. To increase the time

k-Space Sampling Strategies 3D MRA acquired using ultrashort TE sequences, such as projection or spiral acquisition, gives a minimal artifact from phase dispersion of flow across a voxel, even without flow

Magnetization Transfer

48

S.-E. Kim and D.L. Parker

Fig. 2.9 Axial MIP of 3D TOF acquisition of a volunteer at 1.5 T (a) without MT saturation and (b) with MT saturation. Notice that MIP with MT saturation demonstrates more small, distal, middle cerebral arteries’ detections (images courtesy of Dr. Tae-Sub Chung, Gangnam Severance Hospital, Seoul, Korea)

efficiency in 3D TOF, the majority of the MT effect is obtained by applying the MT pulses only around the center of k-space [25]. The use of MT saturation at high field strength becomes problematic because the SAR increases with the square of the RF transmission frequency, which is proportional to the main magnetic field strength. For this reason, MT saturation in TOF has been utilized only at 1.5 or 3.0 T. Because of the chemical shift between the fat and water, there is a small (3 ppm) difference in Larmor frequency between fat and water resonances, and a TE-dependent phase shift in their signal contributions to a given voxel. Choosing a TE at which fat and water are outphase can provide some fat suppression. However, the TE, where water and fat are out of phase, is greater than the minimum TE achieved by the pulse sequence and further increases the signal loss induced by intravoxel phase dispersion. As a general rule, the TE should be minimized at the cost of other considerations because of the rapid increase in intravoxel diphase dispersion that accompanies increased TE.

Multiple Overlapping Thin 3D Slab Acquisition 3D TOF MRA is susceptible to signal loss due to saturation of the moving spins in thick slabs. To reduce flow signal saturation while retaining the high spatial resolution, short echo times, and some of the SNR advantages of 3D techniques, a sequential acquisition of multiple 3D slabs has been developed. MOTSA has all the advantages of the single-volume 3D TOF techniques, and the use of overlapping thin slabs generally overcomes the problem of spin saturation. MOTSA is currently one of the most popular clinical 3D TOF applications [4]. This popularity is due in part to stronger in-flow enhancement and better vessel contrast-tonoise ratio properties of MOTSA compared to other techniques. Multiple thin slab acquisition reduces the signal saturation of slowly flowing blood compared to one thick single-slab acquisition, and overlapping slab acquisition

Fig. 2.10 Schematic representation of a MOTSA acquisition with three slabs and eight slices per slab. There are three overlap slices at each boundary between slabs

eliminates the signal void from each slab boundary region – often called the venetian blind artifact. The slices located in the overlapped region can be acquired twice with an entry slice from the first slab and an exit slice from the next slab [26]. This technique moderates the variation of the final MIP image intensity after taking the maximum of two images using a pixel-by-pixel comparison. Increasing the amount of slab overlap can obviously reduce the slab boundary artifact. Using a slab overlap of 50% can almost completely eliminate the artifact (Fig. 2.10). However, increasing overlap also increases the scan time per unit coverage in the slice direction. A sliding interleaved ky (SLINKY) sequence was proposed as an alternative technique to reduce the slab boundary artifact in multiple slab acquisition [27]. In a typical MOSTA data acquisition, the slab excitation is shifted

2

Time-of-Flight Angiography

after acquisition of all of ky and kz phase encodings. In SLINKY acquisition, the slab excitation is sliding every few views by one slab location. A designated partial set of the ky phase-encoding steps are collected and the acquisition of the partial set of the ky space is interleaved during continuous sliding of a slab along the slice selection direction [28]. SLINKY equalizes the flow enhancement effect across the entire slab dimension and eliminates the slab boundary artifacts while retaining good acquisition time efficiency compared with conventional MOTSA.

49

the extended scan time needed for phase encoding of the larger matrix is not clinically feasible. Parallel imaging (PI) techniques can accelerate scans at the cost of SNR and, thus, can achieve very-high-resolution MRA studies within a reasonable scan time [31]. For example, optimized coils coupled with PI techniques at 3.0 T can yield scan times similar to or shorter than those at 1.5 T. On this basis, intracranial TOF image quality improvements have previously been described in the transition from 1.5 to 3 T and 3 to even 7 T [32].

Summary TOF at High Magnetic Fields 3D TOF MRA is considered to be a safe, fully noninvasive imaging procedure with submillimeter spatial resolution. As such, it is routinely used to screen for cerebrovascular diseases, such as aneurysms and arteriovenous malformations. However, more subtle microvascular disease usually cannot be seen with the resolution capabilities of standard field strength MRA. Increased vessel contrast and spatial resolution are highly desirable for more sensitive detection of small aneurysms and vasculitis and for improved morphological characterization of larger aneurysms. The recent development of ultrahigh-field MRI scanners enables assessment of the cerebral arteries with a spatial resolution previously not achieved with standard MRI scanners. MRA at 7.0 T has shown superior contrast between blood and background tissue mainly because of the increased T1 recovery time of tissues and better suppression of the background signal relative to that of the blood vessels [29]. The increased SNR also yields improved visualization of the microvasculature of the human brain at high spatial resolution [30] (Fig. 2.11). However, at very high field strengths, the frequencyencoding dimension is limited by the TE-dependent artifacts of susceptibility-induced dephasing and pulsatile flow, and

In this discussion, we have presented a general overview of the basics of TOF MRA and provided discussion of the various trade-offs in image acquisition. Many of the concepts of TOF MRA were developed in the late 1980s and early 1990s, and hence are very mature. However, improvements in MRI techniques and technology have also led to progress in TOF MRA. These improvements include improved gradient capabilities, higher magnetic field strength, increased number and quality of RF channels and components, and new pulse sequences and acquisition techniques. In the past 20 years, gradient performance has increased from 1 mT/m and 20 T/m/s to over 40 mT/m and 200 T/m/s. Field strength in clinically used MRI scanners has now reached 3 T, and work is being performed at 7 T. The number of receiver channels has increased from 1 to as many as 128 on commercially available scanners, allowing the design of receiver coil arrays with more elements to allow parallel imaging to reduce image acquisition time and thereby reduce motion artifacts. Finally, new pulse sequences, including 3D radial acquisition, PROPELLER, and many others, have increased the flexibility available for novel TOF MRA techniques. Thus, as it is with many fields of MRI, TOF MRA continues to evolve in capability and ultimate utility.

Fig. 2.11 TOF MRA acquired at 7 T with a custom-built transmit/ receive head coil with eight stripline elements. TR/TE 21/3.4 ms; flip angle 40°; bandwidth 303 Hz/pixel; resolution 0.6 × 0.5 × 0.6 mm3 (noninterpolated); acquisition time 5 min 58 s; parallel imaging with

GRAPPA, reduction factor 2, 40 reference lines. The high spatial resolution enables nice depiction of fine, peripheral vessels (images courtesy of Mark Ladd of the Erwin L. Hahn Institute for MRI, Essen, Germany)

50

References 1. Suryan G. A time-of-flight method. Proc Indian Acad Sci Sect. 1959;A33:107. 2. Hinshaw WS, Bottomley PA, Holland GN. Radiographic thin section image of the human wrist by nuclear magnetic resonance. Nature. 1977;270:272–273. 3. Laub G. Displays for MR angiography. Magn Reson Med. 1990;14:222–229. 4. Parker DL, Yuan C, Blatter DD. MR angiography by multiple thin slab 3D acquisition. Magn Reson Med. 1991;17:434–451. 5. Davis WL, Warnock SH, Harnsberger HR, Parker DL, Chen CX. Intracranial MRA: single volume vs. multiple thin slab 3D time-offlight acquisition. J Comput Assist Tomogr. 1993;17:15–21. 6. Atkinson D, Brant-Zawadzki M, Gillan G, Purdy D, Laub G. Improved MR angiography: magnetization transfer suppression with variable flip angle excitation and increased resolution. Radiology. 1994;190:890–894. 7. Axel L, Shimakawa A, MacFall J. A time-of-flight method of measuring flow velocity by magnetic resonance imaging. Magn Reson Imaging. 1986;4:199–205. 8. Gullberg GT, Wehrli FW, Shimakawa A, Simons MA. MR vascular imaging with a fast gradient refocusing pulse sequence and reformatted images from transaxial sections. Radiology. 1987;165: 241–246. 9. Nishimura DG. Time-of-flight MR angiography. Magn Reson Med. 1990;14:194–201. 10. Vlaardingerbroek, MT, den Boer JA. Magnetic Resonance Imaging: Theory and Practice. Springer Verlag Telos; 1996:191–204. 11. Nielsen HT, Gold GE, Olcott EW, Pauly JM, Nishimura DG. Ultrashort echo-time 2D time-of-flight MR angiography using a halfpulse excitation. Magn Reson Med. 1999;41:591–599. 12. Keller PJ, Drayer BP, Fram EK, Williams KD, Dumoulin CL, Souza SP. MR angiography with two-dimensional acquisition and threedimensional display. Work in progress. Radiology. 1989;173: 527–532. 13. Heiserman JE, Drayer BP, Fram EK, et al. Carotid artery stenosis: clinical efficacy of two-dimensional time-of-flight MR angiography. Radiology. 1992;182:761–768. 14. Urchuk SN, Plewes DB. Mechanisms of flow-induced signal loss in MR angiography. J Magn Reson Imaging. 1992;2:453–462. 15. Keller PJ. Magnetic resonance angiography of the neck. Technical issues. Neuroimaging Clin N Am. 1996; 6:853–861. 16. Priatna A, Paschal CB. Variable-angle uniform signal excitation (VUSE) for three-dimensional time-of-flight MR angiography. J Magn Reson Imaging. 1995;5:421–427. 17. Parker DL, Goodrick KC, Roberts JA, et al. The need for phaseencoding flow compensation in high-resolution intracranial magnetic resonance angiography. J Magn Reson Imaging. 2003;18: 121–127.

S.-E. Kim and D.L. Parker 18. Song HK, Wehrli FW. Variable TE gradient and spin echo sequences for in vivo MR microscopy of short T2 species. Magn Reson Med. 1998;39:251–258. 19. Jeong EK, Parker DL, Tsuruda JS, Won JY. Reduction of flowrelated signal loss in flow-compensated 3D TOF MR angiography, using variable echo time (3D TOF-VTE). Magn Reson Med. 2002;48:667–676. 20. Schmalbrock P, Yuan C, Chakeres DW, Kohli J, Pelc NJ. Volume MR angiography: methods to achieve very short echo times. Radiology. 1990;175:861–865. 21. Glover GH, Lee AT. Motion artifacts in fMRI: comparison of 2DFT with PR and spiral scan methods. Magn Reson Med. 1995;33: 624–635. 22. Glover GH, Pauly MJ. Projection reconstruction techniques for reduction of motion effects in MRI. Magn Reson Med. 1992; 28:275–289. 23. Wilman AH, Riederer SJ, King BF, et al. Fluoroscopically triggered contrast-enhanced three-dimensional MR angiography with elliptical centric view order: application to the renal arteries. Radiology. 1997;205:137–146. 24. Dagirmanjian A, Ross JS, Obuchowski N, et al. High resolution, magnetization transfer saturation, variable flip angle, time-of-flight MRA in the detection of intracranial vascular stenoses. J Comput Assist Tomogr. 1995;19:700–706. 25. Parker DL, Buswell HR, Goodrich KC, Alexander AL, Keck N, Tsuruda JS. The application of magnetization transfer to MR angiography with reduced total power. Magn Reson Med. 1995;34: 283–286. 26. Blatter DD, Bahr AL, Parker DL, et al. Cervical carotid MR angiography with multiple overlapping thin-slab acquisition: comparison with conventional angiography. AJR Am J Roentgenol. 1993;161:1269–1277. 27. Liu K, Rutt BK. Sliding interleaved kY (SLINKY) acquisition: a novel 3D MRA technique with suppressed slab boundary artifact. J Magn Reson Imaging. 1998;8:903–911. 28. Liu K, Lee DH, Rutt BK. Systematic assessment and evaluation of sliding interleaved kY (SLINKY) acquisition for 3D MRA. J Magn Reson Imaging. 1998;8:912–923. 29. Kang CK, Park CW, Han JY, et al. Imaging and analysis of lenticulostriate arteries using 7.0-Tesla magnetic resonance angiography. Magn Reson Med. 2009;61:136–144. 30. Kang CK, Hong SM, Han JY, et al. Evaluation of MR angiography at 7.0 Tesla MRI using birdcage radio frequency coils with end caps. Magn Reson Med. 2008;60:330–338. 31. von Morze C, Purcell DD, Banerjee S, et al. High-resolution intracranial MRA at 7T using autocalibrating parallel imaging: initial experience in vascular disease patients. Magn Reson Imaging. 2008;26:1329–1333. 32. von Morze C, Xu D, Purcell DD, et al. Intracranial time-of-flight MR angiography at 7T with comparison to 3T. J Magn Reson Imaging. 2007;26:900–904.

3

Phase-Contrast MRI and Flow Quantification Bernd Jung and Michael Markl

Introduction Magnetic Resonance Imaging (MRI) techniques provide noninvasive methods for the accurate depiction of the vascular morphology and anatomy. In addition, the intrinsic motion sensitivity of MRI can be used to image vessels as in phasecontrast (PC) MR-angiography or to directly acquire and quantify blood flow [1–5]. Such techniques offer the unique possibility to acquire spatially registered functional information simultaneously with the morphological data within a single examination. Visualization and quantification of blood flow and tissue motion using PC MRI has been widely used in a number of applications. Characterization of the dynamic components of blood flow and cardiovascular function provide insight into normal and pathological cardiovascular physiology and are part of imaging protocols in daily clinical routine. In addition, the same underlying principles can be exploited to evaluate other dynamic processes in the human body such as CSF flow [6] or the cyclic wall motion of the heart [7].

ECG-triggered blood flow measurements as presented in 1986 by Nayler et al. [1]. The basic principle of PC MRI relies on the fact that the MR signal is inherently a vector quantity as illustrated in Fig. 3.1. The MR signal can thus be characterized not only by its magnitude but also by its phase f. As a consequence, a phase image can be reconstructed from any acquired MR data set in addition to the typical magnitude image reflecting the underlying anatomy. The MR-signal phase is affected by motion, which can be used to image vessels as in PC MR-angiography but also to quantify blood flow and motion of tissue. In the presence of a magnetic field gradient, the MR signal originating from a moving object presents with an additional signal phase, which is directly proportional to the velocity of the moving object. The measurement of MR signal phase, therefore, allows for motion quantification of the moving spins. The time-evolution of the MR-signal phase of the transversal magnetization of an object at the location r(t) can be derived from the spatial dependent Larmor frequency (3.1).     w L (r , t ) = g ( B0 + Δ B0 (r ) + r (t )G(t )).

(3.1)

Basic Principle of Phase-Contrast MRI The development of PC MRI initiated in the pre-MR imaging area with the first observation of coherent motion on the MR signal phase reported in 1954 by Carr and Purcell [8]. The basic concept was further developed for by Hahn who proposed to exploit the sensitivity of the MR signal phase to flow or motion for the measurement of sea water motion [9]. With the advent of MR imaging in the 1980s, the theory of phase velocity imaging was first described in 1982 [10] followed by first the presentation of MR velocity map images [11, 12] and clinical applications based on time-resolved B. Jung, PhD () • M. Markl, PhD Department of Radiology, Medical Physics, University Hospital Freiburg, Freiburg, Germany e-mail: [email protected]

In this equation, B0 denotes the static main magnetic field, DB0 reflects contributions by local field inhomogeneities, and G(t) is the time-dependent magnetic field gradient. The resulting signal phase f acquired at echo time TE can be derived by integrating the Larmor equation resulting in the following (3.2): TE     f (r ,TE) = f0 (r ) + g ∫ G(t )r (t )dt. t0

(3.2)

In this equation, the initial signal phase and the effects of field inhomogeneities are combined in the spatially dependent and typically unknown background phase f0(r). To evaluate the effect of flow or motion on the signal phase, the spatial location of a moving object or flowing spins can be    approximated in first order as r (t ) = r0 + v (t ) with constant

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_3, © Springer Science+Business Media, LLC 2012

51

52

B. Jung and M. Markl

Fig. 3.1 During an experiment, rf-excitation generates a precessing transverse magnetization vector, which induces the MR signal in a receiver coil surrounding the object under investigation. Following image acquisition anatomical (top right) and phase images (lower right) can be reconstructed from the raw data which reflect the local magnitude and phase of the transverse magnetization

velocity v, i.e., it is implied that tissue motion or blood flow does not change fast with respect to the temporal footprint (TE) of the data acquisition. Equation (3.2) then simplifies to the following (3.3):    TE   TE  f (r ,TE) = f 0(r ) + g r0 ∫ G(t )dt + g v ∫ G(t )tdt t0 t0     = f 0(r ) + g r0 M 0 + g vM1.

(3.3)

In addition to the background phase f0(r) the signal phase measured at TE is determined by the 0th and 1st order gradient moments M0 and M1. For PC-MRI, a bipolar gradient is typically used in order to quantify velocities, i.e., two gradients with identical amplitude G and duration Dt but with opposite polarity of the amplitude (see Fig. 3.2). The symmetry of this gradient scheme results in a vanishing 0th gradient moment M0 = 0. As a result, stationary spins no longer contribute to the signal phase and (3.3) reduces to the following (3.4):    f (r ,TE) = f0 (r ) + g vM1 .

(3.4)

The remaining non-zero first gradient moment M1 determines the velocity induced signal phase and moving spins experience an additional contribution in the signal phase which is proportional to the velocity of the moving spins as illustrated in Fig. 3.2. However, the measured signal phase f(r, TE) is still offset by the spatially dependent and unknown background phase f0(r). By subtracting the phase fref(r, TE) of an additionally performed reference scan the background phase can be eliminated. The resulting phase difference Df = f − fref finally yields the velocities of the moving spins (3.5):

Fig. 3.2 Bipolar velocity encoding gradient and temporal evolution of the MR signal phase for stationary spins and an object moving with constant velocity v

Δf   . v= g ΔM1

(3.5)

The calculated phase difference Df reflects changes in the MR signal phase associated with the motion component along the direction of the velocity encoding gradient. By adding a bipolar gradient to the MR pulse sequence along the read, phase, or slice direction and by subtracting a reference measurement without encoding gradient, flow or motion along these directions can be directly quantified. Such PC-MRI pulse sequences with single-direction velocity encoding are typically available on all commercial MR systems and permit the reconstruction of the following images (Fig. 3.3): • Magnitude image: the signal intensity represents the MR signal amplitude averaged over the two scans used for one-dimensional velocity encoding. • Complex difference image (flow encoded magnitude image): the signal intensity represents the absolute value of the phase shift and therefore contains no information about the flow direction (e.g., phase shifts of −170° and +170° shows the same pixel intensity). • Phase difference image: the signal intensity represents the phase angle DF between the reference and the motion encoded scan, i.e., the local velocities along the encoding direction. As is evident from (3.5), appropriate control of the first gradient moment (gradient strengths and duration) can be used to control flow or motion encoding. Owing to the 2p periodicity of the MR-signal phase, there exists an upper limit of the velocities that can be encoded using the PC principle. The velocity sensitivity “venc” refers to the maximum

3

Phase-Contrast MRI and Flow Quantification

53

Fig. 3.3 Images that are typically reconstructed from a 2D phase contrast acquisition with through-plane velocity encoding (bipolar encoding gradient along the slice direction). Note that the gray scale intensity for the phase difference image characterizes the motion

direction: the ascending aorta appears white due to the positive blood flow direction (flow foot to head) whereas the descending aorta appears black encoding to (negative) flow in the opposite direction (head to foot)

velocity that can be encoded by the bipolar gradient and is given by the following (3.6): venc =

±π  . γΔM1

(3.6)

Thus, the acquisition of moving objects requires the approximate knowledge of the blood flow or tissue velocities. If the first moment of the bipolar encoding gradient is too large, the signal phase can exceed values above ±p. Velocities positively exceeding the preselected venc of the PC-MRI acquisition result in velocity aliasing in the negative velocity range and vice versa. For example, a velocity of 120 cm/s measured with a venc of 100 cm/s will yield a velocity of −80 cm/s in the phase difference image. Figure 3.4 schematically illustrates the velocity aliasing effect for a parabolic flow profile. Figure 3.5 demonstrates the effect of different choices of venc with respect to the highest occurring velocities vmax in an in-vivo situation. No aliasing and a clear positive/negative blood flow velocities in the ascending/descending aorta can be observed with an optimal venc slightly exceeding the highest occurring velocities during peak systole (top left image). Aliasing in a few pixels can be seen in the ascending aorta if venc is chosen slightly smaller than vmax (lower left image). Aliasing in a considerably larger area of both ascending and descending aorta is evident if venc is far below the highest velocities (lower right image). Note also that if venc is chosen much higher than the systolic peak velocities in the aorta, no aliasing occurs but the velocity noise is substantially increased as indicated by the gray color within both ascending and descending aorta (top right image). It can be shown that the velocity noise s linearly increases with an increasing velocity sensitivity venc and is inversely related to the signal-to-noise ratio (SNR) in the corresponding magnitude images as follows (3.7) [13].

Fig. 3.4 Velocity aliasing for a parabolic flow profile. If the velocity sensitivity venc is smaller compared to the highest occurring flow velocities aliasing towards inverted velocity values occurs

Fig. 3.5 PC-MRI with through-plane velocity encoding in the ascending and descending aorta. The individual phase difference images show the encoded blood flow velocities during peak systole in the same subject for four different velocity sensitivities (venc)

54

B. Jung and M. Markl

Fig. 3.6 PC gradient echo pulse sequences for one-directional velocity encoding along the slice direction. The gray shaded bipolar velocity encoding gradient is added to a standard gradient echo sequence with

s=

2 venc . p SNR

(3.7)

To avoid velocity aliasing while maintaining a low velocity noise, it is thus important to select the velocity sensitivity higher than the expected vmax but as small as possible to achieve optimal velocity-to-noise performance for velocity quantification.

Phase-Contrast Pulse Sequences Typical applications of PC-MRI for measurement of blood flow in the aorta or cranial vessels are limited in scan time and therefore based on fast gradient echo pulse sequences. PC pulse sequences are formed by adding the bipolar velocity encoding gradient lobes to a gradient echo sequence as illustrated in Fig. 3.6 for velocity encoding along the slice (through-plane) direction. The addition of bipolar gradients for motion encoding prolongs the echo time TE which may cause flow artifacts, especially in the presence of pulsatile flow with rapid changes in velocity across a voxel. Therefore, a flow compensated acquisition (M1 = 0) is often used as the reference scan, i.e., an additional gradient is applied that refocus the MR-signal phase at TE independent from the velocity of moving spins (Fig. 3.6, left). For PC measurements a short echo time TE is desirable to minimize the signal loss due to T2* decay and reduce susceptibility and flow artifacts. However, the addition of bipolar gradients and use of flow compensated gradient waveforms results in relatively long echo times (in the order of 4–10 ms)

flow compensation along the slice direction. For improved performance, the bipolar gradient can be combined with the imaging gradient resulting in more time efficient gradient waveforms and reduced echo time TE

compared to a conventional gradient sequence as used for example used in contrast-enhanced measurements (1–3 ms). Therefore, several methods to reduce TE in PC-MRI have been introduced [14, 15]. One method that is implemented on most modern clinical MR systems utilizes the combination of the gradient waveforms of the reference scan with the bipolar gradients used for motion encoding, i.e., the flow encoding gradients are added onto the imaging gradients as depicted in Fig. 3.6 (right) resulting gradient waveforms with improved time efficiency. A further reduction of TE can be achieved by the implementation of two-sided flow encoding [14]. The difference between the two-sided flow encoding and conventional flow encoding with reference scan and flow encoded scan (also called one-sided flow encoding) is illustrated in Fig. 3.7. For one-sided flow encoding, the reference scan must be acquired with the same TE as the flow encoded scan to allow for a consistent subtraction of the background phase. For twosided flow encoding the moment DM1 of the flow encoding gradient is split in equal parts for both scans. The first scan (“up-case”) encodes the moment DM1/2, the second scan (“down-case”) −DM1/2 such that the difference yields the total encoding moment DM1. The corresponding gradients to encode DM1/2 require considerably a shorter duration and thus permit a noticeable reduction in echo time TE. Note that according to (3.5), the PC principle only permits the encoding and measurement of the motion component along the direction of the bipolar encoding gradient as shown in the pulse sequences described above for velocity encoding along the through-plane direction. To encode velocities along all three spatial dimensions at least four scans have to be acquired: one reference scan and three velocity encoded scans with different bipolar motion

3

Phase-Contrast MRI and Flow Quantification

55

Fig. 3.7 Different encoding strategies for one-directional velocity encoding along the slice direction

encoding gradients along read, phase, and slice direction but otherwise identical acquisition parameters [15]. This property results in relatively long scan times of PC-MRI which is one of the major drawbacks of PC measurements.

Implementation and Clinical Protocols To synchronize flow or motion sensitive measurements with periodic tissue motion or pulsatile flow, data acquisition is synchronized with the cardiac cycle and a k-space segmented pulse sequence scheme is used as illustrated in Fig. 3.8 [16]. The ECG signal is used to gate the MR measurement to consistently capture a series of time frame in the cardiac cycle for each heartbeat. Since the MR acquisition is not sufficiently fast to measure all required data during a single heartbeat, the periodic movements or flow are assessed by

Fig. 3.8 Time-resolved ECG synchronized CINE PC-MRI data acquisitions. One-directional through-plane velocity encoding requires the execution of two differently velocity encoded scan for each raw data line

56

reconstructing images from data acquired over several cardiac cycles. For each heartbeat and time frame, only a subset (NSeg) of all required (Ny) phase encoding steps are measured (k-space segmentation). The procedure is repeated for a different subset until the full raw data set is acquired for all time frames. As a result, time resolved (CINE) images can be derived depicting the dynamics of periodic physiological processes during the cardiac cycle. PC acquisitions can be combined with the CINE principle by successively acquiring the different velocity encoded scans for each raw data line (interleaved velocity encoding) [1, 3–5]. For an n-directionally encoded PC-MRI measurement n + 1 raw data lines have to be acquired for each phase encoding step. For a given repetition time TR and cardiac period of TECG, different imaging protocols can be constructed based on a trade-off between temporal resolution (Dt), spatial resolution (Ny phase encoding lines per slice) and total acquisition time Tacq. The selection of the number of phase encoding lines NSeg then determines the temporal resolution Dt = n TR NSeg and a total scan time Tacq = Ny/NSeg TECG of the PC CINE acquisition. From different PC protocols, 2D CINE PC with throughplane velocity encoding is mostly applied in the clinical routine. For a typical TR on the order of 5–10 ms measurements can be performed with temporal resolutions of up to 10–20 ms. Such PC-MRI protocols can be employed for volume flow measurements through a vessel. It should be noted that the CINE acquisition scheme is based on a periodic cardiac cycle. The regularity of the heart rate is important to ensure that data are consistently acquired for all time frames and heartbeats. An irregular heart rate would result in different entries in the data matrix acquired at different period of the heart cycle and thus induce artifacts. To avoid such inaccuracies, arrhythmia rejection algorithms have been introduced in clinical CINE PC-MRI protocols. Thoracic applications of PC-MRI are affected by breathing motions and should be performed at a constant breathing position to avoid respiration artifacts and errors in velocity encoding. For fast acquisitions, this can be realized by breath holding for the time of the acquisition [17]. However, the duration of the total acquisition is then limited by the breathhold capacity of the subject therewith limiting the spatial and/or temporal resolution of the acquisition. Breath hold is often not feasible for longer acquisitions such as high resolution or multidirectional PC-MRI. One possible remedy is to monitor the breathing pattern by a fast one dimensional MR scan without phase encoding (navigator) positioned at the interface between the liver and lung [18–20]. The large contrast between the liver tissue and air is easily tracked and can be used to detect the breathing position and to gate the data acquisition to the respiration cycle.

B. Jung and M. Markl

Applications of 2D CINE PC-MRI Most MR systems offer imaging protocols permitting both in-plane and through-plane velocity encoding. In-plane velocity mapping can be performed in one or two directions and is mainly used for visualizing flow patterns within the imaging plane such as jets through stenosed vessels or valves. The velocity encoding direction should be parallel to the direction of the flow in the vessel of interest. An in-plane acquisition of a high-velocity jet can be helpful for planning a subsequent through-plane scan on the jet location with maximum velocities, especially for the assessment of mitral or tricuspid valve regurgitation. The most common applications for CINE PC MRI focus on valvular, congenital, and other heart diseases. Flow quantification with through-plane velocity encoding can be used to estimate the fraction of regurgitant flow in case of valve defects or to quantify global cardiac function such a stroke volume and cardiac output [5, 21]. In the presence of pulmonary–systemic shunts, the pulmonary/systemic flow ratio can be determined and used to assess the severity of the disease [22–24]. For the quantification of left or right ventricular stroke volumes, cardiac output and regurgitant fractions throughplane velocity encoding is used and the temporal resolution is chosen such that cardiac cycle is covered by approximately 15–20 time frames. In combination with a spatial resolution of about 1–2 mm, these scans are typically performed during breath-hold. Typical velocity sensitivities are venc = 150 cm/s for aortic flow measurements and venc = 100 cm/s for flow in the pulmonary artery. However, it should be noted that in the case of a valve or vessel stenosis blood flow velocities can reach values of up to 800 cm/s. To avoid velocity aliasing the PC-MRI scan should be repeated using a higher velocity sensitivity venc until the phase aliasing has disappeared to allow for an appropriate flow analysis. Following successful data acquisition, the PC-MR data can be used for flow quantification as illustrated in Fig. 3.9 showing the systolic magnitude and phase difference image in the ascending and descending aorta. Vessel lumen segmentation of the ascending aorta for all time-frames throughout the cardiac cycle can be used to obtain the flow-time curve. In patients with aortic valve insufficiency, a slight backflow due to a mild aortic valve insufficiency can be observed that is not present for normal volunteers. The mild increase of flow after the closing of the aortic valve is caused by a contraction of the aortic wall, called the “Windkessel”-effect. Figure 3.10 shows systolic magnitude and the phase difference images in the ascending aorta at the level of the aortic valve for a patient with an aortic valve stenosis. At peak systole and maximum valve opening, blood flow is confined to a small area compared to the full diameter of the ascending

3

Phase-Contrast MRI and Flow Quantification

Fig. 3.9 Blood flow quantification in the aorta using 2D PC-MRI and through-plane velocity encoding. Left: Following data acquisition, the vessel contours of the ascending aorta (AAo) are identified by lumen segmentation to quantify time-resolved blood flow.

57

Right: Flow-time curves in the ascending aorta above the aortic valve in normal controls and two patients. In patients, the different extent of early diastolic regurgitant flow (arrows) indicates mild aortic valve insufficiency

Fig. 3.10 2D PC-MRI and through-plane velocity encoding in a patient with aortic valve stenosis and a dilated ascending aorta (Ao)

aorta as indicated by the dark color in the phase difference image (arrow). A further clinical application provided by CINE PC-MRI is provided by the evaluation of shunts reflecting defects in the septum of the heart that allow transfer of blood from the arterial to the venous system or vice versa (from the system with the higher to the lower pressure) [23]. To quantify the shunt volume two CINE flow measurements are performed. PC-MRI in the ascending aorta and in the pulmonary artery is used to quantify the left and right ventricular stroke volumes QS and QP, respectively. The ratio QP /QS represents the direction of shunt and the shunt volume. For QP /QS = 1 no shunt is present, if QP /QS > 1 a shunt from the aortic circulation to the pulmonary circulation is present and vice versa for QP /QS < 1.

Phase-Contrast Angiography While most clinical angiographic applications rely on the application of Gd contrast agent, 3D PC MR angiography (PC-MRA) based on velocity encoded 3D MRI with three-

directional encoding has proven to be a useful alternative. PC-MRA can provide detailed information on vascular geometry and may offer additional information on flow direction [25–28]. Several strategies exist for calculating a PC-MR angiogram from the data. A possible combination of velocity and magnitude data is shown in Fig. 3.11. The three-directional encoded PC-MRI data is used to calculate absolute velocities |v| for each image voxel which are additionally weighted by the magnitude images for suppression of background signal. Alternatively, complex difference images can be calculated directly from the raw data – as described above – for each individual velocity encoding direction and combined in a sum of squares sense [25, 26]. It is important to note that the depiction of the vessels is determined by the choice of the velocity sensitivity venc. The maximum intensity projections (MIP) of the cranial vessels in Fig. 3.12 demonstrate the effect of the velocity sensitivity on PC-MRA data. Small vessels with slow flow are more clearly visible using a smaller venc-factor (Fig. 3.12c, d). Larger vessels with higher blood flow velocities demonstrate a decreased signal due to (multiple)

58

B. Jung and M. Markl

Fig. 3.11 Derivation of 3D PC-MRA from PC-MRI data with three-directional velocity encoding and maximum intensity projects of PC-MRA data in the head and thorax

Flow-Sensitive 4D MRI

Fig. 3.12 Influence of the velocity sensitivity (venv) of PC MRA. PC MRA Maximum Intensity Projection (MIP) images acquired with venc’s of 80 (a), 40 (b), 20 (c), and 10 (d) cm/s. High venc images highlight the arterial system. As the venc is decreased the visualization of the high velocity arterial system is compromised, while, the visualization of the veins is improved. This is especially true for the slowflowing veins such as the internal cerebral vein (arrows) (Images courtesy of Kevin Johnson and Oliver Wieben, Departments of Medical Physics and Radiology, University of Wisconsin Madison, USA)

aliasing of the signal phase. For increased venc-factors, the visualization of larger arterial vessels is improved and venous signal is suppressed. The quality of the depicted vessels is best if the chosen flow sensitivity represents the physiological situation of the vessel segment of interest. Most PC-MRA implementations used nongated data acquisition which can result in artifacts for pulsatile blood flow. Further drawbacks of the PC-MRA method are long scan times and lack of respiration control, which limited most applications of 3D PC-MRA to static regions with low pulsatile flow such as the cranial vessels. In this context, new implementations based on time-resolved accelerated acquisitions or radial imaging strategies are promising and provide a considerable improved performance regarding flow artifacts, total scan time, and spatial resolution without the need for contrast agent administration [29, 30].

Traditionally, MRI imaging of flow is accomplished using methods that resolve two spatial dimensions (2D) in individual slices. Alternatively, 3D spatial encoding offers the possibility of isotropic high spatial resolution and thus the ability to measure and visualize the temporal evolution of complex flow and motion patterns in a 3D-volume. In this context, ECG synchronized flow-sensitive 3D MRI using three-directional velocity encoding (also termed “flow-sensitive 4D MRI” or “time-resolved 3D velocity mapping”) can be employed to detect and visualize global and local blood flow characteristics in entire targeted vascular regions (aorta, cranial arteries, carotid arteries, etc.) [31, 32]. For thoracic or abdominal applications, the data acquisition needs to be synchronized with the subject’s respiration. Owing to the acquisition of at least four data sets for three-directional velocity encoding, PC MRI inherits a trade-off between spatial/temporal resolution and total scan time. Nevertheless, a number of studies have reported methodological improvements (parallel imaging, adaptive navigator gating with increased efficiency, time-optimized velocity encoding gradients, etc.) permitting the acquisition of flow-sensitive 4D MRI data within the entire aorta or other arterial systems within reasonable scan times of the order of 10–15 min. For the subsequent analysis and visualization of complex, three-directional blood flow within a 3D volume, various visualization tools including 2D vector-fields, 3D streamlines, and time-resolved 3D particle traces have been proposed [33]. A representative data acquisition and data analysis strategy for the 3D visualization of blood flow characteristics in the thoracic aorta is illustrated in Fig. 3.13. Several groups have reported advances in the application of flow-sensitive 4D MRI including the analysis of blood flow through artificial valves [34], ventricular and atrial flow patterns [35, 36], blood flow characteristics in

3

Phase-Contrast MRI and Flow Quantification

59

Fig. 3.13 (a) Data acquisition and visualization of vascular geometry and 3D hemodynamics for flow-sensitive 4D MRI in the aorta using navigator gating and prospective ECG gating. (b) The resulting raw data comprises information along all 3 spatial dimensions, 3 velocity

directions, and time in the cardiac cycle. (c) 3D blood flow visualization permits the depiction of time-resolved 3D vascular hemodynamics within the entire thoracic aorta

the thoracic aorta [37–39], peripheral vessels [38], carotid arteries [40], large intracranial arteries [41], as well as flow in the pulmonary and venous systems [42]. Figure 3.14 illustrates the potential of the methods to assess, visualize and quantify flow characteristics in different vascular regions in the body. Recent studies indicate the potential of flow-sensitive 4D-MRI for the detailed visualization of complex flow patterns associated with vascular pathologies. The complete coverage of the vascular region of interest permits the assessment of spatial and temporal characteristics of 3D blood flow as shown in Fig. 3.15 for a patient with aortic valve stenosis and a dilatation of the ascending aorta. The disturbed complex flow patterns in the aorta illustrate the effect of the pathological valve function on aortic hemodynamics and may help to improve the understanding of the link between valve dysfunction and aortic dilatation often sees in such patients. Since flow-sensitive 4D MRI data reflects the true underlying time-resolved blood flow velocity vector field, it is possible to quantify the directly measured (e.g., flow rates) or derived parameters such as pressure difference maps [43],

wall sheer stress [44], pulse wave velocity [45], and others. Findings in recently reported studies combining the complete spatiotemporal coverage of flow-sensitive 4D MRI and advanced quantification strategies are promising and may help to define new clinical markers for the improved characterization of cardiovascular disease. Examples include relative pressure mapping within the heart and aorta [46] or renal arteries [27], wall shear stress analysis in the thoracic aorta [47], or assessment of onset and dynamics of regional turbulent kinetic energy in the aorta [48]. A disadvantage of PC MRI is related to the need for multiple acquisitions for encoding a single velocity direction, resulting in long scan times. New methods based on the combination of PC MRI and fast sampling strategies, e.g., radial imaging with 3D PC-VIPR as shown in Fig. 3.16, have been reported and are promising for further reduction in total scan time and/or increased spatial or temporal resolution [29, 49]. In this context, a number of studies have already demonstrated the potential of radial imaging techniques for the assessment of vascular function with increased efficiency compared to conventional methods [27, 50, 51].

60

B. Jung and M. Markl

In addition, the total acquisition time or temporal and spatial resolution associated with a specific MR technique may be further improved by using new spatiotemporal imaging acceleration [52, 53].

Sources of Errors Over the past three decades numerous studies have systematically validated the quantitative and qualitative analysis of blood flow using in-vitro and in-vivo experiments. It has to be noted, however, that there are a number of sources of inaccuracies in PC-MRI, which can results in errors in the measured velocities. The major sources of errors include eddy current effects [54], Maxwell terms [55], and gradient field distortions [56]. Appropriate correction strategies should, thus, be applied to ensure accurate flow quantification using PC-MRI data.

Maxwell Terms

Fig. 3.14 Flow-sensitive 4D MRI in different vascular territories. (a) 3D flow visualization in the large intracranial arteries in the circle of Willis using systolic streamlines. The tortuous routes of the different vascular segments and changes in regional velocities can clearly be appreciated (CCA/ICA/ECA: common/internal/external carotid artery). (b) Time-resolved 3D particle traces during peak flow in the carotid bifurcation. Note the typical helical flow pattern in the ICA bulb. (c) Blow flow characteristics in the distal abdominal aorta (Ao) and peripheral arteries in the left and right leg (IA iliac artery). Quantitative flowtime curves demonstrating typical triphasic pulsatile flow can be derived from the same data as shown for the distal abdominal aorta

It can be shown that, as a consequence of the Maxwell equations, a magnetic field gradient cannot be switched on, without generating additional unwanted nonlinear magnetic fields. These concomitant gradient terms or Maxwell’s terms arise whenever a gradient is activated. They are described by Maxwell’s equations and result in magnetic fields with nonlinear spatial dependences. Concomitant gradient crossterms arise when the longitudinal gradient Gz is activated with a transverse gradient (Gx or Gy). They result in phase errors and hence are affecting the velocity measurements using PC MRI. It is possible to reduce or correct these effects

Fig. 3.15 Flow-sensitive 3D MRI and visualization using systolic 3D streamlines in a patient with aortic stenosis and a dilated ascending aorta (AAo). Note the flow acceleration through the stenotic aor-

tic valve which results in a flow jet directed towards the outer wall of the AAo and considerable vortex formation in a large segment of the AAo

3

Phase-Contrast MRI and Flow Quantification

61

during postprocessing by fitting a polynomial onto the data as described by Bernstein et al. [55]. Corrections can be performed during image reconstruction, based on the knowledge of the gradient waveforms that can be used to derive the correction factors needed for Maxwell term compensation.

reflected in the strength or direction in the first order gradient moments and thus the velocity encoding. Therefore, gradient field distortions can lead to considerable deviations between the designed and the actual encoding in PC-MRI. The true gradient field demonstrates not only deviations from the nominal gradient strength but also from the intended gradient direction and thus affects not only the magnitude of encoded velocities but also velocity encoding direction. Dependent upon the spatial location of the velocity measurements, errors in velocity magnitude can be as high as 60%, while errors in the velocity encoding direction can be up to 45°. The true magnitude and direction of the underlying velocities can be recovered from the phase difference images by a generalized PC velocity reconstruction which requires the measurement of full three-directional velocity information. A generalized reconstruction of velocities can then applied using a matrix formalism that includes relative gradient field deviations derived from a theoretical model of local gradient field nonuniformity. In addition, approximate solutions for the correction of one-directional velocity encoding are available [56]. The gradient field model needed for the corrections is vendor specific and depends on the design on the gradient systems. Significant improvement in velocity quantification can be achieved by using the known field distortions to correct the measured phase shifts. Results of a phantom experiment with one-directional velocity encoding illustrating the effect of gradient field distortions are shown in Fig. 3.17. The relative velocity encoding errors predicted by the gradient field model are illustrated by the surface plot which demonstrates the relative deviations from nominal z-gradient strength a coronal plane transecting the flow phantom in longitudinal direction. Since steady flow was used for all experiments, the true mean through-plane velocities are expected to be constant as a function of spatial location (z) along the tube. The deviation of the measured velocities demonstrates the effect of gradient field nonuniformity and corresponded well to the predicted errors predicted by the gradient field model. Although errors associated with gradient field distortions have been known for some time, corrections for these inaccuracies are to date often not part of the standard image reconstruction process and mostly absent from commercial systems.

Gradient Field Nonlinearities

Eddy Currents

It is well known that nonuniformity in magnetic field gradients can cause significant image warping and require correction. In PC-MRI, these imperfections introduce errors in velocity measurements by affecting the first moments used to encode flow or motion [56, 57]. Any error in strength or direction of the local gradient from its ideal value is directly

Switching of imaging and encoding gradients in PC-MRI results in changes in magnetic flux, thereby inducing eddy currents in the conducting parts of the scanner system (coils) or in the patient. These eddy currents can cause alterations of the desired gradient strengths and duration and thus result in spatially varying phase errors in the MR images [58]. For

Fig. 3.16 PC VIPR is based on spoiled gradient echo sequence with bipolar flow encoding gradients (a) and a true 3D radial trajectory (b). Visualization of the flow conditions obtained with a PC VIPR scan of a 18 month-old male with pulmonary venolobar syndrome. Particle traces in various cardiac phases obtained from the PC VIPR scan from a posterior view (c). The vascular system is color coded in blue (veins) and red (arteries). (d) Atrial defect measured at 1.34 L/min. (e) Analomous Pulmonary Venous Return “Scimitar Vein” Flow – 0.42 L/min. (f) Abnormal Systemic Artery showing flow to the right lung (Images courtesy of Kevin Johnson and Oliver Wieben, Departments of Medical Physics and Radiology, University of Wisconsin Madison, USA)

62

Fig. 3.17 2D PC MRI with one-directional velocity encoding: Demonstration of the effect of gradient field inhomogeneities on velocity encoding. Left: PC-MRI velocity measurements in a tube phantom with constant laminar flow parallel to the direction of the main magnetic field (B0, z). Mid: Simulated and normalized z-gradient fields (left) depict the spatial variation of the relative deviation from the ideal

B. Jung and M. Markl

gradient strength in a coronal plane at the phantom locations. Right: Measured through plane velocities are compared to predicted measurement errors (thick black lines). Since the measured errors are in agreement with predictions, a correction can be performed based on the model of the encoding gradient fields

Fig. 3.19 Schematic illustration of a typical correction scheme for PC-MRI data. While Maxell and gradient field corrections can be automatically performed during image reconstruction, eddy current effects often require user interaction to identify regions for phase offset estimation (see also Fig. 3.18)

Fig. 3.18 Image-based eddy current correction of 2D PC-MRI data. The application of bipolar velocity encoding gradients can results in eddy current induced phase shifts as illustrated for a static phantom demonstrating which should ideally include constant and zero phase differences inside the entire object. The phase difference gradient is a result of spatially dependent eddy current induced phase offsets. For in-vivo PC-MRI (right) eddy current effects can be corrected by subtracting eddy current offsets estimated from regions of interest in static tissue

PC-MRI, the different gradient waveforms that are used for the subsequent velocity encodes lead to different eddy current induced phase changes in the phase images of each velocity encoded acquisition. As a result, subtraction of phase images does not eliminate errors related eddy currents and additional data processing is needed to restore the original velocity encoded signal phase. Several correction strategies have been proposed and are typically based on the subtraction of an estimation of the spatially varying eddy current induced phases changes. The spatial variation of the phase difference in regions containing static tissue is used to calculate the eddy current induced offset for the entire image. By fitting a plane or higher order polynomial to the phase difference data, the phase offset characteristics can be estimated and subsequently used to correct the entire image by subtraction (see Fig. 3.18) [54].

Compensation for Maxwell terms and for gradient field nonlinearities do typically not require user interaction. Maxwell corrections can be performed during image reconstruction, based on the knowledge of the gradient waveforms in the PC-MRI pulse sequence used for data acquisition. Similarly, gradient field models describing deviations between the designed and the actual velocity encoding gradients can be employed to automatically correct for gradient field nonlinearities during image reconstruction. For eddy current correction, automatic correction algorithms have been reported. However, for a reliable estimation of background offsets, user inter action may often be required to correctly identify static background signal (Fig. 3.19). It is of note that it is common to all sources of inaccuracies, that error in velocity encoding strength increase with increasing distance from the isocenter of the magnet. Velocity measurements with a single slice placed at or near the center of the magnet and with the vessel of interest close to the center of the FOV are therefore largely insensitive to encoding errors. The situation differs considerably, however, if multiple slices are acquired within a single acquisition or if blood flow is to be analyzed within a larger plane or a 3D imaging volume. For off-center flow quantification corrections for all of the above sources of inaccuracies should be performed to ensure reliable quantitative analysis of flow and derived parameters.

3

Phase-Contrast MRI and Flow Quantification

Acceleration Errors The MR signal phase is not only sensitive to the blood or tissue velocity, but it depends as well on the acceleration and higher-order terms. All those higher order terms are neglected when approximation blood flow or tissue motion in first order as (3.3). Although this assumption is reasonable in most situations, it may induce an error on the measured velocities. This effect may be reduced by optimizing the shape of the gradient waveforms or reducing TE (Oshinski et al. [59]). Although it is difficult to fully correct for this effect, it is possible to further reduce it at the cost of longer imaging times by integrating acceleration encoding [60] or Fourier encoding (Firmin et al. [61]).

References 1. Nayler GL, Firmin DN, Longmore DB. Blood flow imaging by cine magnetic resonance. J Comput Assist Tomogr. 1986;10:715–722. 2. Dumoulin CL, Souza SP, Darrow RD, Pelc NJ, Adams WJ, Ash SA. Simultaneous acquisition of phase-contrast angiograms and stationary-tissue images with Hadamard encoding of flow-induced phase shifts. J Magn Reson Imaging. 1991;1:399–404. 3. Pelc NJ, Herfkens RJ, Shimakawa A, Enzmann DR. Phase contrast cine magnetic resonance imaging. Magn Reson Q. 1991;7: 229–254. 4. Mohiaddin RH, Pennell DJ. MR blood flow measurement. Clinical application in the heart and circulation. Cardiol Clin. 1998;16: 161–187. 5. Chai P, Mohiaddin R. How we perform cardiovascular magnetic resonance flow assessment using phase-contrast velocity mapping. J Cardiovasc Magn Reson. 2005;7:705–716. 6. Naidich TP, Altman NR, Gonzalez-Arias SM. Phase contrast cine magnetic resonance imaging: normal cerebrospinal fluid oscillation and applications to hydrocephalus. Neurosurg Clin N Am. 1993; 4:677–705. 7. Drangova M, Zhu Y, Bowman B, Pelc NJ. In vitro verification of myocardial motion tracking from phase-contrast velocity data. Magn Reson Imaging. 1998;16:863–870. 8. Carr HY, Purcell EM. Effects of diffusion on free precession in nuclear magnetic Resonance Experiments. Phys Rev. 1954;94: 30–638. 9. Hahn EL. Detection of sea water motion by nuclear precession. J Geophys Res. 1960;65:776–777. 10. Moran PR. A flow velocity zeugmatographic interlace for NMR imaging in humans. Magn Reson Imaging. 1982;1:197–203. 11. Bryant DJ, Payne JA, Firmin DN, Longmore DB. Measurement of flow with NMR imaging using a gradient pulse and phase difference technique. J Comput Assist Tomogr. 1984;8:588–593. 12. Van Dijk P. Direct cardiac NMR imaging of heart wall and blood flow velocity. J Comput Assist Tomogr. 1984;171:429–436. 13. Andersen AH, Kirsch JE. Analysis of noise in phase contrast MR imaging. Med Phys. 1996;23:857–869. 14. Bernstein MA, Shimakawa A, Pelc NJ. Minimizing TE in momentnulled or flow-encoded two- and three-dimensional gradient-echo imaging. J Magn Reson Imaging. 1992;2:583–588. 15. Pelc NJ, Bernstein MA, Shimakawa A, Glover GH. Encoding strategies for three-direction phase-contrast MR imaging of flow. J Magn Reson Imaging. 1991;1:405–413.

63 16. Atkinson DJ, Edelman RR. Cineangiography of the heart in a single breath hold with a segmented turboFLASH sequence. Radiology. 1991;178:357–360. 17. Thomsen C, Cortsen M, Sondergaard L, Henriksen O, Stahlberg F. A segmented K-space velocity mapping protocol for quantification of renal artery blood flow during breath-holding. J Magn Reson Imaging. 1995;5:393–401. 18. Ehman RL, Felmlee JP. Adaptive technique for high-definition MR imaging of moving structures. Radiology. 1989;173:255–263. 19. Felmlee JP, Ehman RL, Riederer SJ, Korin HW. Adaptive motion compensation in MRI: accuracy of motion measurement. Magn Reson Med. 1991;18:207–213. 20. Taylor AM, Jhooti P, Wiesmann F, Keegan J, Firmin DN, Pennell DJ. MR navigator-echo monitoring of temporal changes in diaphragm position: implications for MR coronary angiography. J Magn Reson Imaging. 1997;7:629–636. 21. Didier D. Assessment of valve disease: qualitative and quantitative. Magn Reson Imaging Clin N Am.2003;11:115–134, vii. 22. Gatehouse PD, Keegan J, Crowe LA, Masood S, Mohiaddin RH, Kreitner KF, Firmin DN. Applications of phase-contrast flow and velocity imaging in cardiovascular MRI. Eur Radiol. 2005;15: 2172–2184. 23. Beerbaum P, Korperich H, Barth P, Esdorn H, Gieseke J, Meyer H. Noninvasive quantification of left-to-right shunt in pediatric patients: phase-contrast cine magnetic resonance imaging compared with invasive oximetry. Circulation. 2001;103: 2476–2482. 24. Underwood SR, Firmin DN, Rees RS, Longmore DB. Magnetic resonance velocity mapping. Clin Phys Physiol Meas. 1990;11 (Suppl A):37–43. 25. Bernstein MA, Grgic M, Brosnan TJ, Pelc NJ. Reconstructions of phase contrast, phased array multicoil data. Magn Reson Med. 1994;32:330–334. 26. Dumoulin CL, Souza SP, Walker MF, Wagle W. Three-dimensional phase contrast angiography. Magn Reson Med. 1989;9:139–149. 27. Lum DP, Johnson KM, Paul RK, Turk AS, Consigny DW, Grinde JR, Mistretta CA, Grist TM. Transstenotic pressure gradients: measurement in swine--retrospectively ECG-gated 3D phase-contrast MR angiography versus endovascular pressure-sensing guidewires. Radiology. 2007;245:751–760. 28. Dumoulin CL. Phase contrast MR angiography techniques. Magn Reson Imaging Clin N Am. 1995;3:399–411. 29. Johnson KM, Lum DP, Turski PA, Block WF, Mistretta CA, Wieben O. Improved 3D phase contrast MRI with off-resonance corrected dual echo VIPR. Magn Reson Med. 2008;60:1329–1336. 30. Bock J, Frydrychowicz A, Stalder AF, Bley TA, Burkhardt H, Hennig J, Markl M. 4D phase contrast MRI at 3 T: effect of standard and blood-pool contrast agents on SNR, PC-MRA, and blood flow visualization. Magn Reson Med. 2010;63:330–338. 31. Markl M, Harloff A, Bley TA, Zaitsev M, Jung B, Weigang E, Langer M, Hennig J, Frydrychowicz A. Time-resolved 3D MR velocity mapping at 3T: improved navigator-gated assessment of vascular anatomy and blood flow. J Magn Reson Imaging. 2007; 25:824–831. 32. Wigstrom L, Sjoqvist L, Wranne B. Temporally resolved 3D phasecontrast imaging. Magn Reson Med. 1996;36:800–803. 33. Buonocore MH. Visualizing blood flow patterns using streamlines, arrows, and particle paths. Magn Reson Med. 1998;40:210–226. 34. Kozerke S, Hasenkam JM, Pedersen EM, Boesiger P. Visualization of flow patterns distal to aortic valve prostheses in humans using a fast approach for cine 3D velocity mapping. J Magn Reson Imaging. 2001;13:690–698. 35. Fyrenius A, Wigstrom L, Ebbers T, Karlsson M, Engvall J, Bolger AF. Three dimensional flow in the human left atrium. Heart. 2001;86:448–455.

64 36. Kilner PJ, Yang GZ, Wilkes AJ, Mohiaddin RH, Firmin DN, Yacoub MH. Asymmetric redirection of flow through the heart. Nature. 2000;404:759–761. 37. Bogren HG, Buonocore MH. 4D magnetic resonance velocity mapping of blood flow patterns in the aorta in young vs. elderly normal subjects. J Magn Reson Imaging. 1999;10:861–869. 38. Frydrychowicz A, Harloff A, Jung B, Zaitsev M, Weigang E, Bley TA, Langer M, Hennig J, Markl M. Time-resolved, 3-dimensional magnetic resonance flow analysis at 3 T: visualization of normal and pathological aortic vascular hemodynamics. J Comput Assist Tomogr. 2007;31:9–15. 39. Kilner PJ, Yang GZ, Mohiaddin RH, Firmin DN, Longmore DB. Helical and retrograde secondary flow patterns in the aortic arch studied by three-directional magnetic resonance velocity mapping. Circulation. 1993;88(5 Pt 1):2235–2247. 40. Harloff A, Albrecht F, Spreer J, Stalder AF, Bock J, Frydrychowicz A, Schollhorn J, Hetzel A, Schumacher M, Hennig J, Markl M. 3D blood flow characteristics in the carotid artery bifurcation assessed by flow-sensitive 4D MRI at 3T. Magn Reson Med. 2009;61:65–74. 41. Wetzel S, Meckel S, Frydrychowicz A, Bonati L, Radue EW, Scheffler K, Hennig J, Markl M. In vivo assessment and visualization of intracranial arterial hemodynamics with flow-sensitized 4D MR imaging at 3T. AJNR Am J Neuroradiol. 2007;28:433–438. 42. Reiter G, Reiter U, Kovacs G, Kainz B, Schmidt K, Maier R, Olschewski H, Rienmueller R. Magnetic resonance-derived 3-dimensional blood flow patterns in the main pulmonary artery as a marker of pulmonary hypertension and a measure of elevated mean pulmonary arterial pressure. Circ Cardiovasc Imaging. 2008;1:23–30. 43. Tyszka JM, Laidlaw DH, Asa JW, Silverman JM. Three-dimensional, time-resolved (4D) relative pressure mapping using magnetic resonance imaging. J Magn Reson Imaging. 2000;12:321–329. 44. Stalder AF, Russe MF, Frydrychowicz A, Bock J, Hennig J, Markl M. Quantitative 2D and 3D phase contrast MRI: optimized analysis of blood flow and vessel wall parameters. Magn Reson Med. 2008;60:1218–1231. 45. Vulliemoz S, Stergiopulos N, Meuli R. Estimation of local aortic elastic properties with MRI. Magn Reson Med. 2002;47:649–654. 46. Ebbers T, Wigstrom L, Bolger AF, Engvall J, Karlsson M. Estimation of relative cardiovascular pressures using time-resolved three-dimensional phase contrast MRI. Magn Reson Med. 2001;45:872–879. 47. Frydrychowicz A, Stalder AF, Russe MF, Bock J, Bauer S, Harloff A, Berger A, Langer M, Hennig J, Markl M. Three-dimensional analysis of segmental wall shear stress in the aorta by flow-sensitive four-dimensional-MRI. J Magn Reson Imaging. 2009;30:77–84. 48. Dyverfeldt P, Sigfridsson A, Kvitting JP, Ebbers T. Quantification of intravoxel velocity standard deviation and turbulence intensity by generalizing phase-contrast MRI. Magn Reson Med. 2006;56: 850–858.

B. Jung and M. Markl 49. Gu T, Korosec FR, Block WF, Fain SB, Turk Q, Lum D, Zhou Y, Grist TM, Haughton V, Mistretta CA. PC VIPR: a high-speed 3D phase-contrast method for flow quantification and high-resolution angiography. AJNR Am J Neuroradiol. 2005;26:743–749. 50. Turk AS, Johnson KM, Lum D, Niemann D, Aagaard-Kienitz B, Consigny D, Grinde J, Turski P, Haughton V, Mistretta C. Physiologic and anatomic assessment of a canine carotid artery stenosis model utilizing phase contrast with vastly undersampled isotropic projection imaging. AJNR Am J Neuroradiol. 2007;28: 111–115. 51. Moftakhar R, Aagaard-Kienitz B, Johnson K, Turski PA, Turk AS, Niemann DB, Consigny D, Grinde J, Wieben O, Mistretta CA. Noninvasive measurement of intra-aneurysmal pressure and flow pattern using phase contrast with vastly undersampled isotropic projection imaging. AJNR Am J Neuroradiol. 2007;28:1710–1714. 52. Baltes C, Kozerke S, Hansen MS, Pruessmann KP, Tsao J, Boesiger P. Accelerating cine phase-contrast flow measurements using k-t BLAST and k-t SENSE. Magn Reson Med. 2005;54:1430–1438. 53. Jung B, Honal M, Ullmann P, Hennig J, Markl M. Highly k-t-spaceaccelerated phase-contrast MRI. Magn Reson Med. 2008;60: 1169–1177. 54. Walker PG, Cranney GB, Scheidegger MB, Waseleski G, Pohost GM, Yoganathan AP. Semiautomated method for noise reduction and background phase error correction in MR phase velocity data. J Magn Reson Imaging. 1993;3:521–530. 55. Bernstein MA, Zhou XJ, Polzin JA, King KF, Ganin A, Pelc NJ, Glover GH. Concomitant gradient terms in phase contrast MR: analysis and correction. Magn Reson Med. 1998;39:300–308. 56. Markl M, Bammer R, Alley MT, Elkins CJ, Draney MT, Barnett A, Moseley ME, Glover GH, Pelc NJ. Generalized reconstruction of phase contrast MRI: analysis and correction of the effect of gradient field distortions. Magn Reson Med. 2003;50:791–801. 57. Peeters JM, Bos C, Bakker CJ. Analysis and correction of gradient nonlinearity and B0 inhomogeneity related scaling errors in twodimensional phase contrast flow measurements. Magn Reson Med. 2005;53:126–133. 58. McCauley TR, Pena CS, Holland CK, Price TB, Gore JC. Validation of volume flow measurements with cine phase-contrast MR imaging for peripheral arterial waveforms. J Magn Reson Imaging. 1995;5:663–668. 59. Oshinski JN, Ku DN, Bohning DE, Pettigrew RI. Effects of acceleration on the accuracy of MR phase velocity measurements. J Magn Reson Imaging. 1992;2:665–670. 60. Forster J, Sieverding L, Breuer J, Lutz O, Schick F. Fast acceleration-encoded magnetic resonance imaging. Med Phys. 2001; 28:28–35. 61. Firmin DN, Nayler GL, Kilner PJ, Longmore DB. The application of phase shifts in NMR for flow measurement. Magn Reson Med. 1990;14:230–241.

4

Technical Aspect of Contrast-Enhanced MRA Honglei Zhang, Wei Zhang, and Martin R. Prince

Introduction Although MR angiography (MRA) has revolutionized imaging of vascular diseases, the complicated nature of blood/soft tissue contrast mechanisms and the unique artifacts associated with each of the many techniques have made it challenging for referring physicians to become comfortable interpreting noncontrast MRA studies. Fortunately, contrast-enhanced (CE) MRA provides the type of contrast arteriogram that clinicians and radiologists are comfortable interpreting while eliminating the risks of radiation, iodinated contrast, and arterial catheterization [1]. Just like DSA and CTA, CE-MRA provides reliable enhancement of the arterial lumen during the arterial phase of the Gd bolus injection. Although MR imaging is slower than DSA or CTA, advances in magnet technology, gradient performance, pulse sequence design, and MR contrast agents continue to improve CE-MRA image quality such that it rivals DSA in accuracy for diagnosing vascular anomalies and diseases. This chapter describes the basic principles underlying CE-MRA techniques, approaches to optimizing applications throughout the body, and methods of contrast agent bolus timing.

Theory Unlike conventional MRA techniques which rely on blood flow or intrinsic blood relaxation properties to distinguish vasculature from background tissue, CE-MRA uses a contrast agent (e.g., gadolinium) to shorten the T1 (spin lattice) relaxation time of blood so that the intraluminal signal is brighter than that of surrounding tissues [1, 2] on T1-weighted images. Blood can then be directly imaged using fast,

H. Zhang, MD () • W. Zhang, MD • M.R. Prince, MD, PhD Department of Radiology, Weill Medical College of Cornell University, New York, NY 10022, USA e-mail: [email protected]

three-dimensional, spoiled gradient echo or steady-state free precession pulse sequences. The short echo time of these 3D gradient echo sequences minimizes blood motion, e.g., flow artifacts which are problematic on noncontrast techniques. Spoiled gradient echo is preferred over steady-state free precession especially for large, field of view (FOV) applications to eliminate banding and susceptibility artifacts of steadystate free precession. T1 shortening effect of paramagnetic contrast agents, e.g., gadolinium, is independent of blood flow or scan plane. With CE-MRA, in-plane imaging of vessels allows a small number of slices, required in the plane of those vessels, to quickly image an extensive length of vessel at high spatial resolutions. CE-MRA data can be acquired in coronal, sagittal, or oblique planes to encompass the vascular anatomy of interest with a minimal number of slices. Imaging in the plane of the arteries also takes advantage of the tendency for MR to image at higher resolution in plane compared to through plane. Thus, CE-MRA is intrinsically fast, allowing high-resolution, breath-hold MR angiograms. The high resolution and quality of CE-MRA have yielded sensitivity and specificity for evaluating different vascular territories in the high 90% range using conventional X-ray angiography as gold standard of reference [3–5]. CE-MRA captures central k-space data during the peak arterial phase of a contrast agent bolus [2]. More peripheral k-space data may be collected before and/or after the Gd peak to enhance MRA image resolution. More k-space details are described under “Fourier consideration.” This 3D data can be reformatted into any obliquity to unfold tortuous arteries and to view lesions in multiple organs. Compared to computed tomographic angiography (CTA), volume rendering is more straightforward because arteries are the brightest structures; there is no need to cut away bones since calcium has low signal on MR. As long as the echo time (TE) is sufficiently short, less than 3 ms, vascular calcifications do not interfere with depiction of the artery lumen. The short echo time used with 3D Fourier transform (FT) gradient echo sequences also minimizes artifact from metal clips, bowel gas, and other susceptibility sources.

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_4, © Springer Science+Business Media, LLC 2012

65

66

Pulse Sequence CE-MRA primarily utilizes 3D spoiled gradient echo pulse sequences to take advantage of their high speed, short echo time, and T1 weighting with a single center of k-space for the entire volume of data [2]. CE-MRA can also be used with steady-state free precession with more signal-to-noise ratio (SNR) but also more artifacts, especially with larger FOV, where it is difficult to have good field homogeneity. PC MRA may also improve postinjection of paramagnetic contrast agent. Three-dimensional sequences have high spatial resolution with thin slices and intrinsically high SNR. This sequence is made fast by using short RF pulses or completely eliminating slice selection. Spoiling of the transverse magnetization is useful because it accentuates T1 contrast, thereby magnifying the effect of paramagnetic contrast agents (e.g., gadolinium). It also suppresses signal from background tissues to enhance image contrast and reduce visibility of background aliasing. Mask subtraction and fat suppression can also enhance image contrast, but can create artifacts. For example, motion between the mask and arterial phase creates misregistration artifacts. Poor field homogeneity can cause a fat saturation pulse to drift onto the water peak at the edge of the image or near-susceptibility sources giving the false appearance of occlusion. A recent advance is the acquisition of two echoes, in phase and out of phase with each TR and eliminating fat signal with the Dixon method. Repetition times (TR) used for CE-MRA are generally around 5 ms or less, and TE are typically 1–2 ms. Total scan times range from 10 to 30 s, although this can be lengthened further in regions not affected by respiratory motion for greater resolution, coverage, or SNR. It can also be shortened by sacrificing spatial resolution or coverage and repeated multiple times to provide time-resolved MRA showing the passage of contrast through the volume of interest. Using ultrashort TR, parallel imaging, partial Fourier, and/or sharing of peripheral k-space data, sliding window reconstruction with radial or spiral trajectories, time-resolved CE-MRA is possible at subsecond temporal resolution [1, 6–11]. Even faster temporal resolution can be acquired with Cartesian acquisition with projection reconstruction (CAPR), vastly undersampled isotropic projection reconstruction (VIPR), k–t broad-use linear acquisition speed-up technique (BLAST), or using 2D spoiled gradient echo sequences with complex mask subtraction [1, 12–15]. Gradient echo imaging without spoiling can also be useful when tissue perfusion is of interest.

Fourier (k-Space) Considerations MR imaging does not collect data voxel by voxel. Instead, MR scanners collect spatial frequency data, also known as Fourier or k-space data. With 3D MR imaging, the entire 3D

H. Zhang et al.

“Fourier” or “k-space” dataset is collected before reconstructing individual slices [2]. Because k-space maps spatial frequencies rather than spatial data, k-space data does not directly correspond to image space. Instead, different regions of k-space data determine different image features. For example, the center of k-space or “low” spatial frequencies dominates image contrast, whereas the periphery of k-space or “high” spatial frequencies contributes more to fine details, such as edges. Thus, the state of the contrast agent bolus and intravascular T1 in large vessels is captured at the moment corresponding to acquisition of central k-space [1, 2, 11]. If central k-space is collected when the arterial contrast agent concentration is peaking but has not yet reached capillaries and veins (arterial phase), only the arteries will be bright on MRA images. With optimal timing, contrast injection duration does not need to be as long as the acquisition time when the bolus peak in the artery of interest occurs during acquisition of central k-space data. Most protocols use contrast injection durations that are much shorter than the scan acquisition time. Shorter injection duration with the same total Gd dose results in a faster injection with higher blood Gd concentrations which increases SNR. On the other hand, shorter injections must be timed precisely to avoid artifacts which occur when the contrast bolus peak does not coincide perfectly with acquisition of the center of k-space. For a given Gd dose, the injection strategy is a trade-off between a fast injection for higher intravascular signal versus a slow injection for more uniform signal, easier bolus timing and fewer artifacts. As a general rule, injecting the total contrast dose over approximately half of the acquisition time is optimum if the Gd bolus is perfectly timed for the arterial phase to synchronize with the acquisition of central k-space data. Make sure to avoid a rapid increase or decrease in the concentration of the contrast agent during the acquisition of the central k-space as this causes edge-ringing artifacts (Fig. 4.1). Another factor that must be considered is phase-encoding order. Generally, phase encoding is performed “sequentially” such that central k-space is acquired at the midpoint of the scan. Alternatively, phase encoding can be performed in a “centric” fashion such that central k-space is acquired at the beginning of the scan. Although centric phase-encoding order has been shown to be more prone to artifacts [15], it greatly simplifies bolus timing and may be less susceptible to venous enhancement and artifacts from incomplete breath holding. Centric encoding is generally most effective when performed as an elliptical centric variant, as this concentrates the center of k-space into a shorter time at the beginning of the scan [16, 17]. Another choice is partial Fourier acquisition beginning near the k-space center which has some of the benefits of both centric and sequential ordering [18]. The best suppression of venous enhancement is with recessed elliptical centric encoding. Recessing the absolute center of k-space a few seconds from the beginning of the scan allows

4

Technical Aspect of Contrast-Enhanced MRA

Fig. 4.1 Time-resolved 2D MRA images show ringing artifact in the anterior tibial artery on early images, where there is enhancement of the periphery of k-space but not yet in the center of k-space. Ringing

67

disappears on the third image at 22 s indicating that the center of k-space is fully enhanced. Later images show enhancement of the veins and heel ulcer

Fig. 4.2 MR venogram of the legs with 9 mL gadofosvoset trisodium shows extensive acute deep venous thrombosis of the left upper soleus and gastrocnemius veins with surrounding inflammation

data acquisition to begin on the leading edge of the bolus while still synchronizing the center of k-space with the arterial Gd peak [19].

Contrast Agents At present, most CE-MRA examinations are conducted using gadolinium-based MR contrast agents. Gadolinium (Gd3+) is a paramagnetic metal ion that decreases both the spin–lattice (T1) and spin–spin (T2) relaxation times [3]. Because Gd3+ itself is biologically toxic, it is chelated with ligands, such as DTPA (gadopentetate dimeglumine), HP-DO3A (gadoteridol), gadoversetamide, or gadobenate dimeglumine to form low-molecular-weight contrast agents. These small, “extracellular” agents rapidly redistribute from the intravascular compartment into the interstitial space. Typically, 80% of gadolinium chelate leaks into the intravascular space within 5 min. Thus, imaging of arteries must be performed rapidly to exploit the “first pass” of the contrast agent. As compared with iodinated contrast agents, gadolinium chelates have a very low rate of adverse events and no nephrotoxicity, which is a significant advantage in patients with impaired renal

function. One exception is gadobenate dimeglumine which is ionic and has nearly as many reactions as iodinated contrast agents but very low risk of nephrogenic systemic fibrosis (NSF). Some new high-relaxivity contrast agents are large enough, e.g., USPIO, or bind to large, serum molecules, e.g., gadofosvoset trisodium, so that they stay within the intravascular compartment with minimal leaking out of the capillaries [20, 21]. These agents are referred to as blood pool or intravascular contrast agents. The high relaxivity of these blood pool agents makes them ideal for first pass, arterial phase imaging because a large, contrast effect is achieved with safe, low-injection rates. In addition, the longer intravascular half-life of these contrast agents may allow imaging longer for higher resolution or imaging additional vascular territories during the equilibrium, blood pool phase. Pulmonary and coronary 3D MRA may benefit from these new contrast agents. Other uses relate to venous imaging (Fig. 4.2) and “road map” imaging for monitoring vascular interventions. Blood pool agents may also be useful for identifying gastrointestinal (GI) bleeding using delays in a manner analogous to labeled red blood cells. Similarly, they can be used to detect slow or intermittent stent graft leaks [22].

68

Contrast Dose and Injection Rate In order to image bright blood with respect to background tissues during the arterial phase, sufficient contrast agent must be administered to transiently reduce the arterial blood T1 to substantially less than that of the brightest background tissue, fat, which has a T1 of approximately 270 ms [2]. During the “first-pass” arterial phase, blood T1 is more related to infusion rate and contrast agent relaxivity than total contrast dose, as arterial phase imaging occurs well before intravascular contrast equilibration. Total gadolinium dose and rate of delivery is a trade-off between maximizing intravascular signal (higher injection rate) and minimizing artifacts (longer injection duration). In the ideal situation, a large dose at a high rate would be optimal. This, however, must be weighed against safety, practicality, and cost. Gd chelates’ concentration in the arteries is estimated by the injection rate divided by the cardiac output. Higher injection rates give higher arterial Gd concentration and correspondingly higher SNR until the concentration is so high that T2* effects reduce signal. Diminishing return due to these T2* effects starts to occur for injection rates greater than 5 mL/s for standard, extracellular gadolinium chelates. It is also important to avoid Gd concentration changing too rapidly during acquisition of central k-space data as this creates ringing artifacts. But having a high injection rate for a long duration results in a large dose of Gd which can be expensive and may excessively exceed FDA limits. A good strategy for maximizing injection rate while minimizing the dose is to use a relatively rapid (2–3 mL/s) injection with a duration shorter than the scan duration. Typically, an injection duration that is half of the scan duration is optimum. For a 75 kg patient who received 15 mL of Gd (0.1 mmol/kg) at an injection rate of 3 mL/s for 5 s, the scan duration should be about 10-s long. Due to contrast agent dilution at the leading and trailing edges of the bolus, as well as varying transit times through different portions of the pulmonary circulation, the contrast bolus tends to lengthen as it travels from the antecubital vein (where it is typically injected), through the heart and lungs, to the arteries being imaged. This bolus dispersion depends on individual cardiovascular parameters which are difficult to predict, but at least 5–7 s of bolus prolongation occurs in most individuals [2]. Even greater bolus dispersion can be expected for peripheral arteries, whereas only minimal bolus dispersion can be expected for pulmonary arteries. For fast breath-held acquisitions of the systemic arterial system, the total dose of contrast should be administered at a rate that results in an injection time that is at least 5–7 s shorter than the data acquisition time. An injection duration which is about half the scan duration is recommended since it is not necessary to have peak Gd concentration while acquiring the periphery of k-space.

H. Zhang et al.

The optimal Gd dose also depends upon the vascular territory being imaged and the available software and hardware. Generally, the more contrast agent, the better the MRA image quality because larger contrast agent doses allow high injection rate for a longer time. However, the FDA-approved dose is only up to 0.3 mmol/kg body weight. Recently, concern over NSF has led to greater attention to the dose of Gd use for MR procedures because the risk scales with dose. It appears that for ionic and macrocyclic gadolinium chelates, there is negligible risk at 0.1 mmol/kg as the overwhelming majority of NSF cases have occurred at high doses (Fig. 4.3) [23]. For nonionic linear chelates which have the most cases of NSF, the risk of single dose appears to be less than 1 in 10,000 for patients with GFR <30 mL/min. For extremely fast MRA acquisitions, the contrast agent dose also has to be small because of the limit on how fast one can inject intravenously. Recently, ultralow-dose MRA has been developed by Kramer et al. using just a few mL of Gd for time-resolved imaging of the arch and carotid arteries [24]. These dose issues are entirely different with iron oxide agents which have no risk of NSF.

Patient Preparation The more relaxed and informed a patient is, the more easily they can remain still and perform a long breath hold. Reassurance and a brief description of the scan can help. In patients who are particularly anxious, premedication with sedatives, such as diazepam, may be useful. This helps the patient to relax and lie still. Relaxation also decreases cardiac output which yields a higher arterial contrast agent concentration for the same injection rate. Patients on beta blockers and older patients with reduced cardiac output tend to have the best-quality CE-MRA for this reason. Since most CE-MRA is performed in the coronal plane, arm position is important to allow for a smaller FOV without aliasing, particularly important when using parallel imaging (e.g., SENSE) which is less forgiving of wrap-around artifact [25]. For carotid, large, FOV, sagittal, oblique, aortic or peripheral studies, the arms can remain by the patient’s side, as aliasing is not a factor. For pulmonary, small FOV aortic, renal, and mesenteric studies, the arms must either be elevated out of the imaging plane using cushions, extended overhead, or enclosed in Faraday sleeves. Positions in which the arms are elevated have the added benefit of gravity-aided venous return from the IV injection site. This ensures rapid delivery of the entire Gd dose to the central circulation. Recently, a new pulse sequence know as rotated slice encoding (ROSE) allows for excitation of a limited slab in the sagittal plane with data reconstruction in the coronal plane which eliminates wrap around of the arms without requiring elevation overhead.

4

Technical Aspect of Contrast-Enhanced MRA

69

Fig. 4.3 Bolus chase MRA in a patient with GFR = 25 mL/min using thigh compression and only 11 mL gadobenate dimeglumine to reduce NSF risk

If CE-MRA will be performed with the arms by the patient’s side, placing the IV in the antecubital vein is adequate. If the arms are overhead or crossed over the chest, it may be preferable to place the IV above or below the antecubital fossa so it will not kink and obstruct when the patient’s elbow bends. A 20- or 22-gage catheter is optimal. For smalldiameter catheters (greater than 22 gage), warming the contrast to body temperature decreases the viscosity to allow for adequate injection rates. Use of a standardized IV tubing system is helpful to avoid uncertainty about priming volume and to simplify flushing.

Contrast Injection (Hand Versus Power Injection) Hand injection gives the operator more control over bolus administration and allows early detection of IV problems, such as blockage or extravasation. Hand injection is always preferred for pediatric patients (especially babies) when injecting central lines or when there is a tenuous IV line. Hand injections avoid tethering the IV to a mechanical pump, eliminating the problem of inadvertently pulling out the IV when the patient is advanced into the magnet. Problems of battery failure and accidental premature injection are also eliminated with hand injection. With hand injection, it is helpful to always use the same IV tubing system to avoid

uncertainty about flow resistance and how much contrast agent must be injected to fill the IV tubing. A power injector has the advantage of delivering precise infusion of contrast agent at a consistent and predictable injection rate, provided there are no obstructions [26]. It also allows a single operator to run the MRA exam from the control room without having to enter the scanner room. With power injection, it is important to carefully monitor the patient using a pulse oximeter or other monitoring device, as it is more difficult to detect a contrast reaction from the control room as compared to standing next to the patient for hand injections. Either way, the contrast injection must be immediately followed by adequate saline flush to complete delivery of the bolus and help flush the arm veins. Suggested flush volumes range from 15 to 50 mL, with most authors using 20 mL [2]. Improved filling of peripheral k-space has been suggested with a dilute Gd flush composed of 1 part Gd and 9 parts saline.

Bolus Timing Considerations Gd bolus timing relative to acquisition of central k-space is critical for CE-MRA. The contrast travel time, defined as the time required for contrast to travel from the injection site to the vascular territory of interest, is highly variable. It depends

70

on the degree to which the bolus “broadens” as it transits the cardiopulmonary vasculature. Mean transit time to the common femoral artery of 24 ± 6 s, with a range of 13–37 s, has been reported [27]. Older patients, patients with a history of congestive heart disease, patients on beta blockers, and patients with aortic aneurysms tend to have slower flow.

“Best Guess” Technique Several solutions to proper bolus timing have been developed. The simplest is to just make an educated “best guess.” Sequential phase encoding is preferred when using “best guess,” as this encoding strategy as well as longer scan duration are more tolerant of timing errors. The most difficult aspect of best guess technique is estimating the contrast agent travel time. If timing is not accurate, it is better to image slightly late (increased venous phase) than slightly early (ringing/banding artifact). With this in mind as a general guideline, the contrast agent travel time from an antecubital vein to the abdominal aorta would be approximately 15 s for a young healthy patient and a few seconds less for hypertensive or athletic patient. Contrast agent travel times would be approximately 20–25 s for a healthy elderly patient (>70 years old), 25–35 s in patients with cardiac disease or an aortic aneurysm, and up to 40–50 s when severe cardiac failure in conjunction with aortic aneurysm is present. Add 3 s if the IV is in the hand [1]. Best guess technique works well for scan time of at least 40 s or longer (with sequential k-space encoding) when the impact of 5 s of timing error is minimal.

H. Zhang et al.

background signals. Finally, the imaging bolus may not behave identical to the test bolus due to moment-to-moment patient variables, such as venous return and cardiac output, as well as the different total volume of injection. It is especially important to be aware that a test bolus with the arm at the patient’s side may not be accurate if the forearms are then repositioned above the patient’s head for the actual scan as overhead positioning the patient’s arms may stretch and pinch the subclavian veins.

Automatic Triggering Another approach to bolus timing uses a pulse sequence that can automatically detect contrast arrival in the aorta and synchronize 3D MRA data acquisition with the arterial phase of the contrast bolus. The operator selects a region of aorta to be sampled at 20-ms intervals. As contrast arrives in the aorta, the signal within the sampling region increases. A trigger threshold of typically 20% signal increase detects the leading edge of the contrast bolus. This gives the patient time to take in a deep breath and suspend respiration before the actual scan commences 5 s later. By beginning the scan near the center of k-space, i.e., centric or preferably recessed elliptical centric encoding, there is synchronization between the arterial phase of the bolus and elliptical centric acquisition of central k-space data. Commercial versions of this pulse sequence are available as “SmartPrep” (GE Medical Systems) or “Bolustrack” (Philips Medical Systems).

MR Fluoroscopy Test Bolus Technique A bolus-timing acquisition can be performed prior to acquisition of 3D MRA dataset using a small Gd contrast dose of ~2 mL followed by a 20-mL saline flush at the same rate as planned for the actual injection. Multiple, single-slice, fast gradient echo images in the vascular region of interest are acquired as rapidly as possible (every 1–2 s) for approximately 1 min. In order to minimize time-of-flight effects, the 2D test bolus images should either be oriented in the plane of imaged vessel (i.e., sagittal or coronal for the aorta) or alternatively be relatively thick (greater than 1 cm) with a superior saturation band or a blood-nulling inversion prepulse. The time of peak arterial enhancement (contrast travel time) is then determined visually or using ROI analysis. While quite effective, there are several drawbacks to the test bolus technique. Setting up, performing, and analyzing the test bolus lengthen the overall examination time. The test bolus rapidly redistributes into the interstitial space and is excreted by the kidneys into the ureters, thereby increasing

Perhaps, the most popular method of contrast bolus timing is to use extremely rapid MR fluoroscopy [1, 2, 16, 17]. With this technique, 2D gradient-refocused images are rapidly (less than 1 s/image) obtained through the vascular structure of interest, ideally using complex subtraction to improve contrast and decrease artifacts. Images are generated in near real time and updated at greater than once per second. The operator watches the contrast bolus arrive and then switches over to recessed elliptical centric 3D MRA when the desired enhancement is detected. This technique allows real-time, operator-dependent decision making. This may be particularly advantageous in cases with unusual or asymmetric flow patterns, such as unilateral stenoses and slow filling aneurysms. The ability to assess the vasculature “on the fly” under these circumstances is a great asset, as is the time and contrast savings associated with avoiding a test injection. It is useful to shift the MR fluoroscopic monitoring to a proximal region of vascular anatomy to get more advanced warning of contrast agent arrival. Using carotid MRA as an

4

Technical Aspect of Contrast-Enhanced MRA

71

example, MR fluoroscopic monitoring in the chest shows contrast agent arriving in the subclavian vein, then right heart, pulmonary arteries, left heart, arch, and finally proximal great vessels. Tracking the bolus over such a long period makes precise triggering easier. If the flow is very slow, the operator can compensate to allow for greater target vessel enhancement before triggering. Fluoroscopic triggering is improved by recessing the absolute center of k-space a few seconds from the beginning of the 3D acquisition, thus avoiding the previously described “leading-edge” artifact from early triggering.

More recent refinements traverse k-space using radial or spiral projections [1], which allow for sliding window reconstructions at very high temporal and spatial resolutions. VIPR and CAPR are particularly promising. VIPR can be combined with phase-contrast imaging as well. Another strategy that works particularly well with the sparse MRA datasets is to use training data acquired either pre or post contrast injection. Training data and post-Gd steady-state data help calculate high spatial resolution during high temporal resolution undersampling while contrast agent is injected, e.g., k–t BLAST or HYPR TRICKS.

Time-Resolved 3D Contrast MRA

Imaging Different Vascular Phases

Because proper bolus timing is difficult and a timing mistake can ruin the study, time-resolved techniques have been developed whereby multiple 3D datasets are acquired extremely rapidly (typically, 1–10 s per acquisition). Bolus timing is no longer a factor as multiple vascular phases are obtained without any predetermined timing (i.e., inject and begin scanning simultaneously). The operator simply selects the desired image set: pure arterial phase, maximum venous, etc. This is particularly useful in the carotid and pulmonary arteries, where the venous phase is extremely rapid, and in the calf, where variable rate filling may occur due to stenoses, occlusions, or rapid AV shunting due to soft tissue ulcers, cellulitis, and other inflammatory diseases [2]. The most straightforward way to accelerate acquisition time is using some combination of limiting the imaging volume, decreasing the resolution (e.g., fewer, thicker slices and fewer phase-encoding steps), decreasing TR, and partial Fourier parallel imaging. Recent developments in gradient systems allow TRs of <2 ms on many systems. Simultaneous high spatial resolution and high temporal resolution over a large FOV can be obtained using 3D TRICKS (time-resolved imaging of contrast kinetics) [28– 30] and its variants, including TREAT, HYPR-TRICKS, and PR TRICKS. TRICKS divides k-space into multiple blocks. Central k-space (which contributes most to overall image contrast) is collected more frequently (oversampled) than blocks of data corresponding to more peripheral k-space. Images are then “synthesized” at high temporal resolution by piecing together each unique block of central k-space data with an interpolation of the remaining k-space blocks acquired in closest temporal proximity. This technique allows greater temporal and spatial resolution/volume coverage than with a streamlined conventional sequence. k-space discontinuities, in conjunction with varying intravascular Gd concentration, may lead to ringing artifacts. These artifacts, however, can be minimized by making sure that the Gd bolus is not too compact.

With most 3D CE-MRA studies, extensive efforts are made to optimally image the arterial phase. Once the arterial phase dataset is collected, the sequence can (and should) be repeated to obtain venous and equilibrium phases. Later phases are useful in detecting and evaluating dissections, portal venous or other venous structures, parenchymal enhancement patterns, tumors, and perhaps even renal glomerular function. Because these “later”-phase examinations occur more in the equilibrium phase of gadolinium distribution, signal is reduced as compared to arterial phase. In order to maximize signal in these later phases, the flip angle should, if possible, be reduced. Whereas the optimum flip angle for the arterial phase may be 30–40° at a TR of 4–6 ms, the optimal flip angle decreases to 15–25° in later phases. To image highly concentrated Gd excreted into the collecting system and ureter, use a high flip angle (45–60°) with the widest bandwidth and shortest possible echo time.

Postproccessing and Display CE-MRA produces a 3D volume of image data. The default data presentation is a series of contiguous images (slices) in the plane of acquisition (typically, coronal plane). These are referred to as source images. This dataset is best viewed interactively using a computer workstation allowing for thin multiplaner reformatting (MPR). In this manner, thin (typically, 2.0 mm) slices reconstructed at 1-mm intervals with 50% overlap can be viewed in axial, sagittal, coronal, or oblique planes. This eliminates overlapping structures and unfolds tortuous vessels. Because these reformatted slices remain thin and are not projections, they not only provide the best achievable contrast, but also minimize the chance of diagnostic error. Because single-voxel-thick reformations show only short vascular segments, it is advantageous to utilize the maximum intensity projection (MIP) postprocessing technique for

72

combining information from multiple slices to see longer segments of vessels. Using this algorithm, the user specifies the subvolume thickness and obliquity. The algorithm then generates rays perpendicular to the subvolume and uses the maximum value of any voxel encountered along that ray for the corresponding pixel intensity in the output image. This technique is fast, and works extremely well with CE-MRA, since vascular signal is much greater than background signal. It provides projection images that are similar to conventional angiograms. This technique is helpful for displaying complex vascular anatomy, particularly when vessels are not oriented along a single plane. Despite its usefulness, the MIP algorithm is subject to artifacts. Perhaps, the most common artifact arises when stationary tissue has greater signal intensity than the vascular structures of interest, as can occur in the presence of other vessel segments, fat, hemorrhage, metallic susceptibility artifacts, or motion artifacts. This in turn leads to the mapping of nonvascular signal into the projection image and causes a discontinuity in vessel signal, potentially mimicking a stenosis or occlusion. This type of artifact is best overcome by minimizing the thickness of the MIP subvolume, thereby excluding as much extraneous data as possible. Other artifacts inherent to the MIP technique have been described. These mainly consist of underestimating vessel lumen, and are more of a problem with TOF or PC techniques. Zebrastripe artifact from slices too thick can be reduced using zero filling in the slice direction. Because of these potential pitfalls, most authors agree that the MIP images should be used as a roadmap, utilizing the source images for definitive diagnosis. Some authors suggest volume rendering may be more accurate than the MIP and similar to using the MPR, but should still not replace careful evaluation of the source images and MPR. Subtraction techniques are also useful in image evaluation, particularly in vascular regions not subject to significant respiratory motion. These areas include the extremity, pelvic, and carotid and intracerebral arteries. A “digital subtraction” MRA is easily produced by subtracting precontrast from postcontrast magnitude image data. Provided the patient maintains the same position on both studies, subtraction will reduce background signal and improve vessel conspicuity. Subtraction is particularly useful in the extremities because they are stationary and subsequently have less chance for misregistration between the precontrast mask and arterial phase images. An improvement on this technique, particularly with thicker slice or 2D projection MRA exams, involves complex subtraction of the pre- and postcontrast “complex” or “raw” (k-space) datasets. Complex, vector subtraction detects both the phase and magnitude effects of Gd to obtain a larger effect from the contrast agent.

H. Zhang et al.

Whole-Body MRA The systemic nature of vascular disease, particularly atherosclerosis, means that vascular pathology is likely to be present in multiple vascular territories. With faster gradient systems and specialized coils, whole-body MRA to cover from the carotid arteries to the calf can be achieved by adding proximal stations to a peripheral MRA exam. However, it is difficult to scan fast enough with MR to keep up with the rapid flow of contrast down the legs. Excessive venous enhancement may obscure calf arteries and limit diagnostic utility in the lower extremities. A possible solution of this technique is placement of blood pressure cuffs (e.g., SmartTourniquet, TopSpins Inc, Ann Arbor, MI) as high up as possible toward the groin around the thighs for venous compression [31]. By applying subsystolic thigh compression with blood pressure cuffs, the window of arterial enhancement in the calf (only 35 ± 14 s without compression) is lengthened considerably. Venous compression whole-body MRA is simple to implement and is likely to enhance the performance of all other multistation MRA strategies for assessment of the peripheral arterial tree.

References 1. Zhang HL, Zhang W, Juluru K, Prince MR. Body magnetic resonance angiography. Semin Roentgenol. 2009;44:84–98. 2. Prince MR, Grist TM, Debatin JF. 3D Contrast MR Angiography. New York: Springer, 2003. 3. Fain SB, King BF, Breen JF, Kruger DG, Riederer SJ. High-spatialresolution contrast-enhanced MR angiography of the renal arteries: a prospective comparison with digital subtraction angiography. Radiology. 2001;218:481–490. 4. Hathout GM, Duh MJ, El-Saden SM. Accuracy of contrastenhanced MR angiography in predicting angiographic stenosis of the internal carotid artery: linear regression analysis. AJNR Am J Neuroradiol. 2003;24:1747–1756. 5. Schoenberg SO, Rieger J, Weber CH. High-spatial-resolution MR angiography of renal arteries with integrated parallel acquisitions: comparison with digital subtraction angiography and US. Radiology. 2005;235:687–698. 6. Chung T, Muthupillai Rh. Application of SENSE in clinical pediatric body MR imaging. Top Magn Reson Imaging. 2004;15: 187–196. 7. Fink C, Ley S, Kroeker R, Requardt M, Kauczor HU, Bock M. Time-resolved contrast-enhanced three-dimensional magnetic resonance angiography of the chest: combination of parallel imaging with view sharing (TREAT). Invest Radiol. 2005;40:40–48. 8. Tsuchiya K, Honya K, Fujikawa A, Tateishi H, Shiokawa Y. Postoperative assessment of extracranial-intracranial bypass by time-resolved 3D contrast-enhanced MR angiography using parallel imaging. AJNR Am J Neuroradiol. 2005;26:2243–2247. 9. Xu Y, Haacke EM. Partial Fourier imaging in multi-dimensions: a means to save a full factor of two in time. J Magn Reson Imaging. 2001;14:628–635.

4

Technical Aspect of Contrast-Enhanced MRA

10. Hu HH, Madhuranthakam AJ, Kruger DG, Glockner JF, Riederer SJ. Combination of 2D sensitivity encoding and 2D partial fourier techniques for improved acceleration in 3D contrast-enhanced MR angiography. Magn Reson Med. 2006;55:16–22. 11. Zaitsev M ZK, Shah NJ. Shared k-space echo planar imaging with keyhole. Magn Reson Med. 2001;45:109–117. 12. Zhu H, Buck DG, Zhang Z, et al. High temporal and spatial resolution 4D MRA using spiral data sampling and sliding window reconstruction. Magn Reson Med. 2004;52:14–18. 13. Mistretta CA, Wieben O, Velikina J, et al. Highly constrained backprojection for time-resolved MRI. Magn Reson Med. 2006; 55:30–40. 14. Wang Y, Johnston DL, Breen JF, et al. Dynamic MR digital subtraction angiography using contrast enhancement, fast data acquisition, and complex subtraction. Magn Reson Med. 1996;36:551–556. 15. Wilman AH, Yep TC, Al-Kwifi O. Quantitative evaluation of nonrepetitive phase-encoding orders for first-pass, 3D contrastenhanced MR angiography. Magn Reson Med. 2001;46:541–547. 16. Riederer SJ, Fain SB, Kruger DG, Busse RF. Real-time imaging and triggering of 3D contrast-enhanced MR angiograms using MR fluoroscopy. Magma. 1999;8:196–206. 17. Wilman AH, Riederer SJ, King BF, Debbins JP, Rossman PJ, Ehman RL. Fluoroscopically triggered contrast-enhanced threedimensional MR angiography with elliptical centric view order: application to the renal arteries. Radiology. 1997;205:137–146. 18. Goldfarb JW, Prasad PV, Griswold MA, Edelman RR. Dynamic three-dimensional magnetic resonance abdominal angiography and perfusion: implementation and preliminary experience. J Magn Reson Imaging. 2000;11:201–207. 19. Watts R, Wang Y, Redd B, et al. Recessed elliptical-centric viewordering for contrast-enhanced 3D MR angiography of the carotid arteries. Magn Reson Med. 2002;48:419–424. 20. Robert P, Violas X, Santus R, Le Bihan D, Corot C. Optimization of a blood pool contrast agent injection protocol for MR angiography. J Magn Reson Imaging. 2005;21:611–619. 21. Li W, Tutton S, Vu AT, et al. First-pass contrast-enhanced magnetic resonance angiography in humans using ferumoxytol, a novel ultr-

73

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

asmall superparamagnetic iron oxide (USPIO)-based blood pool agent. J Magn Reson Imaging. 2005;21:46–52. Ersoy H, Jacobs P, Kent CK, Prince MR. Blood pool MR angiography of aortic stent-graft endoleak. AJR Am J Roentgenol. 2004;182:1181–1186. Prince MR, Zhang H, Morris M, et al. Incidence of nephrogenic systemic fibrosis at two large medical centers. Radiology. 2008;248:807–816. Kramer U, Fenchel M, Laub G, et al. Low-dose, time-resolved, contrast-enhanced 3D MR angiography in the assessment of the abdominal aorta and its major branches at 3 Tesla. Acad Radiol. 2010;17:564–576. Born M, Willinek WA, Gieseke J, von Falkenhausen M, Schild H, Kuhl CK. Sensitivity encoding (SENSE) for contrast-enhanced 3D MR angiography of the abdominal arteries. J Magn Reson Imaging. 2005;22:559–565. Kopka L, Vosshenrich R, Rodenwaldt J, Grabbe E. Differences in injection rates on contrast-enhanced breath-hold three-dimensional MR angiography. AJR Am J Roentgenol.1998;170:345–348. Prince MR, Chabra SG, Watts R, et al. Contrast material travel times in patients undergoing peripheral MR angiography. Radiology. 2002;224:55–61. Turski PA, Korosec FR, Carroll TJ, Willig DS, Grist TM, Mistretta CA. Contrast-Enhanced magnetic resonance angiography of the carotid bifurcation using the time-resolved imaging of contrast kinetics (TRICKS) technique. Top Magn Reson Imaging. 2001;12:175–181. Naganawa S, Koshikawa T, Fukatsu H, et al. Contrast-enhanced MR angiography of the carotid artery using 3D time-resolved imaging of contrast kinetics: comparison with real-time fluoroscopic triggered 3D-elliptical centric view ordering. Radiat Med. 2001;19:185–192. Du J, Carroll TJ, Wagner HJ, et al. Time-resolved, undersampled projection reconstruction imaging for high-resolution CE-MRA of the distal runoff vessels. Magn Reson Med. 2002;48:516–522. Zhang HL, Ho BY, Chao M, et al. Decreased venous contamination on 3D gadolinium-enhanced bolus chase peripheral MR angiography using thigh compression. AJR Am J Roentgenol. 2004;183: 1041–1047.

wwwwwwwwwwww

5

Time-Resolved, Contrast-Enhanced MR Angiography Using Cartesian Methods Stephen J. Riederer, Clifton R. Haider, Casey P. Johnson, and Petrice M. Mostardi

Introduction and Overview Ever since the introduction of contrast-enhanced MR angiography (CE-MRA) in the mid-1990s [1, 2], there has been steady improvement in the technique. Early investigations identified a number of performance targets, including the desirability for high-spatial-resolution 3D imaging, need to synchronize the acquisition to the arterial phase of the contrast bolus passage, and desirability for minimal enhancement of the venous system. To a great extent, these were addressed with the development of short repetition time (TR) gradient echo pulse sequences, nonreal-time [3] and realtime [4, 5] means for determination of accurate timing, and development of centric view orders which allowed extended acquisition times into the venous phase [6]. Early on, there also was interest in the development of time-resolved techniques to allow visualization of the temporal passage of the contrast bolus through a targeted vascular system. One way to do this was by simply repeating the application of a pulse sequence. For 3DFT acquisition, the total number of repetitions is equal to NY × NZ, where NY and NZ are the number of phase-encoding steps in the y and z directions, respectively. The acquisition time for one image, Tacq, is then given by Tacq = NY × NZ × TR. For repetition times of 10 ms, as available in the mid-1990s, this restricted the NY × NZ values to be in the range of 64 × 32 in order to limit the acquisition times to about 20 s. Times much longer than this would have limited value in portraying contrast bolus passage, particularly for the arterial phase. One way to allow shorter acquisition times was to convert from a 3DFT to a 2DFT pulse sequence, as studied with “MR DSA” [7]. This allowed the portrayal of contrast material passage with acquisition times of 2 s or less, but in a projection or thickS.J. Riederer, PhD () • C.R. Haider, PhD • C.P. Johnson • P.M. Mostardi Department of Radiology, Mayo Clinic, Rochester, MN 55902, USA e-mail: [email protected]

slice format; i.e., the z spatial resolution was sacrificed in exchange for improved time resolution. The above illustrates the fundamental trade-off between spatial and temporal resolution in MRI. During the time the contrast bolus passes through the vasculature, the MR data acquisition time can be spent in sampling as many low- and high-spatial-frequency k-space phase encodings as possible once each or it can be spent in sampling some subset of spatial frequencies multiple times each. The first of these provides a single, high-spatial-resolution image. The latter provides a time series of images having reduced spatial resolution. One way to modify this relationship is to use the method of “view sharing” [8] in which images are reconstructed at time intervals smaller than the intrinsic acquisition time Tacq. A variant of this is to sample the low spatial frequencies in 3D k-space at a higher rate than high spatial frequencies and to then reconstruct a full image at the sampling rate used for the low spatial frequencies. This is the basis of keyhole imaging [9] and the TRICKS technique [10]. Because the appearance of an MR image is dominated by the state of the magnetization during the measurement of the low spatial frequency or “central” views, the resultant images tend to at least superficially represent the contrast material distribution at the times of the central k-space samplings. However, depending on the manner and speed by which k-space is sampled, the resultant images are subject to artifact [11], as to be discussed briefly in this chapter. The above-described trade-off of spatial and temporal resolution was altered by the discovery of methods of “parallel acquisition,” first introduced in the late 1990s. Parallel acquisition is based on the principle that the redundancy of measurements of the magnetization made using multiple vs. a single receiver coil can allow a reduction in the total number of repetitions of the acquisition for some given spatial resolution. With one class of parallel acquisition techniques, the signal values at specific target k-space phase-encoding locations are estimated from samples at nearby measured points and from knowledge of the spatial response functions of the receiver coils. Such k-space-based methods include

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_5, © Springer Science+Business Media, LLC 2012

75

76

“SMASH” [12] and “GRAPPA” [13]. With the second broad class of techniques called “SENSE” [14], aliased reconstructed images formed individually from multiple receiver coils are algebraically combined with the coil spatial response functions to provide the unaliased result. The above-cited works describe parallel acquisition performed along one phase-encoding direction, for which case the typical maximum practical acceleration R is about 4. However, these methods can be applied along the two phase-encoding directions of 3DFT acquisition [15] for which accelerations of R = 8 or more are routinely possible. The use of parallel acquisition has a price: compared to an unaccelerated acquisition, the use of a reduced number of measurements and the additional mathematics required to form the final image adds complexity and causes a reduction of the signal-to-noise ratio (SNR). If severe, the SNR loss may render the accelerated images nondiagnostic. One of the only ways to address this is by the use of a receiver coil array which is designed to perform parallel acquisition effectively. Also, compared to applications in which the magnetization state does not change from one repetition to the next, CE-MRA is more immune to SNR loss. This is because the acceleration in effect causes more of the k-space data to be sampled during the highly enhanced phase of the contrast bolus passage [16]. For these reasons, effective coil design and 2D parallel acquisition are well-matched to the standard acquisition techniques used for CE-MRA. In the remainder of this chapter, a number of the above techniques and relationships are discussed in more detail. Examples are drawn from time-resolved CE-MRA of several vascular territories.

Methods

S.J. Riederer et al.

reconstructed, and ideally is chosen to be just larger than the size of the object along that direction. The total length along a k-space direction over which sampling is performed is proportional to the spatial resolution along that conjugate direction. For example, the further one samples along the kX direction, then the finer the spatial resolution is along x. Once k-space is adequately sampled, the final image is formed by inverse Fourier transform of the accumulated data set. With Cartesian methods, the sampled k-space points are along a rectilinear pattern, with the principal directions of the pattern along kX, kY, and kZ. For 3D acquisition, it is assumed that the kX direction corresponds to frequency encoding, and during one repetition of the acquisition all kX points for a specific combination of (kY, kZ) values are sampled. Because sampling along kX occurs quickly, within msec, it is convenient to characterize the specific manner that k-space is sampled in Cartesian 3D acquisition by simply focusing on the kY–kZ plane. Non-Cartesian methods can be described as those in which the k-space sampling deviates from that described above. The 3D rectilinear sampling pattern is disrupted along one or more of the three directions. Cartesian and non-Cartesian methods both have their relative advantages. Cartesian techniques are useful in general and valuable for 3D CE-MRA for several specific reasons. These include the following: (1) the rectilinear k-space sampling pattern lends itself to straightforward image reconstruction by fast Fourier transformation; (2) truncated FOVs can readily be created along two directions using the processes of slice selection (along z) and bandwidth limitations (along x); (3) the intrinsic rectilinear sampling pattern readily allows the regular undersampling desired for performing parallel acquisition; and (4) the ability to easily and preferentially sample the center of k-space is retained. Description of non-Cartesian methods is outside the scope of this chapter.

Cartesian Sampling Here, a brief review of Cartesian sampling is provided as background for the pulse sequences to be described in the remainder of the chapter. With virtually all MRI acquisition techniques, an image is formed by sampling its Fourier transform. If the spatial coordinates of the 3D image are (x, y, z), then the “conjugate variable” coordinates of its Fourier transform are (kX, kY, kZ). This latter, three-dimensional space is called “k-space.” For each repetition of the data acquisition, a different region of k-space is sampled, the specific region depending upon the gradient waveforms applied. The MR signal is sampled at discrete time points, causing the sampling of k-space to be performed at discrete points. The spacing between consecutive k-space points in a specific direction, e.g., kX, is equal to the reciprocal of the field of view (FOV) along the conjugate (here, x) direction in the resultant image. FOV is the distance over which the image is periodically

k-Space Sampling for Time-Resolved CE-MRA Figure 5.1 shows a progression of sampling patterns for Cartesian 3DFT acquisitions. In all cases, the kY–kZ phaseencoding plane is shown, and the dots indicate the specific possible locations at which samples can be measured. As discussed above, the rectilinear pattern of dots is due to the Cartesian nature of the sampling. Figure 5.1a also shows an orange central disk and black annular region, the black region extending out to radius kM. The disk defined by this radius specifies the region within which actual measurements are made at the discrete points. The radius kM is selected according to the desired spatial resolution: the larger the radius, the finer the spatial resolution. In this figure, sampling of the corners of kY–kZ space is not performed. This is done to make the spatial resolution approximately isotropic within the y–z

5

Time-Resolved, Contrast-Enhanced MR Angiography Using Cartesian Methods

77

Fig. 5.1 Plots of the kY–kZ phase-encoding plane for 3DFT acquisition and various CAPR sampling patterns and readouts. (a) Full sampling out to some maximum k radius, kM. (b) Schematic time ordering of the sampling pattern of (a). (c) Sampling pattern with undersampling allowed in the annular black region. (d) Time ordering used to sample (c). Note the reduced overall duration because of the annular undersampling. (e) Sampling pattern used for CAPR. The vanes of (b) are further apportioned into individual vane sets, each set designated with a different color. (f) Time ordering used for the CAPR acquisition. Sampling alternates between the central orange region and one of the vane sets. (g) Further undersampling applied to the CAPR pattern as allowed with parallel acquisition along both the kY and kZ directions. (h) Time ordering for accelerated CAPR

plane [17]. Also, not sampling the corners provides an approximate 25% (p/4) reduction of acquisition time vs. the case of sampling the full square array of points. It is possible to use different maximum k-space radii along the kY and kZ directions of Fig. 5.1a in case it is desired to have different spatial resolution along y vs. z. Similarly, different FOVs along y and z would cause the spacing between points to differ along the kY and kZ directions in accordance with the 1/ FOV dependence described earlier. Figure 5.1b illustrates a possible time order with which the desired phase-encoding “views” of Fig. 5.1a are sampled. The experimentalist has control of this and may select a specific view order depending on the application. For CE-MRA, the “elliptical centric” view order [6] is commonly used, in which each measured point is first ranked according to its radius from the kY–kZ origin, smallest to largest. The time order of sampling is done using this ranking. In this example of Fig. 5.1a, b, because the points within the orange disk are

closest to the k-space origin, they are sampled first in the sequence. Once these are completely sampled, the process continues to sampling of the points within the black annular region. This is represented schematically in the time ordering of the orange and black blocks of Fig. 5.1b. Similar centric-like samplings are also possible [18]. Figure 5.1c illustrates a variant of the sampling shown in (A) in which the circular sampling region has been decomposed into a central orange disk and a set of vanes. As before, only those points falling within one of these regions are actually sampled. Thus, in the outer annular region, points which lie in a gap between two vanes are not sampled. This reduction in the number of sampled points allows a proportionate reduction in the acquisition time vs. (A), a factor approaching 2.0 in the limit of the central orange region becoming increasingly small. During image reconstruction, the data in these gaps can be filled with zeroes or alternatively it can be estimated using partial Fourier–homodyne techniques [19, 20].

78

In practice, the radius of the central orange disk is typically 0.2–0.4 that of kM, leading to an acquisition time reduction or “acceleration” factor due to this homodyne processing of RHD = 1.6–1.8. Figure 5.1d shows the accompanying temporal playout of the samples of Fig. 5.1c assuming elliptical centric ordering. Note the reduction in acquisition duration depicted schematically vs. (B). Repeating the measurement process of Fig. 5.1d would provide a time-resolved 3D Cartesian sequence for which the frame time or the time between consecutive reconstruction time points would match the overall acquisition time for one image. However, as shown in (E), it is possible to convert the pattern of (C) into one which readily allows view sharing and a reduced frame time. In this case, the vanes comprising the outer annular region are subdivided into sets, each vane set shown in a different color, four in this example. The same points in kY–kZ space are sampled as in (C), but the temporal playout is now altered as shown in (F). This starts with a sampling of the central orange disk region, but is then followed with that of one of the vane sets, here black. The time ordering within all colored blocks is again done using the EC view order. This is followed by a second sampling of the orange disk, followed by sampling of a second vane set, here green. This process repeats until all four vane sets have been sampled, at which time the process repeats itself. Figure 5.2 shows how the measured data are sorted into the image series. Figure 5.2a repeats the playout of data acquisition shown in Fig. 5.1f. Once all four vane sets have been sampled, reconstruction for the first image can be performed. Due to the redundant sampling of the central disk, there is flexibility in how the sorting is done. One method is that shown in Fig. 5.2b. In this case, in addition to the four vane sets, the central orange k-space measurement made just before the last vane set is used in image reconstruction. In studies to be briefly described later in this chapter and described in detail in Mostardi et al. [11], this has been found to provide a good compromise between sharpness in portraying the bolus leading edge and spatial resolution. In Fig. 5.2b, the short vertical arrow indicates the specific time to which the first reconstructed image, “Image 1,” is ascribed. This time corresponds to sampling of the central-most view within the orange region. Again, in Mostardi et al. [11], assignment of this frame time to this point within the data acquisition has been shown to be accurate. Figure 5.2c shows sorting for the second image. In this case, the earliest measured vane set (black) of Image 1, along with central k-space, has been replaced with updated samples while measurements for the three other vane sets have been retained. This process of replacing central k-space and one vane set and sharing views from the others continue for subsequent images in the series as shown. The sampling pattern of Fig. 5.1e and data sorting of Fig. 5.2 are but one specific implementation of the CAPR pulse sequence. Variables include the relative radius of the central

S.J. Riederer et al.

Fig. 5.2 (a) Time ordering of data acquisition for CAPR. The colored blocks refer to the regions of k-space identified in Fig. 5.1e. (b) Selection of data from (a) used to form Image 1. Vertical arrow identifies the time ascribed to the image. (c) Selection of data from (a) used to form Image 2. (d) Time ordering of data acquisition for 2D-accelerated CAPR with colored blocks referenced to Fig. 5.1g. Because of the reduced number of points within each region, the durations of all colored blocks are smaller than in (d) vs. (a). (e, f) Selection of data from (d) used to form Images 1 and 2

orange disk, the number of vane sets, and whether or not 2D homodyne reconstruction is used. In the limit of only one vane set with 2D homodyne sampling disabled, the sequence reverts to a standard, single-phase 3D acquisition (Fig. 5.1a). Alternative patterns for sampling kY–kZ space using a rectilinear Cartesian sampling pattern and for data sorting have also been studied. These include the original TRICKS technique which also used temporal interpolation [10], TREAT [21, 22], TWIST [23, 24], keyhole methods [9, 25, 26], and others [27]. Details of these are available in the respective references.

Parallel Acquisition A method which has proven to be critical in favorably altering the balance between spatial and temporal resolution is

5

Time-Resolved, Contrast-Enhanced MR Angiography Using Cartesian Methods

parallel acquisition. As described earlier in this chapter, the method is based on simultaneously measuring the MRI signal with multiple receiver coils and using the redundant information to estimate unmeasured samples. Avoiding actual measurement of the estimated samples allows a reduction in acquisition time. Parallel acquisition was originally described for reducing the number of samples along the single phase-encoding direction of 2DFT Cartesian acquisition, and acceleration factors R in the range of 2–4 were reported [28, 29]. Accelerations larger than this often lead to unacceptably low SNR. However, it was subsequently recognized that the method can be applied to multiple phase-encoding directions, such as the two directions of 3DFT acquisition [15]. This generally allows a given acceleration at considerably less SNR loss than for 2DFT acquisition. An accelerated acquisition is performed by reducing the number of sampled points in k-space and accounting for this in the reconstruction. For purposes of discussion in this chapter, it is assumed that the “SENSE” parallel acquisition technique is performed, but many aspects of the implementation apply to other acceleration techniques as well. Starting with the original sampling pattern, the undersampling is generally done evenly across all of k-space. This is where the intrinsic, evenly spaced rectilinear pattern of Cartesian sampling readily lends itself to the further undersampling desirable for parallel acquisition. Suppose, for example, it is desirable to undersample by twofold along both kY and kZ, yielding a net acceleration R = RY × RZ = 2 × 2 = 4. The CAPR sequence presented previously (Fig. 5.1e) can be readily converted to this (Fig. 5.1g). Every other measured point of the original pattern has been eliminated along both the kY and kZ directions. The temporal playout of views with this SENSE-accelerated view-shared sequence is schematically shown in Fig. 5.1h. Because of the acceleration, the temporal width of each region is reduced, thus allowing a reduction in the frame rate as well as the overall acquisition time necessary for each image. Figure 5.2d–f shows the view playout and data sorting for the case of accelerated CAPR acquisition of Fig. 5.1g, h. Note that for the same extent of k-space coverage, i.e., the same spatial resolution, the frame time is markedly reduced compared to the unaccelerated case of Fig. 5.2a–c. One can question how far parallel acquisition can be taken to reduce acquisition time. The practical limit depends on the SNR of the resultant image. It has been shown that SNR is lost in parallel-accelerated images in inverse proportion to the product of R and g, where R is the acceleration factor and g is a measure of the noise amplification dependent on the similarity of spatial responses of the individual coil elements comprising the coil array [14]. This latter “g-factor” is a spatially varying function which takes on values at or larger than its ideal value of unity.

79

Temporal Footprint Analysis With view-shared sequences, there is correlation from one image to the next because of the commonality of at least some of the data. That is to say, the temporal resolution of such a sequence is not exactly as small as the time between consecutive images in the time series. Expressed another way, two acquisition techniques which may have the same frame time might not have the same temporal resolution in distinguishing two phenomena. To attempt to elucidate this, the parameter of “temporal footprint” has been proposed [30]. The temporal footprint is defined as the time interval over which the k-space data that contributes to the image is acquired. If views are not shared from one image to the next in a series, then the temporal footprint is simply equal to the acquisition time for each image. For the case of CAPR, the temporal footprint can be seen from the data sorting plot of Fig. 5.2. For Image 1 (Fig. 5.1b), this is the time span from the start of acquisition of the first vane set (black) to the end of the last vane set (red). Note that this is larger than the frame time, the time between consecutive images in the series. Figure 5.3 is a plot of the temporal footprint vs. the frame time for a 3D image of given spatial resolution. The outermost curve is for the case of no parallel acquisition (R = 1), and the rightmost point of the curve is for the case of no view sharing. For this case, the footprint matches the frame time, as expected. Suppose next that additional measurements of the phase-encoding views at and near central k-space are somehow inserted into this process, as shown previously in Fig. 5.1f. This specific case of the use of four vane sets is referred, here, as an “N4” acquisition, and this operating point is identified on the outermost curve. In general, decomposition of the sampling pattern of Fig. 5.1e can be done into N vane sets, leading to the corresponding operating point on the curve in Fig. 5.3 as shown. In general, as N increases, the level of view sharing increases in that the fraction of data maintained from one image to the next increases. Although this leads to a reduction in the frame time and a higher frame rate, it is accompanied by an increase in the temporal footprint. At some point, an extended temporal footprint can lead to an undesirable level of artifact. Consider next the effect of parallel acquisition. An acceleration of R allows the acquisition time to be reduced by R for fixed spatial resolution. In Fig. 5.3, this causes the temporal footprint vs. frame time curve to be shifted down and left toward the origin, as seen for the curves for R = 2, 4, and 8. High accelerations allow a radical shift of the original curve, allowing performance with markedly smaller frame time and smaller temporal footprint compared to that of the R = 1 reference. The innermost curve of Fig. 5.3 is for R = 8, an acceleration which has been shown to be feasible for time-resolved CE-MRA of the peripheral vasculature [31]. At the operating

80

S.J. Riederer et al.

Fig. 5.3 Diagram of temporal footprint vs. frame time. Each point along a given curve corresponds to a specific degree of view sharing. As this degree increases, the frame time is reduced, but the temporal footprint gets longer. Each curve corresponds to a different factor R of acceleration due to some version of parallel acquisition, such as with SENSE

point of N4, a frame time of 5 s is allowed using a temporal footprint of less than 20 s. Both of these times are markedly smaller than the acquisition time of 105 s required for a single-phase image with no parallel acquisition.

Receiver Coils for Parallel-Accelerated, Time-Resolved CE-MRA The typically coronal imaging orientation of CE-MRA readily lends itself to the design of effective coil arrays for parallel acquisition. In the coronal format, the frequency encoding or readout direction is typically chosen to be along the superior/inferior (S/I) direction. This means that the two phaseencoding directions are within the transverse plane, generally left/right (L/R) for phase encoding and anterior/posterior (A/P) for slice encoding. Coil elements can then be formed into an array such that the central axis of each element lies within the transverse plane. This is important because it is along this axis that the spatial sensitivity is greatly variable and can be controlled, thus allowing some control of the g-factor governing noise amplification. The design and optimization of coils for parallel acquisition of the calves are illustrated in Fig. 5.4. Figure 5.4a shows a maximum-intensity-projection (MIP) image of the calves of volunteer acquired with 2D SENSE-accelerated (R = 7.3) CAPR CE-MRA acquisition with a frame time of 5.0 s and temporal footprint of 17.7 s. Spatial resolution is 1-mm isotropic. An eight-element receiver coil array was used, placed circumferentially around the subject’s calves. Although the image quality is considered to be very good, there is still some falloff of signal at the superior and inferior extents of the 40-cm FOV (arrows), there is slight graininess in the result, and although not evident from the figure there was a considerable gap between the subject’s legs and the coil.

Fig. 5.4 Example of effective coil design for 2D-SENSE-accelerated CE-MRA. (a) MIP of calves made with original eight-element receiver coil. Arrows identify falloff at superior and inferior regions. (b) MIP of calves of a different volunteer acquired using modified coil. Note improved S/I coverage and improved SNR vs. (a). (c) Box plots of g-factors for original coil (blue) and modified coil (yellow) several different 2D accelerations (panels A and B reproduced from Haider et al. [31] with permission; panel C reproduced from Johnson et al. [32] with permission)

5

Time-Resolved, Contrast-Enhanced MR Angiography Using Cartesian Methods

81

This was addressed by redesigning the coil array. Specifically, the coil elements were increased from 21.4 to 27.1 cm in length and the two anterior and two posterior elements were all reduced in width from 14.4 to 10.5 cm. This reduced the net circumference to allow a tighter fit of the coil array to the anatomy and reduced the noise amplification due to acceleration in the A/P (z) direction. A result made using the modified coil with the same acquisition but a different subject is shown in Fig. 5.4b. Image quality is markedly improved. Figure 5.4c is a comparison of the range of g-factors occurring over the volume of the calves for four different 2D SENSE acceleration factors for the original coil array (blue plots) and the modified array (yellow). Note the desired reduction in g-values allowed with the new coil. Also note that even for R = 4 × 2 = 8 the 75th percentile g-factor value is still less than 1.4. Similar coil arrays can be developed to facilitate highly 2D-accelerated CE-MRA for other anatomic regions, such as brain, arms and hands, thorax, abdomen, thighs, and feet.

The Signal Enhancement Effect in Accelerated Contrast-Enhanced MRA As stated above, the application of parallel acquisition to MRI extracts a price in lost SNR. However, for the case of CE-MRA, the SNR loss is not as severe as might initially be expected. The above-stated relationship linking SNR loss to the acceleration R and the g-factor assumes that the magnetization level does not change during the acquisition. However, for CE-MRA, the signal of interest due to the contrast bolus is time dependent or transient, and this tempers the degree of SNR loss. This effect has been studied in detail in Riederer et al. [16], but the essence of the result is illustrated in Fig. 5.5. Figure 5.5a shows the hypothetical signal at a point in a blood vessel due to the time-dependent passage of the contrast-enhanced blood. The behavior is modeled here with a gamma variate function, but the exact mathematical dependence is not important other than that the signal rapidly comes to a peak and then wanes over time with a time constant on the order of several tens of seconds. Next, assume that data are acquired for a CE MR angiogram using a 3DFT acquisition with the elliptical centric view order. Furthermore, assume that the acquisition is triggered to start at peak contrast, corresponding to t = 0 in (A). This acquisition matches the time-dependent signal to kY–kZ space repetition by repetition. However, because the k-space values used are ordered according to their k-space radius, this matching can simply be reflected as a scaling of the bolus curve with kr, shown in Fig. 5.5b for the unaccelerated R = 1 case. Here, kr is defined as the distance from a point in the kY–kZ plane to the origin. Next, consider the case of accelerated acquisition with an assumed R = 4. As shown previously in comparing Fig. 5.1e–g,

Fig. 5.5 Graphic illustration of signal enhancement effect in 2D-accelerated CE-MRA. (a) Schematic plot of contrast bolus passage through a region within an artery. CE-MRA acquisition using centric phase-encoding order is assumed to start at t = 0. (b) Plots of the signal of (a) as mapped to kY–kZ space for the cases of no acceleration (R = 1) and assumed acceleration R = 4. For the latter, note that the area under the curve is larger and the shoulder is broader (reproduced from Riederer et al. [16] with permission)

accelerated acquisition causes increased gaps in k-space between the sampled points. In mapping the bolus curve of Fig. 5.5a to k-space for the accelerated case, the regions of peak signal are pushed out to larger kr values because of this increased sampling distance between points. The accelerated case is shown for R = 4 in Fig. 5.5b. The difference between the R = 1 and R = 4 curves in Fig. 5.5b is important in two ways. First, the area under each curve is proportional to the signal level in the final image. Clearly, that for the R = 4 case is larger, leading to an increased relative signal vs. the unaccelerated case. This effect compensates in part for the R- and g-factor-related SNR loss. A second important difference is in the shape of the curves in Fig. 5.5b. The wider shoulder of the R = 4 case leads to a narrower point spread function in y–z space, indicating sharper spatial resolution for an equal extent of k-space sampling. Another way to explain the signal enhancement effect in CE-MRA is that acceleration causes the high-contrast, arterial-phase signal to be assigned over a broader region of k-space. This then leads to an overall signal enhancement. In Riederer et al. [16], the signal enhancement effect is studied in more mathematical and experimental detail.

82

Specifically, the enhancement effect is demonstrated experimentally as being about 20% in R = 4 SENSE-accelerated images of the intracranial venous system for a CE-MRA study using intravenously administered contrast material.

Temporal Fidelity in Contrast-Enhanced MRA Although view sharing allows a reduction in the frame time and parallel acquisition allows a reduction in the overall acquisition time, neither of these methods has driven the absolute acquisition time of 3D CE-MRA down to the subsecond level. Thus, it is important to determine how accurately an image reconstructed for a specific time point using some kind of 3D CE-MRA method represents the actual signal in the contrast-enhanced vasculature at the time point. A number of methods have attempted to do this by simulation; e.g., 33. Starting with some assumed temporal enhancement pattern, such as the gamma variate of Fig. 5.5a, the signal in a blood vessel of some assumed size is simulated and then matched to k-space using the assumed CE-MRA acquisition technique. Fourier transformation of the simulated, sampled k-space gives the final image. In addition to simulation, it is possible to study such temporal fidelity effects experimentally, as in Mostardi et al. [11]. In this work, a phantom was constructed (Fig. 5.6a) in which rods filled with diluted contrast material were translated under computer control through the FOV of the MRI system. The abrupt leading edge of the phantom allows measurement of any temporal blurring or artifact which might precede the advancing edge in a CE-MRA image. Sample results from one experiment are shown in Fig. 5.6b–e. In all cases, the frequency-encoding direction is along the direction of motion, left to right in the results shown. Figure 5.6b is an enlargement of the leading edge of the phantom showing a small level of blur, defined as the width between the 25 and 75% levels of full signal. This was measured as 10 mm or about 10% of the 104-mm distance moved from one time frame to the next. The assumed acquisition was the CAPR method of Fig. 5.1e using no acceleration; i.e., R = 1. Figure 5.6c shows results using a random view ordering in the kY–kZ plane. This order has properties akin to projection reconstruction or radial acquisition in which central k-space is sampled throughout the frame time. All other aspects of the acquisition were identical to that used for (B). As shown, this causes the blur to be 60 mm, considerably larger than in (B). Figure 5.6d, e shows results for the CAPR and random-order acquisitions using 2D SENSE acceleration of R = 7.3. The resultant reduced frame time causes the frame-to-frame advance of the phantom to be reduced from 104 to 22 mm. The accelerated CAPR acquisition shows a reduction in blurring from 10 to 7 mm, still well focusing the leading edge of the phantom. For the random

S.J. Riederer et al.

ordering (E), the blurring is also reduced as expected, but the blur width is still more than half of the distance traveled per update. The above example demonstrates that for a given sampling pattern and view order, the sharpness in portraying the advancing contrast bolus improves as the temporal footprint is reduced using k-space undersampling acceleration techniques. The results also prove that the view order itself can have a major impact on controlling the sharpness of the image. Even with the reduced temporal footprint allowed by R = 7.3 acceleration, the random view ordering of (E) had inferior sharpness to the elliptical centric ordering pattern of (B) for the unaccelerated case.

Fig. 5.6 Experimental phantom studies for assessment of CAPR performance. (a) Schematic of computer-controlled phantom. (b–e) Comparisons of elliptical centric (EC) vs. random phase-encode ordering without (R = 1) and with (R = 7.3) acceleration. Enlargements of leading edge of moving phantom acquired with (b) EC, R = 1; (c) random, R = 1; (d) EC, R = 7.3; (e) random, R = 7.3. Note that EC provides sharper edge depiction than random, and acceleration provides sharper edge depiction than no acceleration (reproduced from Mostardi et al. [11] with permission)

5

Time-Resolved, Contrast-Enhanced MR Angiography Using Cartesian Methods

Results Accelerated, time-resolved, 3D Cartesian CE-MRA has been used in imaging many vascular regions. Here, we present representative results using the CAPR technique.

MRI Data Acquisition 3D CAPR examinations were performed on a number of subjects using a protocol approved by the Institutional Review Board of our institution. For all volunteers, written consent was provided. The examinations were performed using the CAPR sampling pattern discussed previously. A fast gradient echo pulse sequence was used, typically with the following parameters: repetition time (TR) 5.85 ms; echo time (TE) 2.7 ms; flip angle 30°; bandwidth ±62.5 kHz; and a full echo comprising 400 points. SENSE calibration was performed using a similar sequence, except with a flip angle of 10°; bandwidth of ±31.25 kHz; and a fourfold reduction in the Y × Z spatial resolution. Because of the time-resolved nature of the acquisition, no timing bolus is necessary. However, to obtain an accurate subtraction, a fully sampled CAPR image was acquired prior to injection of contrast material. The general contrast administration technique used was injection of 20 mL of gadobenate dimeglumine contrast agent (Multihance; Bracco Diagnostics, Princeton NJ) administered at a rate of 3 mL/s into the antecubital vein using a power injector (Spectris; Medrad; Indianola PA). This was followed by a saline flush of 20 mL administered at 3 mL/s. Automated reconstructions of all images were performed using a custom system interfaced to the MR imager. The time delay from end of acquisition to image display for a CAPR sequence comprising over 30 time frames was typically no longer than two minutes, thus allowing immediate clinical review.

In Vivo Results Figure 5.7 shows results in the calves from three normal subjects and illustrates the intrinsic spatial and temporal resolution of the method. Common temporal enhancement patterns are shown as follows: simultaneous bilateral filling of the arterial vasculature of both calves (A1–A4); early filling of one side (here, the right) vs. the other (B1–B4); and fast arterial transit with fast arterial-to-venous circulation (C1–C4). In B2 and B3, note that venous return is already starting in the proximal region (arrows), but owing to the short 5-s frame time and the high temporal fidelity, this can readily be distinguished from the arterial signal in the same region in (B1). In C1–C4, note the rapid transit of the contrast bolus

83

along nearly the full superior–inferior extent of the FOV in going from C1 to C2. Also, note the prominent venous signals in some structures only one frame (5 s) after initial filling (arrows, C3). The high precision in the temporal and spatial resolution of the CAPR acquisition is illustrated in the study shown in Fig. 5.8. The spatial resolution readily supports magnification and projection at arbitrary angles. The temporal resolution allows clear determination of the sources of the blood for the major vessels. Figure 5.9 shows results from another study of the calves, this from a subject who was first imaged using CT angiography because of suspicion of peripheral vascular disease. This study illustrates how CTA can be limited by calcification interfering with assessment of the vessel lumen. However, CE-MRA does not have this limitation. Figure 5.10 illustrates the application of CAPR to CE-MRA of the hands. In this case, the sequence was modified to provide a somewhat shorter update time (3 vs. 5 s) and temporal footprint (12 vs. 17 s) compared to those used for the calves. Figure 10.5 shows 3 full coronal MIP images calculated 30.0, 33.0, and 57.0 s, after the injection of contrast agent, respectively. Also shown are time-of-arrival (TOA) maps [34] of the left hand with the displayed values approximately limited to the arterial phase (D), and with all arrival times displayed (E–F) at several projection angles. CAPR has also been used to image the vasculature of the brain. In this case, the CAPR acquisition parameters are generally modified to allow for a somewhat faster frame time to account for the reduced arterial-to-venous transit time. The study of Fig. 5.11 was acquired using a 2.3-s frame time and shows the early arterial phase (A) formed 21.3 s post injection, and late arterial (B, 23.6 s), early venous (C, 25.9 s), and subsequent venous (D, 28.2 s) phases. A general, easily assessable measure of the temporal fidelity of a time-resolved MR angiographic sequence is its ability to generate an image of the intracranial arterial system devoid of venous enhancement. Figure 5.12a–c all clearly demonstrates this. Signals in the superior sagittal sinus are not apparent until the next frame 2.3 s later (arrows, D). Finally, Fig. 5.12 illustrates the recent application of accelerated CAPR acquisition to multistation CE-MRA of the lower extremity [35]. In this example, the acquisition was initiated at the level of the thighs, with image reconstruction performed in real time. That is, a coronal MIP of a 3D image was reconstructed and presented to the observer approximately 900 ms after completion of data acquisition for that image. The observer monitored arrival and transit of contrast material, and when it reached the distal extent of the thigh FOV he/she triggered table motion by mouse click. After an approximate 6-s delay to allow table motion, the CAPR acquisition was initiated at a calf station. Separate eightelement coil arrays were used at each station. Figure 5.12a

84

S.J. Riederer et al.

Fig. 5.7 CAPR results from volunteer studies for three common flow patterns. (a1–a4) Symmetric blood flow in left and right calves. (b1–b4) Asymmetric left vs. right filling and early venous return (arrows). (c1–c4) Rapid superior-toinferior contrast transit and rapid arterial-to-venous circulation (reproduced from Haider et al. [31] with permission)

shows an MIP of the full FOV while Fig. 5.12b–e shows an individual region of the distal station identified by the dashed box of (A) at consecutive 5.1-s time intervals.

Discussion In this chapter, we have attempted to describe some of the properties and applications of Cartesian-based, time-resolved 3D CE-MRA as implemented using contemporary techniques. Compared to the practice of 3D time-resolved MRA techniques introduced over 10 years ago, e.g., the TRICKS method [10], there are several notable advances. Perhaps, foremost among these is the reduction in acquisition time allowed using acceleration techniques, particularly 2D acceleration techniques allowed by the two phase-encoding directions of 3D acquisition. 2D sensitivity encoding (SENSE) [15] allows accelerations of 4× to 8× or more, and when coupled with the typical 1.8× acceleration

allowed by 2D homodyne the net acceleration factor is well over an order of magnitude. As shown in Fig. 5.3, this radically reduces the frame time and temporal footprint of timeresolved sequences. In effect, this allows time-resolved imaging at higher spatial resolution than what was possible previously with single-phase methods. A critical technical element contributing to the high quality of accelerated 3D CE-MRA is the receiver coil array. It is important to understand that simply making the element count higher does not in and of itself provide improved image quality. Rather, the element size should be matched to the size of the object, and the spatial orientation should be matched to the direction(s) of any parallel acquisition. For the case of 2D SENSE applied to the transverse plane, which is typically the case in CE-MRA, coil arrays which can be placed circumferentially around the subject provide low g-factors and retain SNR at high acceleration factors. The improvement in speed of acquisition means that image sequences having superior temporal and spatial

Fig. 5.8 Illustration of the high spatial and temporal resolution of CAPR. CAPR results of the calves in a subject with suspected peripheral arterial disease (PAD). (a–c) are coronal MIPs from consecutive frames acquired at 50, 55, and 60 s post injection, respectively. The left anterior tibial artery (ATA) is well-seen (arrow, a), but the other two major vessels in the left calf appear to be occluded. (d–g) MIP of the subvolume identified in the dashed box of (a) projected at 0, 30, 60, and

90° from the A/P direction and corresponding to 50, 55, 60, and 65 s post injection, respectively. These subvolumes clearly show how the left posterior tibial artery (e, arrow) is perfused by a spontaneously recruited geniculate artery (e, arrowhead). Also, the peroneal artery (f, arrow) is fed by an auxiliary artery (f, arrowhead) originating from the left ATA

Fig. 5.9 Comparison of 3D CE-MRA acquired with CAPR to CT angiography (CTA). (a) Subvolume of the circulation of the right calf taken from the CTA showing a stenosis at the origin of right anterior tibial artery (arrow). (b, c) Consecutive frames obtained 5 s apart of the same anatomic region as (a) imaged with the CAPR MRA. Both images clearly show the stenosis (arrow, b) and also portray the progression of

contrast material through the vasculature. (d) Targeted reconstruction of the left leg from the CTA, also portraying a stenosis at the origin of the anterior tibial artery (arrow). (e) Comparison MRA, showing the lesion at the origin (arrow) but in addition a distal region suspicious of narrowing (arrowhead) difficult to appreciate in the CTA exam because of overlying calcium (reproduced from Haider et al. [31] with permission)

86

S.J. Riederer et al.

Fig. 5.10 CAPR CE-MRA of the hands. (a–c) Coronal full field-of-view MIPs of the hands acquired 30.0, 33.0, and 57.0 s post injection. (d) Dorsal time-of-arrival (TOA) map of the left hand with TOA values restricted to the arterial phase. (e–g) Maps of all TOA values of the left hand at progressive oblique angles (reproduced from Riederer et al. [34] with permission)

stricter. For example, the image frame rate of some technique should not be equated with the temporal resolution. It has been shown that the time over which data are generated for image formation affects the portrayal of a time-varying phenomenon [30]. Similarly, if it is desirable to cause a reconstructed image to best freeze the status of the object at the time of the image reconstruction, then the central k-space information should be measured in only one, temporally compact time interval. Accurate portrayal of velocity of an object over a sequence of frames requires that the distribution of k-space sampling be consistently applied during the temporal footprint for each frame of the sequence. Finally, incorporation into an image of appreciable high spatial frequencies measured well after central k-space can lead to “anticipation” artifact in which the vasculature distal to the bolus leading edge position at the time of reconstruction can erroneously appear to be enhanced. Such signals can potentially be reduced with iterative methods, e.g., [36], but Cartesian acquisition with appropriate view ordering can still effectively limit the extent of such artifact. Fig. 5.11 CAPR CE-MRA of the neurovasculature. Full field-of-view MIPs formed at an oblique angle of the brain acquired (a) 21.3 s, (b) 23.6 s, (c) 25.9 s, and (d) 28.2 s post injection. Note that in spite of the relatively long (6.9 s) temporal footprint the signal from the superior sagittal sinus (arrows, d) is not seen in the early frames

resolution are now possible compared to their previous counterparts. However, for adequate assessment of performance, the criteria for evaluation of performance have now become

Conclusion Time-resolved 3D CE-MRA based on Cartesian k-space sampling can now be performed using acceleration techniques, view ordering, and receiver coil arrays which allow high-quality imaging in multiple vascular territories.

5

Time-Resolved, Contrast-Enhanced MR Angiography Using Cartesian Methods

87

Fig. 5.12 CAPR CE-MRA of the lower extremity using a time-resolved multistation technique. (a) Coronal MIP of the full longitudinal FOV, encompassing the thighs and calves. (b–e) Coronal MIPs of the subvolume identified in (a) formed 46.6 s (b), 51.7 s (c), 56.7 s (d), and 61.8 s (e) post injection (reproduced from Johnson et al. [35] with permission)

Acknowledgments The authors acknowledge the assistance of Eric Borisch, Kathy Brown, Norbert Campeau MD, James Glockner MD PhD, Roger Grimm, Thomas Hulshizer, John Huston MD, Thanila Macedo MD, Phillip Rossman, Diane Sauter, Terri Vrtiska MD, and Phillip Young MD.

References 1. Prince MR, Yucel EK, Kaufman JA, Harrison DC, Geller SC. Dynamic gadolinium-enhanced 3D abdominal MR arteriography. J Magn Reson Imaging. 1993;3:877–881. 2. Prince MR. Gadolinium-enhanced MR aortography. Radiology. 1994;191:155–164. 3. Earls JP, Rofsky NM, DeCorato DR, Krinsky GA, Weinreb JC. Hepatic arterial-phase dynamic gadolinium-enhanced MR imaging: optimization with a test examination and a power injector. Radiology. 1997;202:268–273. 4. Foo TKF, Saranathan M, Prince MR, Chenevert TL. Automated detection of bolus arrival and initiation of data acquisition in fast, three-dimensional, gadoliuium-enhanced MR angiography. Radiology. 1997;203:275–280. 5. Wilman AH, Riederer SJ, King BF, Debbins JP, Rossman PJ, Ehman RL. Fluoroscopically-triggered contrast-enhanced threedimensional MR angiography with elliptical centric view order: application to the renal arteries. Radiology. 1997;205:137–146. 6. Wilman AH, Riederer SJ. Improved centric phase encoding orders for three dimensional magnetization prepared MR angiography. Magn Reson Med. 1996;36:384–392. 7. Wang Y, Johnston DL, Breen JF, et al. Dynamic MR digital subtraction angiography using contrast enhancement, fast data acquisition, and complex subtraction. Magn Reson Med. 1996;36:551–556. 8. Riederer SJ, Tasciyan T, Farzaneh F, Lee JN, Wright RC, Herfkens RJ. MR fluoroscopy: technical feasibility. Magn Reson Med. 1988; 8:1–15. 9. van Vaals JJ, Brummer ME, Dixon WT, et al. “Keyhole” method for accelerating imaging of contrast agent uptake. J Magn Reson Imaging. 1993;3:671–675. 10. Korosec FR, Frayne R, Grist TM, Mistretta CA. Time-resolved contrast-enhanced 3D MR angiography. Magn Reson Med. 1996;36: 345–351.

11. Mostardi PM, Haider CR, Rossman PJ, Borisch EA, Riederer SJ. Controlled experimental study depicting moving objects in viewshared time-resolved MRA. Magn Reson Med. 2009;62:85–95. 12. Sodickson DK, Manning WJ. Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med. 1997;38:591–603. 13. Griswold MA, Jakob PM, Heidemann RM, et al. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med. 2002;47:1202–1210. 14. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999;42: 952–962. 15. Weiger M, Pruessmann KP, Boesiger P. 2D SENSE for faster 3D MRI. Magma. 2002;14:10–19. 16. Riederer SJ, Hu HH, Kruger DG, Haider CR, Campeau NG, Huston III J. Intrinsic signal amplification in the application of 2D SENSE parallel imaging to 3D contrast-enhanced elliptical centric MRA and MRV. Magn Reson Med. 2007; 58:855–864. 17. Bernstein MA, Fain SB, Riederer SJ. Effect of zero-filling and windowing of MRI data on spatial resolution and acquisition strategy. J Magn Reson Imaging. 2001;14:270–280. 18. Willinek WA, Gieseke J, Conrad R, et al. Randomly segmented central k-space ordering in high-spatial-resolution contrastenhanced MR angiography of the supraaortic arteries: initial experience. Radiology. 2002;225:583–588. 19. Noll DC, Nishimura DG, Macovski A. Homodyne detection in magnetic resonance imaging. IEEE Trans Med Imaging. 1991;10: 154–163. 20. Madhuranthakam AJ, Hu HH, Barger AV, et al. Undersampled elliptical centric view-order for improved spatial resolution in contrastenhanced MR angiography. Magn Reson Med. 2006;55:50–58. 21. Fink C, Ley S, Kroeker R, Requardt M, Kauczor H-U, Bock M. Time-resolved contrast-enhanced three-dimensional magnetic resonance angiography of the chest. Invest Radiol. 2005;40:40–48. 22. Cashen TA, Carr JC, Shin W, et al. Intracranial time-resolved contrast-enhanced MR angiography at 3T. Am J Neuroradiol. 2006; 27:822–829. 23. Lim RP, Shapiro M, Wang EY, et al. 3D time-resolved MR angiography (MRA) of the carotid arteries with time-resolved imaging with stochastic trajectories: comparison with 3D contrast-enhanced bolus-chase MRA and 3D time-of-flight MRA. Am J Neuroradiol. 2008;29:1847–1854.

88 24. Lim RP, Jacob JS, Hecht EM, et al. Time-resolved lower extremity MRA with temporal interpolation and stochastic spiral trajectories: preliminary clinical experience. J Magn Reson Imaging. 2010;31: 663–672. 25. Hadizadeh DR, Falkenhausen Mv, Gieseke J, et al. Cerebral arteriovenous malformation: Spetzler-Martin classification at subsecondtemporal-resolution four-dimensional MR angiography compared with that of DSA. Radiology. 2008;246:205–213. 26. Ruhl KM, Katoh M, Langer S, et al. Time-resolved 3D MR angiography of the foot at 3 T in patients with peripheral arterial disease. AJR Am J Roentgenol. 2008;190:W360-W364. 27. Frydrychowicz A, Bley TA, Winterer JT, et al. Accelerated timeresolved 3D contrast-enhanced MR angiography at 3T: clinical experience in 31 patients. Magma. 2006;19:187–195. 28. Weiger M, Pruessmann KP, Kassner A, et al. Contrastenhanced 3D MRA using SENSE. J Magn Reson Imaging. 2000;12:671–677. 29. Sodickson DK, McKenzie CA, Li W, Wolff S, Manning WJ, Edelman RR. Contrast-enhanced 3D MR angiography with simultaneous acquisition of spatial harmonics: a pilot study. Radiology. 2000;217:284–289.

S.J. Riederer et al. 30. Haider CR, Hu HH, Campeau NG, Huston III J, Riederer SJ. 3D high temporal and spatial resolution contrast-enhanced MR angiography of the whole brain. Magn Reson Med. 2008;60:749–760. 31. Haider CR, Glockner JF, Stanson AW, Riederer SJ. Peripheral vasculature: high-temporal and high-spatial -resolution three-dimensional contrast-enhanced MR angiography. Radiology. 2009;253:831–843. 32. Johnson CP, Haider CR, Rossman PJ, Hulshizer TC, Borisch EA, Riederer SJ. Coil design for highly accelerated 2D SENSE MRA of the lower legs. In: 16th Meeting, ISMRM. Toronto, 2008;1079. 33. Cashen TA, Jeong H, Shah MK, et al. 4D radial contrast-enhanced MR angiography with sliding subtraction. Magn Reson Med. 2007;58:962–972. 34. Riederer SJ, Haider CR, Borisch EA. Time-of-arrival mapping at three-dimensional time-resolved contrast-enhanced MR angiography. Radiology. 2009;253:532–542. 35. Johnson CP, Haider CR, Borisch EA, Glockner JF, Riederer SJ. Time-resolved bolus-chase MR angiography with real-time triggering of table motion. Magn Reson Med. 2010;64:629–637. 36. Mistretta CA, Wieben O, Velikina J, Block W, Perry J, Wu Y. Highly constrained backprojection for time-resolved MRI. Magn Reson Med. 2006;55:30–40.

6

Flow-Dependent Noncontrast MR Angiography Mitsue Miyazaki, Satoshi Sugiura, Yoshimori Kassai, Hitoshi Kanazawa, Robert Edelman, and Ioannis Koktzoglou

Introduction Noncontrast MR angiography (NC-MRA) techniques such as time-of-flight (TOF) and phase contrast (PC) have been available since the early days of MRI development. However, long scan times and various artifacts have limited their use in the clinical setting over contrast-enhanced (CE) MRA techniques, introduced first around the mid-1990s. At the present time, three-dimensional (3D) TOF remains the main technique for intracranial MRA, while PC has been used as an angiogram technique as well as flow functional technique in various regions of the body. In other regions of the body like the chest, abdomen, and peripheral regions, CE MRA became the main MR technique in those areas. Rapid development of CE MRA was facilitated by various developments in system hardware, parallel imaging, sequence improvements, and the introduction of time-resolved MRA. Since the recent FDA black-box warning associated with the use of gadoliniumbase contrast agent in 2007 and the risk of developing nephrogenic systemic fibrosis (NSF), there is renewed interest in NC-MRA as an alternative to CE MRA. In this chapter, we focus on the following NC-MRA techniques, flow-dependent angiography techniques: TOF, PC, fresh blood imaging (FBI) using electrocardiographically (ECG) or peripheral pulse gating (PPG)-triggered half-Fourier fast spin-echo (FSE), and an arterial spin labeling (ASL) technique, time-spatial labeling inversion pulse (time-SLIP) combined using ECG- or PPG-triggered half-Fourier FSE and bright blood balanced steady-state free precession techniques. M. Miyazaki (*) MRI Department, Toshiba Medical Research Institute, Vernon Hills, IL, USA S. Sugiura • Y. Kassai • H. Kanazawa MRI Systems Development Department, Toshiba Medical Systems Corporation, Otawara, Tochigi, Japan R. Edelman • I. Koktzglou Northshore University Health System Evanston, IL, USA

Flow-dependent MRA techniques can be classified into (1) TOF, based on flow-related enhancement, (2) PC using the phase difference images, (3) triggered FBI using the intrinsic bright blood of FSE, and (4) an ASL, time-SLIP, and balanced steady-state free precession (bSSFP). A brief explanation of TOF and PC is provided, but we emphasize FBI and time-SLIP techniques with brief inclusion of the bSSFP approaches including various clinical applications.

Time-of-Flight Time-of-flight (TOF) utilizes flow-related enhancement which is an inflow effect whereby blood flowing in vessels perpendicularly the slice plane or slab appears bright. The technique relies upon the signal difference between the inflow blood and the stationary background signals to define the lumen of the blood vessels. The MRI signal arising from stationary tissue is saturated by the application of repeated RF excitation pulses and a relatively short repetition time (TR). Blood flowing into the imaging plane experience fewer RF excitations pulses, appear brighter that static tissue. For intracranial MRA, a 3D gradient echo (GE) or field echo (FE) pulse sequence is used for image acquisition and the data is processed into maximum intensity projections (MIPs). Figure 6.1 shows good contrast between the stationary background and the vessels in the intracranial MIP images obtained using 3D TOF at 3 T. As compared to 1.5 T, 3 T provides a higher signal-to-noise ratio (SNR), but also benefits from longer T1. The longer T1 results in more effective saturation of the signal from stationary background tissue. Originally developed at 1.5 T to enhance inflow effects, TOF technical advancements include tilted optimized nonsaturated excitation (TONE) [1], which progressively increase flip angles through the slab to lessen the effects of blood saturation within the slab and multiple overlapping thin slab acquisition (MOTSA) [2], where each thin slab benefits from the inflow of fresh unsaturated blood. Magnetization transfer (MT) pulses can be applied to improve

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_6, © Springer Science+Business Media, LLC 2012

89

90

M. Miyazaki et al.

Fig. 6.2 (a) 3D time-of-flight (TOF) sequence diagram using a slice selective off-resonance since (SORS) magnetization transfer (MT) pulse. Note that the slice selective gradient (A) is applied along with the SORS MT pulse. (b) 3D TOF images acquired with two types of MT pulses performed on a healthy volunteer with a metal clip placed on the right side of his neck at 1.5 T. Note the loss of vessel signals on the right hemisphere using a nonselective MT pulse, whereas the application of the SORS MT pulse maintains blood signal in the right hemisphere

Fig. 6.1 Intracranial 3D time-of-flight performed on a healthy male volunteer at 3 T. Maximum intensity projection in the axial (left) and oblique sagittal directions (right) demonstrates excellent blood to background contrast and depiction of small peripheral intracranial vessels

suppress the brain and improves the blood to background contrast in intracranial MRA [3]. The MT pulse excites short T2 restricted water protons, which are bound to macromolecules, and through their chemical exchange, signals of the free unbound water protons are reduced. There is a notable amount of MT effects in blood as well. In order to additionally improve the contrast between the blood and the brain background signals, a slice selective off-resonance sinc (SORS) pulse was developed as an MT pulse [3]. Figure 6.2a shows a pulse sequence diagram of 3D TOF with a slice

selective gradient A as indicated as the SORS pulse. Figure 6.2b shows 3D MRA images with and without the slice selective MT pulse on a volunteer with a metal clip placed on the right side of his neck at 1.5 T. Note the loss of arteries in the right hemisphere in the 3D TOF MIP image observed using the nonselective MT pulse. With the application of the slice selective MT pulse, SORS, the right intracranial vessels are visualized and less affected by the metal clip. The nonselective MT pulse applied across the entire anatomy, including the blood vessels near the metal clip, results in loss of blood signals in the right hemisphere; whereas the SORS pulse located at the top of the brain has less affect on the blood traveling from the heart. The effect of the SORS pulse is schematically explained in Fig. 6.3. The SORS pulse placed on the top of the head provides a large MT effect to the brain parenchyma because it is closer to the center frequency. With the nonselective MT pulse, blood traveling from the heart experiences less MT effect as blood entering into the coil sensitivity has a long distance to travel into the center of the imaging slab. The farther the travel distance of blood to the center of the imaging slab, the smaller the MT effect in the blood. Results that show higher blood to background contrast can be obtained using the SORS pulse rather than applying a spatial nonselective pulse. In peripheral MRA, 2D TOF sequences using a presaturation pulse to saturate venous signals are utilized to depict arterial vessels. However, peripheral run-off imaging requires a large coverage, which leads to long acquisition times. In addition, the 2D TOF technique is limited in depicting the vessels that are perpendicular to the orientation of the vessels. Vessels that are oriented through the slice plane have less inflow effect and lose signal, making it difficult to image slow tortuous peripheral vessels. In order to improve the

6

Flow-Dependent Noncontrast MR Angiography

91

Fig. 6.3 (a) Schematic of 3D time-of-flight scan using an SORS MT pulse. The yellow box shows the slice selective MT (SORS) pulse and the red box shows the imaging plane. The distance between the SORS MT pulse and imaging slab can be translated to the frequency. (b) MT effect in frequency domain. The curve shows the MT effect on frequency that the closer to center frequency, F0, the greater the MT effect.

Since the SORS MT pulse is slice selective, and positioned at the top of the head, it gives a constant frequency offset, providing a strong MT effect to the background signals, whereas the blood traveling from the heart experiences various frequency offsets on its way to the center of the imaging slab, which give less MT effect to the blood signals in average (Modified from Miyazaki et al. [3], with permission)

depiction of slow flow peripheral vessels, especially in patients with peripheral arterial diseases (PAD), ECG or PPG is applied to trigger at the peak systolic period, which lengthens scan time.

to straightforward flowing vessels. Figure 6.4 shows a sagittal view of a brain MRA obtained using the 3D PC technique and the PC cine images of aortic outflow track and its color map display. The PC cine images allow calculations of flow velocity distribution, maximum velocity, and flow quantity.

Phase Contrast Phase contrast (PC) technique utilizes the phase information of moving objects by applying pairs of bipolar (flowcompensated and uncompensated) gradient pulses to generate flow sensitive images [4]. The phase data is used to reconstruct and generate MR angiographic images and velocity-encoded flow quantification images. To produce PC MRA images with flow sensitivity in the x, y, and z directions, four acquisitions are required, one flow uncompensated and three (x, y, and z) with different flow compensation directions. Despite the long acquisition time, the strength of this technique beyond imaging the flowing blood is the quantitative measurements of blood flow that can be obtained with the same data. To obtain flow quantification data, velocity-encoded images are created by subtracting the phase shifts that are generated by the flow-encoded and flow-compensated acquisitions. The signal intensity in the velocity-encoded images contain phase shift ranges from +180° to −180°, where the positive phase shifts provide bright signals and the negative phase shifts produce dark signals. Unlike the TOF method, PC MRA required a priori knowledge of the blood flow velocity. Velocity information is used to set the value of the velocity-encoding (VENC) gradient which controls contrast of the flowing blood. Lengthy scan times and potential occurrence of artifacts from incorrect VENC settings limit this technique

Fresh Blood Imaging: ECG-Gated 3D PartialFourier Fast Spin-Echo Background Originally described by Wedeen et al. in the 1980s, the ECGgated spin-echo technique demonstrated pulsatile flow and presented 2D projection images as noncontrast-enhanced MRA in lower extremities [5, 6]. Improvements in both hardware and software have made it possible to extend this technique to 3D acquisitions and further shorten scan time by using a single-shot partial-Fourier FSE pulse sequence. The shortening of the echo train spacing (ETS) in 3D partialFourier FSE has made this technique clinically feasible on commercially available MRI scanners. Below we describe the distinct features and intrinsic properties that allow partial-Fourier FSE pulse sequences to be optimized for angiography. In FSE acquisitions, T2-mediated decay of the MR signal that results in blurring effect in the phase-encode (PE) direction is unavoidable. This results since FSE-based sequences sample multiple echoes at different echo times, unlike spin-echo sequences collecting an echo per line [7, 8]. Table 6.1 shows the T2 burring effect of blood simulated using a point spread function (PSF). The shorter the T2 value, one pixel signal spreads to a couple pixels, as indicated by the increasing of full-width at

92

M. Miyazaki et al.

Fig. 6.4 (a) Intracranial sagittal maximum intensity projection view image on a healthy volunteer using the 3D phase contrast (PC) technique. (b) PC cine images of the aortic outflow track (top) and the color map image (bottom) on a healthy volunteer

Table 6.1 Signal intensity in arbitrary units and full-width at halfmaximum (FWHM) simulated using a point spread function (PSF) with several T2 values (Modified from Miyazaki et al. [8] with permission) T2 200 ms (artery) 100 ms (vein) 50 ms (muscle)

Signal intensity (arbitrary units) 42.1 12.0 1.8

FWHM (pixels) 2.2 4.1 8.1

half-maximum (FWHM). The FWHM can be improved by increasing the number of shots and the application of parallel imaging to reduce T2 blurring. Increasing the data collection to two shots allows the first shot and the second shot to be interleaved while filling k space for each partition, which results in sharper images by reducing the data sampling period. However, the scan time will be twice as long as a single-shot method. The application of parallel imaging permits shorter data sampling periods, which results in reduced T2 blurring; however, it does not affect the scan time because of a fixed repetition time (TR).

Principles The NC-MRA technique, FBI, relies on ECG-gated 3D partial-Fourier FSE sequences, triggered for systolic and/or diastolic acquisitions. In order to obtain successful bright blood MRA images, essential concepts are discussed [9, 10]. The T2 relaxation time of blood causes T2 blurring effects in the phase-encoding direction on fast spin-echo type sequences [7]. In order to reduce T2 blurring where the PE direction is perpendicular to the vessel orientation, applying parallel imaging helps reduce the data sampling period, thus reducing the T2. Also, shortening of the ETS in partialFourier FSE reduces the total data sampling duration, which

reduces T2 blurring, motion-related artifacts, and minimizes susceptibility artifacts. Acquiring from near the center of the k space with a rectilinear filling permits intrinsically less flow-dephasing in the PE direction compared to the readout (RO) direction [11]. Additional prepulses like a short-tau inversion recovery (STIR) can be used in conjunction with the 3D partial-Fourier FSE technique to achieve fat suppressed images. FSE-based sequences allow the selection of effective TE (TEeff) to control the contrast. With 3D partial-Fourier FSE acquisitions, appropriate ECG delay(s) for diastolic and/or systolic triggering need to be selected for various applications. Figure 6.5a shows the relationship between the flow velocities of the ascending aorta, the descending aorta, and the superior vena cava (SVC) and the various ECG-triggering phases or delays measured using the PC technique, and Fig. 6.5b shows the corresponding signals intensities during each ECG delay obtained using partial-Fourier FSE. A preparatory “ECGprep scan” that produces 2D single-shot images at incremental triggering times of arbitrary steps can be used to find the specific trigger delay (TD) for systole (Fig. 6.6a), when arteries are black blood (flow voids), and for diastole when arteries are bright, as shown in Fig. 6.6b [9]. Each singleshot 2D acquisition is acquired in two or three R–R intervals at incremental delays to produce different phase images. In a peripheral run-off study, a second 2D preparatory scan provides the amount of spoiler gradient that provides the best selective dephasing of the arteries of interest [10]. In the spoiler preparation, each single-shot projection image is repeated by varying spoiler gradients. Presetting the amount of spoiler gradient accordingly to the regions is preferred to reduce total scan time. In general, the slower the vessel, the stronger the spoiler gradient strength is to differentiate slow flow arteries.

6

Flow-Dependent Noncontrast MR Angiography

Fig. 6.5 (a) Flow velocity measurement of the ascending aorta, the descending aorta, and the superior vena cava obtained throughout the cardiac cycle using the PC technique. (b) The signal intensities of the corresponding vessels using partial-Fourier FSE at effective TE (TEeff) of 60 ms on a healthy volunteer (Modified from Wedeen et al. [5] with permission)

Before moving to the 3D acquisition methods, it is noteworthy to mention the by-product of the 2D ECG-prep scan; subtraction of the systolic- from diastolic-triggered images provides noncontrast perfusion images. Figure 6.7 shows lung perfusion images obtained using ECG-gated partialFourier FSE on a healthy volunteer and a patient with a pulmonary thromboembolism. The signal difference of pulmonary vessels between diastole and systole provides noncontrast pulmonary perfusion MRA (Fig. 6.7a), and the hypointensity defect is due to the pulmonary thromboembolism, which matches with DSA image (Fig. 6.7b). This signal difference of systole and diastole was intensively investigated in both human and canine imaging [13, 14].

Noncontrast MRA Applications in Abdomen The noncontrast-enhanced MRA technique relies on an ECG-gated 3D partial-Fourier FSE sequence which is triggered during diastole to obtain bright blood signals and during systole to produce a flow void or black blood signals [9]. In abdominal MRA, since venous enhancement may not be a serious detriment to image quality, a single diastolic-

93

triggering acquisition can be performed without a systolic subtraction. For the relatively fast flow of the thoracic aorta, the PE direction should be oriented to be parallel to the direction of flow (craniocaudal) with the application of presaturation bands superior and inferior to avoid wrap-around artifacts. Clinical evaluation of noncontrast FBI images was performed in a prospective study of 75 patients with arterial diseases referred for noncontrast FBI (34 dissection, 27 aneurysm, 4 arterial occlusion, and 10 surgical bypass); image quality of the noncontrast FBI images graded excellent in 45, satisfactory in 25, and poor in 5 patients; with all 34 dissection cases, the intimal flap was consistently visualized [15]. In addition, the noncontrast FBI technique provides clear depiction of the aorta in postoperative patients with metal wires, clips, or implants. FBI provides images with minimal artifacts that unambiguously delineate the anastomosis between the bypass graft and the native vessel. Figure 6.8 shows a typical noncontrast diastolic-triggered FBI images on a patient with a dissecting aortic arch using the ECG-gated 3D partial-Fourier FSE method compared to a contrast-enhanced (CE) MRA (Fig. 6.8b). The FBI source images clearly demonstrate the intimal flap and tear and some pleural and pericardial effusions by synchronizing the ECG triggering. The corresponding CE MRA source images acquired after the FBI scans reveal the patent lumen and the intimal flap and tear by the difference in the contrast arrival timing.

Peripheral MRA: Separation of Arteries from Veins As discussed previously, the orientation of the PE direction relative to the vessels of interest can be selected to optimize arterial signal for the thoracic cavity. For relatively slower moving protons in the peripheral arteries, the PE direction can be oriented perpendicular to the direction of flow and flow-spoiler gradient pulses applied in the readout (RO) or frequency direction to increase the flow-dephasing effect during systole. The percentage of the flow-spoiling gradient is defined as a percentage relative to one-half of the entire RO gradient area [10]. Flow-spoiling gradient pulses accentuate the differences in signal between systolic and diastolic phases, without affecting slower flowing venous blood and the stationary background signals. Therefore, in peripheral MRA, the signal between systole and diastole is emphasized by adding flow-spoiling or partial flow compensation for optimizing flow void in systolic acquisitions [10]. In peripheral MRA, systolic- and diastolic-triggering timing becomes crucial to have optimal signal difference. We use the ECG-prep scan using similar scan parameters as in the 3D acquisition, such as R–R interval, effective TE, orientation of the PE direction, application of a prepulse such as STIR for fat suppression, and a single-shot acquisition window duration. The ECG-prep acquisition is typically incremented by a 50 or 100 ms with repetition time (TR) of

94

M. Miyazaki et al.

Fig. 6.6 (a) Sequence diagram of an ECG-prep scan (single slice with multiple phases). (b) Corresponding images of the aortic vessels obtained using the ECG-prep scan. Note that the systolic-triggered image shows the aorta with dark blood (yellow arrow), whereas the

diastolic image shows the aorta with bright blood ( red arrow ) . The superior vena cava (SVC) (blue arrows) shows bright signal throughout the cardiac cycle (Modified from Wedeen et al. [5] with permission)

Fig. 6.7 (a) The signal difference of pulmonary vessels from the subtraction of the systolic- from diastolic-triggered 2D projection images on a healthy volunteer. (b) DSA image (top) shows the right pulmonary thromboembolism (red open arrow), whereas the diastole–systole subtraction 2D projection image shows the corresponding region with hyposignal intensities (red arrows) (Courtesy of Dr. K. Nakamura of Kyoritsu Hospital, Japan)

two or three R–R intervals to observe the single-shot 2D images across the entire cardiac cycle. The artery of interest with lowest signal will be selected as a systolic triggering delay and the subtraction of the systolic image from all the ECG-prep images will provide high signal intensity of the artery of interest, which will be the diastolic triggering delay. These manual calculations sometimes are cumbersome and difficult to determine the systolic and diastolic triggering delays, especially in small vessels in the 2D projection images. This cumbersome procedure of determining the triggering delays is even more difficult when the signal intensities of diastole are similar in images. The development of graphic display software, FBI-Navi, is utilized to provide a graph of signal intensity versus ECG phases [16], as shown in Fig. 6.9. The FBI-Navi manipulates all the phase images of the ECG-prep scan and automatically calculates the signals in the cardiac phases of the artery. Figure 6.9a shows a variation of the arterial signal intensity presented in a twodimensional graph to select the delays, the hypointensity for

6

Flow-Dependent Noncontrast MR Angiography

95

Fig. 6.8 (a) Contiguous source images of noncontrast FBI MRA and (b) contrast-enhanced (CE) MRA of the thoracic arch of a patient with an aortic dissection. The FBI source images clearly demonstrate the intimal flap and tear (arrow) and some pleural and pericardial effusions (arrowheads) due to synchronizing the ECG triggering. The CE MRA source images acquired after the FBI scan reveal the patent lumen and the intimal flap and tear by the difference in the contrast arrival timing (Modified from Urata et al. [15] with permission)

Fig. 6.9 (a) The result of the FBI-Navi plot shows signal intensity in arbitrary units and the ECG phases. The lowest signal area should be the systolic triggering delay (red circle) and the high signal plateau area is the diastolic triggering period (red circle). (b) The diastolic source image

shows both arteries and veins in bright blood, whereas the systolic source image shows black blood arteries and bright blood veins. The subtraction of the systolic source images from the diastolic images generates an arteriogram after the maximum intensity projection processing

systolic, and the hyperintensity for diastolic triggering delays. The determination of diastole should be the beginning to the middle of a plateau area, since the single-shot acquisition duration is about 150–350 ms for a 256 × 256 matrix, depending on the number of shots and parallel imaging factor. Figure 6.9b shows typical images of diastolic and

systolic MIP images and the subtraction of systolic from diastolic images that generate the arteriogram after the MIP procession. In order to reduce misregistration caused by two separate systolic and diastolic 3D acquisitions, a continuous acquisition is enabled to collect systolic and diastolic images

96

M. Miyazaki et al.

Fig. 6.10 (a) A schematic of continuous acquisitions of systole and diastole using ECG-gated 3D partial-Fourier FSE. Each slice is acquired in one single-shot acquisition using partial-Fourier FSE. One trigger delay (d1) is timed for systole for one 3D acquisition, while a second (d2) is timed for diastole. Acquisitions are performed every other or every

third R–R intervals. A short-tau inversion recovery (STIR) pulse is generally applied for fat suppression. To generate a bright blood angiogram, the systolic images (where arterial flow appears dark) are subtracted from diastolic images (Reproduced with permission from Ito et al. [25]). (b) Corresponding peripheral run-offs MRA on a healthy volunteer

without interrupting the scan. Figure 6.10a shows the schematics of continuous systolic scans with each slice encoding with the same delay from an R wave, followed by the diastolic scans. At each slice encoding, a single-shot partialFourier FSE is acquired with a STIR pulse for fat suppression. Operators can preset the system to subtract systolic from diastolic triggering scans, this subtraction of source images provides an arteriogram after the MIP processing. Typically, the RO flow compensation or spoiler gradient is present on patients and healthy volunteers. Generally, older patients with slower flow require a stronger RO spoiler pulse. Vessels in the hand and feet with slower flowing blood are also imaged with this approach using stronger flow-spoiling pulses to differentiate systolic signal from diastolic signal. Three separate groups clinically evaluated the ECG-gated partial-Fourier FSE MRA technique on patients with peripheral diseases. In the preliminary study in 2002, Urata et al. [17] reported that the ECG-gated FBI method was comparable to CE MRA in 24/44 regions, while 15 were inferior and 5 were superior, when tested in a total of 56 regions (18 iliac, 20 femoral, and 18 calf area) in 26 patients who were diagnosed or had suspected arterial occlusive disease. To avoid the overestimation of stenoses, the investigators in this study suggested that diastolic images, depicting both arteries and veins, always be evaluated in addition to the subtracted MIP images. Nakamura et al. used a similar strategy and compared noncontrast MRA with 16-detector CTA and found in their study of 13 patients with 56 diseased segments from the iliac to calf regions that the MR method resulted in a sensi-

tivity of 94%, a specificity of 94%, and an accuracy of 94% for the detection of ³50% stenosis [18]. In a separate study, Lim et al. compared the noncontrast MRA method in the distal station (calf and pedal arteries) with the conventional bolus chase imaging and 3D timeresolved contrast-enhanced imaging in 36 patients [19], where the reference standard was a combined consensus interpretation of all three sequences. When all subjects were studied, the noncontrast technique demonstrated accuracy of 79.4% (1,083/1,364), sensitivity 85.4% (437/512), and specificity 75.8% (646/852), with a high negative predictive value of 92.3% (646/700). Serious artifacts lead to poor diagnostic confidence in 17 patients (47.2%). Among the patients with satisfactory diagnostic confidence, accuracy, sensitivity, and negative predictive values were 92.2% (661/717), 92.4% (158/171), and 97.5 (503/516), respectively. One limitation of the study was that only the subtracted MIP images were evaluated. Further clinical studies using similar settings are required.

Noncontrast MR Venography In the area of noncontrast MR venography, there have been few reports outside the 2D TOF technique which can use saturation bands to suppress arterial flow. New modifications to the FBI technique were developed for faster flow in the iliac region using a swap phase-encode arterial double-subtraction elimination (SPADE) technique and flow-refocusing pulses (FR-FBI) in the RO direction [17] for slower flow in the femoral and calf regions. The FR-FBI technique uses a

Fig. 6.11 (a) A schematic of the swap phase-encode artery doublesubtraction elimination (SPADE) technique to visualize the iliac veins. The SPADE technique utilizes three acquisitions (two diastolictriggered acquisitions, one with the head–feet and the other with the right–left PE directions, and one systolic acquisition FR-FBI) to depict the iliac veins. (b) A 61-year-old female with an extensive deep vein thrombosis (DVT) 20 days after surgery for tibial fracture. The iliac region was acquired using the SPADE technique. The femoral and calf regions were acquired using the FR-FBI. (A) Conventional venography shows thrombi in the peroneal, popliteal, superficial, and common femoral veins (arrowheads). The proximal extent of throm-

bosis is not visualized (arrows). (B) maximum intensity projection (MIP) images of SPADE for pelvis, and FR-FBI for thigh and calf show the thrombi as hypointense signals in a wide range from the popliteal to common iliac venous lumen (arrowheads) and high signal intensity in the IVC (arrow). In the calf region, the presence of elevated signals due to edema obscures the appreciation of calf veins in the MIP images. (C) Source images of (B) reveal the presence of thrombus in peroneal vein as a hypointense signal defect within the venous lumen, as well as popliteal, superficial femoral, and iliac veins (arrows) (Reproduced with permission from Furudate and Miyazaki [16])

98

M. Miyazaki et al.

15% flow compensation pulse, the percentage of FR relative to one-half the entire RO gradient area, in the RO direction. The SPADE technique utilizes three acquisitions (two diastolic-triggered acquisitions, one with the head–feet and the other with the right–left PE directions, and one systolic acquisition FR-FBI) to depict the iliac region, as shown in Fig. 6.11a. Figure 6.11b shows an example of clinical cases presenting a comparison of DSA and noncontrast venography using SPADE for the iliac veins and FR-FBI for the femoral and calf veins [17]. The clinical evaluation was reported on 41 legs of 32 patients (25 of the 32 patients with nonmagnetizing metal implants) suspected of having deep vein thrombosis (DVT) examined using SPADE (iliac) and FR-FBI (femoral and calf), compared to conventional X-ray venography [18]. The results of the FR-FBI and SPADE for diagnosing thrombus study provided sensitivities of 100% (53 of 53) for two reviewers and the corresponding specificities of 100% (for one reviewer) and 99.6% (for the other reviewer). There were greater numbers of nondiagnostic scores in the conventional X-ray angiograms as compared to those in the noncontrast SPADE and FR-FBI exams.

Time-Spatial Labeling Inversion Pulse with 2D/3D Partial-Fourier FSE Background In an ASL technique, a time-spatial labeling inversion pulse (time-SLIP) can be applied in conjunction with 2D/3D partial-Fourier FSE or bSSFP. The FSE-based methods have the advantage of less sensitivity to susceptibility artifacts compared with gradient-type bSSFP method. However, bSSFP has the advantage of depicting fast flow and multidirectional flow over the FSE-type method. These merits and pitfalls of the FSE and bSSFP methods are discussed in detail, along with their applications to specific regions and purposes are introduced [22] Here, we discuss the three main approaches and applications of time-SLIP with 2D/3D partial-Fourier FSE methods. Depending upon the applications, the three types of techniques, flow-in, flow-out, and tag-on and tag–off alternative subtraction, were selected to visualize the vessels of interest and relationship to the surrounding background signals [23]. Principle of Techniques Figure 6.12a shows a sequence diagram of a time-SLIP with 3D partial-Fourier FSE. Sequence can be converted to a 2D acquisition by removing the slice encoding. There are three different approaches to studying the vessels of interest; flowout, flow-in, and an alternate tag-on and tag-off subtraction. The flow-out technique utilizes both nonselective (A) and the spatially selective pulse (B) to make the marked bright blood

Fig. 6.12 (a) A sequence diagram of time-spatial labeling inversion pulse (time-SLIP) with 3D partial-Fourier FSE. The A pulse is nonselective and the B pulse is spatially selective followed by the 3D partialFourier FSE. Note that the selective pulse allows rotating to any orientation to freely mark the area. The flow-out technique, the marked blood flow-out, utilizes both a nonselective (A) and the spatially selective pulses (B) to bring the magnetization of tagged area blood signals to flow out from the area during the inversion time (TI). The flow-in technique utilizes only the selective (B) pulse without applying nonselective pulse (A). The area of the A pulse experiences a 180 pulse and follows a recovery path to a null point, the fresh unperturbed blood flow into the A pulse area. (b) Schematics of a tag-on and tag-off alternate acquisition and subtraction technique. Each tag-on and tag-off data was separately Fourier transformed, followed by the subtraction (Reproduced with permission from Kanazawa and Miyazaki [23])

flow-out within inversion time (TI). The flow-in technique utilizes only the selective (B) pulse without applying nonselective pulse (A). The area of the A pulse experiences a 180 pulse and follows a recovery path to a null point, the fresh unperturbed blood flow into the A pulse area. Figure 6.12b shows the acquisition, subtraction, and reconstruction of the alternate tag-on and tag-off acquisition and subtraction technique. Flow-In Technique A flow-in technique utilizes a selective pulse to mark the area of interest to invert the signals. During the TI time, fresh unperturbed blood flows into the tagged area, while the tagged area is at a null point. Figure 6.13a, b shows a 3D partial-Fourier FSE (FBI) without time-SLIP and with timeSLIP. In the image without time-SLIP, not only are the portal veins bright but the long T2 component of bile fluid is also

6

Flow-Dependent Noncontrast MR Angiography

99

Fig. 6.13 Portal veins images on a healthy volunteer using diastolictriggered 3D partial-Fourier FSE (FBI) without time-SLIP (a) and with time-SLIP (b). The FBI without time-SLIP depicts not only the portal veins bright but also the long T2 component of bile duct fluid which hinders the depiction of portal veins. The time-SLIP image suppresses the unwanted bile signals with the selective inversion pulse to null the background at 800 ms, and the superior vena cava and splenic veins

flow into the portal veins, which results in depiction of only portal veins. Pulmonary vessels using FBI without (c) and with time-SLIP (d). The FBI without time-SLIP shows all vessels in pulmonary vessels, whereas the FBI with time-SLIP using a TI of 800 ms shows arterial flow. (e) Positioning of 3D imaging plane and the time-SLIP pulse to visualize the pulmonary arteries (Reproduced with permission from Kanazawa and Miyazaki [23])

depicted which hinders the depiction of only the portal veins. With application of the time-SLIP tag pulse, the liver area was tagged in an oblique position so that SVC and splenic veins flow into the tagged area. During the TI time, the tagged area experiences a null point around 800 ms. Another example of 3D partial-Fourier FSE (FBI) without time-SLIP and with time-SLIP in the pulmonary arteries is shown in Fig. 6.13c, d. Without time-SLIP, FBI image shows all of the pulmonary vessels. Pulmonary arteries are visualized by placing the tagged pulse on the pulmonary region below the heart so freshly pumped blood reaches the pulmonary arteries during the TI time. FBI with time-SLIP shows only the pulmonary arteries at a TI of 800 ms.

the flow-out technique is shown in the cerebrospinal fluid (CSF) movement in Fig. 6.14b [24]. The time-SLIP tag pulse was applied at a right angle in a region of interest that covers the lateral and third ventricles. Note that the untagged CSF is dark around TI of 2,000 and 4,000 ms, whereas the tagged CSF flows out from the third ventricle to the fourth ventricle through the aqueduct. As TI time is increased, the marked CSF travels farther in distance. Another example of noncontrast flow-out was conducted to study the distribution of blood to the portal veins, which utilizes both a nonselective and selective pulses [24]. In this study, the SVC blood was tagged in a coronal plane to study the distribution to right and left portal veins, as well as the splenic vein was tagged to find the blood distribution as well on healthy volunteers.

Flow-Out Technique The flow-out technique comprises both a nonselective and selective tag pulses. As shown in Fig. 6.14, the double inverted region of both of the pulses has high signals at +Mz, whereas the background signal follows the T1 recovery relaxation. The best contrast between the tagged area and the background signal is around the null point. An example of

Tag-On and Tag-Off Alternate Subtraction When acquiring the tag-on and tag-off alternate subtraction technique, acquisition time is doubled by acquiring both the tag-on and tag-off scans; as shown in Fig. 6.12b, the technique does not require to consider the return of background

100

M. Miyazaki et al.

Fig. 6.14 A flow-out technique showing cerebrospinal fluid movement. (a) A sequence diagram of a time-SLIP with 2D partial-Fourier FSE and the magnetization states of tagged region and background signals. An initial nonselective (A) pulse inverts all magnetizations to the −Mz direction within the radiofrequency (RF) coil and followed by a selective (B) pulse that restores the magnetization within the selective region to the +Mz direction. The untagged region follows the T1 recovery relaxation and returns to +Mz. When untagged region signals reach a null point, tagged and untagged signals have good contrast. (b) Midsagittal brain time-SLIP images of a 25-year-old healthy male volunteer with various TI times. (a) Images of time-SLIP or pulsed cerebrospinal fluid (CSF) at the TI times indicated. The labeling pulse is 1-cm thick

(dotted lines) and is applied at a right angle in a region of interest that covered the third ventricle. Note that the nonlabeled CSF at TI times between 2,000 and 4,000 ms is dark or null, whereas the pulse-labeled CSF shows high signal intensity over the same time period. As the TI times increase, the labeled CSF moves from the third ventricle to the fourth ventricle via the aqueduct, as indicated by the arrows. (b) Using the time-SLIP technique with a much wider labeled section (dotted lines) that covered the posterior fossa, one can observe movement of CSF from the prepontine subarachnoid space (SAS) into the spinal SAS through the cisterna magna as indicated by arrows. Images in (a) and (b) demonstrate that the width of the time-SLIP labeled CSF can be varied (Reproduced with permission from Ito et al. [25])

signals unlike the flow-in and flow-out techniques. Because the background signal is canceled out, one can freely select any TI times and thus any position of movement can be depicted. Figure 6.15a shows an FBI angiogram of foot using

ECG-gated partial-Fourier FSE. With an application of a strong flow-dephasing pulse, distal branches of the foot arteries are well depicted without venous contamination. The time-SLIP image of the foot shows even further distal

6

Flow-Dependent Noncontrast MR Angiography

101

tissue, the steady-state signal magnetization (i.e., signal) for bSSFP can be estimated by the relation:

M ss = M0

sin a , 1 + cos a + (1 - cos a)(T1 / T2)

where M0 is the spin-density and a (alpha) is the flip angle of the RF pulse. The flip angle aopt (alpha opt) for a tissue with T1 and T2, a opt = a cos (T1 / T2 - 1) / (T1 / T2 - 1) results in a the optimal signal amplitude of

M ss =

Fig. 6.15 (a) Foot MRA using FBI using the RO spoiler of +35% and (b) using time-SLIP with a TI of 1,000 ms. Time-SLIP MRA of the hand with varying TI times, 600, 800, and 1,000 ms. As the TI time is increased, marked blood vessels flow farther in distance (Image courtesy to Dr. J Isogai of Hasuda Hospital, Japan)

branches in toe arteries, as shown in Fig. 6.15b [26]. Figure 6.15c shows a hand MRA of a patient with Raynaud’s syndrome (asymptomatic state) obtained using the tag-on and tag-off alternate subtraction technique [27]. As TI increases, the marked blood travels farther in distance.

Noncontrast MRA Using BSSFP Imaging Balanced steady-state free precession (bSSFP) imaging, also referred to as true fast imaging with steady-state precession (trueFISP), fast imaging employing steady-state excitation (FIESTA), or balanced fast field echoes (FFE), is another method for performing flow-independent MRA. Unlike the methods described above which can only be collected in segmented manner, with segments of acquisition interspersed with periods of no imaging activity, bSSFP naturally supports continuous noninterrupted acquisition. Although the technique was first applied to humans for noncardiovascular applications nearly 20 years ago [29–31], improvements in the gradient systems of the MR scanners have only made bSSFP imaging practical for cardiovascular imaging in the last decade. The most common application of bSSFP imaging is for cine imaging of heart function, nonetheless, the method can also be used for noncardiac flow-independent MRA, as will be described. The pulse sequence timing diagram for bSSFP imaging is shown in Fig. 6.16. When the repetition time of the imaging sequence is substantially shorter than the T1 and T2 of the

1 T2 Mo . 2 T1

Tissues with large T2/T1, such as blood and fluid, are enhanced in bSSFP images while others having short T2, such as muscle, are suppressed. Thus, the method naturally depicts both veins and arteries. Fat, however, is enhanced with the bSSFP technique as well due to its large T2/T1, which renders the method unsuitable for producing MIP views of the vasculature. The image contrast provided by bSSFP imaging is shown in Fig. 6.17. Two-dimensional bSSFP imaging has been reported for detecting deep venous thrombosis in the lower extremities [32]. In that study, the technique was found to be highly sensitive and specific for detecting thrombi in the iliac and femoral veins, but was only moderately sensitive for demonstrating thrombi in the calf. The bSSFP acquisition can also be segmented to allow for the incorporation of magnetization preparation modules such as spectrally selective fat saturation and muscle suppression. Frequently coupled with centric (i.e., low-high) k space ordering, application of an additional T2 preparation module accentuates the appearance of arterial blood pool in relation to background signals such as muscle and fat [33]. Several methods for suppressing fat signal during bSSFP imaging have been reported [35–37]. The most frequent use of noncontrast flow-independent MRA with segmented bSSFP has been used for visualizing the coronary arteries [31].

Subtractive Methods Using BSSFP Imaging Subtraction-based techniques using a bSSFP acquisition have also been proposed for flow-independent angiography. The advantage of subtractive bSSFP techniques relative to standard bSSFP acquisitions is their ability to provide near complete suppression of unwanted signal from background tissue. One such subtractive technique proposed by Edelman and Koktzoglou is signal targeting alternative radiofrequency and flow-independent relaxation enhancement (STARFIRE) [38–40]. The STARFIRE technique has potential applications for venography as well as angiography. Figure 6.18

102

M. Miyazaki et al.

Fig. 6.16 Timing diagram for 3D balanced steady-state free precession magnetic resonance imaging. The net gradient area over all gradient axes equals zero. The repetition time (TR) is typically short, on the order of 3–5 ms. RF radiofrequency pulse, Gss slice-select gradient, Gpe phase-encoding gradient, Gro readout gradient, ADC analog to digital converter

shows a generic pulse sequence diagram of STARFIRE flow-independent MRA. The technique consists of an interleaved acquisition of two bSSFP readouts, in which one of the readouts is preceded by an inversion RF pulse. With appropriate selection of the inversion time (TI) (normally ranging between 900 and 1,200 ms at 1.5 T for a TR between 2 and 3 s), subtraction of the data suppresses the appearance of fat and muscle signal, while leaving signal from the vascular pool with longer T1 and T2. A three-dimensional acquisition is normally used; however, two-dimensional STARFIRE is also feasible. Figure 6.19 shows images acquired with the STARFIRE technique in the lower extremities. With STARFIRE, subtraction of complex image data is generally preferred over subtraction of magnitude images since the former reduces partial volume artifacts and reduces the appearance of off-resonance banding artifacts. Magnitude subtraction, however, can be used to suppress the appearance of fluids with a mild alteration of the STARFIRE pulse sequence [39].

Quiescent Interval Single-Shot MRA

Fig. 6.17 Axial bSSFP image obtained through the neck of a volunteer. Veins, arteries, fluids, and fat appear hyperintense relative to muscle tissue

More recently, Edelman and collaborators [40–42] have developed a cardiac-gated noncontrast MRA pulse sequence based on steady-state free precession readout. The technique, quiescent interval single shot (QISS), combined the bright

6

Flow-Dependent Noncontrast MR Angiography

103

Fig. 6.18 (Top) Pulse sequence for STARFIRE flow-independent MRA showing the interleaved acquisition of two data sets which are later subtracted to produce a background-suppressed angiogram. (Bottom) Repetitive saturation RF pulses may be applied between

bSSFP readouts to provide selective arteriography or venography (Figure courtesy of Edelman et al., Radiology, 2009; reprinted with the permission of the Radiological Society of North America)

Fig. 6.19 (Left) Maximum intensity projection (MIP) of a 3D STARFIRE angiogram displaying the deep and superficial venous systems. Axial slab prescription and superior repetitive presaturation was used to suppress the appearance of the arteries. (Right) MIP of a 2D time-of-

flight venogram in the same subject. Note the improved depiction of small and superficial veins with STARFIRE, owing to its flow-independent nature (Figure courtesy of Edelman et al., Radiology, 2009; reprinted with the permission of the Radiological Society of North America).

blood contrast of bSSFP with simultaneous suppression of fat, venous inflow, and static background tissue. Figure 6.20 shows a timing diagram of the QISS acquisition. Venous and background saturation pulses are played with an intervening “quiescent time interval” (QI). During the QI time, arterial flow is replenished with a venous saturation pulse which tracks the 2D image acquisition. QISS is a 2D ECG-gated single-shot bSSFP acquisition which uses an initial in-plane saturation pulse to suppress background tissues and a quiescent inflow time period (QI). The QI overlaps the period of

rapid systolic flow, ensuring maximal inflow of unsaturated spins into the slice even in the setting of very slow flow. Imaging parameters are fixed irrespective of heart rate or other factors, without variation from patient to patient. QISS nonenhanced MRA does not require modification of the sequence based on the absence, presence, or severity of peripheral arterial disease. Consequently, it is particularly easy to use and thus of widespread clinical utility. Moreover, it uses a subsecond single-shot data acquisition that makes the technique particularly insensitive to patient motion.

104

M. Miyazaki et al.

Fig. 6.20 Quiescent interval single-shot pulse sequence timing diagram. Slice selective saturation pulses suppress are coordinated with the cardiac cycle to the signal from static background tissue and veins. After a quiescent time to allow for replenishment of arterial signal into the slice, a fat saturation pulse, a/2 prep pulse, and single-shot bSSFP readout are performed. This approach allows for a 2D slice to be acquired with each heartbeat, allowing for whole body coverage in less than 7 min

In a recent study, QISS was evaluated in a two-center trial of 53 patients referred for lower extremity MRA (Fig. 6.21) for suspected or known chronic PAD using CE MRA as the noninvasive standard of reference. The diagnostic accuracy of QISS was found to be nearly equivalent to CE MRA and DSA. Non-GEE segment based analysis demonstrated diagnostic accuracy of QISS for two reviewers as follows: sensitivity, 89.7% (436/486) – 87.0% (423/486); specificity, 96.5% (994/1,030) – 94.6% (973/1,028). It was concluded that QISS MRA was an accurate nonenhanced alternative to CE MRA and would be of particular value for patients with impaired renal function = 833 ms, 9 stations, 48 slice/station give a typical scan time of 6.4 min.

Summary

Fig. 6.21 Representative quiescent interval single shot nonenhanced (left) and contrast-enhanced (right) MR angiograms show occluded right and left superficial femoral arteries with collateral vessels

In this chapter, flow-dependent noncontrast MRA techniques, TOF, PC, ECG-gated partial-Fourier FSE, timeSLIP with ECG-gated partial-Fourier FSE and bSSFP noncontrast MRA were discussed. As each of them have their primary clinical applications, the latest advances of noncontrast, ECG-gated partial-Fourier FSE with and without time-SLIP have really expanded the noncontrast clinical applications beyond the head and neck. In time-SLIP with partial-Fourier FSE, to make the bright blood, tagged images with partial-Fourier FSE, the readout should be during diastole, which lengthens the scan time compared to the bSSFP method. However, in examinations of patients with clips or implants, ECG-gated partial-Fourier FSE provides better image quality as it is less affected to susceptibility artifacts. Therefore, depending on the application, flow-dependent techniques have strong impact on various areas of the body.

6

Flow-Dependent Noncontrast MR Angiography

105

References 19. 1. Pike GB, Hu BS, Glover GH, Enzmann DR. Magnetization transfer time-of-flight magnetic resonance angiography. Magn Reson Med 1992:25:372–379. 2. Atkinson D, Brant-Zawadzki M, Gillan G, Purdy D, Laub G. Improved MR angiography: magnetization transfer suppression with variable flip angle excitation and increased resolution. Radiology 1994:190:890–894. 3. Miyazaki M, Kojima F, Ichinose N, Onozato Y, Igarashi H. A novel saturation transfer contrast method for 3D time-of-flight magnetic resonance angiography: a slice-selective off-resonance sinc pulse (SORS) technique. Magn Reson Med. 1994:32:52–59. 4. Dumoulin CL, Souza SP, Walker MF, Wagle W. Three-dimensional phase contrast angiography. MRM 1989:9:139–149. 5. Wedeen VJ, Meuli RA, Edelman RR, et al. Projective imaging of pulsatile flow with magnetic resonance. Science 1985: 230: 946–948. 6. Meuli RA, Wedeen VJ, Geller SC, et al. MR gated subtraction angiography: evaluation of lower extremities. Radiology 1986:159:411–418. 7. Constable RT, Gore JC. The loss of small objects in variable TE imaging: implications for FSE, RARE, and EPI. Magn Reson Med 1992:28:9–24. 8. Miyazaki M, Ichinose N, Sugiura S, Kassai Y, Kanazawa H, Machida Y. A novel MR angiography technique: swap phase encode extended data (SPEED) acquisition using half-Fourier RARE. JMRI 8:505–507, 1998. 9. Miyazaki M, Sugiura S, Tateishi F, Wada H, Kassai Y, Abe H. Noncontrast-enhanced MR angiography using 3D ECG-synchronized half-Fourier fast spin echo. JMRI 2000:12:776–783. 10. Miyazaki M, Takai H, Sugiura S, Wada H, Kuwahara R, Urata J. Peripheral MR angiography: separation of arteries from veins with flow-spoiled gradient pulse in electrocardiography-triggered threedimensional half-Fourier fast spin-echo imaging. Radiology 2003:227(3): 890–896. 11. Hinks RS, Constable RT. Gradient moment nulling in fast spin echo. Magn Reson Med 1994:32:698–706. 12. Hayashi S and Miyazaki M. Thoracic duct: visualization at nonenhanced MR lymphography-initial experience. Radiology 1999:212:598–600. 13. Kawanami S, Nakamura K, Miyazaki M, Sugiura S, Yamamoto S, Nakata H. Flow-weighted MRI of the Lungs with the ECG-gated half-Fourier FSE technique: Evaluation of the Effect of the Cardiac Cycle. Magn Reson in Medical Sciences 2002:1:3 137–147. 14. Suga K, Ogasawara N, Okada M, Tsukuda T, Matsunaga N, Miyazaki M. Lung perfusion impairments in pulmonary embolic and airway obstruction with noncontrast MR imaging. J Appl Physiol. 2002:92: 2439–2451. 15. Urata J, Miyazaki M, Wada H, Nakaura T, Yamashita Y, Takahashi M. Clinical evaluation of aortic diseases using non-enhanced MRA with ECG-triggered 3D half-Fourier FSE. JMRI 2001:14(2): 113–119. 16. Furudate N and Miyazaki M. FBI-Navi for easy determination of diastolic and systolic triggering phases in non-contrast fresh blood imaging (FBI). Presented at the ISMRM 16th Annual Meeting, Toronto, 2008: p 2902. 17. Urata J, Wada H, Kuwahara R, Miyazaki M, Nishiharu T, Yamashita Y. Evaluation of lower extremity arterial diseases using flowdephasing spoiler gradient pulses in ECG-triggered 3D half-Fourier FSE: initial experience. In: International Society for Magnetic Resonance in Medicine. Honolulu, 2002; 1750. 18. Nakamura K, Kuroki K, Yamamoto A, Hiramine A, Miyazaki M, Matsufuji Y. Fresh blood imaging (FBI) of peripheral arteries:

20.

21.

22. 23.

24.

25.

26.

27.

28. 29.

30.

31.

32.

33.

34.

35.

Comparison with 16-detector row CT angiography. In: International Society for Magnetic Resonance in Medicine. Seattle, 2006; 1929. Lim RP, Hecht EM, Xu J, et al. 3D Non-gadolinium enhanced ECG-gated MRA of the distal lower extremities: Preliminary clinical experience. J Magn Reson Imaging, 2008:28:181–189. Ono A, Murase K, Taniguchi T, Shibatani O, Takata S, Kobashi Y, Miyazaki M. Deep vein thrombosis using non-contrast-enhanced MR venography with electrocardiographically-gated three-dimensional half-Fourier FSE: preliminary experience. Magn Reson Med 2009:61: 907–917. Ono A, Murase K, Taniguchi T, Shibatani O, Takata S, Kobashi Y, Miyazaki M. Deep Venous Thrombosis: Diagnostic. Value of Noncontrast-enhanced MR Venography Using Electrocardiographytriggered Three-dimensional half-Fourier FSE. Magn Reson Med, in press. Miyazaki M and Lee VS. Non-contrast-enhanced magnetic resonance angiography: state-of-the-art. Radiology 2008:248:20–43. Kanazawa H and Miyazaki M. Time-spatial labeling inversion tag (t-SLIT) using a selective IR-tag on/off pulse in 2D and 3D halfFourier FSE as arterial spin labeling. Presented at the ISMRM 10th Annual Meeting, Hawaii, p140, 2002. Yamada S, Miyazaki M, Kanazawa H, Higashi M, Morohoshi Y, Bulml S, McComb JG. MRI visualization of cerebrospinal fluid movement with spin labeling: Preliminary results in normal and pathophysiological conditions. Radiology 249:644–652, 2008. Ito K et al: Intraportal venous flow distribution: evaluation with single breath-hold ECG-triggered three-dimensional half-Fourier fast spin-echo MR imaging and a selective inversion-recovery tagging pulse. AJR 2002:178(2): 343–348. Isogai J, Miyazaki M, Shimada T, Hatakeyama H, Yamada T, Matsuo N, Maejima S, Yodo K, Miyata T. Non-contrast MRA of the toes using time-spatial labeling inversion pulse (time-SLIP) and optimization of flow-spoiled gradient pulses for the assessment of foot arteries in flow-spoiled fresh blood imaging (FBI) Presented at the ISMRM 16th Annual Meeting, Toronto, 2008: p2901. Isogai J. Kobayashi Y, Ogawa Y, Nakajima Y, Maejima S, Hatakeyama H, Yamada T, Matsuo N, Yodo K, Miyata T, Miyazaki M. Presented at the ISMRM 15th Annual Meeting, Berlin, p3152, 2007. Oppelt A, Graumann R, Barfuß H, Fischer H, Hartl W, Schajor W. FISP – A New Fast MRI Sequence. Electromedica 1986;51:15–18 Carr JC, Simonetti O, Bundy J, Li D, Pereles S, Finn JP. Cine MR angiography of the heart with segmented true fast imaging with steady-state precession. Radiology 2001;219:828–834. Barkhausen J, Ruehm SG, Goyen M, Buck T, Laub G, Debatin JF. MR evaluation of ventricular function: true fast imaging with steady-state precession versus fast low-angle shot cine MR imaging: feasibility study. Radiology 2001;219:264–269. Deshpande VS, Shea SM, Laub G, Simonetti OP, Finn JP, Li D. 3D magnetization-prepared true-FISP: a new technique for imaging coronary arteries. Magn Reson Med 2001;46:494–502. Cantwell CP, Cradock A, Bruzzi J, Fitzpatrick P, Eustace S, Murray JG. MR venography with true fast imaging with steady-state precession for suspected lower-limb deep vein thrombosis. J Vasc Interv Radiol 2006;17:1763–1769. Cukur T, Lee JH, Bangerter NK, Hargreaves BA, Nishimura DG. Non-contrast-enhanced flow-independent peripheral MR angiography with balanced SSFP. Magn Reson Med 2009;61: 1533–1539. Shea SM, Deshpande VS, Chung YC, Li D. Three-dimensional true-FISP imaging of the coronary arteries: improved contrast with T2-preparation. J Magn Reson Imaging 2002;15:597–602. Scheffler K, Heid O, Hennig J. Magnetization preparation during the steady state: fat-saturated 3D TrueFISP. Magn Reson Med 2001;45:1075–1080.

106 36. Huang TY, Chung HW, Wang FN, Ko CW, Chen CY. Fat and water separation in balanced steady-state free precession using the Dixon method. Magn Reson Med 2004;51:243–247. 37. Stafford RB, Sabati M, Mahallati H, Frayne R. 3D non-contrastenhanced MR angiography with balanced steady-state free precession Dixon method. Magn Reson Med 2008;59:430–433. 38. Edelman RR, Koktzoglou I. Unenhanced flow-independent MR venography by using signal targeting alternative radiofrequency and flowindependent relaxation enhancement. Radiology 2009;250: 236–245. 39. Edelman RR, Koktzoglou I, Ankenbrandt WJ, Dunkle EE. Cererbral venography using fluid-suppressed STARFIRE. Magn Reson Med 2009;62:538–543

M. Miyazaki et al. 40. Kotzoglou I, Edelman RR. STAR and STARFIRE for flowindependent and flow-dependent noncontrast carotid angiography . Magn Reson Med 2009; 61:117–124. 41. Edelman RR, Sheehan JJ, Dunkle E, Schindler N, Carr J, Koktzoglou I. Quiescent-interval single-shot unenhanced magnetic resonance angiography of peripheral vascular disease: Technical considerations and clinical feasibility Magn Reson Med 63(4), 951–8. 2010. 42. Hodnet PA, Koktzoglou I, Davarpanah A, et al. Evaluation of peripheral arterial disease with nonenhanced quiescentinterval single-shot MR angiography. Radiology 260(1). 2011 282–293

7

Low-Dose Contrast-Enhanced MR Angiography Kambiz Nael, Roya Saleh, Gerhard Laub, and J. Paul Finn

Background

Association of Gadolinium with NSF

Today, three-dimensional (3D) contrast-enhanced magnetic resonance angiography (CE-MRA) is a widely accepted and powerful diagnostic tool for assessment of almost all vascular territories. Its noninvasive nature and flexibility make CE-MRA an appealing alternative to digital subtraction angiography (DSA) or computed tomography angiography (CTA). In addition, compared to the iodinated contrast agents used in CTA and DSA, gadolinium (Gd)-based contrast agents (GBCAs) have long enjoyed an excellent safety record. In the past, there has been little motivation for, or emphasis on, dose reduction strategies. However, recent reports linking high-dose GBCA to the development of nephrogenic systemic fibrosis (NSF) [1–3] have raised concerns over the safety of CE-MRA. As a result, many investigators have focused attention on gadolinium dose reduction strategies [4, 5]. This chapter reviews some strategies to reduce contrast dose for CE-MRA applications and summarizes current applications and the authors’ clinical experience to date. It also highlights evolving techniques, which the authors feel are likely to enhance the future impact of CE-MRA.

The association of NSF with GBCA first appeared in the literature in 2006 [6]. While it is still speculative how GBCA can trigger NSF, renal function impairment is a prerequisite, and most cases have been associated with end-stage renal failure. The risk of NSF with Gd use is most likely dose dependent and may be related to the residence time of gadolinium within the body [7, 8]. The rate of renal Gd excretion is exponential such that the higher the injected dose, the faster the rate of elimination, but the longer it takes for the blood concentration to fall below a given threshold. The exposure to an extracellular Gd agent is proportional to the area under the blood concentration–time curve, which is in turn determined by the ratio of injected dose to glomerular filtration rate. In renal failure, the elimination rate constant for extracellular contrast agents falls proportionately to the degree of renal impairment. More than 90% of the injected dose of extracellular gadolinium agent is removed via the kidneys in normal individuals within 24 h. However, in patients with severe renal impairment, this time can be prolonged up to 7 days [9]. Although the exact cause–effect relationship between gadolinium dose and risk of NSF remains to be established, it seems logical to use the minimum effective dose for CE-MRA in at-risk patients. Fortunately, in many instances, dramatic dose reduction is possible with little if any penalty in diagnostic image quality [4, 5].

K. Nael, MD () • R. Saleh, MD Department of Radiological Sciences, David Geffen School of Medicine at University of California Los Angeles, 10945 Le Conte Avenue, Suite # 3371, Los Angeles, CA 90095-7206, USA e-mail: [email protected] G. Laub, PhD MR Division, MR R&D West, Siemens Healthcare USA, Pleasanton, CA, USA J.P. Finn, MD Department of Radiology, Ronald Reagan UCLA Medical Center, Los Angeles, CA, USA

Alternatives to Conventional CE-MRA To avoid all this discussion, one may consider noncontrast MRA techniques, such as time-of-flight, phase-contrast [10, 11], and bright blood steady-state free precession (SSFP) [12–14]. However, all of the noncontrast techniques are to some extent flow sensitive and are less robust and often less practical than CE-MRA. Also, the majority of noncontrast

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_7, © Springer Science+Business Media, LLC 2012

107

108

Fig. 7.1 (a–c): Selected images of timing run (a), time-resolved MRA (b), and high-spatial-resolution CE-MRA (c) of 39-year-old research subject with Tetralogy of Falot and Glenn shunt at 1.5 T. Patient has single functional ventricle (very large VSD) with pulmonary atresia. The right lung is perfused from the Glenn shunt (arrowhead in image b) and the left lung receives systemic blood supply. Timing run and time-

MRA techniques require longer scan times since multiple arterial and venous phases are necessary to complete the data acquisition. An alternative approach to noncontrast MRA is to use low Gd doses, which we discuss in detail below. Low-dose CE-MRA can be performed at both 1.5 and 3 T, although the trade-off in vascular signal-to-noise ratio (SNR) when reducing the contrast dose is less noticeable at higher field, such as 3.0 T. Relative to 1.5 T, the inherently higher SNR at 3.0 T [15, 16] can be used to reduce acquisition time, improve spatial resolution, or both [15, 17, 18]. In addition, since the longitudinal relaxation time (T1) of unenhanced blood increases with field strength [19], sensitivity to injected gadolinium agents for CE-MRA is heightened, which allows further dose reduction. Radiofrequency chain developments, such as the introduction of arrays of more sensitive imaging coils and multiple receiver channels, have further improved image quality and performance at both 1.5 and 3.0 T [20–22]. Combining large arrays of surface coils with multiple independent detectors [23, 24] allows collection of signals from several object regions in parallel, supporting parallel imaging and sparse k-space sampling methods [25–30]. As a result, improvements in SNR, field of view (FOV), or a combination of both are possible [23, 24, 31, 32]. Resulting improvement in SNR and signal reception can be used to reduce the contrast dose for CE-MRA applications if needed. Extended FOV imaging may be useful, where pathology extends beyond the coverage of a single body array coil; for example, thoraco-abdominal aortic aneurysms or aortic dissections (Fig. 7.1) may extend above and below the diaphragm.

K. Nael et al.

resolved MRA were acquired with injection of 2 mL half-strength Gd solution and 6 mL half-strength Gd solution, respectively. High-spatialresolution MRA was acquired with injection of 28 mL of half-strength Gd solution at the same injection rate of timing run (2 mL/s) in a 20-s breath-hold period (TR/TE: 2.28/0.95, FA = 30°). The high-spatialresolution MRA study was timed to show the systemic arteries

Fast Imaging Tools Time-resolved (TR) MRA can generate high-frame-rate 3D MR images, providing graphic visualization of the first pass of injected contrast agents. One of the major advantages of TR-MRA is the requirement for substantially less gadolinium in comparison to conventional CE-MRA [25, 26]. Recently, Lohan et al. described a carotid TR-MRA protocol as a screening tool using 1.5–3.0 cc of gadolinium [27]. We recently documented the advantage of low-dose TR-MRA to diagnose central venous occlusive disease among the growing population of patients with end-stage renal disease [28]. With a sensitivity similar to conventional MRA, TR-MRA has the potential for initial screening and diagnosis of a number of vascular diseases [28]. Although the specifics of protocol and contrast dose may vary depending on the clinical scenario, our TR-MRA protocols generally requires much lower contrast dose (on the order of 2–6 mL) in comparison to our conventional MRA protocols (8–15 mL).

Contrast Agent Type Recently, a variety of new GBCA have been investigated to enhance SNR and image quality. For example, blood pool contrast agents with increased T1 relaxivity (shorter T1), such as gadofosveset [29] and gadomer-17 [30], have been described. These blood pool agents have been shown to boost and prolong vascular SNR with promising results for

7

Low-Dose Contrast-Enhanced MR Angiography

venography and steady-state high-resolution imaging [33]. Gadofosveset has recently been approved by the FDA for aorto-iliac imaging and is marketed in North America as “Ablavar” (Lantheus Medical Imaging, Inc.). The introduction of macrocyclic Gd chelates, such as Gadobutrol, is also promising. As a result of their lower osmolality and viscosity, macrocyclics can be prepared in higher concentration (1.0 M formulation) than currently used ionic Gd chelates (0.5 M formulation) [34, 35]. This twofold increase in the Gd concentration can be used to inject a smaller and more concentrated volume when compared with a formulation of 0.5 M. Additionally, the macrocyclic Gd chelates are more stable than linear chelates and potentially less prone to dissociate free Gd ions [36–38]. Because most current theories implicate free Gd ions as a likely pathway for development of NSF, ultra-stable Gd formulations have theoretical advantages in reducing the risk of dissociation virtually to zero. However, only clinical experience defines the overall safety profiles of the various Gd formulations.

Contrast Timing and Injection Protocol An important factor for CE-MRA is the injection protocol, which may change with Gd dose. Ideally, the injection protocol would result in a sustained intravascular contrast peak, encompassing the data acquisition period. The peak concentration of Gd in the blood following intravenous injection is determined by the rate of injection (mmols Gd/s), and the duration over which the peak persists (the “plateau”) is determined by the duration of the injection. Therefore, a high, sustained blood concentration requires a fast injection rate over a sustained period, which means delivering a high total dose (mmols Gd). Injecting a smaller dose at the same rate shortens the bolus peak and it may not overlap the entire data acquisition window, potentially causing “low-pass” or “high-pass” filtering. Low-pass filtering occurs if the bolus peak is short but coincides with the central k-space lines. The resulting images are usually high in contrast, but may appear blurred due to loss of high spatial frequency information. High-pass filtering occurs if only the peripheral portion of k-space coincides with the bolus peak. The resulting images may be nondiagnostic or may even mimic vascular thrombosis due to artifactual signal loss in large vessels. To avoid these problems while decreasing dose, we preserve the original bolus duration by injecting contrast more slowly (mmols Gd/s) while settling for a lower mean and peak arterial concentration. In practice, we implement this by diluting the native gadolinium formulation at the time of administration so that the timing and infusion duration of the (diluted) contrast solution are identical to what they would be for an equal volume of the native (undiluted) gadolinium formulation. So, whereas the injection rate in mmols Gd/s is lower with reduced dose, the volume infusion rate in mL/s is identical

109

Fig. 7.2 (a, b): Volume rendered reconstructed images from two 3D high-spatial-resolution CE-MRA studies (TR/TE = 512/288, FA = 30°) at 1.5 T in two patients with severe pectus excavatum deformity. (a) 20-year-old male patient received 15 mL of Gd diluted to 30 mL with saline injected at 2 mL/s. (b) 16-year-old male patient received 30 mL of undiluted Gd contrast injected at 2 mL/s

with both dose regimens. The result is that the peak intravascular concentration of Gd is lower with the diluted solution, but occurs at the same time as with the original protocol. Therefore, the shape and duration of the concentration–time curves are identical. We preserve volume injection rate by dilution because it had been our anecdotal experience that with infusion rates less than 1 mL/s in adults bolus timing becomes unreliable due to the variable capacitance of the central veins. Figure 7.2a, b shows thoracic MRA studies in two patients with pectus excavatum deformity using two different contrast dose dilutions at the same volume injection rate.

Our Experience A wide range of image acquisition parameters and contrast dose and injection protocols for CE-MRA have been reported by various investigators and institutions. Below, we summarize our current approach for CE-MRA in several vascular territories.

Head and Neck We always perform an initial “4D” MRA with 2 mL Gd (diluted to 8 mL with saline) at a rate of 2 mL/s. This study serves as our “timing bolus,” which is used to plan the subsequent high-spatial-resolution CE-MRA. Using a combination of head, neck, and spine coils, we normally use an FOV of 450 mm (frequency) × 270 mm (phase). This extended FOV includes the aortic arch, the origins of the great vessels, the carotid arteries, and the intracranial vessels to the vertex in one single acquisition. Parallel imaging (GRAPPA) with acceleration factor of 4 is currently standard. Typically, we acquire 128 slices with near-isotropic resolution

110

K. Nael et al.

(voxel size = 0.8 × 0.7 × 0.8 mm) in a scan time of 22 s. We perform our protocol with breath holding, which minimizes respiratory motion in the upper thorax and aortic arch. For contrast injection at 3 T, a typical dose of 8 mL Gd is diluted with normal saline by a factor of four and the dilute solution is infused at 2 mL/s for a 15-s duration. For patients weighing 70 kg or less, this dose corresponds to 0.05 mmol/kg or less.

Chest Most of our chest MRA studies are performed at 1.5 T as they are generally combined with cardiac functional assessment, often in patients with adult congenital heart disease (Fig. 7.1). We generally perform a coronal TR-MRA using parallel imaging (GRAPPA × 3) and the TWIST sequence. Simultaneously, with injection of 6 mL of half-strength contrast at a rate of 3 mL/s, TR-MRA is acquired with a voxel size of 1.3 × 1.3 × 6 mm and temporal resolution of 1.5 s, during a 25-s breath holding. For high-spatial-resolution chest CE-MRA, we typically acquire 112 slices with voxel size = 1 × 1 × 1.2 mm in a scan time of 22 s. Parallel imaging (GRAPPA = 3) is used routinely. Patients must hold their breath during the scan to minimize respiratory motion artifact. A dose of 15 mL Gd is diluted with equal amount of normal saline at 1.5 T to create a 50% solution and injected at 2 mL/s, so the infusion period is 15 s (Figs. 7.2 and 7.3).

Abdomen–Pelvis For high-spatial-resolution abdominal CE-MRA at 3.0 T, we typically use about 12 mL Gd, diluted to quarter strength. Typically, we acquire 128 slices with near-isotropic resolution (voxel size = 0.9 × 0.8 × 0.9 mm, FOV is 500 × 300 mm) in a scan time of 22 s. Parallel imaging (GRAPPA = 3) is used. Patients must hold their breath during the scan to minimize motion artifact in the abdomen.

Lower Extremity Runoff At our institution, we acquire a three-station, dual-injection protocol with image acquisition performed first in the calves before the pelvis and thighs. A dual-injection protocol is used to assure that the calf station, often the most technically challenging, can be imaged without venous contamination [39, 40]. For our 1.5-T protocol, the total amount of contrast material injected is approximately 0.2 mmol/kg in comparison to a lower dose of 0.1 mmol/kg at 3.0 T. Again as mentioned earlier, we keep the injection rate of the contrast solution constant (approximately 1.5 mL/s infusion rate)

Fig. 7.3 (a–c): Volume rendered reconstructed images from 3D highspatial-resolution CE-MRA (TR/TE = 512/288, FA = 30°) with low contrast dose prescription at 1.5 T. 47-year-old female with a history of aortic arch repair was referred for evaluation of distal aortic arch aneurysm, mimicking pseudocoarctation. Note (arrow) the persistent narrowing of the arch at the proximal graft anastomosis to 1.1 cm in maximum diameter. 33 mL of half-strength Gd contrast was injected at 2 mL/s

among the various dose ranges by diluting the original formulation with normal saline. This allows maintaining a constant infusion period during image acquisition for corresponding stations. A 500-mm FOV is used for each station, covering a total length of 1,350 mm along the patient axis. There is typically a 10-cm overlap between the abdominal and thigh stations and a 5-cm overlap between the thigh and calf stations. We typically acquire 128 slices with a voxel size of 0.9 × 0.8 × 0.9 mm in a scan time of 24 s. Parallel imaging (GRAPPA = 4) is used. For the abdominal–pelvic station, patients are requested to hold their breath to minimize motion in the lower abdomen. In anticipation of potential asymmetry in contrast arrival in each calf, which can occur with vascular diseases, three sequential postcontrast MRA data sets are acquired routinely per calf station and two sets per thigh station. To eliminate stationary background signal, the precontrast images are used as masks and are subtracted from the CE-MRA images in the calves and thighs. At 3.0 T, we perform high-spatial-resolution CE-MRA of the entire lower extremity arterial tree with a contrast dose in the order of 0.08–0.1 mmol/kg in divided doses (Fig. 7.4). Our initial results suggested that low-dose protocols provide comparable image quality to high-dose runoff studies (Fig. 7.5)

7

Low-Dose Contrast-Enhanced MR Angiography

111

Fig. 7.4 (a–c): 3.0 T high-spatial-resolution MRA of 62-year-old male patient who presented with claudication. A total of 40 mL of undiluted Gd contrast was injected using a dual-injection protocol. (a) Abdominal station (TR/TE = 2.54/0.97, FA = 15°) shows a long stent in the right common iliac artery (note the signal loss). Right internal iliac artery is occluded. Left external iliac artery is occluded and internal iliac artery

is diffusely irregular. (b) Thigh station (TR/TE = 3.74/1.34, FA = 40°) shows right and left common femoral and superficial femoral arteries are occluded. Right and left profunda femoris are both stenotic. (c) Calves station (TR/TE = 3/1.15, FA = 17°) shows that popliteal arteries are reconstructed from collaterals through profunda femoris, bilaterally. Left posterior tibial artery shows mutifocal stenosis

Fig. 7.5 (a–c): 3.0 T high-spatial-resolution, lower extremity MRA in a 75-year-old female patient with several peripheral vascular disease. 20 mL Gd was diluted to 60 mL with saline. 20 mL of the solution was infused for imaging the calves and 33 mL for the combined thigh and abdominal stations. (a) Abdominal station (TR/TE = 2.54/0.97, FA = 15°) shows a long stent in the right common iliac artery (note the signal loss). Right internal iliac artery is occluded. Left external iliac artery is

occluded and internal iliac artery is diffusely irregular. (b) Thigh station (TR/TE = 3.74/1.34, FA = 40°) shows right and left common femoral and superficial femoral arteries are occluded. Right and left profunda femoris are both stenotic. (c) Calves station (TR/TE = 3/1.15, FA = 17°) shows that popliteal arteries are reconstructed from collaterals through profunda femoris, bilaterally. Left posterior tibial artery shows mutifocal stenosis

and have the potential to diminish sensitivity to contrast dose-dependent complications [4, 41]. At this point, we have been employing the above reduced dose strategies for over 3 years with entirely consistent results.

and abdomen and lower extremities, and in our practice the results have been uniformly positive such that this is now our routine clinical practice. To facilitate contrast dose reduction while maintaining image quality, various strategies can be employed, including the use of high magnetic field systems and sensitive receiver arrays with multiple RF channels. The increasing availability of contrast agents with higher relaxivity and persistence holds promise for more flexible protocols. Finally, TR-MRA has a significant role to play low-dose applications and continues to drive clinical and technical research.

Conclusion Today, low-dose CE-MRA with Gd dose in the range of 0.01–0.1 mmol/kg can be readily performed over a wide spectrum of vascular territories, including carotids, chest,

112

References 1. Sadowski EA, Bennett LK, Chan MR, et al. Nephrogenic systemic fibrosis: risk factors and incidence estimation. Radiology. 2007;243: 148–157. 2. Marckmann P, Skov L, Rossen K, et al. Nephrogenic systemic fibrosis: suspected causative role of gadodiamide used for contrastenhanced magnetic resonance imaging. J Am Soc Nephrol. 2006;17:2359–2362. 3. Clorius S, Technau K, Watter T, et al. Nephrogenic systemic fibrosis following exposure to gadolinium-containing contrast agent. Clin Nephrol. 2007;68:249–252. 4. Habibi R, Krishnam MS, Lohan DG, et al. High-spatial-resolution lower extremity MR angiography at 3.0 T: contrast agent dose comparison study. Radiology. 2008;248:680–692. 5. Tomasian A, Salamon A, Lohan DG, Jalili M, Villablanca JP, Finn JP. Supraaortic arteries: contrast material dose reduction at 3.0-T high-spatial-resolution MR angiography--feasibility study. Radiology. 2008;249:980–990. 6. Grobner T. Gadolinium--a specific trigger for the development of nephrogenic fibrosing dermopathy and nephrogenic systemic fibrosis? Nephrol Dial Transplant. 2006;21:1104–1108. 7. Perazella MA, Rodby RA. Gadolinium-induced nephrogenic systemic fibrosis in patients with kidney disease. Am J Med. 2007;120:561–562. 8. Collidge TA, Thomson PC, Mark PB, et al. Gadolinium-enhanced MR imaging and nephrogenic systemic fibrosis: retrospective study of a renal replacement therapy cohort. Radiology. 2007;245: 168–175. 9. Morcos SK, Thomsen HS, Webb JA. Dialysis and contrast media. Eur Radiol. 2002;12:3026–2030. 10. Prieto C, Uribe S, Razavi R, Atkinson D, Schaeffter T. 3D undersampled golden-radial phase encoding for DCE-MRA using inherently regularized iterative SENSE. Magn Reson Med. 2010;64: 514–526. 11. Kang CK, Kim SH, Lee H, Park CA, Kim YB, Cho ZH. Functional MR angiography using phase contrast imaging technique at 3T MRI. Neuroimage. 2010;50:1036–1043. 12. Wyttenbach R, Braghetti A, Wyss M, et al. Renal artery assessment with nonenhanced steady-state free precession versus contrastenhanced MR angiography. Radiology. 2007;245:186–195. 13. Francois CJ, Tuite D, Deshpande V, Jerecic R, Weale P, Carr JC. Unenhanced MR angiography of the thoracic aorta: initial clinical evaluation. AJR Am J Roentgenol. 2008;190:902–906. 14. Francois CJ, Tuite D, Deshpande V, Jerecic R, Weale P, Carr JC. Pulmonary vein imaging with unenhanced three-dimensional balanced steady-state free precession MR angiography: initial clinical evaluation. Radiology. 2009;250:932–939. 15. Campeau NG, Huston J 3rd, Bernstein M, Lin C, Gibbs GF. Magnetic resonance angiography at 3.0 Tesla: initial clinical experience. Top Magn Reson Imaging. 2001;12:183–204. 16. Willinek WA, Born M, Simon B, et al. Time-of-flight MR angiography: comparison of 3.0-T imaging and 1.5-T imaging--initial experience. Radiology. 2003;229:913–920. 17. Thulborn KR. Clinical rationale for very-high-field (3.0 Tesla) functional magnetic resonance imaging. Top Magn Reson Imaging. 1999;10:37–50. 18. Robitaille PM, Abduljalil AM, Kangarlu A. Ultra high resolution imaging of the human head at 8 tesla: 2K x 2K for Y2K. J Comput Assist Tomogr. 2000;24:2–8. 19. Rinck PA, Muller RN. Field strength and dose dependence of contrast enhancement by gadolinium-based MR contrast agents. Eur Radiol. 1999;9:998–1004. 20. Bodurka J, Ledden PJ, van Gelderen P, et al. Scalable multichannel MRI data acquisition system. Magn Reson Med. 2004;51: 165–171.

K. Nael et al. 21. King SB DG, Peterson D, Varosi S, Molyneaux DA. A comparison of 1, 4, and 8 channel phased array head coils at 1.5 T. Presented at: 9th Annual Meeting of ISMRM, Glasgow, Scotland, 2001. 22. de Zwart JA, Ledden PJ, van Gelderen P, Bodurka J, Chu R, Duyn JH. Signal-to-noise ratio and parallel imaging performance of a 16-channel receive-only brain coil array at 3.0 Tesla. Magn Reson Med. 2004;51:22–26. 23. Hayes CE, Hattes N, Roemer PB. Volume imaging with MR phased arrays. Magn Reson Med. 1991;18:309–319. 24. Roemer PB, Edelstein WA, Hayes CE,Souza SP, Mueller OM. The NMR phased array. Magn Reson Med. 1990;16:192–225. 25. Wu Y, Wieben O, Mistretta CA, Korosec FR. Evaluation of temporal and spatial characteristics of 2D HYPR processing using simulations. Magn Reson Med. 2008;59:1090–1098. 26. Frayne R, Grist TR, Swan JS, Peters DC, Korosec FR, Mistretta CR. 3D MR DSA: effects of injection protocol and image masking. J Magn Reson Imaging. 2000;12:476–487. 27. Lohan DG, Tomasian A, Saleh RS, Singhal A, Krishnam MS, Finn JP. Ultra-low-dose, time-resolved contrast-enhanced magnetic resonance angiography of the carotid arteries at 3.0 tesla. Invest Radiol. 2009;44:207–217. 28. Nael K, Krishnam MS, Ruehm SG, Michaely HJ, Laub G, Finn JP. Time-resolved MR angiography in the evaluation of central thoracic venous occlusive disease. AJR Am J Roentgenol. 2009;192: 1731–1738. 29. Port M, Corot C, Violas X, Robert P, Raynal I, Gagneur G. How to compare the efficiency of albumin-bound and nonalbumin-bound contrast agents in vivo: the concept of dynamic relaxivity. Invest Radiol. 2005;40:565–573. 30. Dong Q, Hurst DR, Weinmann HJ, Chenevert TL, Londy FJ, Prince MR. Magnetic resonance angiography with gadomer-17. An animal study original investigation. Invest Radiol. 1998;33:699–708. 31. Hayes CE, Roemer PB. Noise correlations in data simultaneously acquired from multiple surface coil arrays. Magn Reson Med. 1990;16:181–191. 32. Constantinides CD, Westgate CR, O’Dell WG, Zerhouni EA, McVeigh ER. A phased array coil for human cardiac imaging. Magn Reson Med. 1995;34:92–98. 33. Nael K, Saleh R, Nyborg GK, et al. Pulmonary MR perfusion at 3.0 Tesla using a blood pool contrast agent: initial results in a swine model. J Magn Reson Imaging. 2007;25:66–72. 34. Olukotun AY, Parker JR, Meeks MJ, Lucas MA, Fowler DR, Lucas TR. Safety of gadoteridol injection: U.S. clinical trial experience. J Magn Reson Imaging. 1995;5:17–25. 35. Herborn CU, Lauenstein TC,Ruehm SG, Bosk S, Debatin JF, Goyen M. Intraindividual comparison of gadopentetate dimeglumine, gadobenate dimeglumine, and gadobutrol for pelvic 3D magnetic resonance angiography. Invest Radiol. 2003;38:27–33. 36. Idee JM, Port M, Medina C, et al. Possible involvement of gadolinium chelates in the pathophysiology of nephrogenic systemic fibrosis: a critical review. Toxicology. 2008;248:77–88. 37. Morcos SK. Extracellular gadolinium contrast agents: differences in stability. Eur J Radiol. 2008;66:175–179. 38. Frenzel T, Lengsfeld P, Schirmer H, Hutter J, Weinmann HJ. Stability of gadolinium-based magnetic resonance imaging contrast agents in human serum at 37 degrees C. Invest Radiol. 2008;43:817–828. 39. Morasch MD, Collins J, Pereles FS, et al. Lower extremity steppingtable magnetic resonance angiography with multilevel contrast timing and segmented contrast infusion. J Vasc Surg. 2003;37:62–71. 40. Pereles FS, Collins JD, Carr JC, et al. Accuracy of stepping-table lower extremity MR angiography with dual-level bolus timing and separate calf acquisition: hybrid peripheral MR angiography. Radiology. 2006;240:283–290. 41. Nael K, Krishnam N, Nael A, Ton A, Ruehm SG, Finn JP. Peripheral contrast-enhanced MR angiography at 3.0T, improved spatial resolution and low dose contrast: initial clinical experience. Eur Radiol. 2008;18:2893–2900.

8

Vessel Wall Imaging Techniques Rui Li, Niranjan Balu, and Chun Yuan

Introduction Atherosclerosis-related cardiovascular disease (CVD) is the leading cause of morbidity and mortality all over the world [1]. Atherosclerosis can occur in the aorta, carotid, and coronary arteries causing stroke, myocardial infarction, and sudden death without prior symptoms. The culprit lesions are mainly caused from the rupture of the plaque surface which then triggers thrombotic complications. Thus, a challenge for imaging is to identify vulnerable plaques that are likely to rupture. Extensive pathological studies link plaque morphology and tissue composition as major contributors to plaque instability [2]. Thus, raises the need for vessel wall imaging. Noninvasive magnetic resonance, through its high-resolution and multicontrast imaging has proven its ability to detect vulnerable plaque features. Magnetic resonance angiography (MRA), discussed in previous chapters, is focused on the depiction of vessel lumen narrowing, which is a clinical need but an indirect manifestation of the pathology of the vessel wall. In contrast, vessel wall MRI is intended to directly visualize the vessel wall, and especially the composition of the wall. A clear delineation of the vessel wall entails the use of effective measures to suppress signals from flowing blood, usually referred to as black-blood (BB) techniques. With BB techniques, the vessel lumen is seen as black, giving good contrast to the bright vessel wall. By combining BB techniques with the different contrast mechanisms, characterization of the components of the atherosclerotic plaque is possible. In this chapter, we discuss the various techniques for black-blood imaging and briefly discuss their applications in vessel wall imaging. We focus our description on the carotid artery because atherosclerosis in the bifurcation region of the carotid is a cause of stroke and also because the technical R. Li, PhD () • N. Balu, PhD • C. Yuan, PhD Department of Radiology, University of Washington, Seattle, WA, USA e-mail: [email protected]

challenges and considerations for the carotids can be extended to other large arteries.

Black-Blood Techniques In order to identify the morphology and the components in plaque, signals arising from flowing blood must be suppressed. This allows the inner wall of the artery to be distinguished from the blood pool. In this section, several black-blood approaches are discussed including: saturation band; double inversion recovery (DIR); quadruple inversion recovery (QIR); and diffusion preparation.

Flow and Relaxation Properties of Blood Flow and relaxation times are the two major differences between blood and static tissues which can be exploited to suppress the signal from blood while retaining the signal from the vessel wall. For the purposes of MR imaging, blood flow can be separated into two models: laminar flow and turbulent flow. Laminar flow is the normal blood flow pattern throughout most vascular beds. The blood velocity is distributed concentrically with the highest velocity in the center of vessel. When blood moves at high speeds or the lumen shape becomes irregular, blood is no longer laminar with smooth velocity transitions between adjacent layers, but instead becomes chaotic or turbulent. Turbulent flow often occurs at the ascending aorta or the bifurcation of the carotid artery. It is a challenge for black-blood imaging to suppress turbulent flow, especially where blood reenters the region it flowed over, for example in carotid bulb. Blood flow velocities also vary depending on the vascular bed and the cardiac cycle. During systole, blood velocity in normal carotid arteries is about 100 cm/s. At diastole, it will decrease to less than 40 cm/s in the normal carotid artery. Stenosis of the artery also causes increased flow velocities in excess of 300 cm/s in

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_8, © Springer Science+Business Media, LLC 2012

113

114

a 90% stenotic carotid artery. Almost all blood suppression techniques depend on flow properties of blood. Therefore, the timing and other parameters of the saturation band, DIR, and gradient moments of diffusion-related methods must be optimized according to the blood velocity. Blood with slow or turbulent flow have velocities that fall out of the range to which the sequence was optimized, remaining bright, resulting in poor visualization of the arterial wall. The knowledge of blood T1 relaxation time is critical for sequence design and parameter optimization in black-blood preparations, such as inversion recovery. T1 values of 1,664 ± 14 ms in arterial and 1,584 ± 5 ms in venous blood are reported at 3 T [3], which are longer than the T1 values of adipose tissue (240–250 ms) and muscle (860–900 ms). Interestingly, the same research group showed a linear relationship between 1/T1 and hematocrit. Another issue to be kept in mind is that in general T1 values are 30% shorter in a 1.5 T magnetic field, however, still longer than in other tissues.

Blood Suppression with Spin Echo Spin echo sequences, including fast spin echo sequences, provide a black-blood appearance of blood vessels by the requirement of “through flow.” Only spins exposed to both the excitation signal and a refocusing RF pulse at the same location can produce signals. In spin echo imaging, blood excited by the RF pulse flows out of the imaging plane and cannot be refocused after the TE delay time resulting in blood suppression. However, this method is insufficient for complete blood suppression with short echo times or slow blood flow. Blood suppression results can be improved with longer echo times and a thinner slice thickness. Considering that the time between the 90° and 180° pulse is TE/2 in a single-echo spin echo pulse sequence, excited blood flows out of the imaging slice (thickness Dz) completely when the velocity perpendicular to the slice exceeds v = 2Dz/TE. Assuming that TE = 40 ms with a slice thickness of 5 mm, blood is suppressed when the perpendicular velocity is larger than 25 cm/s a common finding in the major arteries of healthy human subjects. However, to use such a long TE in T1-weighted sequences is not realistic. Intrinsic spin echo blood suppression often fails in slow and turbulent flow conditions, such as the triphasic flow seen in the femoral artery and recirculating flow distal to the carotid bulb. Therefore, spin echo acquisitions are normally used in combination with the other flow suppression methods for vessel wall imaging.

Saturation Band A saturation band placed upstream and parallel to the imaging plane suppresses the signal from in-flowing blood. Blood, which flows into the imaging slice, is saturated by the

R. Li et al.

Fig. 8.1 DIR prepulse diagram

saturation RF pulse and spoiled by the following gradient. No magnetization in the +Z direction from blood will be tipped down into the transverse plane when the exciting RF pulse is imposed. The gap between saturation band and imaging plane remains constant for all slices to ensure uniform flow suppression for different slices, called traveling saturation band. The saturation band should be thick enough to ensure that all blood flowing into the imaging plane will be saturated. Bearing in mind the longitudinal relaxation of blood magnetization in the time it takes for saturated blood to flow into the imaging plane, saturation RF pulses with flip angles larger than 90° should be used. Saturation band thickness, distance from imaging plane, delay time, and saturation RF pulse flip angle can be adjusted according to different flow conditions. This technique is commonly used in BB MRI because of its simplicity and relative low specific absorption rate (SAR). However, blood with slow and turbulent flow velocities can degrade the blood suppression effect since it cannot flow into the imaging plane in the short delay time after the saturation band.

Double Inversion Recovery As its name implies, DIR employs two inversion RF pulses [4]. Shown in Fig. 8.1, the first pulse is a nonselective 180° RF pulse, which inverts all magnetization in the effective volume of the excitation RF coil. The second is a selective 180° RF pulse, which inverts only the magnetization in the imaging plane. It restores the magnetization in the imaging slice to the original state but keeps the magnetization in other regions inverted. After a selected delay time TI, blood with inverted magnetization flows into the imaging plane as it realigns itself along the main field. The longitudinal component of the magnetization follows a T1-mediated regrowth curve. If acquisition of the MR image is timed to occur when the longitudinal magnetization passes through zero, no blood magnetization in Z direction can be excited by the host sequence, so the lumen appears black in the acquired images.

8

Vessel Wall Imaging Techniques

115

This method is similar to the short TI inversion recovery (STIR) method used to suppress fat. The optimum delay time TI depends on the T1 value of blood (8.1): 2 æ ö TI = T1,blood ´ In ç . (8.1) - TR / T1,blood ÷ 1 + e è ø When TR is greater than 3 × T1, (8.1) can be simplified to the following (8.2): TI = T1,blood ´ In(2).

(8.2)

The general host sequences (such as SE, FSE) are applied after TI. Since T1 of blood is above 1,000 ms, the delay time TI is long enough to ensure flow of inverted blood into the imaging plane. A hard pulse or an adiabatic inversion pulse can be chosen as the nonselective RF pulse in the DIR prepulse, and a sliceselective adiabatic pulse, such as a hyperbolic secant, can be employed as the selective RF pulse to ensure that magnetization flips back into the whole imaging plane. The slice thickness of the selective RF pulse should be slightly larger than the actual imaging plane to prevent the signal loss caused by an imperfect slice profile. DIR is much more effective than saturation band for creating flow voids for slow flow because of the prolonged period for inflow of inverted protons [4]. For blood to produce signal, it must be affected by the second, slice-selective inversion RF pulse and remain within the imaging plane for entire TI period. Similar to saturation band, DIR flow suppression efficiency is still dependent on the flow direction because it is assumed all the blood in the imaging plane is replaced by the inflow of inverted blood. DIR can suppress only through-plane flow, so only cross-sectional black-blood images can be acquired using a DIR prepulse. Another intrinsic drawback of DIR prepulse is that DIR requires a single slice acquisition, which results in long scan time for the T2- and proton density-weighted images that requires long TR. Although DIR has several disadvantages, it is still used on most state-of-art scanners as the main black-blood technique.

Improvement of DIR One method to solve DIR’s single slice acquisition problem is to apply multiselective 180° RF pulses, each corresponding to a different slice location during a single repetition time [5]. This configuration creates an unequal situation for the magnetization of different slices, as the delay between nonselective inversion RF pulse and selective inversion RF pulse depends on the actual slice number and host sequence length. Another strategy, named multislice DIR (MDIR), replaces a

Fig. 8.2 QIR and SFQIR prepulse diagram

slice-selective inversion RF pulse with a thicker slabselective inversion, which is applied to the entire set of sliced to be imaged in one TR, and employs a modified DIR prepulse before each slice readout [6]. A shorter TI should be chosen according to the acquisition time of a single slice and the postacquisition time after each slice. This MDIR acquisition improves the efficiency of image acquisition time, at the expense of blood suppression, due to the slice increased slice thickness. The use of T1 shortening contrast agents can improve the detection of neovasculature and inflammation in atherosclerotic plaques [7]. After an injection of MRI-based contrast agent, the T1 of blood is typically under 300 ms. The challenge is to maintain black-blood contrast between lumen and vessel walls despite considerable shortening of T1 in blood after contrast injection. Obviously, using the same TI in DIR will not suppress the blood signal before and after contrast injection because blood T1 changes. A technique that employs the QIR preparative pulse sequence shown in Fig. 8.2a can efficiently suppress flow signals with T1 in a range 200–1,200 ms [8]. Time intervals for prepulses can be calculated using an algorithm based on minimization of the variation of a signal equation over an entire range of T1 occurring in blood before and after contrast agent injection. The time inefficiency caused by the requirement that QIR to be acquired as a single slice scan is the main drawback of this approach. Another version of QIR named small-FOV quadruple inversion recovery (SFQIR) [9] substitutes the nonselective inversion RF pulse to the slab-selective RF pulse, which is orthogonal to the imaging plane with the thickness equal to the FOV size in the phase-encoding direction shown in Fig. 8.2b. Each double inversion results in the reinversion of magnetization in the central part of the FOV while leaving

116

R. Li et al.

the magnetization of outer areas of the FOV and inflowing blood inverted. This method can suppress the signal from the outer volume and inflowing blood simultaneously by optimizing two blood inversion times (TI1b, TI2b) and two outer area inversion times (TI1o, TI2o). Since the imaging time is mainly dependent on the number of phase encoding steps, the scan time can be shortened dramatically. However, SNR, which is reduces in proportion to the square root of the imaging time, should be considered when reducing scan time. The method has been tested on aorta and carotid arteries to show the ability to eliminate motion and flow artifacts. Although QIR and SFQIR suppress the flowing in blood with large range of T1, they inherit the limitations of DIR. Therefore, time efficiency and imaging plane should be carefully considered.

pulse. Since the original meaning of “diffusion preparation” refers to molecular diffusion and not blood motion, some researchers suggest using motion-sensitized driven-equilibrium (MSDE) instead [10]. The moment of flow-sensitizing gradient pairs should be optimized according to the flow speed and direction of the target vessel. MSDE prepulse is time efficient for two reasons (1) The host sequence can be executed immediately after the prepulse because MSDE does not depend on blood T1; (2) Multiple slices can be acquired in one TR. In contrast to DIR, the blood suppression effect is only related to the direction of gradient pair, so it can suppress in-plane flow by applying the gradient at the in-plane direction. The main drawback of MSDE is that it brings an increased T2- and diffusion-weighting into the acquired images. The advantages and disadvantages of these BB techniques are briefly concluded in Table 8.1.

Diffusion Preparation Diffusion-weighted magnetization preparation adds three nonselective RF pulses with flip angles 90°–180°–90° and symmetric gradients around the 180° pulse shown in Fig. 8.3. The flow-sensitizing gradient pair introduces the phase dispersion among moving spins while the magnetization of static tissue is fully refocused and then restored by a 90° flip-back

Contrast Weightings Currently, MRI contrast can be used for vessel wall imaging in three different ways (1) multicontrast techniques used to identify key structure, such as the lipid-rich necrotic core (LRNC), fibrous cap, and intraplaque hemorrhage (IPH); (2) commercially available nonspecific contrast agents to increase the accuracy of plaque component identification and evaluate vessel wall permeability caused by inflammation and neovasculature; (3) molecular imaging examining biochemical and/or cellular targets that indicate plaque vulnerability. Molecular imaging is briefly reviewed in future directions section while the multicontrast technique and general contrast agent enhancement are detailed below.

Multicontrast Techniques

Fig. 8.3 MSDE prepulse diagram

Based on histological analysis of carotid endarterectomy specimens, the atherosclerotic plaque is a complex structure, containing such components as calcium, fibrous tissue, IPH,

Table 8.1 Advantages and disadvantages of BB techniques BB techniques Saturation band DIR

Advantage Simplicity, low RF energy deposition Effective for slow flow

MDIR QIR

Multislice acquisition, shorter TI than DIR Insensitive to large range of T1 before and after contrast agent injection Effective to different flow velocities, time efficient, valid for in-plane flow

MSDE

Disadvantage Susceptible to slow flow and turbulent flow, not effective to in plane flow Only useful for through-plane flow, single slice acquisition strategy, long scan time Shorter scan time than DIR, but sensitive to slow flow Same as DIR Certain amount of T2 and diffusion weighting

8

Vessel Wall Imaging Techniques

117

Table 8.2 MRI protocols used for the carotid at 1.5 T (Reprinted with permission from Yuan et al. [12]) Technique TR (ms) TE (ms) NEX FOV (mm) Matrix Slice Num SliceTH (mm) ScanTime (min)

T1WI 2D FSE, QIR 800 11 2 160 × 120 256 × 192 12 2 ~7.5

PDWI 2D FSE, MDIR 2,400 9 2 160 × 120 256 × 192 12 2 ~4

T2WI 2D FSE, MDIR 3,000 50 2 160 × 120 256 × 192 12 2 ~4

3D TOF 3D SPGR 23 3 2 160 × 120 256 × 192 24 1 ~4

LRNC, thrombus, loose matrix and dense fibrous matrix. An MRI image with single MRI contrast weighting cannot distinguish between different components of the atherosclerotic plaque. By taking the advantage of intrinsic differences in proton density and T1, T2 relaxation times multicontrast MRI can identify atherosclerotic plaque components. The MR properties of plaque components were determined directly from ex vivo carotid endarterectomy (CEA) specimens [11]. Comparison to histology from CEA specimens was also used as the gold standard for validation of in vivo MR identification of plaque components on images acquired prior to the CEA. The T1-weighted image (T1WI), Proton density-weighted image (PDWI), and T2-weighted image (T2WI) are used to distinguish plaque components, such as LRNC, fibrous cap, loose matrix, IPH, and calcium. The gradient echo based bright blood sequence (TOF) is helpful in the discrimination of lumen morphology and composition in plaques with ulceration and IPH, therefore it is also integrated into the protocols. Typical parameters for a 1.5 T scanner are shown in Table 8.2 [12]. A relatively shorter echo time is proposed in the T2W image sequence because of a shorter T2 value in plaque components than that of other tissues. Other investigators have used similar criteria for differentiating plaque components in ex vivo and in vivo experiments, shown in Table 8.3 [13]. Multicontrast images and corresponding histology results are shown in Fig. 8.4. Two more issues should be carefully considered when determining multicontrast criteria. First, no “pure” components exist, that is, all biochemical compositions, such as free cholesterol, cholesterol ester, fibrin, collagen, red blood cells, and calcium, are mixed together to some extent. Second, components of the plaque in different stages of plaque progression vary greatly. For example, fresh hemorrhage (less than 1 week), recent hemorrhage (1–6 weeks)

Table 8.3 Tissue classification criteria (Reprinted with permission from Saam et al. [13]) LRNC with No or little hemorrhage Fresh hemorrhage Recent hemorrhage Calcification Loose matrix Dense (fibrous) tissue

TOF

T1WI

PDWI

T2WI

0 + + − 0 −

0/+ + + − −/0 0

0/+ −/0 + − + 0

−/0 −/0 + − + 0

The classification into the subgroups is based on the signal intensities relative to adjacent muscle: + hyperintense, 0 isointense, − hypointense

and old hemorrhage (more than 6 weeks) have different signal characteristics on MRI [14]. Other specific sequences, such as MRDTI [15] or MP-RAGE with heavy T1 weightings, were proposed to detect IPH. Hemorrhage appears very bright on these images. Diffusion-weighted images have also been studied to identify LRNC [16].

Contrast Enhancement Techniques Commercially available MRI contrast agents, such as gadopentic acid (Gd-DTPA), can be used to detect plaque components [7]. The imaging methods relative to contrast agents can be divided into two categories: pre- vs. postcontrast enhancement T1-weighted images (CE-MRI), and dynamic contrast enhanced (DCE) images. Since the T1 value of blood changes greatly before and after contrast administration, T1 insensitive flow suppression methods, such as QIR, should be used for CE-MRI. Plaque components, especially fibrous cap and LRNC, can be identified by the signal intensity changes of T1-weighted images before and after contrast injection acquired by the same sequence, shown in Table 8.4 [7]. Quantitative measurement of the fibrous cap and LRNC is feasible using in vivo highresolution CE-MRI shown in Fig. 8.5 [17]. DCE imaging inspects the intensity change of the plaque at different time points after the contrast agent is injected. Usually, a T1-weighted gradient echo sequence is employed to acquire the images in a very short time. For example, a cross-sectional two-dimensional (2D) spoiled gradientrecalled-echo sequence (TR/TE: 100/3.5 ms, FA: 60°, FOV: 16 cm × 12 cm, Matrix: 256 × 144) acquired 7 slices in 15 s, and the sequence can be repeated 10 times to acquire the dynamic images [18]. The parameters of the kinetic model,

118

R. Li et al.

Fig. 8.4 An example of the histological validation of MRI at four consecutive locations spanning the bifurcation. Multiple histological sections (at 0.5–1.0 mm separation) generally correspond to each 2 mm thick image. Contours have been drawn for lumen (red), outer wall

Table 8.4 Percent increase in signal intensity (Mean ± SD) of the three plaque tissue types (fibrous tissue, necrotic core, and calcification) and of the adjacent, nonarterial sternocleidomastoid muscle (Reprinted with permission from Yuan et al. [7]) Tissue type Fibrous tissue Necrotic core Calcification Sternocleidomastoid muscle

Signal intensity increase 79.5 ± 29.1 28.8 ± 20.1 47.4 ± 39.6 21.4 ± 10.8

Number of ROIs 119 39 20 48

such as the plasma volume (Vp) and transfer constant (K trans), are calculated from the dynamic signal curve during contrast injection. Inflammation [18], neovasculature volume [19], and adventitial vasa vasorum [20] in carotid atherosclerosis were found to correlate with these parameters.

(cyan), lipid-rich/necrotic core (yellow), calcification (black), loose fibrous matrix (pink/white), and hemorrhage (orange) (Reprinted with permission from Saam et al. [13])

SNR-Related Issues SNR optimization is a challenge for black-blood vessel wall imaging because thin slice thickness, high in-plane resolution, and short scan times are competing imaging parameters. RF coils and imaging parameters must be carefully considered to improve SNR. Surface coils should be used to increase SNR and must be matched to the anatomy. A general principle for the coil selection is that the coil with the smallest diameter and with the closest fit to the imaged anatomy should be used because the SNR is inversely proportional to the diameter of the coil and the distance between the coil and the imaging object. Neck coils or head–neck combined coils can be used for carotid arteries. Since the carotid arteries are close to surface, another choice is a dedicated carotid surface coil shown in Fig. 8.6 that can be placed as close as possible to the carotid arteries. Some home-made and commercial versions of this design are available for various scanner platforms [21].

Practical Considerations Field Strength Besides the general techniques used for black-blood imaging, several issues, such as SNR, field strength, fat suppression, motion artifacts, localization, and image processing, must be considered.

Predictably, higher magnetic field strengths provide higher signal and contrast to noise ratio. Based on a recent study that compares carotid plaque images acquired by 1.5 T and

8

Vessel Wall Imaging Techniques

Fig. 8.5 Corresponding pre- and postcontrast MR and histological images, showing the delineation of the fibrous cap (FC; green contour) and the lipid-rich necrotic core (LRNC; yellow contour) in the left carotid artery. (a) TOF; (b) Precontrast T1W; (c) PDW; (d) T2W; and (e and f) T1W with contrast. f illustrates the measurement method for the length of the FC (orange line a). The FC shows strong enhancement. The histology image (g) shows a matched section with green and yellow contours. High-power photomicrograph shows necrotic debris

119

and cholesterol clefts taken from an area in the LRNC (h) with no enhancement and loose matrix (LM) taken from an area in the FC (I) with corresponding strong enhancement in postcontrast MR images e and f. In d, the T2W signal intensity of the LRNC is heterogeneous, and the border between the LRNC and FC is unclear. Calcification is visible on all images (arrowhead). Asterisk indicates lumen; JV jugular vein, H&E hematoxylin-eosin staining (Reprinted with permission from Jiangming et al. [17])

considered is SAR increase with field strength. However, the use of 3.0 T scanners in vessel wall imaging is increasing due to improved SNR and CNR.

Fat Suppression Since vessel wall signals may be affected by the bright intensity from subcutaneous fat, a spectral selective saturation RF pulse is usually employed to suppress the fat signal. The fat saturation RF pulse has little effect on plaque contrast because the composition of plaque fat is primarily cholesterol not the triglyceride that exists in subcutaneous fat.

Motion Artifact Reduction

Fig. 8.6 Dedicated carotid surface coil

3.0 T scanners [22], wall SNR and lumen/wall CNR increased by 1.5-fold for T1WI and 1.7/1.8-fold for PDWI/ T2WI. The imaging criteria at 1.5 T for carotid artery wall interpretation are also applicable at 3.0 T. Susceptibility artifacts caused by calcification and paramagnetic ferric iron in hemorrhage may affect quantification and detection, especially for higher field application. Another issue to be

Both periodic motion, such as respiratory or cardiac motion and nonperiodic motion, such as swallowing [23], can affect carotid plaque imaging. While navigator gating and cardiac triggering can help with periodic motion, special strategies are required for swallowing compensation. Navigator motion detection of the tongue and pharyngeal wall only accept data during the period when these structures are relatively motionless. Placing external coils on the larynx to detect the swallowing motion has also been proposed [24]; however, such techniques are still in the experimental stages. Applying phase encoding in an anterior and posterior direction may be helpful in preventing artifacts from overlapping with the

120

R. Li et al.

Vessel Wall Imaging in Other Vascular Beds The same techniques are applicable to other vascular beds with appropriate adjustments for local conditions, such as flow velocities, need for cardiac or respiratory gating and motion artifacts. In particular, arteries such as coronaries, aorta and to a much lesser extent the carotids and peripheral arteries benefit from cardiac gating. Arteries within the thorax (coronaries and aorta) require an additional respiratory gating.

Image Processing

Fig. 8.7 The anatomy of carotid arteries on oblique 2D black-blood MRA images processed by multiplanar reformation (MPR, left) and minimal-intensity projection (MinIP, right). Images were obtained from the left side of a patient with moderate atherosclerotic disease. Abbreviations: CCA common carotid artery, ICA internal carotid artery, ECA external carotid artery, JV jugular vein, P plaque. Reformatted view (left) shows a fragment of plaque (P), whereas minimal intensity projection (right) confirms that the lumen is unobstructed (Reprinted with permission from Chun et al. [12])

region of interest. Saturation bands to suppress signals from the throat may also reduce motion artifacts.

Localization Precise localization of arteries is difficult due to their inherent tortuosity and variable location. However, it is extremely important to find a point of reference and to align the center slice of the set of 2D slices. This allows the potential for follow-up studies. As an example the carotid artery localization procedure is described below [12]. In order to acquire cross-sectional black-blood images of the carotid artery, the carotid bifurcation is the most convenient landmark for localization. The precise position of the bifurcation requires that a routine scout images and 2D TOF images are acquired to confirm center of the bifurcation. Obliquely oriented black-blood images are then acquired to cover the internal and external carotid artery, as shown in Fig. 8.7. A three-point localization method is useful to find the oblique plane. The center slice for the following multicontrast imaging sequences is placed at the bifurcation. The types of multicontrast imaging sequences can be selected according to the aim of the study. A T1WI is enough for plaque burden evaluation which only measures the lumen and vessel wall size. Multicontrast T1WI, T2WI, PDWI, and dedicated IPH detection protocol, such as MP-RAGE, should be used for the characterization of the plaque components. DCE and pre- and postcontrast T1WI are added to the protocol when using a contrast agent.

The importance of quantification in evaluating plaque features by MRI suggests a role for image postprocessing techniques. Automated measurement techniques can reduce analysis time, reduce reader-dependent bias, and improve measurement reproducibility [25]. Lumen and outer wall detection, multicontrast registration, plaque segmentation, plaque measurement, and display are the most popular questions in plaque image postprocessing. Many image processing techniques, such as snake, Bayes classification, and 3D visualization, can be utilized for more precise analysis. A customer-designed software platform called the computer-aided system for cardiovascular disease evaluation (CASCADE) was developed, including the functions of boundary detection, registration of multiple contrast weightings, segmentation of internal plaque components, and a three-dimensional (3D) display of the results. CASCADE was found to provide quantitative information for the sensitivity and specificity of detecting components in plaque that was similar to manual outlining [25].

Applications Vessel Morphology Black-blood MRI is an effective tool used to analyze vessel wall morphology. Interestingly, BB MRI is sometimes referred to as black-blood angiography (BB MRA) because it provides lumen visualization which is similar to bright blood MRA. Lumen area measurements by BB MRI correlate well with those by contrast-enhanced (CE) MRA (r = 0.77) showing that they provide similar kinds of lumen information [26]. In contrast to bright blood MRA, since blood is suppressed, BB MRI is not biased by poor luminal area determination resulting from flow dephasing. Accordingly, lumen area measurements are larger on BB MRI compared to CE-MRA [26]. However, the improvement in stenosis measurements by BB MRI has to be balanced against the considerably longer time to image the same span of vasculature.

8

Vessel Wall Imaging Techniques

121

Fig. 8.8 Transverse carotid plaque images and a maximum intensity projection (MIP) image of the left carotid artery were obtained from a 60-year-old woman. On the left side, MIP of the CE-MRA identifies a 0% stenosis at the left CCA. The horizontal lines indicate the level of the transverse carotid plaque images as shown on the right side. On the transverse

images, a surface disruption (white arrow) is noted as hyperintense on the TOF and hypointense on the pre- and postcontrast T1-weighted, T2-weighted, and proton-attenuation images. This surface disruption is not identified on CE-MRA. The asterisk indicates the lumen of either the CCA or ICA (Reprinted with permission from Dong et al. [27])

Another important aspect of vessel wall imaging is to identify the diseased artery that has not caused luminal narrowing. It has been shown that newly formed atherosclerotic lesions can be occult by luminography due to the phenomenon of arterial remodeling whereby the outer wall of the vessel expands to accommodate plaque growth so there is no overt stenosis. Only in later stages does this outward remodeling fail and luminal compromise occur. Li et al. showed that atherosclerotic lesions were present in carotids, where no luminal narrowing is observed on MR angiography, as shown in Fig. 8.8 [27]. Furthermore, atherosclerotic plaque was present in carotids with multiple plaque components. Identification and quantification of plaque compositional characteristics, such as lipid rich necrotic core, calcification, IPH, and fibrous cap rupture, are important in risk stratification. Vessel wall imaging can identify the true extent of plaque in the arterial wall, especially in early disease while there is a higher chance of reversing the disease process with adequate therapeutic intervention.

thin-walled plaques are referred to as “vulnerable plaques.” Ischemic events secondary to thrombi or emboli from ruptured plaques are manifested as stroke, myocardial ischemia, or gangrene depending on the vascular bed involved. Identification and treatment of vulnerable plaques can prevent these catastrophic events and MR vessel wall imaging currently offers the best prospect for vulnerable plaque characterization. Knowledge about plaque vulnerability has hitherto been mainly through histological studies [28]. With the advent of carotid vessel wall imaging, plaque components, such as thin fibrous caps, IPH, and a large necrotic core, have been identified as features characteristic of vulnerable plaque. One utility of MRI, especially by combining both bright and black-blood, is to determine the integrity of the fibrous cap. Identification of thin, thick, and ruptured fibrous caps on 3D time-of-flight images [29] showed a good agreement with histology [30, 31]. Gadolinium contrast-enhancement can also allow quantification of fibrous cap length and area [31]. Yuan et al. showed that fibrous cap status detected on MRI can distinguish between symptomatic and asymptomatic subjects. Subjects with a ruptured fibrous cap on MRI were 23 times more likely to have had a recent TIA or stroke compared to subjects with thick fibrous caps [32]. These studies show that the identification of disrupted or thin fibrous cap is a key characteristic of a vulnerable plaque. A case showing

Vulnerable Plaque Imaging Atherosclerotic plaques with thin fibrous caps are at risk of rupture and likely to cause thromboembolic events by exposing the thrombogenic core to the bloodstream. These

122

R. Li et al.

Fig. 8.9 Plaque progression subsequent to fibrous cap rupture: Baseline images (top panel) demonstrate plaque with LRNC (arrowheads) and minimal lumen narrowing. There is an identifiable band of tissue separating the LRNC (white arrowheads, hypointense signal intensity on T2) from the lumen on T2, which is consistent with the presence of a thick fibrous cap. At 3 years (bottom panel), there is an identifiable surface disruption (white arrow) on both axial images and the longitudinal scout image with visible increase of plaque especially noticeable in column D. The dashed line on the longitudinal image represents the imaging location of the axial image in column B. Asterisk indicates the lumen; black arrowhead indicates calcification (Reprinted with permission from Underhill et al. [33])

an increase in plaque size following fibrous cap rupture is shown in Fig. 8.9 [33]. The presence of IPH signifies an advanced plaque and can be identified by MRI by its bright signal on T1-weighted images, such as MP-RAGE, precontrast T1 and TOF. MRIdetected IPH has been shown to be predictive of neurological symptoms [34]. Baseline IPH was more likely to be associated with subsequent IPH compared with controls (43% vs. 0%) and 12 carotid cerebrovascular events occurred ipsilateral to the index carotid artery with extended follow-up. Singh et al. followed patients over 1 year using an MP-RAGE sequence. Six cerebrovascular events occurred in patients with IPH compared to no events in controls [34]. Similarly, in symptomatic patients with mild to moderate (30–69%) carotid stenosis, five ipsilateral strokes and 14 ipsilateral ischemic events occurred in a 2-year period in patients with IPH [35]. Collectively, these studies show the strong association of IPH with plaque vulnerability. Other plaque components, such as a large necrotic core [36], have also been implicated as indicative of plaque vulnerability. A plaque vulnerability scoring system (Fig. 8.10) has been developed based on MRI features [36] that include plaque

composition in addition to plaque morphology. Patient risk stratification using vessel wall imaging-based scoring systems may complement angiographic clinical imaging in future.

Natural History Studies In addition to vulnerable plaque detection, MRI of plaque can also be useful for understanding the evolution of atherosclerotic disease. Carotid plaque imaging has proven to be an effective technique in longitudinal studies designed to evaluate the natural history of atherosclerotic disease progression. The PRIMARI (Plaque Rupture In MAgnetic. Resonance Imaging) study followed subjects with 50–79% carotid stenosis by duplex ultrasound examination with serial imaging every 18 months. IPH was found to be a strong driver of plaque progression (Fig. 8.11) with higher percent change in wall volume (6.8% vs. −0.15%) and LRNC volume (28.4% vs. −5.2%) in patients with IPH compared to controls [37]. Carotid and coronary vessel wall imaging were included in the multicenter multi-ethnic study of atherosclerosis (MESA) study to investigate the prevalence, correlates, and progression

8

Vessel Wall Imaging Techniques

123

Fig. 8.10 Carotid atherosclerosis scoring (CAS) system combines both morphological and compositional features to stratify risk of plaque vulnerability. Three levels of CAS correspond to increasing risk with maximum wall thickness (MWT) and maximum percent LRNC (MPL) used for classification into the three CAS levels: CAS 1: MWT <2 mm, No LRNC, low risk subjects, CAS 2: MWT >2 mm,

MPL £ 20%, medium risk, CAS 3: MPL > 40%. Examples of matched cross-sectional images from three contrast weightings (TOF, T1WI, and CE-T1WI) for each category are provided. Arrowheads indicate the outer wall boundary; asterisk indicates the lumen and single arrows indicate LRNC. Reprinted with permission from Underhill et al. [36]

Fig. 8.11 Representative T1-weighted images of progression of atherosclerosis with intraplaque hemorrhage in right carotid artery. Each column presents matched cross-sectional locations in the carotid artery from baseline MRI (a) and MRI obtained 18 months later

(b). Lumen area was decreased, and wall area was increased in each section at second examination. CCA indicates common carotid artery; Bif bifurcation, ICA internal carotid artery, ECA external carotid artery (Reprinted with permission from Takaya et al. [37])

124

of subclinical CVD. There was a high prevalence of LRNC (71%) in subjects with wall thickness greater than 1.5 mm [38]. Vessel wall imaging is important to identify atherosclerosis considering the high prevalence of carotid lesions in patients with minimal stenosis. The feasibility of coronary vessel wall imaging was also demonstrated in this study. Coronary maximum wall thickness measured by black-blood MRI was greater for subjects with two or more cardiovascular risk factors compared to subjects with one or no risk factors [39]. Positive remodeling of the coronaries in response to atherosclerosis was also demonstrated [40].

Clinical Trials Safe noninvasive monitoring of change in atherosclerotic plaque size and composition is possible using MRI. This is useful for both patient follow-up and to assess the efficacy of drug treatment in reducing atherosclerotic disease burden. Currently, a number of clinical trials of statin therapy use MRI-based plaque measurements for assessing the efficacy of the treatment. Serial MRI studies monitoring a change in either plaque morphology or composition require high measurement reproducibility to ensure that the study endpoint can be assessed using a small number of subjects. At 1.5 T, Saam et al. [41] showed high measurement reproducibility denoted by low coefficients of variation (CV) of 5.8% for wall volume, 4.3% for lumen volume, and 3.2% for the percent atheroma volume. Compositional measurements, such as maximum percent LRNC had a larger CV (15%). Using MRI with this level of reproducibility will only require 14 subjects per treatment arm to identify a 5% treatment effect if the percent atheroma volume is used as an endpoint. Recent studies [42] established the corresponding CVs to be similar at 3 T. Based on these studies the lowest sample size can be achieved for studies using plaque burden change as the primary endpoint resulting in considerable time and cost savings. Accordingly, several studies using plaque burden as the primary endpoint have demonstrated significant reduction in plaque burden with treatment. With a 2-year simvastatin treatment, carotid and aortic vessel wall area (VWA) reduced by 14% and 10%, respectively, at 12 months and 18% and 15%, respectively, at 24 months [43]. Similarly, a high-dose atorvastatin (20 mg/day) treatment reduced vessel wall thickness and VWA of thoracic aorta plaques compared to lowdose atorvastatin (5 mg/day) [44]. In the carotid plaque composition (CPC) Study [45] of subjects with coronary artery disease or carotid disease treated with atorvastatin alone or combined with niacin and/or colesevelam, percent wall volume also decreased by 3.8% over the 3-year course of the study. The reductions in the first, second, and third years were 0.3%, 3.6%, and 0.1%, respectively, showing that

R. Li et al.

plaque burden can be used as the primary endpoint with a 2-year treatment. Plaque composition, such as the percent lipid composition, is increasingly used in clinical drug trials to monitor the efficacy of drug treatment. In a case control study of eight patients with coronary artery disease on intensive lipid-lowering therapy for 10 years, the main treatment effects were related to plaque composition with treated patients having a smaller core lipid area (0.7 vs. 10.2 mm2, respectively; P = 0.01) and percent lipid composition (1% vs. 17%, respectively) compared to untreated patients. In the Outcome of Rosuvastatin treatment on carotid artery atheroma: a magnetic resonance Imaging ObservatioN (ORION) trial of rosuvastatin treatment effects on carotid plaque, percent LRNC decreased by 41.4% in 24 months in all patients with LRNC at baseline [46]. A decrease in percent LRNC (8.4% to 5.2 over 3 years) was also noted in the CPC study. When normal wall was excluded percent LRNC decreased by 3.2%, 3.0%, and 0.91% in the first, second, and third years, respectively. Thus, using LRNC as the primary endpoint, a 1-year study can be used to test the efficacy of drug treatment. The ongoing carotid MRI substudy of the Atherothrombosis Intervention in Metabolic Syndrome with Low HDL/High Triglycerides and Impact on Global Health Outcomes (AIM-HIGH) comparing intensive LDLlowering plus HDL-raising therapy against LDL-lowering alone also monitors the effect on LRNC by serial MRI.

Future Directions Coronary Vessel Wall Imaging While vessel wall imaging of the carotids and aorta have advanced to the point where clinical studies are feasible, direct vessel wall imaging of the coronaries is still challenging because of the demands of high-resolution and high SNR due to the thin vessel walls and their position within the thorax in addition to the need for cardiac and navigator gating. Although coronary vessel walls have been visualized with DIR [47] and local pencil beam reinversion [48] image quality depends on effective gating and respiratory compensation. Imaging in the mid-diastolic quiescent period or using a subject-specific trigger delay measured from a cine scout can provide better image quality [49]. Breath-holding [47] or navigator monitoring of diaphragmatic motion [48] are generally used for respiratory compensation.

Molecular Imaging In addition to morphologic and compositional information MRI can provide functional information with the use of exogenous contrast agents. Ultrasmall superparamagnetic

8

Vessel Wall Imaging Techniques

125

particles of iron oxides (USPIO) have a large R2/R1 ratio and can be visualized based on their T2 or T2* effects. USPIO accumulation in macrophages in the carotid plaque has been validated by comparison against histology [50]. Plaques from symptomatic patients showed a decrease in signal intensity by 16.4% after USPIO infusion when compared to an 8.4% increase in plaques from asymptomatic patients [51]. Further studies have shown that asymptomatic arteries can also exhibit signal loss after USPIO infusion [52] suggestive of occult inflammation. Plaque activity monitoring by USPIO agents was used in the Atorvastatin Therapy: Effects on Reduction of Macrophage Activity (ATHEROMA) study [53] to monitor the effects of atorvastatin therapy. A significant reduction of USPIO-defined inflammation was observed in as early as 6 weeks. Gadolinium-based targeted molecular contrast agents have been demonstrated for thrombus imaging using a fibrintargeted peptide, inflammation targeting matrix metalloproteinases with specific contrast agents, gadofluorine or polyethylene glycol (PEG)-micelles incorporated with gadolinium DTPA for LRNC. Plaque neovessels can be targeted using alpha(v)beta3-integrin-targeted, paramagnetic nanoparticles. Details about these preclinical contrast agents can be found in the review by Briley-Saebo et al. [54].

carotid artery with black-blood MR images [59]. In-let and out-let flow rates were also acquired by phase contrast MRI. Finite element computational fluid dynamic simulation according to lumen boundary and flow rates was proposed to calculate the specific flow pattern and wall shear stress. Low and oscillating shear was found at the wall thickening place. Noninvasive vessel wall imaging is ideal to provide the boundary and flow information for hemodynamic studies.

Higher Resolution

1. Yusuf S, Reddy S, Ounpuu S, Anand S. Global burden of cardiovascular diseases: part I: general considerations, the epidemiologic transition, risk factors, and impact of urbanization. Circulation. 2001;104:2746–2753. 2. Naghavi M, Libby P, Falk E, et al. From vulnerable plaque to vulnerable patient: a call for new definitions and risk assessment strategies: Part I. Circulation. 2003;108:1664–1672. 3. Lu H, Clingman C, Golay X, van Zijl PC. Determining the longitudinal relaxation time (T1) of blood at 3.0 Tesla. Magn Reson Med. 2004;52:679–682. 4. Edelman RR, Chien D, Kim D. Fast selective black blood MR imaging. Radiology. 1991;181:655–660. 5. Parker DL, Goodrich KC, Masiker M, Tsuruda JS, Katzman GL. Improved efficiency in double-inversion fast spin-echo imaging. Magn Reson Med. 2002;47:1017–1021. 6. Yarnykh VL, Yuan C. Multislice double inversion-recovery blackblood imaging with simultaneous slice reinversion. J Magn Reson Imaging. 2003;17:478–483. 7. Yuan C, Kerwin WS, Ferguson MS, et al. Contrast-enhanced high resolution MRI for atherosclerotic carotid artery tissue characterization. J Magn Reson Imaging. 2002;15:62–67. 8. Yarnykh VL, Yuan C. T1-insensitive flow suppression using quadruple inversion-recovery. Magn Reson Med. 2002;48:899–905. 9. Yarnykh VL, Yuan C. Simultaneous outer volume and blood suppression by quadruple inversion-recovery. Magn Reson Med. 2006;55:1083–1092. 10. Wang J, Yarnykh VL, Hatsukami T, Chu B, Balu N, Yuan C. Improved suppression of plaque-mimicking artifacts in black-blood carotid atherosclerosis imaging using a multislice motion-sensitized driven-equilibrium (MSDE) turbo spin-echo (TSE) sequence. Magn Reson Med. 2007;58:973–981. 11. Clarke SE, Beletsky V, Hammond RR, Hegele RA, Rutt BK. Validation of automatically classified magnetic resonance images for carotid plaque compositional analysis. Stroke. 2006;37:93–97.

Current vessel wall MRI protocols have high spatial resolution in the image plane but low resolution (2–3 mm) in the slice direction which can limit detection of small plaque components [55]. 3D imaging can improve resolution in the partition encoding direction but blood suppression can be compromised if DIR-based techniques are used due to the large imaging slab thickness [55]. Diffusion preparation is well suited for 3D vessel wall imaging owing to its flow direction-independent blood suppression. Recently, new 3D vessel wall imaging approaches have been demonstrated using diffusion preparation [56]. Isotropic voxels can improve both accuracy and reproducibility of plaque component measurement. 3D isotropic black-blood imaging [56–58] can potentially improve the utility of vessel wall imaging in clinical studies.

Hemodynamic Study Previous sections are focused on the components of atherosclerotic plaque. However, the morphology of the lumen, which is clearly depicted by vessel wall imaging, is also used to evaluate the connection between mechanical forces and atherosclerotic disease. Steinman et al. reconstructed threedimensional models of the lumen and wall boundaries of

Conclusion Vessel wall imaging with black-blood prepulse provides not only morphological information about the lumen, but also the components of plaque, which are critical to evaluate plaque vulnerability, monitor plaque progression, and observe drug efficacy. Comprehensive understanding of atherosclerosis with the additional vessel wall imaging could improve the clinical management of vascular disease. Acknowledgment We would like to thank Marina S. Ferguson and Zach Miller for editing this chapter.

References

126 12. Yuan C, Kerwin WS, Yarnykh VL, et al. MRI of atherosclerosis in clinical trials. NMR Biomed. 2006;19:636–654. 13. Saam T, Ferguson MS, Yarnykh VL, et al. Quantitative evaluation of carotid plaque composition by in vivo MRI. Arterioscler Thromb Vasc Biol. 2005;25:234–239. 14. Chu B, Kampschulte A, Ferguson MS, et al. Hemorrhage in the atherosclerotic carotid plaque: a high-resolution MRI study. Stroke. 2004;35:1079–1084. 15. Moody AR. Magnetic resonance direct thrombus imaging. J Thromb Haemost. 2003;1:1403–1409. 16. Kim SE, Jeong EK, Shi XF, Morrell G, Treiman GS, Parker DL. Diffusion-weighted imaging of human carotid artery using 2D single-shot interleaved multislice inner volume diffusion-weighted echo planar imaging (2D ss-IMIV-DWEPI) at 3T: diffusion measurement in atherosclerotic plaque. J Magn Reson Imaging. 2009;30:1068–1077. 17. Cai J, Hatsukami TS, Ferguson MS, et al. In vivo quantitative measurement of intact fibrous cap and lipid-rich necrotic core size in atherosclerotic carotid plaque: comparison of high-resolution, contrast-enhanced magnetic resonance imaging and histology. Circulation. 2005;112:3437–3444. 18. Kerwin WS, O’Brien KD, Ferguson MS, Polissar N, Hatsukami TS, Yuan C. Inflammation in carotid atherosclerotic plaque: a dynamic contrast-enhanced MR imaging study. Radiology. 2006;241:459–468. 19. Kerwin W, Hooker A, Spilker M, et al. Quantitative magnetic resonance imaging analysis of neovasculature volume in carotid atherosclerotic plaque. Circulation. 2003;107:851–856. 20. Kerwin WS, Oikawa M, Yuan C, Jarvik GP, Hatsukami TS. MR imaging of adventitial vasa vasorum in carotid atherosclerosis. Magn Reson Med. 2008;59:507–514. 21. Balu N, Yarnykh VL, Scholnick J, Chu B, Yuan C, Hayes C. Improvements in carotid plaque imaging using a new eight-element phased array coil at 3T. J Magn Reson Imaging. 2009;30:1209–1214. 22. Yarnykh VL, Terashima M, Hayes CE, et al. Multicontrast blackblood MRI of carotid arteries: comparison between 1.5 and 3 tesla magnetic field strengths. J Magn Reson Imaging. 2006;23: 691–698. 23. Boussel L, Herigault G, de la Vega A, Nonent M, Douek PC, Serfaty JM. Swallowing, arterial pulsation, and breathing induce motion artifacts in carotid artery MRI. J Magn Reson Imaging. 2006;23:413–415. 24. Chan CF, Gatehouse PD, Hughes R, Roughton M, Pennell DJ, Firmin DN. Novel technique used to detect swallowing in volumeselective turbo spin-echo (TSE) for carotid artery wall imaging. J Magn Reson Imaging. 2009;29:211–216. 25. Kerwin W, Xu D, Liu F, et al. Magnetic resonance imaging of carotid atherosclerosis: plaque analysis. Top Magn Reson Imaging. 2007;18:371–378. 26. Babiarz L, Astor B, Mohamed M, Wasserman B. Comparison of gadolinium-enhanced cardiovascular magnetic resonance angiography with high-resolution black blood cardiovascular magnetic resonance for assessing carotid artery stenosis. J Cardiovasc Magn Reson. 2007;9:63–70. 27. Dong L, Underhill H, Yu W, et al. Geometric and compositional appearance of atheroma in an angiographically normal carotid artery in patients with atherosclerosis. AJNR Am J Neuroradiol. 2010;31:311–316. 28. Falk E. Why do plaques rupture? Circulation. 1992;86(6 Suppl): III30–42. 29. Hatsukami TS, Ross R, Polissar NL, Yuan C. Visualization of fibrous cap thickness and rupture in human atherosclerotic carotid plaque in vivo with high-resolution magnetic resonance imaging. Circulation. 2000;102:959–964. 30. Trivedi RA, U-King-Im JM, Graves MJ, et al. MRI-derived measurements of fibrous-cap and lipid-core thickness: the potential for identifying vulnerable carotid plaques in vivo. Neuroradiology. 2004;46(9):738–743.

R. Li et al. 31. Cai JM, Hatsukami TS, Ferguson MS, et al. In vivo quantitative measurement of intact fibrous cap and lipid-rich necrotic core size in atherosclerotic carotid plaque - Comparison of high-resolution, contrast-enhanced magnetic resonance imaging and histology. Circulation. 2005;112:3437–3444. 32. Yuan C, Zhang SX, Polissar NL, et al. Identification of fibrous cap rupture with magnetic resonance imaging is highly associated with recent transient ischemic attack or stroke. Circulation. 2002;105:181–185. 33. Underhill HR, Yuan C, Yarnykh VL, et al. Predictors of surface disruption with MR imaging in asymptomatic carotid artery stenosis. AJNR Am J Neuroradiol. 2010;31:487–493. 34. Singh N, Moody A, Gladstone D, et al. Moderate carotid artery stenosis: MR imaging-depicted intraplaque hemorrhage predicts risk of cerebrovascular ischemic events in asymptomatic men. Radiology. 2009;252:502–508. 35. Altaf N, MacSweeney ST, Gladman J, Auer DP. Carotid intraplaque hemorrhage predicts recurrent symptoms in patients with highgrade carotid stenosis. Stroke. 2007;38:1633–1635. 36. Underhill H, Hatsukami T, Cai J, et al. A noninvasive imaging approach to assess plaque severity: the Carotid Atherosclerosis Score. AJNR Am J Neuroradiol. 2010;31:1068–1075. 37. Takaya N, Yuan C, Chu B, et al. Presence of intraplaque hemorrhage stimulates progression of carotid atherosclerotic plaques - A high-resolution magnetic resonance Imaging study. Circulation. 2005 2005;111:2768–2775. 38. Wasserman BA, Sharrett R, Lai S, et al. Risk factor associations with the presence of a lipid core in carotid plaque of asymptomatic individuals using high-resolution MRI - The Multi-Ethnic Study of Atherosclerosis (MESA). Stroke. 2008;39:329–335. 39. Macedo R, Chen S, Lai S, et al. MRI Detects Increased Coronary Wall Thickness in Asymptomatic Individuals: The Multi-Ethnic Study of Atherosclerosis (MESA). J Magn Reson Imaging. 2008;28:1108–1115. 40. Miao CL, Chen SG, Macedo R, et al. Positive Remodeling of the Coronary Arteries Detected by Magnetic Resonance Imaging in an Asymptomatic Population MESA (Multi-Ethnic Study of Atherosclerosis). J Am Coll Cardiol. 2009;53:1708–1715. 41. Saam T, Kerwin WS, Chu B, et al. Sample size calculation for clinical trials using magnetic resonance imaging for the quantitative assessment of carotid atherosclerosis. J Cardiovasc Magn Reson. 2005;7: 799–808. 42. Li F, Yarnykh V, Hatsukami T, et al. Scan-rescan reproducibility of carotid atherosclerotic plaque morphology and tissue composition measurements using multicontrast MRI at 3T. J Magn Reson Imaging. 2010;31:168–176. 43. Corti R, Fayad ZA, Fuster V, et al. Effects of lipid-lowering by simvastatin on human atherosclerotic lesions - A longitudinal study by high-resolution, noninvasive magnetic resonance imaging. Circulation. 2001;104:249–252. 44. Yonemura A, Momiyama Y, Fayad ZA, et al. Effect of lipid-lowering therapy with atorvastatin on atherosclerotic aortic plaques detected by noninvasive magnetic resonance imaging. J Am Coll Cardiol. 2005;45:733–742. 45. Zhao X, Phan B, Chu B, et al. Testing the hypothesis of atherosclerotic plaque lipid depletion during lipid therapy by magnetic resonance imaging: study design of Carotid Plaque Composition Study. Am Heart J. 2007;154:239–246. 46. Underhill H, Yuan C, Zhao X, et al. Effect of rosuvastatin therapy on carotid plaque morphology and composition in moderately hypercholesterolemic patients: a high-resolution magnetic resonance imaging trial. Am Heart J. 2008;155:584.e581-588. 47. Fayad ZA, Fuster V, Fallon JT, et al. Noninvasive in vivo human coronary artery lumen and wall imaging using black-blood magnetic resonance imaging. Circulation. 2000;102:506–510. 48. Botnar RM, Won YK, Peter Br, Matthias S, Elmar S, Warren JM. 3D coronary vessel wall imaging utilizing a local inversion

8

49.

50.

51.

52.

53.

Vessel Wall Imaging Techniques technique with spiral image acquisition. Magn Reson Med. 2001;46:848–854. Kim W, Stuber M, Kissinger K, Andersen N, Manning W, Botnar R. Impact of bulk cardiac motion on right coronary MR angiography and vessel wall Imaging. J Magn Reson Imaging. 2001 2001;14:383–390. Trivedi RA, Mallawarachi C, U-King-Im J-M, et al. Identifying inflamed carotid plaques using in vivo USPIO-enhanced MR imaging to label plaque macrophages. Arterioscler Thromb Vasc Biol. 2006;26:1601–1606. Tang T, Howarth S, Miller S, et al. Comparison of the inflammatory burden of truly asymptomatic carotid atheroma with atherosclerotic plaques in patients with asymptomatic carotid stenosis undergoing coronary artery bypass grafting: an ultrasmall superparamagnetic iron oxide enhanced magnetic resonance study. Eur J Vasc Endovasc Surg. 2008;35:392–398. Howarth S, Tang T, Trivedi R, et al. Utility of USPIO-enhanced MR imaging to identify inflammation and the fibrous cap: a comparison of symptomatic and asymptomatic individuals. Eur J Radiol. 2009;70:555–560. Tang T, Howarth S, Miller S, et al. The ATHEROMA (Atorvastatin Therapy: Effects on Reduction of Macrophage Activity) Study. Evaluation using ultrasmall superparamagnetic iron oxide-enhanced magnetic resonance imaging in carotid disease. J Am Coll Cardiol. 2009;53:2039–2050.

127 54. Briley-Saebo KC, Mulder WJM, Mani V, et al. Magnetic resonance imaging of vulnerable atherosclerotic plaques: current imaging strategies and molecular imaging probes. J Magn Reson Imaging. 2007;26:460–479. 55. Balu N, Chu B, Hatsukami T, Yuan C, Yarnykh V. Comparison between 2D and 3D high-resolution black-blood techniques for carotid artery wall imaging in clinically significant atherosclerosis. J Magn Reson Imaging. 2008;27:918–924. 56. Koktzoglou I, Li D. Submillimeter isotropic resolution carotid wall MRI with swallowing compensation: imaging results and semiautomated wall morphometry. J Magn Reson Imaging. 2007; 25:815–823. 57. Zhang Z, Fan Z, Carroll TJ, et al. Three-Dimensional T2-Weighted TSE MRI of the Human Femoral Arterial Vessel Wall at 3.0Tesla. Paper presented at: Proceedings of the 17th ISMRM Scientific Meeting and Exhibition2009; Honolulu. 58. Balu N YV, Chu B, Wang J, Hatsukami T, Yuan C. Carotid Plaque Assessment using Fast 3D Isotropic-Resolution Black-Blood MRI. Paper presented at: Proceedings 17th ISMRM Scientific Meeting and Exhibition2009; Honolulu. 59. Steinman DA, Thomas JB, Ladak HM, Milner JS, Rutt BK, Spence JD. Reconstruction of carotid bifurcation hemodynamics and wall thickness using computational fluid dynamics and MRI. Magn Reson Med. 2002;47:149–159.

wwwwwwwwwwww

9

Noncontrast Coronary Artery Imaging Allison Hays, Robert G. Weiss, and Matthias Stuber

Coronary Magnetic Resonance Angiography Introduction Despite advances in the diagnosis of coronary artery disease (CAD), there still exists a strong need for safer and costeffective techniques to improve visualization of the coronary lumen and vessel wall. Currently, the gold standard for the diagnosis of CAD is coronary X-ray angiography, however, this technique is invasive, costly, and not without risk to the patient. Furthermore, this test does not provide any information about early atherosclerotic disease progression that precedes luminal narrowing. Since its introduction, coronary magnetic resonance angiography (MRA) has been able to overcome some of the disadvantages of X-ray angiography and shows promise in the evaluation of CAD, particularly for proximal coronary disease. Coronary MRA provides a good alternative for patients because it is noninvasive, cost-effective, and without exposure to ionizing radiation. In addition, magnetic resonance techniques can detect abnormalities in the coronary vessel wall before luminal narrowing occurs. The ability to detect early atherosclerotic changes in the vessel wall, assess coronary function, and identify stenotic luminal disease are advantages of coronary MRA that make it a potentially powerful and comprehensive tool for the evaluation of CAD.

A. Hays, MD Division of Cardiology, Department of Medicine, Johns Hopkins Hospital, Baltimore, MD, USA R.G. Weiss, MD Department of Medicine, Johns Hopkins Hospital, Baltimore, MD, USA M. Stuber, PhD () Department of Radiology, Centre Hospitalier Universitaire Vaudois, Center for Biomedical Imaging and University of Lausanne, Lausanne, Switzerland e-mail: [email protected]

However, for successful coronary MRA and coronary vessel wall data acquisition, there are many challenges which must be overcome. The heart is subject to intrinsic and extrinsic motion due to its periodic contraction and relaxation as well as the effect of respiration. Both of these motion components exceed the coronary artery dimensions by a significant degree, making high-resolution data acquisition technically challenging. Therefore, efficient motion suppression strategies must be implemented. In addition, enhanced contrast between the coronary lumen and the surrounding tissue is essential for the visualization of both coronary lumen and the coronary vessel wall. This chapter reviews the technical developments and recent advances that have contributed to the current state of coronary MRA imaging. Furthermore, we discuss the clinical applications as well as the benefits and limitations of current MR approaches. Finally, we address future applications and recent technical developments to improve visualization of the coronary lumen and vessel wall.

Overcoming the Technical Challenges of Coronary MRA Cardiac Motion Suppression The main sources of image artifacts in coronary MRA include cardiac and respiratory motion. To address the issue of cardiac motion, data acquisition is cardiac triggered with the R-wave of the surface electrocardiogram. Data are collected over multiple cardiac cycles (k-space segmentation, Fig. 9.1), and in order to minimize intrinsic cardiac motion, the segmented data are typically acquired in a short acquisition window and during a period of minimal myocardial motion. The duration of the acquisition window must be long enough for an acceptable scan time and short enough to minimize cardiac motion during data acquisition. Although the time of least cardiac motion, or “trigger delay” may be estimated from a patient’s heart rate [1], a more accurate

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_9, © Springer Science+Business Media, LLC 2012

129

130

A. Hays et al.

Fig. 9.1 Suppression of intrinsic myocardial motion is obtained by ECG triggering, image data collection in late diastole during period of minimal myocardial motion and image data collection in a narrow

acquisition window. Segments of the k-space are filled in subsequent cardiac cycles (=k-space segmentation). FFT Fast Fourier Transform

means is to determine a patient-specific trigger delay with a high-resolution cine scout scan [2]. The following coronary MRA scans can then be tailored for data acquisition to occur only during this predefined quiescent period. Typically, a period of minimal myocardial motion occurs in late diastole [3]; however, in patients with high heart-rates, end-systolic imaging may also be beneficial [4]. An alternate approach is with the use of automated algorithms that predict the period of minimal myocardial motion [5–7].

patients often have comorbidities preventing them from complying with prolonged breath-holds, limiting the spatial resolution. Finally, the use of techniques, such as fold-over suppression or signal averaging for signal enhancement, is greatly limited by the achievable breath-hold duration. Therefore, the broad applicability of breath-holds is more restricted in sick or noncompliant patients. As a result, the vast majority of the recent coronary MRA studies did not include breath-holding as a mechanism to suppress respiratory motion artifacts [15–17].

Respiratory Motion Compensation Free-Breathing Coronary MRA Breath-Hold Coronary MRA The effect of respiratory motion on coronary MRA poses a major challenge for coronary imaging. Two-dimensional (2D) breath-hold techniques were implemented early to suppress respiratory motion artifacts for coronary imaging [8]. The goal of this 2D approach was to acquire contiguous images of the proximal segments of the coronary arteries during serial breath-holds. Recently, with the initiation of steady-state with free precession (SSFP) in combination with parallel imaging, 3D data collection during a single breathhold became technically feasible [9–13]. In general, the advantage of the breath-hold approach is that rapid imaging is possible, and it is relatively easy to perform in wellmotivated subjects. However, there are several limitations associated with the breath-hold strategy. The position of the diaphragm may vary significantly during repeated breathholds and during a sustained breath-hold, and there is frequently upward diaphragmatic drift of approximately 1 cm [14]. Data acquisition during serial breath-holds may also result in mis-registration gaps that may appear as signal voids in the coronaries and lead to misinterpretation. In addition,

Alternate methods to breath-hold approaches include freebreathing coronary MRA with the use of respiratory navigators [18–21]. Using this method, many ECG-gated data are acquired but only those where the navigator-identified lung–liver interface position falls within a prespecified endexpiratory gating window are included for image reconstruction. Navigator gating can be enhanced with prospective adaptive slice tracking. With this technology, the imaged slice position is adjusted in real-time to account for the residual diaphragmatic displacement within the gating window [22]. This helps to overcome some of drawbacks associated with prolonged scan times and narrow gating windows and has resulted in equivalent or improved image quality with submillimeter spatial resolution and abbreviated scan times [14, 22, 23]. Typically, respiratory data collection occurs at end expiration when the diaphragm position is most consistent within the respiratory cycle (Fig. 9.2). Therefore, respiratory gating lengthens the scan time because no data are collected in the remainder of the respiratory cycle and on average data are collected during 50% of the R-R intervals [2].

9

Noncontrast Coronary Artery Imaging

Fig. 9.2 The navigator signal is typically obtained from the lung–liver interface. A computer algorithm detects the position of that interface (lung–liver interface) in real-time. If the computer identified lung–liver interface position is found inside of the gating window (window width can be adjusted by the user), the k-space segment that is collected

131

immediately after the navigator is accepted for reconstruction. However, if the lung–liver interface position is found outside of the gating window, that k-space segment has to be remeasured in a later cardiac cycle. Adding navigator gating to an imaging sequence typically increases scanning time by a factor of ~2

One study directly compared the quality of 3D coronary MRA acquired during a single breath-hold versus a realtime navigator-gated free breathing technique in patients with suspected CAD [24]. In this study, it was found that navigator gated and corrected coronary MRA was improved with respect to diagnostic accuracy of stenosis quantification and image quality.

Contrast Enhancement Coronary MRA examinations are typically performed without the addition of intravenously administered contrast agents. The relative signal of the coronary artery lumen is augmented by taking advantage of the natural T2 differences between blood and the surrounding myocardium by using fat-saturation prepulses [8], magnetization transfer contrast prepulses [25] or T2 preparatory pulses (Fig. 9.3) [26, 27]. With these techniques, the coronary lumen appears bright while the surrounding myocardium appears dark with reduced signal intensity. Novel intravascular contrast agents offer the ability to improve spatial resolution and SNR in coronary imaging. The advantages of these blood pool contrast agents are that the plasma half-life is longer and that they do not extravasate as quickly into the extracellular space as Gadolinium does [28, 29]. These factors result in reduced myocardial signal, greater blood pool enhancement, and allow image acquisition over longer time periods after intravenous administration of the contrast agent. Therefore, intravascular contrast agents are well-suited to be combined with navigator technology and prolonged scanning times. However, the use of specific intravascular or extracellular contrast agents are discussed in another chapter.

Fig. 9.3 Baseline images (a, c) and T2prep-enhanced images (b, d) of 3D coronary data set. Application of T2prep (b, d) suppressed cardiac muscle as well as skeletal muscle in the chest and back. Due to enhanced blood-to-muscle contrast in T2prep images (b, d), visualization of LAD and LCx is improved compared with reference images (a, c). Suppression of venous blood additionally helps to differentiate between great cardiac vein (GCV), LAD, and LCx, which is difficult to distinguish in image (a) (Reprinted with permission from Botnar et al. [27])

Whole-Heart Coronary MRA One of the challenges of coronary MRA is that extensive planning and scout scanning is required prior to imaging the coronaries adding to the complexity of the scan and

132

reducing the ease-of-use. An alternative approach that circumvents this issue is whole-heart imaging [30] which utilizes a volumetric approach to improve coverage (Fig. 9.4). Arbitrary views can then be reconstructed during postprocessing without compromising image quality and can enable visualization of more distal coronary segments. Whole-heart coronary MRA is typically performed during free breathing with navigator gating technology, and has progressed largely due to advances in parallel imaging and SSFP, which has high endogenous contrast. Contrast enhancement is achieved using spatially selective fat saturation and T2 preparation [30, 31]. One of the limitations of whole-heart imaging is the relatively prolonged scan times (approximately 10–15 min) that may be susceptible to slow diaphragmatic drift during the relatively long scanning time. However, there are currently mechanisms in place to compensate for such diaphragmatic drift that can occur with time [32, 33]. The use of the whole-heart technique was investigated in a study of 20 CAD patients and reported a sensitivity of 82% and a specificity of 91% for the identification of significant CAD when compared to X-ray coronary angiography in a single-center setting [34]. The average scan time for the study was less than 15 min during free breathing. Further improvements in image quality and scan time are anticipated with the development of larger cardiac coil arrays for coronary imaging and with high field imaging [35, 36].

Fig. 9.4 Example of the right and left coronary arteries using whole heart bright blood SSFP imaging in a 71-year-old healthy volunteer. Imaging parameters used were: TR = 279 ms, TE = 1.6 ms, slice thickness 0.7 mm, in plane resolution of 1.1 × 1.1 mm with an imaging time of 10 min (Image courtesy of Kai Lin and Debiao Li, Northwestern University, Chicago, IL)

A. Hays et al.

Visualization of CAD Identification of Coronary Stenosis Current breath-hold coronary MRA techniques were shown to accurately identify proximal coronary stenoses in several clinical studies. Gradient-echo techniques depict focal stenoses as signal voids (Fig. 9.5). In an early patient study, a segmented k-space 2D breath-hold ECG-gated gradientecho sequence was used to compare coronary MRA to the gold standard of X-ray coronary angiography [37]. This single-center 2D coronary MRA technique yielded a sensitivity of 90% and a specificity of 92% for correctly classifying individual vessels with significant CAD defined as greater than 50% diameter stenosis on X-ray angiography. Other studies have subsequently reported variable sensitivity and specificity values for the detection of significant CAD [38–43]. The variability in these studies may be due to differences in the MR sequences employed, patient selection, or arrhythmias which may degrade image quality. Newer breath-hold [44] and nonbreath-hold approaches for 3D coronary MRA have also demonstrated the ability of this technique to detect coronary stenoses. An international multicenter trial prospectively compared X-ray coronary angiography with coronary MRA using common hardware, software, and methodology [45]. This trial showed that freebreathing submillimeter 3D coronary MRA can accurately identify significant proximal and mid coronary disease while nonsignificant coronary disease can be excluded with high certainty. Although the specificity of the technique remains to be improved, it is likely that advances in hardware and software in combination with a higher magnetic field strength may reduce the false positive rate and improve accuracy for

Fig. 9.5 (a) Radiograph coronary angiogram of a patient with a 50% luminal stenosis of the mid-LAD (arrow). (b) Transverse multiplanar reformatted free-breathing 3D Balanced FFE coronary MRA acquired in the same patient. The location of the lesion (dotted arrow) corresponds to that of the radiograph image (Reprinted with permission from Spuentrup et al. [88])

9

Noncontrast Coronary Artery Imaging

133

the identification of more distal stenoses. Clearly, the above mentioned whole-heart approach is ready to be tested in a multicenter setting.

Coronary MRA for Coronary Artery Bypass Graft Assessment The utility of coronary MRA for the assessment of coronary bypass graft patency has been investigated since the 1980s and has made significant improvement in recent years. In early studies using ECG-triggered spin-echo [46] and gradient echo techniques [46, 47], visualization of grafts and graft patency was limited by cardiac and respiratory motion artifacts. As a result, this restricted the ability to accurately quantify graft stenosis. More recently, data collection during a single breath-hold became possible [48]. However, while all occluded grafts were successfully identified using breathholding techniques [48], only 67% of the patent grafts were correctly diagnosed. Some of the limitations of these earlier techniques occurred because of problems with the patient’s ability to perform breath-holds and the inherent dependence of the maximum achievable spatial resolution on the breathhold duration, or because of diaphragmatic drift. More advanced approaches for respiratory motion compensation, such as the use of retrospective respiratory navigators [20] has enabled 3D acquisition with high in-plane spatial resolution and removed the constraints related to breath-holds [49]. This method increased both sensitivity and specificity for the assessment of graft patency (87% and 100%, respectively); however, the detection of luminal stenosis was not as reliable. More recently, navigator gated and corrected 3D coronary MRA has been shown to accurately graft identify occlusion or stenosis [50]. Using this method, a sensitivity of 83% and a specificity of 98% for the definition of graft occlusion were obtained, and a reasonable diagnostic accuracy for the assessment of vein graft stenosis severity was reported. A subsequent study of bypass graft patency with steady-state free-precession angiography found a comparable sensitivity although with reduced specificity for graft stenosis severity compared to other techniques [51]. A practical limitation of coronary MRA bypass graft assessment is related to local signal loss/artifacts due to nearby metallic objects, such as hemostatic clips, stainless steel graft markers, and sternal wires. The limited ability to consistently identify severely diseased yet patent grafts is also a hindrance to clinical utility and acceptance.

Visualization of Anomalous Coronary Arteries and Coronary Aneurysms Traditionally, X-ray angiography has been the imaging test of choice for the diagnosis of coronary artery anomalies, a rare cause of myocardial ischemia and sudden death among

Fig. 9.6 A young woman with a history of Kawasaki disease. Multiplanar reformatted coronary MRA images of the left (a) and right (c) coronary arteries are compared with selective X-ray angiograms of the left (b) and right (d) coronary arteries. Two coronary artery aneurysms (CAA) in the left coronary and one in the right coronary (a through d, black arrows), as well as a stenosis between the two left CAAs (a and b, white arrows) are shown (Reprinted with permission from Greil et al. [61])

young adults. However, this technique is limited in its ability to define the course of the anomalous coronaries particularly with regards to the great vessels. Several published series [52–55] have documented a good correlation between coronary MRA with X-ray angiography in the characterization of anomalous coronaries. Early coronary MRA studies often used a 2D breath-hold gradient echo approach [52–58]. These 2D coronary MRA studies uniformly reported high accuracy, including several studies in which coronary MRA was determined to be superior to X-ray angiography [53, 54]. Most centers currently use 3D coronary MRA because of superior reconstruction capabilities with similar results [59]. As a result, coronary MRA is now the preferred imaging modality for the evaluation of anomalous coronary arteries [60]. Though coronary artery aneurysms are relatively uncommon, recent studies indicate an important role for coronary MRA for the assessment of this condition [39] and in the evaluation of coronary ectasia. 3D coronary MRA has shown to be valuable in the characterization of coronary aneurysms in pediatric patients with Kawasaki disease (Fig. 9.6), a rare small vessel vasculitic disease [61]. A strong correlation between coronary MRA and X-ray coronary angiography has also been reported for ectatic coronary arteries [62].

134

Fig. 9.7 Black-blood 3D RCA vessel wall image showing the anterior and the posterior coronary walls (arrows). The dotted arrows point to the contrast between the tissue in the path of the cylindrical pulse and the surrounding tissue by use of the local inversion prepulse (Reprinted with permission, Desai et al. [69])

A. Hays et al.

window and therefore, less susceptibility to intrinsic cardiac motion. In one study using a free-breathing 3D approach, the coronary vessel wall could be visualized with an inplane resolution of 0.78–1.0 mm in both healthy adults and those with nonsevere CAD [64]. In the CAD patients, the mean vessel wall thickness was significantly increased when compared to that of normal volunteers, whereas as the lumen diameter was similar, illustrating positive arterial remodeling. In another study of patients with type I diabetes, coronary vessel wall imaging using a similar blackblood approach revealed significantly greater wall thickness in diabetic patients with nephropathy than those with normal kidney function [68]. Therefore, black-blood coronary MRA represents a powerful investigative tool to noninvasively quantify subclinical atherosclerosis in a variety of patient populations. Preliminary studies of coronary vessel wall imaging at 3 T are promising and show the potential to detect preclinical disease and monitor treatment effects over time [69, 70].

Coronary Flow Imaging Coronary Vessel Wall Imaging Using Black-Blood Coronary MRA Early changes in the development of atherosclerosis include outward “Glagov” arterial remodeling with relative preservation of the lumen [63]. Because thickening of the vessel wall precedes luminal narrowing, MRI has the ability to detect early coronary atherosclerosis (Fig. 9.7). Conventional imaging of the coronary lumen, such as with X-ray angiography, may significantly underestimate the degree of subclinical atherosclerosis. Using black-blood MRA imaging techniques, the coronary vessel wall can be visualized and vessel wall thickness and plaque volume quantified [64]. Most black-blood imaging techniques involve a dual inversion approach to suppress luminal blood based on both its T1 properties and flow, and an initial inversion pulse is immediately followed by a spatially selective reinversion pulse to restore magnetization along the vessel. Initially, single cross-sectional slices of the coronary vessel wall were obtained during breath-holds and vessel wall thickness was measured in a subset of cases [65]. Subsequently, this technique has been refined using respiratory navigators for free-breathing data acquisition [66]. Using a localinversion 3D spiral technique [67], a large anatomical coverage of the coronaries can be obtained with thinner reconstructed slices than previously possible with 2D approaches. Although scan times are prolonged using this free-breathing technique (approximately 10–15 min), the advantage includes data acquisition in a very short acquisition

MR flow mapping has been validated as a noninvasive means to assess coronary flow velocity and velocity reserve, and has a strong correlation to measurements obtained using the gold standard, Doppler guidewire [71]. In response to stress, prior invasive coronary artery studies documented similar increases in peak diastolic coronary flow velocity as that obtained using MRI [71, 72]. The clinical utility of MR flow studies of the coronaries has been examined in specific populations of CAD patients, including after bypass grafting or stent placement. In one study, MR flow measurements using phase contrast methods permitted the noninvasive evaluation of coronary flow in CAD patients after percutaneous coronary intervention and showed similar results to Doppler guidewire measurements [73]. In this study, the authors reported a diagnostic accuracy of 86% for the detection of significant in-stent restenosis defined as >50% arterial narrowing. In a study of 69 coronary bypass patients, velocity-encoded flow mapping by MR was performed at baseline and with vasodilator stress to assess flow in the bypass grafts [74]. Using X-ray angiography as the standard, a sensitivity of 96% and specificity of 92% for the identification of stenosis >70% was reported. One limitation, however, was that flow scans could only be obtained in 80% of grafts because of suboptimal image quality. The measurement of coronary flow using MRI may provide a valuable noninvasive alternative to monitor the hemodynamic significance of coronary stenoses in a variety of patients with CAD.

9

Noncontrast Coronary Artery Imaging

135

Newer Technical Developments Radial and Spiral Imaging Techniques Non-Cartesian methods of k-space sampling, such as radial and spiral trajectories, have gained more widespread use in coronary MRA because of advantages in reducing motion artifacts and improved sampling efficiency. One benefit with radial imaging is a low inherent sensitivity to motion. In addition, it is a robust technique with respect to undersampling, yielding adequate image quality during a single breathhold for a targeted volume approach [75]. 3D Radial sampling applied to whole-heart imaging may lead to a loss in SNR because image reconstruction entails a nonuniform weighting of data. However, with a 3D radial approach, isotropic spatial resolution for whole-heart imaging has been achieved [31]. With further developments in surface coils to generate the high signal required and advanced reconstruction algorithms, this may potentially improve radial sampling techniques for coronary applications. Spiral sampling techniques have also been applied to generate high-quality submillimeter resolution coronary images using 2D techniques (Fig. 9.8) [76]. Advantages of this approach are that flow artifacts can be minimized and SNR increased because k-space sampling is more efficient. Spiral sampling algorithms have been applied successfully for acquiring high-resolution coronary MRA using a freebreathing real-time navigator approach [77], 3D acquisitions [78], and a segmented interleaved strategy [79]. One study directly compared 3D spiral coronary MRA to 3D Cartesian acquisition and found superior results in image quality, SNR and CNR with the spiral approach [77]. Therefore, spiral sampling techniques show promise for high-resolution coronary imaging and warrant further study.

Arterial Spin Labeling To improve the visualization of the vessel lumen and suppress signal from surrounding structures, an MR subtraction technique called arterial spin labeling can be employed to exclusively image the coronary lumen [80], analogous to images obtained using conventional X-ray angiography. A 2D spatially selective inversion pulse is applied to tag blood in the aortic root [81] and after a short-time delay for wash-in, the labeled blood in the proximal coronaries may be imaged using a volume slab approach (Fig. 9.9). An advantage of arterial spin-labeling techniques is that they enable the depiction of the lumen of the coronary tree alone at varying angles. Furthermore, no postprocessing is required to facilitate analysis. However, a drawback to this technique includes the need for two separate acquisitions, thus prolonging scan time [82, 83]. Recently, an arterial

Fig. 9.8 Coronary MRA images. Reformatted images obtained from different subjects are shown. Examples for the mid RCA and proximal RCA and LM/LAD/LCX portions of the vessels are well visualized. Additionally, some anatomical structures are indicated (RV right ventricle, LV left ventricle, Ao aorta, and PA pulmonary artery). The left column represents data sets obtained using the Cartesian segmented k-space gradient-echo method. The middle column shows data obtained using the single-interleaf spiral imaging approach. The measuring times for the Cartesian and single-interleaf spiral approaches were similar. The right column shows data sets measured using the double-interleaf spiral approach obtained in half of the total measuring time used for a Cartesian scan (Reprinted with permission from Bornert et al. [77])

spin labeling with local reinversion was developed that takes advantage of inflow and natural differences in T1, and avoids subtraction, thus abbreviating scan time [84, 85]. In the future, arterial spin-labeling techniques may benefit from the improved SNR and reduced decay of labeled product (due to prolonged T1 values) inherent to higher field imaging.

136

Fig. 9.9 Local reinversion coronary MR angiography obtained in a healthy adult subject. A segment of the right coronary artery (RCA) is displayed with high visual signal intensity, contrast, and vessel delineation. Simultaneously, a cross-section of the left main (LM), a short proximal segment of the left anterior descending (LAD), and a proximal segment of the left circumflex coronary artery (LCX) are visualized with high signal intensity. AO aorta, RV right ventricle (Reprinted with permission from Katoh et al. [84])

Steady-State Free-Precession Coronary MRA The application of SSFP imaging of the heart enhances contrast between the ventricular blood pool and the myocardium without the requirement of exogenous contrast enhancement. SSFP imaging has been employed widely using either using a single breath-hold [86] or free-breathing with respiratory navigators [87]. Given the inherent properties of SSFP imaging with high SNR, this sequence may be potentially useful for contrast enhancement in 3D coronary MRA, in which in-flow effects are in general decreased due to thick slab excitations [86]. When combined with fat saturation and T2 preparation techniques, SSFP approaches lead to images with a high contrast and vessel sharpness (Fig. 9.10) [88]. An initial study performed using an animal model [87] observed that free-breathing SSFP imaging resulted in high-quality coronary MRA when compared with standard T2-prepared gradient-echo imaging with substantial improvements in SNR, CNR, and vessel sharpness. Although spiral imaging demonstrated the highest SNR, SSFP imaging yielded the highest vessel definition and image quality score. In another study of healthy volunteers, reliable fat suppression and a significant increase in the blood-myocardial CNR was achieved postcontrast using SSFP techniques [89]. At 3 T, both SSFP and segmented k-space gradient echo imaging have been investigated. In an early study, segmented k-space gradient echo imaging was found to be superior to SSFP imaging which was attributed in part to greater magnetic field susceptibility at 3 T as well as limitations in specific

A. Hays et al.

Fig. 9.10 Navigator gated and corrected free-breathing coronary MRA data acquired in a healthy subject (male, 40 years) using a segmented 3D Balanced FFE imaging sequence (TR = 4 ms, TE = 2 ms). (a) Image displays a double-oblique view of the right coronary artery (RCA) and a left circumflex (LCX) with a high visual vessel definition (dashed arrows). (b) Transverse imaging plane demonstrating a left coronary system, including the left main (LM), the left anterior descending (LAD), the left circumflex (LCX) and a great cardiac vein (GCV). A signal attenuated “dark rim” delineating the coronary arteries is readily apparent (dashed arrows) (Reprinted with permission from Spuentrup et al. [88])

absorption rate (SAR) [90]. However, it should be noted that the combination of SSFP and parallel imaging techniques has made it possible to acquire whole-heart data as noted above [30]. This major step forward has significantly improved ease-of-use over volume-targeted approaches and allowed access to more distal vessels.

Parallel Imaging for Coronary MRA Parallel imaging techniques decrease acquisition time and enable an entire data set to be collected in one cardiac cycle. The development of such rapid parallel imaging approaches, such as “SENSE” [91], “SMASH” [92], or GRAPPA [93, 94], have been shown to reduce scan time for cardiac MRI substantially [95] and offer great potential to enhance coronary imaging. However, the trade-off is reduced SNR, which may be an important consideration in some applications. Parallel imaging techniques have been employed at higher magnetic field strengths, and have resulted in excellent image quality with abbreviated scanning times [96].

High Field Coronary MR Imaging Imaging at higher field strengths has the potential of higher SNR, higher spatial resolution, and shorter scanning times compared to lower magnetic field strengths. Three Tesla systems have now been in wide clinical use, and several reports

9

Noncontrast Coronary Artery Imaging

137

design will likely yield continued progress and address many of the imaging challenges specific to 7 T imaging.

Conclusion

Fig. 9.11 Representative examples of MR images of the RCA obtained at 7 T. Proximal (a) and more distal (b) segments of the RCA are visualized. The 13-cm RF transmit and receive coil provides sufficient penetration depth to visualize a considerable RCA segment length. Sharply delineated contiguous coronary segments with good contrast between the coronary lumen blood pool and the epicardial fat are shown (Reprinted with permission from van Elderen et al. [104])

have shown promising results using similar techniques as at 1.5 T [96, 97]. However, scanning at 3 T and higher field strengths pose several technical challenges, including increased susceptibility artifacts particularly at tissue borders and an increased magnetohydrodynamic effect affecting ECG triggering. Despite these limitations, recent advances in hardware and software (i.e., vector ECG [98], higher order shimming, B1 shimming [99, 100]) have enabled high-resolution free-breathing scans of at least comparable image quality to that obtained at lower field strengths. Coronary MRA may be improved at higher field strengths and an initial study of coronary angiography at 3 T in healthy adults reported a higher spatial resolution compared to that achievable at 1.5 T [97]. Sommer and coworkers directly compared early 3 T to 1.5 T coronary MRA to assess the accuracy of diagnosing CAD compared to the standard of X-ray coronary angiography [101]. Using navigator-corrected, 3D segmented k-space gradient-echo techniques at both field strengths, they found comparable image quality with a 30% increase in SNR and a 22% increase in CNR at 3 T. Overall, the diagnostic accuracy at both field strengths was equivalent, with sensitivity for the detection of CAD of 82% for both, and a specificity of 89% and 88% for 3 T and 1.5 T, respectively. However, newer techniques that were not employed at the time, such as adiabatic T2 preparation pulses [102], parallel imaging [96], or advanced shimming algorithms [103], will likely contribute to superior results of coronary MRA at higher field strengths. More recently, commercial human 7 T MR systems have become available and represent a potentially powerful tool for coronary imaging. Initial studies demonstrate the feasibility of in vivo human coronary MRA at 7 T using custombuilt coils and vector ECG hardware to address some of the obstacles found with high field imaging (Fig. 9.11) [104]. Further developments in contrast enhancement and coil

Coronary MRA, because of its noninvasive nature and the capacity for soft tissue characterization, has emerged as a powerful modality to evaluate the coronary lumen and vessel wall. Current MRA techniques can reliably identify both anomalous coronary arteries and coronary aneurysms; however, there is still limited multicenter data to suggest that MRA is comparable to X-ray angiography for the detection of stenotic disease. Despite this, it has a high negative predictive value for the assessment of both proximal and multivessel coronary disease. A further increase in both spatial resolution and contrast to noise ratio is needed before coronary MRA can be used to screen asymptomatic patients or to precisely characterize focal coronary stenosis. Although computed tomography angiography is an alternative imaging modality with higher spatial resolution than MRI, it has several limitations including the exposure of patients to ionizing radiation and nephrotoxic contrast agents. Its application to studies of heavily calcified vessels and coronary function is limited, as coronary blood-flow velocity cannot be quantified. In addition, the radiation and contrast doses limit studies in low-risk subjects, repeated studies in patients over time or with stress, and evaluation of patients with renal disease, all of which are important for screening populations and monitoring responses to therapy. Because of its multipurpose nature, coronary MRA is a well-suited imaging modality for the comprehensive characterization of CAD. With further refinement of techniques, including improvements in hardware and high field imaging, coronary MRA will likely emerge as the premier modality to characterize tissue in the coronary vessel wall and to evaluate coronary function.

References 1. Kim WY, Stuber M, Kissinger KV, Andersen NT, Manning WJ, Botnar RM. Impact of bulk cardiac motion on right coronary MR angiography and vessel wall imaging. J Magn Reson Imaging. 2001;14:383–390. 2. Stuber M, Botnar RM, Danias PG, et al. Double-oblique freebreathing high resolution three-dimensional coronary magnetic resonance angiography. J Am Coll Cardiol. 1999;34:524–531. 3. Hofman MB, Wickline SA, Lorenz CH. Quantification of in-plane motion of the coronary arteries during the cardiac cycle: implications for acquisition window duration for MR flow quantification. J Magn Reson Imaging. 1998;8:568–576. 4. Duerinckx A, Atkinson DP. Coronary MR angiography during peak-systole: work in progress. J Magn Reson Imaging. 1997; 7:979–986.

138 5. Jahnke C, Paetsch I, Nehrke K, et al. A new approach for rapid assessment of the cardiac rest period for coronary MRA. J Cardiovasc Magn Reson. 2005;7:395–399. 6. Roes SD, Korosoglou G, Schar M, et al. Correction for heart rate variability during 3D whole heart MR coronary angiography. J Magn Reson Imaging. 2008;27:1046–1053. 7. Wang Y, Grimm RC, Felmlee JP, Riederer SJ, Ehman RL. Algorithms for extracting motion information from navigator echoes. Magn Reson Med. 1996;36:117–23. 8. Edelman RR, Manning WJ, Burstein D, Paulin S. Coronary arteries: breath-hold MR angiography. Radiology. 1991;181:641–643. 9. Bornert P, Jensen D. Coronary artery imaging at 0.5 T using segmented 3D echo planar imaging. Magn Reson Med. 1995;34:779–785. 10. Goldfarb JW, Edelman RR. Coronary arteries: breath-hold, gadolinium-enhanced, three-dimensional MR angiography. Radiology. 1998;206:830–834. 11. Stuber M, Botnar RM, Danias PG, Kissinger KV, Manning WJ. Breathhold three-dimensional coronary magnetic resonance angiography using real-time navigator technology. J Cardiovasc Magn Reson. 1999;1:233–238. 12. Wang Y, Grimm RC, Rossman PJ, Debbins JP, Riederer SJ, Ehman RL. 3D coronary MR angiography in multiple breath-holds using a respiratory feedback monitor. Magn Reson Med. 1995; 34:11–16. 13. Wielopolski PA, van Geuns RJ, de Feyter PJ, Oudkerk M. Breathhold coronary MR angiography with volume targeted imaging. Radiology. 1998;209:209–219. 14. Danias PG, McConnell MV, Khasgiwala VC, Chuang ML, Edelman RR, Manning WJ. Prospective navigator correction of image position for coronary MR angiography. Radiology. 1997;203:733–736. 15. Lai P, Bi X, Jerecic R, Li D. A respiratory self-gating technique with 3D-translation compensation for free-breathing whole-heart coronary MRA. Magn Reson Med. 2009;62:731–738. 16. Sakuma H, Ichikawa Y, Chino S, Hirano T, Makino K, Takeda K. Detection of coronary artery stenosis with whole-heart coronary magnetic resonance angiography. J Am Coll Cardiol. 2006; 48:1946–1950. 17. Yang Q, Li K, Liu X, et al. Contrast-enhanced whole-heart coronary magnetic resonance angiography at 3.0-T: a comparative study with X-ray angiography in a single center. J Am Coll Cardiol. 2009;54:69–76. 18. Ehman RL, Felmlee JP. Adaptive technique for high-definition MR imaging of moving structures. Radiology. 1989;173:255–263. 19. Ehman RL, Felmlee JP. Flow artifact reduction in MRI: a review of the roles of gradient moment nulling and spatial presaturation. Magn Reson Med. 1990;14:293–307. 20. Li D, Kaushikkar S, Haacke EM, et al. Coronary arteries: threedimensional MR imaging with retrospective respiratory gating. Radiology. 1996;201:857–863. 21. Oshinski JN, Hofland L, Mukundan S, Jr., Dixon WT, Parks WJ, Pettigrew RI. Two-dimensional coronary MR angiography without breath holding. Radiology. 1996;201:737–743. 22. McConnell MV, Khasgiwala VC, Savord BJ, et al. Prospective adaptive navigator correction for breath-hold MR coronary angiography. Magn Reson Med. 1997;37:148–152. 23. Stuber M, Botnar RM, Danias PG, Kissinger KV, Manning WJ. Submillimeter three-dimensional coronary MR angiography with real-time navigator correction: comparison of navigator locations. Radiology. 1999;212:579–587. 24. Jahnke C, Paetsch I, Nehrke K, et al. Rapid and complete coronary arterial tree visualization with magnetic resonance imaging: feasibility and diagnostic performance. Eur Heart J. 2005;26: 2313–2319.

A. Hays et al. 25. Li D, Paschal CB, Haacke EM, Adler LP. Coronary arteries: threedimensional MR imaging with fat saturation and magnetization transfer contrast. Radiology. 1993;187:401–406. 26. Brittain JH, Hu BS, Wright GA, Meyer CH, Macovski A, Nishimura DG. Coronary angiography with magnetizationprepared T2 contrast. Magn Reson Med. 1995;33:689–96. 27. Botnar RM, Stuber M, Danias PG, Kissinger KV, Manning WJ. Improved coronary artery definition with T2-weighted, freebreathing, three-dimensional coronary MRA. Circulation. 1999;99:3139–3148. 28. Huber ME, Paetsch I, Schnackenburg B, et al. Performance of a new gadolinium-based intravascular contrast agent in free-breathing inversion-recovery 3D coronary MRA. Magn Reson Med. 2003;49:115–121. 29. Li D, Zheng J, Weinmann HJ. Contrast-enhanced MR imaging of coronary arteries: comparison of intra- and extravascular contrast agents in swine. Radiology. 2001;218:670–678. 30. Weber OM, Martin AJ, Higgins CB. Whole-heart steady-state free precession coronary artery magnetic resonance angiography. Magn Reson Med. 2003;50:1223–1228. 31. Stehning C, Bornert P, Nehrke K, Eggers H, Dossel O. Fast isotropic volumetric coronary MR angiography using free-breathing 3D radial balanced FFE acquisition. Magn Reson Med. 2004; 52:197–203. 32. Huber ME, Hengesbach D, Botnar RM, et al. Motion artifact reduction and vessel enhancement for free-breathing navigatorgated coronary MRA using 3D k-space reordering. Magn Reson Med. 2001;45:645–652. 33. Jhooti P, Gatehouse PD, Keegan J, Bunce NH, Taylor AM, Firmin DN. Phase ordering with automatic window selection (PAWS): a novel motion-resistant technique for 3D coronary imaging. Magn Reson Med. 2000;43:470–480. 34. Sakuma H, Ichikawa Y, Suzawa N, et al. Assessment of coronary arteries with total study time of less than 30 minutes by using whole-heart coronary MR angiography. Radiology. 2005; 237:316–321. 35. Buehrer M, Huber ME, Wiesinger F, Boesiger P, Kozerke S. Coil setup optimization for 2D-SENSE whole-heart coronary imaging. Magn Reson Med. 2006;55:460–464. 36. Niendorf T, Sodickson DK, Hardy CJ, et al. Towards whole heart coverage in a single breath-hold: Coronary artery imaging using a true 32-channel phased array MRI system. In: International Society for Magnetic Resonance Imaging; 2004; Kyoto, Japan: ISMRM; 2004. p. 703. 37. Manning WJ, Li W, Edelman RR. A preliminary report comparing magnetic resonance coronary angiography with conventional angiography. N Engl J Med. 1993;328:828–832. 38. Cheng L, Gao Y, Guaricci AI, et al. Breath-hold 3D steady-state free precession coronary MRA compared with conventional X-ray coronary angiography. J Magn Reson Imaging. 2006;23: 669–673. 39. Duerinckx AJ, Urman MK. Two-dimensional coronary MR angiography: analysis of initial clinical results. Radiology. 1994;193:731–738. 40. Müller MF, Fleisch M, Kroeker R, Chatterjee T, Meier B, Vock P. Proximal coronary artery stenosis: three-dimensional MRI with fat saturation and navigator echo. J Magn Reson Imaging. 1997;7:644–651. 41. Pennell DJ, Bogren HG, Keegan J, Firmin DN, Underwood SR. Assessment of coronary artery stenosis by magnetic resonance imaging. Heart. 1996;75:127–133. 42. Post JC, van Rossum AC, Hofman MB, de Cock CC, Valk J, Visser CA. Clinical utility of two-dimensional magnetic resonance angiography in detecting coronary artery disease. Eur Heart J. 1997;18:426–433.

9

Noncontrast Coronary Artery Imaging

43. Post JC, van Rossum AC, Hofman MB, Valk J, Visser CA. Threedimensional respiratory-gated MR angiography of coronary arteries: comparison with conventional coronary angiography. AJR Am J Roentgenol. 1996;166:1399–1404. 44. van Geuns RJ, Wielopolski PA, de Bruin HG, et al. MR coronary angiography with breath-hold targeted volumes: preliminary clinical results. Radiology. 2000;217:270–277. 45. Kim WY, Danias PG, Stuber M, et al. Coronary magnetic resonance angiography for the detection of coronary stenoses. N Engl J Med. 2001;345:1863–1869. 46. White RD, Caputo GR, Mark AS, Modin GW, Higgins CB. Coronary artery bypass graft patency: noninvasive evaluation with MR imaging. Radiology. 1987;164:681–686. 47. Aurigemma GP, Reichek N, Axel L, Schiebler M, Harris C, Kressel HY. Noninvasive determination of coronary artery bypass graft patency by cine magnetic resonance imaging. Circulation. 1989;80:1595–1602. 48. von Smekal A, Knez A, Seelos KC, et al. A comparison of ultrafast computed tomography, magnetic resonance angiography and selective angiography for the detection of coronary bypass patency [in German]. Rofo. 1997;166:185–191. 49. Kessler W, Achenbach S, Moshage W, et al. Usefulness of respiratory gated magnetic resonance coronary angiography in assessing narrowings > or = 50% in diameter in native coronary arteries and in aortocoronary bypass conduits. Am J Cardiol. 1997;80: 989–993. 50. Langerak SE, Vliegen HW, de Roos A, et al. Detection of vein graft disease using high-resolution magnetic resonance angiography. Circulation. 2002;105:328–333. 51. Bunce NH, Lorenz CH, John AS, Lesser JR, Mohiaddin RH, Pennell DJ. Coronary artery bypass graft patency: assessment with true as imaging with steady-state precession versus gadoliniumenhanced MR angiography. Radiology. 2003;227:440–446. 52. McConnell MV, Ganz P, Selwyn AP, Li W, Edelman RR, Manning WJ. Identification of anomalous coronary arteries and their anatomic course by magnetic resonance coronary angiography. Circulation. 1995;92:3158–3162. 53. Taylor AM, Thorne SA, Rubens MB, et al. Coronary artery imaging in grown up congenital heart disease: complementary role of magnetic resonance and x-ray coronary angiography. Circulation. 2000;101:1670–1678. 54. Post JC, van Rossum AC, Bronzwaer JG, et al. Magnetic resonance angiography of anomalous coronary arteries. A new gold standard for delineating the proximal course? Circulation. 1995;92:3163–171. 55. Vliegen HW, Doornbos J, de Roos A, Jukema JW, Bekedam MA, van der Wall EE. Value of fast gradient echo magnetic resonance angiography as an adjunct to coronary arteriography in detecting and confirming the course of clinically significant coronary artery anomalies. Am J Cardiol. 1997;79:773–776. 56. Machado C, Bhasin S, Soulen RL. Confirmation of anomalous origin of the right coronary artery from the left sinus of Valsalva with magnetic resonance imaging. Chest. 1993;104:1284–1286. 57. Doorey AJ, Wills JS, Blasetto J, Goldenberg EM. Usefulness of magnetic resonance imaging for diagnosing an anomalous coronary artery coursing between aorta and pulmonary trunk. Am J Cardiol. 1994;74:198–199. 58. Razmi RM Meduri A, Chun W, Rathi VK, Pohost GM. Coronary magnetic resonance angiography (CMRA): The gold standard for determining the proximal course of anomalous coronary arteries [abstract]. J Am Coll Cardiol. 2001:37:380. 59. Bunce NH, Lorenz CH, Keegan J, et al. Coronary artery anomalies: assessment with free-breathing three-dimensional coronary MR angiography. Radiology. 2003;227:201–8. 60. Bluemke DA, Achenbach S, Budoff M, et al. Noninvasive coronary artery imaging: magnetic resonance angiography and multidetector

139

61.

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76. 77.

computed tomography angiography: a scientific statement from the American Heart Association committee on cardiovascular imaging and intervention of the council on cardiovascular radiology and intervention, and the councils on clinical cardiology and cardiovascular disease in the young. Circulation. 2008;118:586–606. Greil GF, Stuber M, Botnar RM, et al. Coronary magnetic resonance angiography in adolescents and young adults with kawasaki disease. Circulation. 2002;105:908–911. Mavrogeni S MA, Papadakis E, Foussas S, Douskou M, Baras P, et al. Correlation between magnetic resonance angiography (MRA) and quantitative coronary angiography (QCA) in ectatic coronary vessels. J Cardiovasc Magn Reson. 2004;6:17–23. Glagov S, Weisenberg E, Zarins CK, Stankunavicius R, Kolettis GJ. Compensatory enlargement of human atherosclerotic coronary arteries. N Engl J Med. 1987;316:1371–1275. Kim WY, Stuber M, Bornert P, Kissinger KV, Manning WJ, Botnar RM. Three-dimensional black-blood cardiac magnetic resonance coronary vessel wall imaging detects positive arterial remodeling in patients with nonsignificant coronary artery disease. Circulation. 2002;106:296–299. Fayad ZA, Fuster V, Fallon JT, et al. Noninvasive in vivo human coronary artery lumen and wall imaging using black-blood magnetic resonance imaging. Circulation. 2000;102:506–510. Botnar RM, Stuber M, Kissinger KV, Kim WY, Spuentrup E, Manning WJ. Noninvasive coronary vessel wall and plaque imaging with magnetic resonance imaging. Circulation. 2000;102:2582–2587. Botnar RM, Kim WY, Bornert P, Stuber M, Spuentrup E, Manning WJ. 3D coronary vessel wall imaging utilizing a local inversion technique with spiral image acquisition. Magn Reson Med. 2001;46:848–854. Kim WY, Astrup AS, Stuber M, et al. Subclinical coronary and aortic atherosclerosis detected by magnetic resonance imaging in type 1 diabetes with and without diabetic nephropathy. Circulation. 2007;115:228–235. Desai MY, Lai S, Barmet C, Weiss RG, Stuber M. Reproducibility of 3D free-breathing magnetic resonance coronary vessel wall imaging. Eur Heart J. 2005;26:2320–2324. Priest AN, Bansmann PM, Mullerleile K, Adam G. Coronary vessel-wall and lumen imaging using radial k-space acquisition with MRI at 3 Tesla. Eur Radiol. 2007;17:339–346. Hundley WG, Lange RA, Clarke GD, et al. Assessment of coronary arterial flow and flow reserve in humans with magnetic resonance imaging. Circulation. 1996;93:1502–1508. Nagel E, Bornstedt A, Hug J, Schnackenburg B, Wellnhofer E, Fleck E. Noninvasive determination of coronary blood flow velocity with magnetic resonance imaging: comparison of breath-hold and navigator techniques with intravascular ultrasound. Magn Reson Med. 1999;41:544–549. Nagel E, Thouet T, Klein C, et al. Noninvasive determination of coronary blood flow velocity with cardiovascular magnetic resonance in patients after stent deployment. Circulation. 2003; 107:1738–1743. Langerak SE, Vliegen HW, Jukema JW, et al. Value of magnetic resonance imaging for the noninvasive detection of stenosis in coronary artery bypass grafts and recipient coronary arteries. Circulation. 2003;107:1502–1508. Wang Y, Rossman PJ, Grimm RC, Riederer SJ, Ehman RL. Navigator-echo-based real-time respiratory gating and triggering for reduction of respiration effects in three-dimensional coronary MR angiography. Radiology. 1996;198:55–60. Meyer CH, Hu BS, Nishimura DG, Macovski A. Fast spiral coronary artery imaging. Magn Reson Med. 1992;28:202–213. Bornert P, Stuber M, Botnar RM, et al. Direct comparison of 3D spiral vs. Cartesian gradient-echo coronary magnetic resonance angiography. Magn Reson Med.2001;46:789–794.

140 78. Thedens DR, Irarrazaval P, Sachs TS, Meyer CH, Nishimura DG. Fast magnetic resonance coronary angiography with a threedimensional stack of spirals trajectory. Magn Reson Med. 1999;41:1170–1179. 79. Bornert P, Aldefeld B, Nehrke K. Improved 3D spiral imaging for coronary MR angiography. Magn Reson Med. 2001;45:172–175. 80. Reeder SB, Atalay MK, McVeigh ER, Zerhouni EA, Forder JR. Quantitative cardiac perfusion: a noninvasive spin-labeling method that exploits coronary vessel geometry. Radiology. 1996;200:177–184. 81. Stuber M, Bornert P, Spuentrup E, Botnar RM, Manning WJ. Selective three-dimensional visualization of the coronary arterial lumen using arterial spin tagging. Magn Reson Med. 2002; 47:322–329. 82. Wang SJ, Hu BS, Macovski A, Nishimura DG. Coronary angiography using fast selective inversion recovery. Magn Reson Med. 1991;18:417–423. 83. Wang SJ, Nishimura DG, Macovski A. Multiple-readout selective inversion recovery angiography. Magn Reson Med. 1991;17: 244–2451. 84. Katoh M, Spuentrup E, Barmet C, Stuber M. Local re-inversion coronary MR angiography: arterial spin-labeling without the need for subtraction. J Magn Reson Imaging. 2008;27:913–917. 85. Katoh M, Spuentrup E, Stuber M, et al. Flow targeted 3D steadystate free-precession coronary MR angiography: comparison of three different imaging approaches. Invest Radiol. 2009;44: 757–762. 86. Deshpande VS, Shea SM, Laub G, Simonetti OP, Finn JP, Li D. 3D magnetization-prepared true-FISP: a new technique for imaging coronary arteries. Magn Reson Med. 2001;46:494–502. 87. Spuentrup E, Buecker A, Stuber M, et al. Navigator-gated coronary magnetic resonance angiography using steady-state-freeprecession: comparison to standard t2-prepared gradient-echo and spiral imaging. Invest Radiol. 2003;38:263–268. 88. Spuentrup E, Bornert P, Botnar RM, Groen JP, Manning WJ, Stuber M. Navigator-gated free-breathing three-dimensional balanced fast field echo (TrueFISP) coronary magnetic resonance angiography. Invest Radiol. 2002;37:637–642. 89. Deshpande VS, Li D. Contrast-enhanced coronary artery imaging using 3D trueFISP. Magn Reson Med. 2003;50:570–577. 90. Kaul MG SA, Bansmann PM, et al. Evaluation of balanced steadystate free precession (TrueFISP) and K-space segmented gradient echo sequences for 3D coronary MR angiography with navigator gating at 3 Tesla. Rofo. 2004;176:1560–1565.

A. Hays et al. 91. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999; 42:952–962. 92. Sodickson DK, Griswold MA, Jakob PM. SMASH imaging. Magn Reson Imaging Clin N Am. 1999;7:237–254, vii–viii. 93. Huang F, Akao J, Vijayakumar S, Duensing GR, Limkeman M. k-t GRAPPA: a k-space implementation for dynamic MRI with high reduction factor. Magn Reson Med. 2005;54:1172–1184. 94. Lai P, Huang F, Larson AC, Li D. Fast four-dimensional coronary MR angiography with k-t GRAPPA. J Magn Reson Imaging. 2008;27:659–665. 95. Weiger M, Pruessmann KP, Boesiger P. Cardiac real-time imaging using SENSE. SENSitivity Encoding scheme. Magn Reson Med. 2000;43:177–184. 96. Huber ME, Kozerke S, Pruessmann KP, Smink J, Boesiger P. Sensitivity-encoded coronary MRA at 3 T. Magn Reson Med. 2004;52:221–227. 97. Stuber M, Botnar RM, Fischer SE, et al. Preliminary report on in vivo coronary MRA at 3 Tesla in humans. Magn Reson Med. 2002;48:425–429. 98. Fischer SE, Wickline SA, Lorenz CH. Novel real-time R-wave detection algorithm based on the vectorcardiogram for accurate gated magnetic resonance acquisitions. Magn Reson Med. 1999;42:361–370. 99. Schar M, Vonken EJ, Stuber M. Simultaneous B(0)− and B(1)+ − map acquisition for fast localized shim, frequency, and RF power determination in the heart at 3 T. Magn Reson Med. 2010;63: 419–426. 100. Kubach MR, Bornstedt A, Hombach V, et al. Cardiac phasespecific shimming (CPSS) for SSFP MR cine imaging at 3 T. Phys Med Biol. 2009;54:N467–478. 101. Sommer T, Hackenbroch M, Hofer U, et al. Coronary MR angiography at 3.0 T versus that at 1.5 T: initial results in patients suspected of having coronary artery disease. Radiology. 2005; 234:718–725. 102. Nezafat R, Stuber M, Ouwerkerk R, Gharib AM, Desai MY, Pettigrew RI. B1-insensitive T2 preparation for improved coronary magnetic resonance angiography at 3 T. Magn Reson Med. 2006;55:858–864. 103. Schar M, Kozerke S, Fischer SE, Boesiger P. Cardiac SSFP imaging at 3 Tesla. Magn Reson Med. 2004;51:799–806. 104. van Elderen SG, Versluis MJ, Webb AG, et al. Initial results on in vivo human coronary magnetic resonance angiography at 7 Tesla. Magn Reson Med. 2009;62:1379–1384.

Contrast-Enhanced MR Angiography of the Coronary Arteries

10

Qi Yang and Debiao Li

Introduction

Technical Considerations for CMRA

X-ray angiography is the current gold standard for the diagnosis of coronary artery disease (CAD). However, X-ray angiography is an invasive and costly procedure associated with a small but definite morbidity and mortality. There is a clear need for a noninvasive and more reliable method of directly detecting functionally significant CAD. Magnetic resonance imaging (MRI) overcomes a lot of the problems associated with X-ray angiography and has shown great advantages for the diagnosis of CAD. In addition to being noninvasive, radiation free, and cost-effective, it can provide functional, hemodynamic, and metabolic data as well as anatomical images in the same setting for a comprehensive exam of CAD. Over the past 15 years, substantial advances have been made in the field of coronary MR angiography (CMRA). However, for successful CMRA imaging, a series of technical challenges have to be overcome, including small vessel diameter, intrinsic and extrinsic motion of the heart, and complex geometry of the arterial trees. In addition, sufficient contrast between the coronary lumen and the surrounding tissue is crucial for the visualization of the coronary vessels. In this chapter, we briefly review the historical development of noncontrast CMRA, starting from the motion compensation techniques, with subsequent developments, including SSFP sequences, whole-heart acquisitions approach, and contrast-enhanced CMRA. Then, the current status of technological developments of contrast-enhanced CMRA at 3 T and its clinical role for the evaluation of CAD are also considered.

Motion Compensation

Q. Yang, MD, PhD () Department of Radiology, Xuanwu Hospital, Capital Medical University, Beijing, China D. Li, PhD Cedars-Sinai Medical Center, Biomedical Imaging Research Institute, Los Angeles, CA, USA

The major obstacles for obtaining CMRA images are the two types of motion: cardiac motion related to myocardial contraction/relaxation and respiratory motion attributable to diaphragm and chest wall movement. The extent of motion exceeds the diameter of the coronary artery, blurring artifacts will occur unless adequate motion compensation techniques are applied. To account for cardiac motion, ECG signal is commonly used to synchronize data acquisition to the quiescent period of each heart beat [1] and CMRA data were collected over multiple cardiac cycles. To deal with respiratory motion two possible solutions have been used: breath-hold and free-breathing imaging. Breath-hold is a straightforward approach and is easy to implement. However, the spatial coverage and resolution are limited by the patient’s ability to hold his/her breath. The implementation of navigator techniques for coronary artery imaging has enabled free breathing CMRA and allows higher resolution and larger coverage [2]. Navigator methods involve the detection of a signal at the interface of the dome of the diaphragm and lung tissue. The navigator signal can be produced by a slice-selective 90–180° radiofrequency (RF) pulse pair, where the rectangular slices excited by each pulse are oriented to produce a diamondshaped intersection. Imaging time can be extended from previous one breath-hold to several minutes with free breathing.

Spatial Resolution Insufficient spatial resolution prevents the consistent visualization of the distal and branch vessels and submillimeter spatial resolution is desirable in CMRA [3]. However, higher spatial resolution is associated with decreased signal-tonoise ratio (SNR), longer imaging time. In addition, precise motion compensation is required. Therefore, it is important

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_10, © Springer Science+Business Media, LLC 2012

141

142

Q. Yang and D. Li

Fig. 10.1 Three-dimensional free-breathing coronary MRI using (a) an inversion recovery prepulse and novel intravascular contrast agent (B-22,956, Bracco Spa, Milan) and comparison with (b) conventional

noncontrast T2 preparation coronary MRI. Note the improved contrast with the intravascular agent. (From Huber ME et al. [12].)

to get appropriate balance between spatial resolution, SNR and imaging time when determining imaging parameters of CMRA.

making it sensitive to phase error from complex flow in region with stenosis or local field inhomogeneities. Moreover, the sequence uses a combination of very short TR and large flip angles so that power deposition is high. Power deposition becomes a limiting factor at 3 T.

Contrast Between Blood and Surrounding Tissue The coronary arteries are surrounded by fat proximally and by myocardial tissue along their entire course. It is natural to generate contrast between coronary luminal blood and the surrounding tissue based on refreshing of the inflow blood with gradient echo techniques. Breath-hold two-dimensional (2D) segmented k-space gradient echo sequence was the first robust approach for CMRA using in-flow effect [4]. The breath-hold 2D approach was easy to implement and had been used in the visualization of the proximal coronary arteries. This, in conjunction with frequency-selective saturation pulse, fat suppression can be achieved [5]. Thereafter, the introduction of three-dimensional (3D) gradient echo CMRA eliminates some drawbacks of 2D CMRA related to long scan time and low SNR [5, 6]. These features of 3D CMRA permit 3D postprocessing of the entire coronary arteries. Nevertheless, the signal from the stationary or slow moving blood also gets saturated in 3D acquisitions due to excitation of a thick slab, and thus reduces the contrast. Therefore, magnetization preparation is necessary to suppress the fat and myocardial signal and enable clear delineation of the coronary arteries [6, 7]. With the improvement in gradient systems, steady-state free precession (SSFP) imaging techniques have been extensively used in cardiac MRI [8]. The signal intensity on SSFP sequence is primarily determined by the T2/T1 ratio which makes it intrinsically beneficial for cardiac imaging. Transverse magnetization is largely preserved between RF pulses, allowing for higher flip angle and SNR than conventional spoiled gradient echo sequences. With SSFP sequence, relatively high spatial resolution images with adequate SNR and CNR can be acquired reliably. However, the balanced gradient structure accumulates phases in each readout train,

Benefits of Contrast Agents for CMRA The use of T1-shortening contrast agents has revolutionized MRA of the entire body [9]. It dramatically improved blood SNR and permitted the use of short repetition times. Vessel contrast for CMRA can be further improved by administering T1-shortening contrast agents [10]. It is less important how the imaging plane is selected (i.e., parallel or perpendicular to the flow direction) since the contrast is generated by T1 differences in CMRA. In addition, the blood signal becomes largely flow independent, which is important for the depiction of slow-flowing blood and the reliable detection of coronary artery stenosis. Various paramagnetic T1-shortening contrast agents have been used to generate “bright blood” CMRA images. Based on the capability of diffusing to interstitial space, these agents are typically classified as intra- or extravascular agents. For CMRA, extravascular contrast agent is typically administered in a short time to assure adequate T1-shortening of blood pool. The spatial resolution and/or 3D coverage for each contrast-enhanced scan are limited by confining data acquisition within a short time-frame that coincides with the arterial phase of contrast media. In order to overcome the inherent limitations of extracellular contrast agents, MR blood pool agents have been developed due to the much higher T1 relaxivity and longer half-life in the blood pool [11]. Several researchers have demonstrated that combining with an inversion prepulse to suppress myocardial signal intravascular agent allows for improved arterial contrast on 3D gradient echo CMRA with thick 3D volume [12] (Fig. 10.1).

10

Contrast-Enhanced MR Angiography of the Coronary Arteries

Methods of Contrast Agent Administration for CMRA MR Smartprep and fluoroscopic triggering have been proposed for real-time triggering of contrast-enhanced MRA in pulmonary arteries, aorta, abdominal arteries, and peripheral arteries [13, 14]. For contrast-enhanced CMRA, currently there are two approaches. The first approach to coronary imaging with extracellular contrast agents is to image during breath-holding and the first pass of a contrast agent. One must image rapidly after injection to preserve good contrast between blood in the coronary arteries and the myocardium. The second is slow injection of the contrast agent in conjunction with a free breathing; respiratory-gated sequence. The major advantage of slow-injection, respiratory-gated coronary artery imaging is the relative flexibility in choosing TR, spatial resolution, and coverage. One major limitation is the reduced contrast agent concentration in the blood pool, resulting in longer T1 as compared with dynamic scans with faster injection. Gadobenate dimeglumine (Gd-BOPTA, Bracco Imaging SpA, Milan, Italy) has a roughly twofold higher T1 relaxivity in blood compared to other clinical extracellular contrast agents currently in widespread use. The higher in vivo relaxivity and prolonged half-life of Gd-BOPTA make it more suitable for whole-heart CMRA with a gradient echo imaging sequence.

Advanced Methods in CMRA k-Space Trajectories The MR sampling employed to acquire the actual MR data has an impact on the motion sensitivity and on the image quality. A variety of k-space trajectories have been tested in an effort to maximize spatial resolution and scan efficiency. Due to the limited system and reconstruction requirements, most coronary artery imaging protocols use Cartesian k-space sampling for image acquisition. Cartesian trajectories acquire one or a series of phase-encoding lines within each R–R interval efficiently. Non-Cartesian trajectories, which include multishot echo planar, spiral, and radial, each offer advantages and disadvantages for CMRA. Advantages of spiral acquisitions include a more efficient filling of k-space, a higher SNR, and favorable flow properties [15]. Drawbacks include marked sensitivity to off-resonance effects and the need for specialized image reconstruction algorithms. Radial methods acquire a series of phase-encoding lines of data that are oriented in a radial pattern. Each radial line of data passes through the center of k-space. Thus, radial imaging methods naturally are resistant to motion and flow-related artifacts for

143

CMRA and offer the benefit of more rapid acquisitions [16]. As in spirally sampled MRI, radial k-space sampling depends on specialized reconstruction algorithms to produce images. These reconstruction algorithms are generally not available on commercial MRI scanners and therefore restrict radial MRI to few specialized research groups.

SSFP Whole-Heart CMRA Due to the intrinsically tortuous course of coronary arteries, it has always been of interest to cover the entire heart to depict the long segments of all major coronary arteries. Li et al. [2] first reported using multiple overlapped 3D slabs to cover the whole-heart with retrospective respiratory gating. Advances in SSFP sequences make it feasible to perform whole-heart imaging with higher blood signal intensity as compared to 3D gradient echo sequences. SSFP sequences permit acquisition of large 3D axial volume that encompasses whole-heart without losing arterial contrast in 10–15 min at 1.5 T [17, 18].

Contrast-Enhanced EPI Whole-Heart Technique Imaging speed is important for whole-heart CMRA as reduced scan time leads to improved study success rate and higher image quality. Echo planar imaging (EPI) is one of the data acquisition strategies in which several lines of k-space are acquired following a single RF excitation to reduce the total imaging time. Thick slab coverage (40 slices with 2 mm thickness) within a single breath-hold is possible by using a segmented EPI image acquisition. Conventional EPI techniques suffer from low SNR and spatial resolution. T1 shortening contrast agent can be used to boost the blood signal intensity and imaging contrast [19]. A recently proposed interleaved GRE-EPI acquisition scheme has been used to speed up contrast-enhanced whole-heart CMRA [20]. The preliminary results showed excellent delineation of all the major coronary arteries with scan time reduced by a factor of 2 compared with the SSFP acquisition.

Benefits of Parallel Imaging Parallel imaging techniques, such as the k-space-based technique (GRAPPA: generalized autocalibrating partially parallel acquisitions) and the image space-based technique (SENSE: sensitivity encoding) enable substantial scan time reduction in cardiac imaging. The major drawback of using parallel imaging techniques is the loss of SNR. Therefore, it is important to have an appropriate balance between speed

144

Q. Yang and D. Li

Fig. 10.2 Sample reformatted right coronary MR angiography: (a) acquired with no T2 Prep; (b) with two MLEV-weighted composite T2 Prep; (c) with adiabatic T2 Prep. Arrows in (b) point to the artifacts

resulting from T2 Prep sequence; note the suppression of the banding artifacts in (c) and also the homogeneity of the signal. (From Nezafat R et al. [23].)

and image quality. Fortunately, for higher fields like 3.0 T, the natural SNR gain can be used to trade the acquisition time. The development of 32- or even 128-channel phased-array coils which are designed for 2D parallel imaging, higher acceleration factors will permit higher acceleration and scan time reduction for whole-heart CMRA.

Contrast-Enhanced Whole-Heart CMRA at 3 T

Advantages and Disadvantages of High-Field Coronary Imaging The signal strength in MRI increases in proportion to the strength of the main magnetic field. The higher signal produced by 3.0 T or even 7.0 T MRI scanners results in a proportionate increase in SNR. In other words, the SNR of an image acquired at 3.0 T is roughly twice that of the same image acquired on a 1.5 T scanner. This SNR gain allows images with reduced imaging time or improved spatial resolution, with the same clinically acceptable SNR as the analogous 1.5 T image. However, CMRA at high-field strength system has several major challenges, particularly with the SSFP technique, including more off-resonance image artifacts due to increased B0 field inhomogeneity, greater image intensity variations due to B1 field inhomogeneity, and limited flip angles due to the higher RF power deposition. Many measures have been developed to address the problems with SSFP imaging at 3 T, such as shifting the synthesizer frequency to reduce off-resonance-related image artifacts, or improving field homogeneity by applying localized linear or second-order shimming [21, 22]. The use of conventional T2-preparation is more challenging, thus, a new adiabatic refocusing T2 Prep sequence was developed which exploits the insensitivity of the adiabatic pulses to field inhomogeneity, without exceeding the SAR limitations [23] (Fig. 10.2). Recently, the first human coronary MR images were successfully obtained at 7 T, and other studies that take advantage of new high-field specific improvements are ongoing [24].

Noncontrast whole-heart CMRA approach at 1.5 T necessitates the use of SSFP sequences, which has superior SNR to gradient echo sequences. However, due to increased image artifacts at 3 T with SSFP technique, further SNR gain has not been directly translated into improved coronary artery delineation. Conventional spoiled gradient echo sequences are relatively insensitive to the increased field inhomogeneities at 3 T. Contrast-enhanced whole-heart CMRA at 3 T with slow infusion of contrast agent with high T1 relaxivity has recently been proposed [25]. An inversion recovery prepared spoiled gradient echo sequence, which is relatively insensitive to the increased field inhomogeneities at 3 T, was employed. Slow infusion of a high relaxivity agent allows prolonged blood enhancement time required for whole-heart MRA. Ultrashort TR and high acceleration factor allows significantly reduced acquisition time using contrast-enhanced CMRA at 3 T. Previous comparison study performed in the same volunteers has proved that contrast-enhanced CMRA at 3 T provides higher coronary artery contrast-to-noise ratio (CNR), better vessel depiction, and shorter imaging time than the SSFP approach at 1.5 T [26] (Fig. 10.3). In a later study, the blood pool contrast agent gadofosveset was shown to improve the overall CNR and the delineation of distal coronary segments for contrast-enhanced CMRA at 3 T in comparison to noncontrast SSFP CMRA at 1.5 T [27].

Practical Recommendations for Whole-Heart Coronary MRA at 3 T Patient Training Patient training and practice before data acquisition for maintaining regular breathing is useful to improve the gating efficiency and image quality of CMRA. Patients were trained

10

Contrast-Enhanced MR Angiography of the Coronary Arteries

145

Contrast Injection For 3 T CMRA, 0.2 mmol/kg body weight of Gadobenate dimeglumine (MultiHance; Bracco Imaging SpA, Milan, Italy) was slowly infused using a power injector at a rate of 0.3 ml/s, immediately followed by 20 ml saline at the same rate. Sixty seconds after the initiation of contrast administration, whole-heart CMRA data acquisition was started.

Survey Scanning

Fig. 10.3 Maximum intensity projection images of a 45-year-old healthy male volunteer demonstrate that contrast-enhanced coronary MRA at 3.0 T (b) has better CNR than SSFP coronary MRA at 1.5 T (a). Original axial images show that the mid LAD (c, arrowhead) and LCX (c, arrow) are buried in pericardial fluid at 1.5 T, whereas the arteries and D2 are clearly delineated at 3.0 T (d). LM, LAD, LCX, D2, AO, LV, and RV indicate left main artery, left anterior descending artery, left circumflex, the second diagonal branch, aorta, left ventricle, and right ventricle, respectively. (Reproduced with permission from Liu X et al. [26].)

to perform regular, shallow breathing and to avoid changes in depth of breathing during the data acquisition. Abdominal belt was rolled tightly around the upper abdomen during deep inspiration to suppress the motion of the diaphragm.

Vector Electrocardiogram at 3 T Regular rhythm and accurate ECG synchronization and R-wave detection are imperative for coronary CMRA. Use of a patient-specific quiescent period is recommended, this can be identified by the acquisition of high temporal resolution ECG-triggered cine images. However, under the influence of higher field strength, magneto-hydrodynamic effect led to considerable artificial augmentation of the T-wave of the ECG. The augmentation of the T-wave may mislead the R-wave detection so that triggering is performed on the T-wave instead of the R-wave. The vector ECG allowed reliable R-wave triggering and has been found to be very robust for R-wave detection at 3 T. Carefully positioning of the ECG leads may impact image quality.

For localization of the coronary arteries and for identification of the period of minimal myocardial motion, three scout scans are recommended: Scout 1: A low-resolution 2D scout images were first obtained in three orthogonal orientations to identify the position of the heart and diaphragm. The scan is performed during free breathing. Scout 2: Retrospective ECG-triggered cine images (50 cardiac phases reconstructed) were acquired in a four-chamber view using a fast low-angle shot (FLASH) sequence during free breathing. The global cardiac motion was visually assessed from cine images to determine the patient-specific triggerdelay time and duration of data acquisition window per heartbeat.

Navigator Efficiency A major challenge for CMRA remains to be respirationinduced motion artifacts. Adaptive navigator-gating and motion correction is an effective method for reducing respiratory motion artifacts. However, drift of the diaphragm and fluctuations of the breathing pattern during this period can decrease the respiratory gating efficiency, increase respiratory motion artifacts, or even lead to failure of the measurement. Patient training and practice before data acquisition for maintaining regular breathing should be useful to improve the gating efficiency and image quality of CMRA.

Recent Clinical Applications of 3 T ContrastEnhanced Whole-Heart Coronary MRA 3 T contrast-enhanced whole-heart CMRA now represents the current state-of-the-art technique. In our recent singlecenter study, we have prospectively examined the diagnostic value of contrast-enhanced whole-heart CMRA at 3 T on patients suspected of CAD [28] (Figs. 10.4 and 10.5). Our study demonstrated that acquisition of CMRA at 3 T was successful in 62 of 69 (90%) patients, with the averaged

146

Q. Yang and D. Li

Fig. 10.4 Reformatted images (a) of MRA in a 55-year-old female patient demonstrate significant coronary stenoses at proximal LCX (arrow) and normal RCA which were consistent with the findings (arrow) of conventional coronary angiography (b, c)

Fig. 10.5 3 T contrast-enhanced whole-heart CMRA images of a 75-year-old male patient with atypical chest pain. CMRA MIP images (a and b) show a significant stenosis in the proximal LCX and a nonsignificant stenosis in the middle RCA (arrows), respectively. VR images (syngo InSpace, Siemens AG Healthcare, Erlangen, Germany) (c and d) have the same findings in LCX and RCA, which were consistent with the conventional coronary angiography (e and f). (Reproduced with permission from Yang Q et al. [28].)

acquisition duration of 9.0 ± 1.9 min. 3 T contrast-enhanced whole-heart CMRA allows for ruling out significant CAD with high sensitivity and moderate specificity. The sensitivity, specificity, and accuracy of whole-heart CMRA for detecting significant stenoses were 91.6% (87/95), 83.1% (570/686), 84.1% (657/781), respectively, on a per-segment basis. The results obtained in our study compare favorably with other single-center studies of whole-heart coronary MRA at 1.5 T [29]. These results are also quite similar to the recent experience with 64-slice multidetector CT angiography in a multicenter study [30]. However, a recent published meta-analysis suggests that coronary CTA has better sensitivity and specificity than CMRA and is therefore advantageous for detecting and ruling out clinically relevant coronary stenoses [31]. Despite the excellent diagnostic accuracy, coronary CTA has several disadvantages of requiring rapid injection of iodinated contrast medium and of exposing patients to ionizing radiation. In addition, blooming artifact from calcification leads to false positive diagnosis in many cases. MR does not suffer from these artifacts caused by calcification; and CMRA can potentially depict the lumen of calcified coronary arteries (Fig. 10.6). Both coronary CTA and CMRA can provide lumenographic information to determine the presence and extent of CAD. However, the functional implications of the lesion are more important. The combination of CMRA with tissue perfusion and viability provides a comprehensive assessment of the patient with known or suspected CAD. Due to the use of contrast agent, high-resolution 3D delay enhancement MR images can be reconstructed from the CMRA images. This facilitates the 3D reformation in any slice orientation as well as precise quantification of the damaged tissue and the direct association of the infracted territory to the respective coronary artery lesion. The major advantage is that it allows for a fast and comprehensive assessment of both coronary artery stenosis and myocardial tissue damage in a single noninvasive and radiation free test.

10

Contrast-Enhanced MR Angiography of the Coronary Arteries

147

Fig. 10.6 Maximum intensity projection (MIP) image of CTA in a 59-year-old man (a) shows a mixed plaque with a severe stenosis at proximal LAD. MRA MIP image (b) detects the stenosis with good correlation with CTA and X-ray angiography (c)

Summary and Future Directions in ContrastEnhanced Coronary MRA 4D (3D cine, or cardiac-motion resolved) coronary artery imaging represents one of the future directions for coronary MRA. Such technique can potentially provide high temporal and spatial resolution coronary artery images. The combination of contrast agent and spoiled gradient-echo sequence (e.g., FLASH) at 3 T is the method of choice for coronary MRA. This will permit a major step forward for the clinical use of CMRA. The use of blood pool contrast agent might open the door to further improve the diagnostic accuracy of contrast-enhanced CMRA at 3 T. The Holy Grail for CMRA is to obtain entire whole-heart data set in one cardiac cycle.

References 1. Wang Y, Vidan E, Bergman GW. Cardiac motion of coronary arteries: variability in the rest period and implications for coronary MR angiography. Radiology. 1999;213:751–758. 2. Li D, Kaushikkar S, Haacke E, et al. Coronary arteries: threedimensional MR imaging with retrospective respiratory gating. Radiology. 1996;201:857–863. 3. Schar M, Kim WY, Stuber M, Boesiger P, Manning WJ, Botnar RM. The impact of spatial resolution and respiratory motion on MR imaging of atherosclerotic plaque. J Magn Reson Imaging. 2003;17:538–544. 4. Edelman R, Manning W, Burstein D, Paulin S. Coronary arteries: breath-hold MR angiography. Radiology. 1991;181:641–643. 5. Li D, Paschal CB, Haacke EM, Adler LP. Coronary arteries: threedimensional MR imaging with fat saturation and magnetization transfer contrast. Radiology. 1993;187:401–406. 6. Botnar RM, Stuber M, Danias PG, Kissinger KV, Manning WJ. Improved coronary artery definition with T2-weighted, free-breathing, three-dimensional coronary MRA. Circulation. 1999;99: 3139–3148. 7. Brittain JH, Hu BS, Wright GA, Meyer CH, Macovski A, Nishimura DG. Coronary angiography with magnetization-prepared T2 contrast. Magn Reson Med. 1995;33:689–696.

8. Carr JC, Simonetti O, Bundy J, Li D, Pereles S, Finn JP. Cine MR angiography of the heart with segmented true fast imaging with steady-state precession. Radiology. 2001;219:828–834. 9. Prince MR, Yucel EK, Kaufman JA, Harrison DC, Geller SC. Dynamic gadolinium-enhanced three-dimensional abdominal MR arteriography. J Magn Reson Imaging. 1993;3:877–881. 10. Zheng J, Li D, Bae KT, Woodard P, Haacke EM. Three-dimensional gadolinium-enhanced coronary magnetic resonance angiography: initial experience. J Cardiovasc Magn Reson. 1999;1:33–41. 11. Anzai Y, Prince MR, Chenevert TL, et al. MR angiography with an ultrasmall superparamagnetic iron oxide blood pool agent. J Magn Reson Imaging. 1997;7:209–214. 12. Huber ME, Paetsch I, Schnackenburg B, et al. Performance of a new gadolinium-based intravascular contrast agent in free-breathing inversion-recovery 3D coronary MRA. Magn Reson Med. 2003; 49:115–121. 13. Foo TK, Saranathan M, Prince MR, Chenevert TL. Automated detection of bolus arrival and initiation of data acquisition in fast, threedimensional, gadolinium-enhanced MR angiography. Radiology. 1997;203:275–280. 14. Wilman AH, Riederer SJ, King BF, Debbins JP, Rossman PJ, Ehman RL. Fluoroscopically triggered contrast-enhanced threedimensional MR angiography with elliptical centric view order: application to the renal arteries. Radiology. 1997;205:137–146. 15. Meyer CH, Hu BS, Nishimura DG, Macovski A. Fast spiral coronary artery imaging. Magn Reson Med. 1992;28:202–213. 16. Stehning C, Bornert P, Nehrke K, Eggers H, Dossel O. Fast isotropic volumetric coronary MR angiography using free-breathing 3D radial balanced FFE acquisition. Magn Reson Med. 2004;52:197–203. 17. Sakuma H, Ichikawa Y, Suzawa N, et al. Assessment of coronary arteries with total study time of less than 30 minutes by using wholeheart coronary MR angiography. Radiology. 2005;237:316–321. 18. Weber OM, Martin AJ, Higgins CB. Whole-heart steady-state free precession coronary artery magnetic resonance angiography. Magn Reson Med. 2003;50:1223–1228. 19. Deshpande VS, Wielopolski PA, Shea SM, Carr JC, Zheng J, Li D. Coronary artery imaging using contrast-enhanced 3D segmented EPI. J Magn Reson Imaging. 2001;13:676–681. 20. Bhat H, Zuehlsdorff S, Bi X, Li D. Whole-heart contrast-enhanced coronary magnetic resonance angiography using gradient echo interleaved EPI. Magn Reson Med. 2009;61:1388–1395. 21. Deshpande VS, Shea SM, Li D. Artifact reduction in true-FISP imaging of the coronary arteries by adjusting imaging frequency. Magn Reson Med. 2003;49:803–809. 22. Schar M, Kozerke S, Fischer SE, Boesiger P. Cardiac SSFP imaging at 3 Tesla. Magn Reson Med. 2004;51:799–806.

148 23. Nezafat R, Stuber M, Ouwerkerk R, Gharib AM, Desai MY, Pettigrew RI. B1-insensitive T2 preparation for improved coronary magnetic resonance angiography at 3 T. Magn Reson Med. 2006;55:858–864. 24. van Elderen SG, Versluis MJ, Webb AG, et al. Initial results on in vivo human coronary MR angiography at 7 T. Magn Reson Med. 2009;62:1379–1384. 25. Bi X, Carr JC, Li D. Whole-heart coronary magnetic resonance angiography at 3 Tesla in 5 minutes with slow infusion of Gd-BOPTA, a high-relaxivity clinical contrast agent. Magn Reson Med. 2007;58:1–7. 26. Liu X, Bi X, Huang J, Jerecic R, Carr J, Li D. Contrast-enhanced whole-heart coronary magnetic resonance angiography at 3.0 T: comparison with steady-state free precession technique at 1.5 T. Invest Radiol. 2008;43:663–668. 27. Prompona M, Cyran C, Nikolaou K, Bauner K, Reiser M, Huber A. Contrast-enhanced whole-heart MR coronary angiography at 3.0 T

Q. Yang and D. Li

28.

29.

30.

31.

using the intravascular contrast agent gadofosveset. Invest Radiol. 2009;44:369–374. Yang Q, Li K, Liu X, et al. Contrast-enhanced whole-heart coronary magnetic resonance angiography at 3.0-T: a comparative study with X-ray angiography in a single center. J Am Coll Cardiol. 2009;54:69–76. Sakuma H, Ichikawa Y, Chino S, Hirano T, Makino K, Takeda K. Detection of coronary artery stenosis with whole-heart coronary magnetic resonance angiography. J Am Coll Cardiol. 2006;48:1946–1950. Miller JM, Rochitte CE, Dewey M, et al. Diagnostic performance of coronary angiography by 64-row CT. N Engl J Med. 2008;359:2324–2336. Schuetz GM, Zacharopoulou NM, Schlattmann P, Dewey M. Metaanalysis: noninvasive coronary angiography using computed tomography versus magnetic resonance imaging. Ann Intern Med. 2010;152:167–177.

MR Angiography and High Field Strength: 3.0 T and Higher

11

Harald H. Quick and Mark E. Ladd

Introduction The motivation for performing MR angiography (MRA) at higher magnetic field strength can be appreciated by answering a few simple questions: Would you like to increase your spatial resolution in time-of-flight (TOF) or contrast-enhanced (CE) MRA, increase your temporal resolution in dynamic MRA applications, decrease your contrast agent dose in CE-MRA, or even use new imaging contrasts for MRA not available at lower field strength? Higher static magnetic field strengths open opportunities in all of these areas. Doubling the static magnetic field strength from the clinical standard 1.5 T to the clinically already established 3.0 T is associated with doubling the inherent signal-to-noise ratio (SNR) in the MR experiment, and thus is a starting point to realizing parts of the above-mentioned wish list. While the benefits of 3.0 T over 1.5 T have already been successfully demonstrated in the literature for certain clinical MRA applications, the potential further gain in SNR cannot simply be extrapolated to higher field strength, such as the still young and emerging 7.0 T technology. In this chapter, the benefits of high-field MRI (i.e., 7.0 T) for angiographic applications are described. Since nothing comes for free in MR, we first have to take into account the physical parameters that change with increasing field strength and that ultimately bolster or undermine specific MRA applications. We begin by comparing 1.5 vs. 3.0 T MRA examples, take a look at TOF and non-TOF MRA alternatives in intracranial MRA at 7.0-T field strength, and then slowly

H.H. Quick, PhD () Institute of Medical Physics (IMP), Friedrich-Alexander-University Erlangen-Nürnberg, Henkestr. 91, 91052 Erlangen, Germany e-mail: [email protected] M.E. Ladd, PhD Erwin L. Hahn Institute for MRI, University Duisburg-Essen, Arendahls Wiese 199, 45141 Essen, Germany

move down below the neckline to explore MRA applications in non-neuro body regions at 7.0 T.

Increasing the Field Strength Physics Background and Practical Consequences for MRA One of the major benefits when moving to high-field MR is the potentially higher achievable overall SNR, whereas the specific absorption rate (SAR) is one of the greatest limiting factors. While the SNR increases linearly with the field strength, the SAR increases quadratically (Fig. 11.1), often practically limiting the use of radiofrequency (RF)-intense imaging sequences. Consequently, MRI pulse sequences that run without limitations at 1.5 T might encounter SAR limits when ported to 3.0 T without modification. In many cases, simple modifications to the image acquisition parameters are required. It may be sufficient to alter sequence parameters, such as lowering the excitation flip angle, increasing the repetition time (TR), and/or reducing the number of slices. The use of RF-based flow saturation pulses as is common in TOF MRA for the suppression of either venous or arterial flow might be limited as well in high-field applications due to SAR constraints. The longitudinal relaxation times T1 of most stationary tissues are significantly longer at higher field strength [1, 2]. This leads to improved background suppression in CE-MRA applications. Associated with this is an improved vessel-totissue contrast in T1-weighted CE-MRA at high field, and in conjunction with the higher MR signal sensitivity, CE-MRA has shown superior results in images acquired on 3.0-T scanners when compared to images acquired on 1.5-T scanners [3–5] (Fig. 11.2). The increase in SNR that goes along with scanning on 3.0-T scanners used compensated for decrease in contrast agent, and reduced risk to the patient dose in CE-MRA (Fig. 11.3). A direct comparison of CE-MRA images acquired at 1.5 T to images acquired at 3.0 T with one half of the contrast agent dose showed no appreciable decease in image quality [6–8].

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_11, © Springer Science+Business Media, LLC 2012

149

150

Fig. 11.1 Signal-to-noise ratio (SNR) and specific absorption rate (SAR) in arbitrary units as function of the magnetic field strength. While the SNR increases approximately linearly with increasing field strength, SAR increases quadratically. Thus, doubling the field strength from 1.5 to 3.0 T would theoretically double the SNR and quadruple the SAR. The desirable increase in SNR at high field strength can be traded for increased spatial and/or temporal resolution and/or for reduced contrast dose in CE-MRA. The undesirable side effect of increasing SAR may practically restrict certain imaging parameters (e.g., limited flip angle or number of slices, prolonged TR, constrained RF saturation pulses, etc.)

Fig. 11.2 In this field strength comparison (1.5 T (a) vs. 3.0 T (b)) of 3D CE-MRA of the supraaortal arteries, the SNR gain was used to increase the spatial resolution. In direct comparison to the 1.5-T image (a), the 3.0-T CE-MRA (b) reveals increased vessel detail and much better visualization of fine arteries. (a) Imaging parameters for 1.5 T were: no parallel imaging, flip angle 30°, voxel size 1.0 × 0.8 × 0.8 mm3, acquisition time 19 s. (b) Imaging parameters for 3.0 T were: parallel imaging with GRAPPA acceleration factor 3, flip angle 16°, voxel size 0.8 × 0.7 × 0.9 mm3, acquisition time 20 s (courtesy of Marc Saake, MD, and Arnd Dörfler, MD, University of Erlangen, Germany)

H.H. Quick and M.E. Ladd

Fig. 11.3 In this comparison of 3D CE-MRA of the supraaortal arteries performed at 3.0 T (a, b), the SNR gain of high-field MRA was used to reduce the dose of the contrast agent while maintaining spatial resolution and acquisition time. The CE-MRA in image (a) was acquired by administering 8 mL of contrast agent (Gadobutrol) while the CE-MRA in image (b) was acquired using only half of the contrast volume (4 mL of Gadobutrol). This comparison study in the same individual shows that bisecting the dose in high-field MRA can qualitatively lead to comparable results in 3D CE-MRA with regards to vessel display and contrast to noise (CNR) while maintaining spatial resolution and acquisition time. (a, b) Imaging parameters for 3.0 T were: parallel imaging with GRAPPA acceleration factor 3, flip angle 16°, voxel size 0.8 × 0.7 × 0.9 mm3, acquisition time 20 s (courtesy of Marc Saake, MD, and Arnd Dörfler, MD, University of Erlangen, Germany)

Analogous to CE-MRA, TOF MRA of the intracranial arterial vasculature at 3.0 T also benefits from the prolonged T1 relaxation times, resulting in good background signal suppression of stationary brain tissue and excellent visualization of the vasculature. Conspicuity of distal vessel segments is greatly improved due to increased spatial resolution, as has been shown in initial studies comparing TOF MRA of the intracranial vessels at 1.5 and 3.0 T [9–11] (Fig. 11.4). The proton resonance frequency increases linearly with field strength. While the Larmor frequency at 1.5 T is approximately 64 MHz, it doubles to 128 MHz at 3.0 T. At field strength of 7.0 T, the excitation RF frequency is slightly less than 300 MHz. Associated with the increase in Larmor frequency is a linear reduction in RF wavelength from about 52 cm in humans at 1.5 T to about 26 cm at 3.0 T to a further reduced value of about 11 cm at 7.0 T. This reduction in RF wavelength can result in RF inhomogeneities during RF excitation at high field. Aside from resulting in signal voids and limited RF signal penetration depth into the tissue, the measured signal response is then a function of the inhomogeneous

11

MR Angiography and High Field Strength: 3.0 T and Higher

Fig. 11.4 In this field strength comparison (1.5 T (a) vs. 3.0 T (b)) of 3D TOF MRA of the intracranial vessels (Circle of Willis), the SNR gain and enhanced TOF signal were used to increase the spatial resolution. In direct comparison to the 1.5-T image (a), the 3.0-T TOF MRA (b) reveals increased vessel detail and much better visualization of fine arteries with simultaneous diminished background signal. (a) Imaging parameters for 1.5 T were: four overlapping slabs, no parallel imaging, flip angle 25°, voxel size 0.9 × 0.4 × 1.0 mm3, acquisition time 6:38 min. (b) Imaging parameters for 3.0 T were: three overlapping slabs, parallel imaging with GRAPPA acceleration factor 2, flip angle 15°, voxel size 0.7 × 0.4 × 0.7 mm3, acquisition time 5:37 min (courtesy of Marc Saake, MD, and Arnd Dörfler, MD, University of Erlangen, Germany)

RF signal distribution during excitation rather than due to inherent tissue contrast. Furthermore, inhomogeneous RF excitation might lead to locally increased SAR values with associated risk of RF tissue heating. Single-channel, whole-body RF transmit coils which are commonly used in 1.5- and 3.0-T MR scanners are no longer an option for signal excitation at 7.0 T. For 7.0-T MR imaging, new multichannel RF transmit technology is an active area of research and development. In multichannel RF transmit, rather than driving the RF body coil with only a single RF channel, the signal-exciting RF coil consists of multiple individual RF transmit elements that can be independently driven with different RF signal phases and amplitudes. This technique has been termed RF shimming, since the RF field within a certain region of interest can be homogenized by shifting and manipulating regions of signal cancellation to be outside of the body habitus of the patient through selection of the appropriate RF signal parameters. Thus, RF inhomogeneities induced by the shorter RF wavelengths can be counteracted and the problem of inhomogeneous RF signal distribution can be successfully alleviated [12–14]. A detailed discussion of various physical parameters in 7-T MRI can be found in Ladd [15], including the impact on T1, T2, and T2* relaxation times and enhancements in sensitivity to magnetic susceptibility differences.

Intracranial Time-of-Flight MRA at 7 T Intracranial TOF MRA is one of the first MRA applications that has demonstrated potential benefits of 7.0-T MRI scanners. As has been shown in intracranial TOF investigations

151

performed at 3.0-T field strength [9–11], prolonged T1 relaxation time leads to improved suppression of MRI signal associated with extravascular brain parenchyma. Better suppression of background is responsible for visualization of distal vascular segments. However, intracranial TOF at 7.0 T has not exhibited the same benefit. The additional factor of 2.3 in field strength imposes SAR limitations that limit the frequency at which flow saturation RF pulse can be applied. In a study exploring the potential for intracranial 7.0-T TOF MRA, the group of Cho et al. [16, 17] demonstrated the benefits of high field strength to provide excellent spatial resolution intracranial TOF MRA (Fig. 11.5). In these studies, the investigators were able to depict the lenticulostriate arteries, which cannot be reliably visualized at lower field strength with TOF MRI. Other groups have also demonstrated that intracranial TOF at 7.0 T is at this early stage, already an attractive technique for transferring the inherent gain in sensitivity of 7.0 T into increased spatial resolution as well as into exquisite vessel detail [18, 19] (Fig. 11.6). It has even been observed that MRA at 7 T can be used to directly visualize arterial dilation in response to specific neural activity associated with task performance, providing a new type of functional MRA [20]. Beyond these initial research endeavors demonstrating the feasibility and potential of 7.0-T TOF MRA in healthy subjects, a possible clinical application is the detection of aneurysms. It is known that small aneurysms pose a risk of bleeding and the devastating effects of subarachnoid hemorrhage. Even small intracranial aneurysms are clinically significant and must therefore be reliably detected with any diagnostic imaging modaility. So far, however, TOF MRA at lower field strength is limited in that aneurysms smaller than 3 mm may not always be detected. A higher resolution and improved contrast would be very helpful. It must be remembered, though, that TOF MRA at high field strength is subject to increased artifacts. Initial results indicate that there may be advantages of ultrahigh field in the detection of aneurysms, but further investigations are needed to evaluate the clinical relevance of these benefits [21] (Fig. 11.7). One advantage of high-field TOF in the detection of aneurysms may be higher sensitivity to slow flow, ensuring more uniform depiction of slow-moving blood within the aneurysmal dome.

To TOF or Not to TOF? High Field Alternatives for Intracranial MRA TOF imaging is an established imaging technique for noncontrast-enhanced 3D MRA of intracranial vessels at 1.5 and 3.0 T. Imaging protocols that are optimized for field strengths of 1.5 or 3.0 T, however, cannot be transferred directly for imaging at 7.0 T; SAR limitations, the altered tissue T1 and T2 times at 7.0 T, and new image artifacts, such as those

152

H.H. Quick and M.E. Ladd

Fig. 11.5 Left–right: Maximum intensity projections (MIPs) of 3D TOF MRA of the intracranial vessels. The SNR gain and TOF signal at 7.0 T were used to systematically increase the spatial resolution. (1) A 3D TOF MRA image obtained at 1.5-T field strength. (2–4) 3D TOF MRA images obtained at 7.0-T field strength. The nominal imaging resolutions are given below the respective images. The 7.0-T TOF MRA

in this example offers sufficient SNR and vessel-to-background contrast to allow for a twofold increase in spatial resolution in all three spatial dimensions, thus reducing the voxel size by a factor of 8. Note the differences (dotted boxes) between the high-resolution image of 7.0 T in (4) and the low-resolution image of 1.5 T in (1). Acquisition times in both examples (1 and 4) were about 10 min (from Kang CK et al. [17].)

Fig. 11.6 This 3D TOF MRA study of intracranial vessels performed on a 7.0-T system in a normal volunteer impressively shows the inherently good background signal suppression (a) due to prolonged T1 relaxation times of static brain tissues. In combination with the strong TOF in-flow signal and high SNR, this renders intracranial TOF MRA in high-field applications a powerful MRA method to display even the finest vessels with high resolution and contrast (b, c). This 3D MRA data set was acquired using a 24-channel radiofrequency (RF) transmit/

receive coil. Due to SAR limitations, no RF flow saturation pulses were used. Consequently, the images (a–c) show both arteries and veins. Imaging parameters were: flip angle 25°, ten overlapping slabs with 48 partitions each to cover the full volume of the brain, spatial resolution 0.3 × 0.3 × 0.4 mm3, acquisition time 19 min for the full volume (courtesy of Markus Thormann, MD, and Oliver Speck, PhD, University of Magdeburg, Germany)

Fig. 11.7 1.5-T (a) vs. 7.0-T TOF MRA source images (b) vs. DSA (c) showing a 11-mm aneurysm of the distal right ICA of a 43-year-old woman. The aneurysm dome and the right A1 branch of the ACA and the M1 segment of the right MCA (arrows) are better delineated at TOF MRA source images at 7 T (b) in comparison to 1.5 T (a). This

finding was attributed to the increased signal intensity and higher spatial resolution of ultrahigh-field MRA. Lateral DSA (c) depicts the parent vessel and neighboring branches of the MCA and ACA (courtesy of Christoph Mönninghoff, MD, University Hospital Essen, Germany)

11

MR Angiography and High Field Strength: 3.0 T and Higher

153

Fig. 11.8 Noncontrast-enhanced images of a healthy volunteer: axially oriented source images of (a) TOF; sagittally oriented source images of (b) VIBE, and (c) MPRAGE. Corresponding maximum intensity projections (MIPs) of the intracranial system: (d) TOF, (e) VIBE, (f) MPRAGE. Note that the vessel-to-background contrast in TOF images (a, d) is high due to good background signal suppression while VIBE (b, e) and MPRAGE images (c, f) show residual anatomic background signal.

The VIBE and the MPRAGE sequences at 7.0 T open up new possibilities for noncontrast-enhanced vascular imaging in the diagnosis of intracranial vascular lesions. For example, in addition to the vessels themselves, which are hyperintense without application of contrast agent as in the TOF MRA, the perivascular structures in the MPRAGE are depicted in the source images with good resolution and can be evaluated with perfect registration to the vasculature (from Maderwald M, et al. [22])

encountered through increased susceptibility, necessitate optimization of the existing sequences. Against this background, we and other groups [22, 23] were motivated to consider other fast gradient echo sequences that might be potential alternatives for MR angiographic imaging at 7.0 T. Magnetization-prepared rapid gradient echo (MPRAGE), for example, has been shown to provide high signal intensity of blood vessels on contrast-enhanced images at 1.5 T [24]. When imaging at 7.0 T, it has been observed that the MPRAGE sequence – even without the administration of contrast agent – provides high signal in the arterial vasculature while the background shows intermediate signal, resulting in high vessel-to-background contrast potential [22, 23, 25]. Moreover, volume-interpolated breath-hold examination (VIBE), a fast T1-weighted 3D spoiled gradient echo sequence that was designed for short acquisition times and high spatial resolution through the use of asymmetric k-space sampling and pixel interpolation [26], has been used for CE-MRA of the abdominal vasculature [26] and intracranial vessels [27]. Similar to MPRAGE imaging, VIBE images acquired at 7.0 T show hyperintense intracranial vascular signal even without administration of contrast agent [22] (Fig. 11.8).

Increased resolution (approximately 0.5 × 0.5 × 0.5 mm3) at 7 T enables the intracranial vessels to be traced far into the periphery (Fig. 11.8d–f), and the use of MPRAGE to depict the intracranial vessels at this resolution without contrast agent is only possible with ultrahigh field strength. This option opens up new possibilities in the diagnosis of intracranial vascular changes. In addition to the vessels themselves, which appear hyperintense as in TOF MRA, the perivascular structures are depicted with good resolution and can be assessed in the source images (Fig. 11.8b, c). Here, further evaluations are needed to show, for example, how this can be helpful in the workup of vascular stenosis. It is possible that this approach can make the additional acquisition of a CT scan to visualize wall calcifications superfluous. Such calcifications are important to depict when considering an endovascular therapy.

Contrast-Enhanced MRA at 7 T As attractive as the approach of using T1-weighted fast gradient echo sequences (such as FLASH, VIBE, and MPRAGE) for non-CE-MRA may seem, it exposes a potential problem

154

H.H. Quick and M.E. Ladd

Fig. 11.9 Magnitude (left column (a, b)) and corresponding phase images (right column (c, d)) of an SWI data set acquired at 7.0 T (top row (a, c) single slice; bottom row (b, d) minimum intensity projection (MIP) over five slices). Note the high contrast even for very small anatomical structures visible on the magnitude and phase images, such as intracortical veins and the layered structure of gray matter. Although generally the magnitude images show more details, some

structures are only visible on the phase images. (e) Venogram at 7.0 T. The SWI data (TR/TE 22/15 ms; acquisition time 10 min; matrix 512 × 384 × 72; resolution 0.4 × 0.4 × 1.5 mm3) were processed by multiplying the unwrapped and filtered phase images with the magnitude images. In addition, the 3D median-filtered SWI data set was subtracted and inverted ((a–d) from Koopmans et al. [34]) ((e) from Rauscher et al. [35])

for CE-MRA applications. Such T1-weighted fast gradient echo sequences are the basis for CE-MRA that are often used in conjunction with image mask subtraction to improve background signal suppression, whereby the precontrast images are subtracted from the postcontrast images. With arterial vessels already displaying brightly in noncontrastenhanced images at 7.0 T, however, mask subtraction ends up subtracting high nonenhanced arterial signal from high contrast-enhanced arterial signal, consequently reducing the arterial vessel-to-background signal. It remains to be seen whether appropriate contrast-enhanced protocols can be adapted for 7.0 T. The relaxivity parameters, R1 and R2, of contrast agents are highly field dependent. The relaxivity ratios of gadopentetate dimeglumine are, for example, 3.7/4.1 Ls−1 mmol−1 and 4.6/3.7 Ls−1 mmol−1 at 1.5 and 3.0 T, respectively [28]. Early results indicate further changes in R1 and R2 relaxivity at 7 T [29]. It is difficult to fully predict which properties will dominate at higher field strengths because many of the effects depend on the details of the injection protocol, including differentiation between first-pass bolus techniques and steadystate techniques. Numerous studies have shown that gadolinium CE-MRA is feasible at 3.0 T and, in many cases, delivers results superior to those at 1.5 T [3–5]. At higher field strengths, the advantages are not as clear because of the

increasing sensitivity to, for instance, R2*. Certainly, other contrast agents, such as those based on superparamagnetic iron oxide or ultrasmall superparamagnetic iron oxide, also need to be considered [30]. A review of contrast agent results at 3.0 T is presented by Trattnig et al. [7].

MRA Techniques Unique to High-Field MR Because of the susceptibility difference between deoxygenated venous blood and the surrounding tissue, venous imaging at high field strength can produce extraordinary results. The use of a gradient echo sequence with a relatively long echo time (TE) to achieve T2* weighting produces excellent contrast in the venous system. Venules with a diameter of 100 mm can be readily visualized. The susceptibility differences between tissue types also lead to differences in signal phase between the tissues. By optimizing the echo time to ensure that the signals are out of phase, contrast can be optimized in the phase image. This information can be combined with the magnitude image to enhance tissue contrast – a technique termed susceptibilityweighted imaging (SWI) [31–34] (Fig. 11.9a–d). Fig. 11.9e shows a high-field venogram acquired at 7 T after taking both magnitude and phase into account [35].

11

MR Angiography and High Field Strength: 3.0 T and Higher

155

to observe how further developments in RF technology and imaging applications can bring us closer to the assessment of body MRA at 7.0-T field strength.

Conclusion

Fig. 11.10 7T: Non-neuro MRA applications are a relatively new and emerging research field for 7.0-T MR imaging. This is due to the lack of commercially and scientifically available multichannel transmit/ receive RF coils for body applications. First research groups are exploring the inherent diagnostic potential of high-field MRA in the body. The example shows TOF MRA of the liver vasculature, acquired in a normal volunteer at 7.0-T field strength. (a) Twenty imaging slices with a spatial resolution of 1.6 × 0.8 × 2.5 mm3 acquired in a 30-s breath hold. (b) Selected MIP of an axial slab with 33-mm slab thickness. Note the residual B1 inhomogeneity leading to periaortal signal loss

Below the Neckline: 7 T Non-neuro MRA Techniques All the above-mentioned techniques and examples for MRA at 7.0 T are aimed toward neurological applications. This narrow focus is largely a result of the limited availability of whole-body RF coils and RF excitation architectures for 7.0-T scanners. Initially, human whole-body high-field systems have been available with head-only transmit coils only. Investigators have been compelled to build their own RF systems and RF transmit/receive coils to access the rest of the human body and to explore the benefits and limits of 7.0-T imaging in non-neurological applications. The development of multichannel RF technology with the ability to perform B1 shimming and associated RF signal homogenization inside the body tissue has been a precondition for realization of body MRI at 7.0 T. The first attempts at construction of 7.0-T whole-body MRI have recently been published, demonstrating the potential of ultrahigh-field imaging and the need for further coil and sequence optimization [36, 37]. With the development of a custom-built 8-channel RF B1 shimming system [38] in conjunction with a custom-built 8-channel transmit/receive RF body array coil [39], our group has begun to explore 7.0-T MRA applications in the human torso. Feasibility studies in normal volunteers have been performed to investigate the potential for non-CE-MRA of the renal arteries as well as of the vasculature of liver [40]. First results of TOF MRA in the human liver at 7.0 T are shown in Fig. 11.10 demonstrating relatively good B1 signal penetration and homogeneity as well as good vessel-to-background contrast. Of course, this can only be considered a very first step in this rather young but fastevolving research field. Seen in this light, it will be exciting

Increasing the field strength from the clinical standard 1.5 T to the already clinically established 3.0 T has brought more SNR to MRA. According to numerous studies, this doubling in SNR has successfully been transformed into improved display of vascular detail through finer spatial resolution in both TOF and CE-MRA; it has accelerated time-resolved 3D CE-MRA to gain additional dynamic information; and it has been used to reduce the contrast agent dose in selected CE-MRA applications. Beyond the clinically established field strength of 3.0 T, the still young but very fast-developing field of 7.0 T is gaining attention. Early research studies have demonstrated superb image quality for intracranial 3D TOF at 7.0 T owing to improved background tissue suppression and higher spatial resolution which is possible due to the increase in SNR. Furthermore, 7.0 T allows for MRA applications not previously available at lower field strength thanks to inherently changed tissue contrasts. Physical and technical challenges, such as RF inhomogeneities due to the short RF wavelength at 7.0 T, are still to be overcome before the full potential of 7.0-T MRA can be assessed in the remainder of the human body. However, technical solutions, such as multichannel RF transmit coils and associated RF shimming approaches, are under development, and great progress is being made in extending the range of 7-T body applications, including angiography.

References 1. Fischer HW, Rinck PA, Van Haverbeke Y, Muller RN. Nuclear relaxation of human brain gray and white matter: analysis of field dependence and implications for MRI. Magn Reson Med. 1990;16:317–334. 2. Rooney WD, Johnson G, Li X, Cohen ER, Kim SG, Ugurbil K, Springer CS. Magnetic field and tissue dependencies of human brain longitudinal 1H2O relaxation in vivo. Magn Reson Med. 2007;57:308–318. 3. von Falkenhausen MM, Lutterbey G, Morakkabati-Spitz N, Walter O, Gieseke J, Blömer R, Willinek WA, Schild HH, Kuhl CK. Highfield-strength MR imaging of the liver at 3.0 T: intraindividual comparative study with MR imaging at 1.5 T. Radiology. 2006; 241:156–166. 4. Ba-Ssalamah A, Nobauer-Huhmann IM, Pinker K, Schibany N, Prokesch R, Mehrain S, Mlynárik V, Fog A, Heimberger K, Trattnig S. Effect of contrast dose and field strength in the magnetic resonance detection of brain metastases. Invest Radiol. 2003;38:415–422. 5. Fenchel M, Nael K, Deshpande VS, Finn JP, Kramer U, Miller S, Ruehm S, Laub G. Renal magnetic resonance angiography at 3.0 Tesla using a 32-element phased-array coil system and parallel imaging in 2 directions. Invest Radiol. 2006;41:697–703.

156 6. Krautmacher C, Willinek WA, Tschampa HJ, Born M, Träber F, Gieseke J, Textor HJ, Schild HH, Kuhl CK. Brain tumors: full- and half-dose contrast-enhanced MR imaging at 3.0 T compared with 1.5 T: initial experience. Radiology. 2005;237:1014–1019. 7. Trattnig S, Pinker K, Ba-Ssalamah A, Nöbauer-Huhmann IM. The optimal use of contrast agents at high field MRI. Eur Radiol. 2006;16:1280–1287. Review. 8. Kramer U, Fenchel M, Laub G, Seeger A, Klumpp B, Bretschneider C, Finn JP, Claussen CD, Miller S. Low-dose, time-resolved, contrast-enhanced 3D MR angiography in the assessment of the abdominal aorta and its major branches at 3 Tesla. Acad Radiol. 2010;17:564–576. 9. Willinek WA, Born M, Simon B, Tschampa HJ, Krautmacher C, Gieseke J, Urbach H, Textor HJ, Schild H. Time-of-flight MR angiography: comparison of 3.0-T imaging and 1.5-T imaginginitial experience. Radiology. 2003;229:913–920. 10. Al-Kwifi O, Emery DJ, Wilman AH. Vessel contrast at three Tesla in time-of-flight magnetic resonance angiography of the intracranial and carotid arteries. Magn Reson Imaging. 2002;20:181–187. 11. Gibbs GF, Huston J 3rd, Bernstein MA, Riederer SJ, Brown RD Jr. Improved image quality of intracranial aneurysms: 3.0-T versus 1.5-T time-of-flight MR angiography. Am J Neuroradiol. 2004;25:84–87. 12. Abraham R, Ibrahim TS. Proposed radiofrequency phased-array excitation scheme for homogenous and localized 7-Tesla wholebody imaging based on full-wave numerical simulations. Magn Reson Med. 2007;57:235–242. 13. Mao W, Smith MB, Collins CM. Exploring the limits of RF shimming for high-field MRI of the human head. Magn Reson Med. 2006;56:918–922. 14. Van de Moortele PF, Akgun C, Adriany G, Moeller S, Ritter J, Collins CM, Smith MB, Vaughan JT, Uğurbil K. B1 destructive interferences and spatial phase patterns at 7 T with a head transceiver array coil. Magn Reson Med. 2005;54:1503–1518. 15. Ladd ME. High-field-strength magnetic resonance: potential and limits. Top Magn Reson Imaging. 2007;18:139–152. Review. 16. Cho ZH, Kang CK, Han JY, et al. Observation of the lenticulostriate arteries in the human brain in vivo using 7.0 T MR angiography. Stroke. 2008;39:1604–1606. 17. Kang CK, Park CW, Han JY, et al. Imaging and analysis of lenticulostriate arteries using 7.0-Tesla magnetic resonance angiography. Magn Reson Med. 2009;61:136–144. 18. von Morze C, Xu D, Purcell DD, et al. Intracranial time-of-flight MR angiography at 7 T with comparison to 3 T. J Magn Reson Imaging. 2007;26:900–904. 19. Heverhagen JT, Bourekas E, Sammet S, Knopp MV, Schmalbrock P. Time-of-flight magnetic resonance angiography at 7 Tesla. Invest Radiol. 2008;43:568–573. 20. Cho ZH, Kang CK, Han JY, et al. Functional MR angiography with 7.0 T Is direct observation of arterial response during neural activity possible? Neuroimage. 2008;42:70–75. 21. Monninghoff C, Maderwald S, Theysohn JM, et al. Evaluation of intracranial aneurysms with 7 T versus 1.5 T time-of-flight MR angiography - initial experience. Rofo. 2009;181:16–23. 22. Maderwald S, Ladd SC, Gizewski ER, et al. To TOF or not to TOF: strategies for non-contrast-enhanced intracranial MRA at 7 T. MAGMA. 2008;21:159–167. 23. Zwanenburg JJ, Hendrikse J, Takahara T, Visser F, Luijten PR. MR angiography of the cerebral perforating arteries with magnetization prepared anatomical reference at 7 T: comparison with time-offlight. J Magn Reson Imaging. 2008;28:1519–26. 24. Liang L, Korogi Y, Sugahara T, et al. Evaluation of the intracranial dural sinuses with a 3D contrast-enhanced MP-RAGE sequence:

H.H. Quick and M.E. Ladd

25.

26.

27.

28.

29.

30.

31.

32. 33.

34.

35.

36. 37. 38.

39.

40.

prospective comparison with 2D-TOF MR venography and digital subtraction angiography. AJNR Am J Neuroradiol. 2001;22: 481–492. Van de Moortele PF, Auerbach EJ, Olman C, Yacoub E, Uğurbil K, Moeller S. T1 weighted brain images at 7 Tesla unbiased for Proton Density, T2* contrast and RF coil receive B1 sensitivity with simultaneous vessel visualization. Neuroimage. 2009;46:432–46. Rofsky NM, Lee VS, Laub G, et al. Abdominal MR imaging with a volumetric interpolated breath-hold examination. Radiology. 1999;212:876–884. Wetzel SG, Law M, Lee VS, Cha S, Johnson G, Nelson K. Imaging of the intracranial venous system with a contrast- enhanced volumetric interpolated examination. Eur Radiol. 2003;13:1010–1018. Weinmann HJ, Bauer H, Griesinger C, et al. Early distribution dynamics of polymeric magnetic resonance imaging contrast agents in rats. Acad Radiol. 2002;Suppl 2:S412-S416. No abstract available. Noebauer-Huhmann IM, Kraff O, Juras V, et al. MR contrast media at 7 Tesla: preliminary study on relaxivities. Proc Intl Soc Mag Reson Med. 2008;16:1457. Allkemper T, Heindel W, Kooijman H, Ebert W, Tombach B. Effect of field strengths on magnetic resonance angiography: comparison of an ultrasmall superparamagnetic iron oxide blood-pool contrast agent and gadopentetate dimeglumine in rabbits at 1.5 and 3.0 Tesla. Invest Radiol. 2006;41:97–104. Reichenbach JR, Barth M, Haacke EM, Klarhöfer M, Kaiser WA, Moser E. High-resolution MR venography at 3.0 Tesla. J Comput Assist Tomogr. 2000;24:949–957. Haacke EM, Xu Y, Cheng YC, Reichenbach JR. Susceptibility weighted imaging (SWI). Magn Reson Med. 2004;52:612–618. Deistung A, Rauscher A, Sedlacik J, Stadler J, Witoszynskyj S, Reichenbach JR. Susceptibility weighted imaging at ultra high magnetic field strengths: theoretical considerations and experimental results. Magn Reson Med. 2008;60:1155–1168. Koopmans PJ, Manniesing R, Niessen WJ, Viergever MA, Barth M. MR venography of the human brain using susceptibility weighted imaging at very high field strength. MAGMA. 2008; 21:149–158. Rauscher A, Barth M, Herrmann KH, Witoszynskyj S, Deistung A, Reichenbach JR. Improved elimination of phase effects from background field inhomogeneities for susceptibility weighted imaging at high magnetic field strengths. Magn Reson Imaging. 2008;26:1145–1151. Snyder CJ, DelaBarre L, Metzger GJ, et al. Initial results of cardiac imaging at 7 Tesla. Magn Reson Med. 2009;61:517–524. Vaughan JT, Snyder CJ, DelaBarre LJ, et al. Whole-body imaging at 7 T: Preliminary results. Magn Reson Med. 2009;61:244–248. Bitz A, Brote I, Orzada S, et al. An 8-channel add-on RF shimming system for wholebody 7 Tesla MRI including real-time SAR monitoring. Proceedings of the 17th Annual Meeting of the International Society for Magnetic Resonance in Medicine, Honolulu, HI, 2009, Abstract 4767. Orzada S, Quick HH, Ladd ME, et al. A flexible 8-channel transmit/ receive body coil for 7 T human imaging. Proceedings of the 17th Annual Meeting of the International Society for Magnetic Resonance in Medicine, Honolulu, HI, 2009, Abstract 2999. Umutlu L, Lauenstein TC, Kraff O, Maderwald S, Orzada S, Kinner S, Heilmaier C, Antoch G, Ladd ME, Quick HH. Non-enhanced vs. contrast-enhanced MRA at 7Tesla: A feasibility and comparison trial. Proceedings of the 18th Annual Meeting of International Society for Magnetic Resonance in Medicine, Stockholm, Sweden 2010; Abstract 1421.

Susceptibility Weighted Imaging and MR Angiography

12

Samuel Barnes and E. Mark Haacke

Introduction Magnetic resonance angiography (MRA) has undergone tremendous development since its inception [1–3]. To image the vessels in the brain at high fields clinically, there is no longer the need to use contrast agents thanks to the increased signal-to-noise at 3 T and the rapid scanning that is possible. Conventional time-of-flight [4] with or without magnetization transfer contrast (MTC) can give excellent coverage with high-resolution images. Similarly, susceptibility weighted imaging (SWI) can be used to create venographic images of vessels as small as 200–300 mm [5, 6]. In this chapter, we discuss the potential to image both arteries and veins in an SWI single or multiecho time-of-flight (TOF)like sequence. For the last 100 years, the arterial system has played the major role in the study of the brain’s hemodynamics and, from a surgical point of view, in the study and treatment of atherosclerosis. However, the other half of the story is told by the venous system. SWI, like blood oxygen level dependent (BOLD) imaging, is sensitive to deoxyhemoglobin in the veins and is able to generate exquisite images of the veins in the brain. SWI is also sensitive to nonheme iron in the form of ferritin or hemosiderin as well and can be used to quantify iron [7] along with T2* maps [8]. These two features make SWI a powerful means by which to study neurovascular diseases such as cerebral amyloid angiopathy (CAA), multiple sclerosis (MS), stroke, traumatic brain injury (TBI) and tumors [9].

S. Barnes, MS () • E.M. Haacke, PhD Department of Radiology, Loma Linda University Medical Center, Loma Linda, CA, USA Department of Radiology, Harper Hospital/Wayne State University, Detroit, MI, USA e-mail: [email protected]

Technical Issues with SWI and MRA in a Single Sequence Basic SWI Concepts The actual SWI sequence itself is akin to the usual flow compensated gradient echo sequence [5, 6]. More recent efforts have focused on developing a multiecho version of SWI with the potential to simultaneously measure T2* as well as phase [10]. Gradient echo imaging is particularly sensitive to local changes in magnetic field. This variation of field changes the spin phase over time and so, at longer echoes, the signal rapidly disappears (thanks to the additional T2¢ dephasing). The total relaxation rate is given by R2* = R2 + R2¢ where R2* = 1/T2*, R2 = 1/T2, and R2¢ = 1/T2¢. In addition to the magnitude image, SWI uses the phase data where, for a righthanded system, phase is given by j(phi) = −g (gamma) × D(delta) B×TE with g the gyromagnetic ratio of hydrogen (42.6 MHz/T) and D(delta)B is the local change in magnetic field caused by iron or calcium or other structural effects in the tissue. The phase image is high pass filtered to remove low spatial frequency phase variations caused by a variety of background field inhomogeneities such as those from poorly shimmed fields or air–tissue interface field effects [6, 11, 12]. Although these SWI filtered phase images are of great interest in and of themselves [6, 13], the phase can also be manipulated to create a phase mask that highlights positive or negative phases or both on the magnitude image. This phase mask is then used to create a new type of image that accentuates both the T2* contrast, from the magnitude image, and phase information into a single SW image (Fig. 12.1). It is this form of SWI data that has become best known [9, 14]. SWI data are collected with a long-TE, fully flowcompensated gradient echo scan; this can be in the form of a single-echo, multiple-echo [10, 15, 16], or segmented echoplanar approach [17]. Imaging with long echoes with reasonable signal-to-noise became possible by using 3D gradient echo imaging [11]. This allowed for thinner slices (1–2 mm),

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_12, © Springer Science+Business Media, LLC 2012

157

158

S. Barnes and E.M. Haacke

Fig. 12.1 Single slice comparing SWI and other images at 4 T, (a) SWI magnitude image (b) SWI phase image (c) SW Image (d) proton density image. Note that the SW image (c) combines features from both the magnitude (a) and phase (b)

which reduced dephasing across the slice and improved image quality. Flow compensation in all directions is useful in SWI because of the long echo times required and because phase is being used as a measure of susceptibility. Velocity compensation is usually enough to reduce flow related signal loss in the magnitude image, and flow induced changes in the phase image. If one considers two tissues with different susceptibilities D(delta)c(chi), changes in the phase image are generated according to the formula (for a right-handed system): Δj = −g ⋅ ΔB ⋅ TE, (12.1) (delta)(phi) = − (gamma)×(delta)B×TE where Δj = −g ( Δc B0 G + ΔBCS + ΔBgeometry + ΔBmain field )TE. (12.2) (delta)(phi) = − (gamma)((delta)(chi)B 0 G + (delta) BCS + (delta)Bgeometry + (delta)Bmain field)TE Here, G represents a constant dependent on the geometry of the object, CS refers to chemical shift, and Bgeometry refers to the geometry of the brain and air–tissue interfaces. The last two terms represent unwanted field effects. The first two terms are of particular interest to us and are meant to represent the local changes in field, such as those that might be caused by iron in tissue. In 12.2, we have rewritten D(delta)

B = D(delta)c(chi)B0 where D(delta)c(chi) represents the local susceptibility change between tissues. The last two terms tend to be slowly varying spatial terms and can be mostly removed using a high-pass spatial frequency filter. Ideally, we can isolate the first two terms −g (gamma)GD(delta)c (chi) B0TE and −g (gamma)D(delta)BCSTE. Both of them lead to similar phase results inside the object of interest. A paramagnetic object (such as iron) causes a local increase in field and therefore a negative phase change relative to surrounding tissues, while a diamagnetic substance (such as calcium) causes a local decrease in field and therefore a positive phase change. The phase filter is designed as a homodyne filter whereby a low spatial frequency phase image is divided into the original phase to leave behind high spatial frequency phase. The size of the low pass filter Nf is usually quoted as something like 64 × 64 for example. However, it might make more sense to refer to the size of the smallest object df that is basically removed by the filter. This is given by df = FOV/Nf. As an example, consider the FOV = 256 voxels and Nf = 64. In this case, df = 4 voxels. The implications are that objects greater than or equal to 4 voxels will be suppressed from the filtered phase image. When aliasing is particularly severe, the phase image can be unwrapped, prior to high pass filtering, to improve the results of the filtering [18]. In some cases, it may be possible to directly remove background phase effects caused by the gross geometry (D(delta)Bgeometry) in 12.2 with minimal filtering. When the geometry and average susceptibility of the object itself (e.g., the brain) and its surroundings (e.g., the sinuses) are roughly known, their magnetic field effects can be calculated and removed using a forward modeling approach [19]. These calculated phase effects are then divided out from the measured phase image, prior to high-pass filtering, removing many artifacts and leaving the local changes in susceptibility and phase unaltered, improving the results of the filtering. Whichever filtering technique is used, the resultant filtered phase image is used as described below in all subsequent steps and will be referred to as the SWI filtered phase image. As mentioned above, a mask must be created from the filtered phase image and then applied to the magnitude image to generate the SWI data. This mask focuses on certain phase values that will enhance the contrast of the original magnitude image. For example, if areas with increased iron are the subject of interest, then the mask is designed to enhance information related to negative phase (in a right-handed system) as follows: ⎧ p + j ( x) ⎪ f ( x) = ⎨ p ⎪⎩ 1

for − p < j ( x ) < 0

(12.3)

otherwise

(pi) + (phi)(x)/(pi) for −(pi) < (phi)(x) < 0 where the phase values can range from −p(pi) to p(pi), j(phi)(x) is the phase at location x, and f(x) is the phase

12 Susceptibility Weighted Imaging and MR Angiography

159

mask. This phase mask can be multiplied by the original magnitude image an integer m number of times to create the SW image: r¢ ( x ) = f m ( x ) ⋅ r( x ). (rho)¢(x) = f m(x) (rho)(x)

(12.4)

The number of times the mask is applied will change the contrast in the SW image. It has been shown that four multiplications produces good CNR for a wide range of phase values [6]. If the aliasing in the phase image has not been fully removed, which is nearly inevitable in the regions near the frontal sinuses, the residual phase artifacts will destroy the existing magnitude contrast in the SW image rather than enhancing it. This can be overcome by improving the phase filtering, using one of the more advanced techniques discussed above, or alternatively not applying the phase mask in those areas. Problem areas can be identified by calculating the local field gradients, the phase mask can then be adjusted accordingly to remove problem areas with high local field gradients [20]. This preserves existing magnitude contrast in problem areas while still getting regular SWI contrast in the rest of brain. Once this SWI data has been created, it is possible to highlight veins by performing a minimum intensity projection (mIP) over a number of slices. This shows venous connectivity in the same way a maximum intensity projection (MIP) shows arterial connectivity in MRA. A disadvantage of using mIPs is that the dark background surrounding the brain will mask out the brain if it is included in the projection. The slice included in the mIP having the smallest visible brain area will dictate how much of the brain is visible in the final projection. This can be problematic at the top and bottom of the brain where the size changes very rapidly. This problem can be partly overcome by using a brain extraction algorithm, such as a complex threshold approach [21], to set the noise values outside the brain to a value much higher than the brain during the mIP processing and then back to zero afterward. Clinically, mIPs are usually limited to projections over 4–8 mm, although it is certainly possible to project over more slices near the center of the brain or if the background has been removed to get a better visualized of a larger portion of the venous vasculature.

Fig. 12.2 A single sagittal slice from 0.5 mm isotropic data showing (a) the high-pass filtered phase image and (b) the SWIM image. The phase image has been inverted to match contrast on the SWIM image. Notice the dark phase located outside of the veins that is removed in the SWIM image

susceptibility values in turn depend directly on the local tissue composition such as iron content, deoxyhemoglobin levels, and calcium content. While the phase image is directly influenced by the changes in tissue susceptibility, it is also influenced by many other factors, most noticeably the shape and distribution of those susceptibility changes (see 12.2 above). Susceptibility mapping seeks to remove all of these influences by taking the phase image and calculating what susceptibility distribution could create that phase image. Unfortunately, this inversion is ill-posed as the mathematical kernel used in the inversion process becomes zero at certain orientations (the so-called magic angle 54.7°). This has been overcome with various techniques such as restrictions to certain geometric shapes [22, 23], multiple scans at different orientations [24], and various regularization schemes ranging from very simple to quite complicated [25–27]. In our own work, we used a simple regularization approach. Although the term susceptibility mapping is well known, we prefer to call this approach SWIM for susceptibility weighted imaging and mapping as we used the highpass filtered phase image in the process. In this approach, the phase is converted into a magnetic field distribution by dividing the phase by g (gamma)TE. To create the susceptibility map, the usual inverse filter is applied and regularized by preventing the inverse from dividing by zero by setting a threshold [27]. We have found that this approach works best for high resolution 0.5 × 0.5 × 0.5 mm isotropic data (Figs. 12.2 and 12.3).

Susceptibility Mapping and Measuring Oxygen Saturation Simultaneous SWI and MRA Several groups have recently started exploring the possibility of creating susceptibility maps. Susceptibility maps are to SWI what T2 maps are to T2 weighted images, albeit much more difficult to calculate. They are a quantitative map of the actual magnetic susceptibility values of the tissue. These

Both SWI and time-of-flight MRA rely on a flow compensated gradient echo sequence. SWI uses a long echo time and MRA uses a very short echo time. This fact enables SWI and MRA sequences to be combined into a single sequence with

160

S. Barnes and E.M. Haacke

Fig. 12.3 A MIP over 8 mm for (a) the high-pass filtered phase image and (b) the SWIM image. The contrast on the phase image has been inverted to match the SWIM image. Notice both show good vein connectivity and gray matter–white matter contrast. The MIP hides many of the extravascular dipole effects seen in Fig. 12.2 as they are dark and so are excluded from the MIP. The hidden artifacts in the phase image cause the images to look very similar in a MIP, but the SWIM image is still quantifiable, while the phase is not. (From Haacke EM et al. [60]; with permission.)

Fig. 12.4 A multiecho SWI sequence with five echoes was used to get both MRA and traditional SWI contrast. (a) First echo with TE = 6.1 ms MIP over 32 mm clearly depicts the arteries. (b) Fifth echo with TE = 22.1 ms mIP over 12 mm of SW images shows the veins and traditional SWI contrast

two echoes. The first echo is optimized for MRA contrast and the second echo is optimized for SWI contrast. One of the problems in implementing a simultaneous MRA and SWI scan occurs in trying to flow compensate all echoes, which takes considerable time. One way around this is to discard the flow compensation in the phase encoding direction for the first echo (MRA contrast), where it is most important to keep the echo time as short as possible. The second or last echo (SWI contrast) should be flow compensated in all directions as one does not want phase from flow to obscure phase from susceptibility. It is fairly easy to accomplish this given the fact that the SWI sequence itself is usually run with a very long echo to enhance phase effects. The real problem lies in the choice of excitation flip angles so as not to oversuppress the cerebrospinal fluid signal (CSF) and maintain good SWI venous contrast inside the CSF in the gray matter, but still get good excitation of the quickly refreshed arterial blood. A contrast agent can be used to enhance results. In this case, the first echo will show excellent arteries and veins while the SWI long echo will still suppress the veins unlike conventional MRA (see contrast agents section for more details) [28]. The possibility of using multiple echoes in SWI to achieve different types of contrast was first explored by Du et al. [16]. They used a flow compensated short echo to achieve time-of-flight MRA contrast, and a second long echo to achieve SWI contrast. In their implementation, the second echo was not flow compensated in the phase encode directions. This was shown to cause some artifacts in the phase image near flowing arteries and Deistung et al. [29] was able to implement a second echo that was fully flow compensated to remove these artifacts. The flow compensation was

achieved by fully rewinding the phase encode gradients to zero before performing a second flow compensated phase encode for the next echo. Further modifications such as adjusting the k-space ordering to optimize flip angle choice for each type of contrast and using multiple thin slabs have also been proposed [15]. With these techniques it is possible to achieve both a high quality TOF MRA and a SWI in a single scan that takes approximately the same amount of time as either of those run separately. These two-echo, two-contrast techniques suggest the possibility of using a general flow compensated multiecho sequence to achieve a variety of contrasts in both magnitude and phase. As discussed above, the short echoes can be used as an MRA and the long echoes for SWI contrast (Fig. 12.4). The echoes in the middle or after the optimal SWI contrast also have some value. The multiple echo data can be used to calculate T2* maps especially in slow flowing vessels such as the veins and sagittal sinus. They can also be useful for various phase filtering techniques. Finally, having a variety of echo times allows the echo with the optimal susceptibility contrast to be used. Shorter echoes are more optimal for hemorrhages that contain lots or iron, while long echoes are more optimal for imaging tiny venules.

Single Echo Approach While the multiecho approaches can produce excellent angiographic and venographic images, the possibility remains open to achieve similar results with a single echo [30]. SWI provides a natural separation of the vasculature, with the arteries being bright from inflow enhancement due

12 Susceptibility Weighted Imaging and MR Angiography

to the short TR and the veins dark from the long TE. By using a higher bandwidth and high isotropic resolution to reduce flow dephasing, it is possible to obtain a decent angiogram without significantly degrading the SWI venogram. Generally, by using a long TE and short TR (typically 20 and 30 ms at 3 T, respectively) the veins can be suppressed and the arteries brightened, allowing a very good separation of arteries and veins in a single echo scan. The arteries can be visualized using a standard maximum intensity projection (MIP) of magnitude information, and the veins visualized with SWI processing and a minimum intensity projection (mIP). The choice of flip angle, resolution, and echo time is very important for determining image quality of both the MRA and MRV. Higher flip angles improve the angiogram due to increased background suppression and better TOF inflow effect, but in turn degrade the venogram by oversuppressing the CSF. Choosing a medium flip angle (15–20°) and a slightly longer TR appears to be optimal. The longer TR allows more inflow enhancement for the angiography and keeps the CSF from being oversuppressed. Shorter echo times improve the angiography by reducing uncompensated higher-order flow losses, but this decreases venous contrast. At 3 T, decreasing the echo time below 20 ms substantially degrades venous contrast and is not recommended; an echo time of 20 ms is preferred. For this reason, this technique shows promise at higher fields (>3 T) as shorter echo times can then be used without degrading venous contrast. It is possible to reduce uncompensated higher-order flow losses by increasing the read bandwidth and acquiring with high isotropic resolution. A high read bandwidth will reduce the time between spatial encoding and the echo readout; this is referred to as the field echo. Reducing the field echo improves the flow compensation by minimizing higher order flow effects which are proportional to higher powers of the field echo time (Fig. 12.5). Keeping the echo time long but reducing the field echo achieves good flow compensation while maintaining T2* contrast. This is only true, however, if you have a homogeneous field. If you have field inhomogeneities, from a poorly shimmed field, air–tissue interfaces, or air–bone interfaces, flow through these inhomogeneities will cause the blood to start collecting phase immediately after the excitation pulse, making the flow losses dependant on the echo time, not the field echo time. Unfortunately the carotid and parts of the MCA lie in regions of poor field homogeneity due to the sinuses causing both to experience nearly complete signal loss at long echo times despite very short field echoes. This can be seen in Fig. 12.6, where only the part of the MCA that is within the inhomogeneous field (as seen with the phase image) experiences bad signal loss. High isotropic resolution reduces dephasing across a voxel and thus can reduce flow losses. This does reduce the quality of the SWI phase image as an isotropic aspect ratio of 1:1 in-plane to through-plane resolution is not ideal for SWI

161

Fig. 12.5 MIP images from a long echo scan (20 ms) at different bandwidths (a) 60 Hz/pixel (b) 120 Hz/pixel (c) 235 Hz/pixel and (d) 465 Hz/pixel. Note that the flow compensation improves at higher bandwidths due to the shortened field echo time as noted by the decreased flow losses in the MCA (arrows)

[31, 32]. The isotropic aspect ratio causes the phase for veins of certain sizes and orientations to have opposite signs. This has the potential to confuse clinical interpretation and could reduce contrast in the SWI processed images. This lost contrast can be completely recovered, however, by postprocessing the images and applying a down-sampling filter to generate a more ideal aspect ratio of 1:4. A simple k-space crop can be used, which would be equivalent to a lower resolution acquisition with 1:4 aspect ratio. Alternatively, a sliding window complex average can be performed, which takes advantage of the fact that the higher resolution was collected. The sliding window filter takes the average of the complex signal (magnitude and phase) over four slices, advances a single slice, calculates the next average, and repeats until all slices are processed. In this way, the through-plane resolution is reduced but the same number of slices as in the original series is maintained (actually the slices will be reduced by a small amount, the collapsing factor – one or three slices in this case). By reconstructing the thick slabs in an overlapping pattern, optimal partial voluming of small structures is guaranteed, increasing their visibility. This reconstruction also offers a distinct advantage over the original k-space data in that it uses the high-pass filtered phase in the complex downsampling process, and not the original phase. This reduces the dephasing in areas of rapid phase change (air– tissue interfaces), improving the quality of the downsampled images. The single echo SWI contrast and venography, as shown in Fig. 12.7, are of good quality with the veins being well depicted. Likewise, nearly all of the arteries are well depicted; however, there is some signal loss in parts of the fast-flowing MCA due to flow dephasing from the long TE. The more distal arteries are well depicted with little to no signal loss.

162

Fig. 12.6 (a) 7 ms magnitude (b) 7 ms phase (c) 24 ms magnitude. Notice the more proximal part of the MCA (arrow) that is visible at 7 ms is significantly reduced in amplitude at 24 ms while the more dis-

Fig. 12.7 Single echo SWI dataset with TE = 20 ms, TR = 35 ms, FA = 15°, BW = 160 Hz/pixel (a) MIP over 64 mm, (b) mIP over 8 mm. (Images adapted or reprinted with permission of Haacke et al. [61].)

Blood Properties at Different Field Strengths For brain parenchyma, the T1 values have been found to generally increase with field strength [33] and the T2 values do not change much until you get above 3.0 T and then they start to fall precipitously [34]. The T2* values of all tissues behave differently than the T2 values. They also fall with field strength but they begin to decrease at lower fields and change significantly between 1.5 and 3.0 T [35]. Relaxation values for blood are difficult to measure accurately as blood is constantly flowing and ex vivo measurements can be challenging due to changes in oxygen saturation. However, T2* values for venous blood do decrease dramatically across commonly used field strengths and range from 97 ms at 1.5 T to 7.4 ms at 7.0 T [36]. The decreasing values of T2* for all tissues as field strength increases has important implications for SWI.

S. Barnes and E.M. Haacke

tal part is still very visible. The proximal part is exposed to field inhomogeneities from the sinuses (indicated with dashed line) causing them to dephase despite the very short field echo time

At 1.5 T relatively long echo times (40 ms) are demanded to give time for significant T2* contrast to develop. These long echo times are also needed to develop adequate contrast in the phase image according for 12.2. But as field strength increases the echo times can be shorted, approximately linearly with the increase in field strength, while still maintaining both T2* contrast and susceptibility contrast in the phase image. Phase contrast is maintained as 12.2 shows it is dependent on the product of echo time and field strength, and T2* contrast from the veins in the magnitude image is maintained due to the falling T2* values with field strength. The reduced echo time allows a significant time savings in SWI as one moves to higher field strengths. If the ratio of field strength remains the same as the ratio of the R2* relaxivities, then the SWI data can be produced identically between field strengths by simply keeping the product of B0TE constant. In practice, this is not the case as R2* values do not behave linearly, although going from 1.5 to 3 T, and dropping the echo from 40 to 20 ms, still generates excellent contrast SWI data.

The Role of MRI Contrast Agents T1 shortening contrast agents, such as gadolinium, can substantially improve the quality of SWI venography. The T1 shortening of the contrast agent causes an increase in the available signal for blood. The deoxyhemoglobin in the blood causes a slight frequency shift in the venous blood (this is what causes the phase contrast); this frequency shift allows the echo time to be chosen such that venous blood and brain parenchyma are out of phase and the signals will cancel. This cancelation allows the shortened T1 and increased signal of the blood to improve contrast in the

12 Susceptibility Weighted Imaging and MR Angiography

163

Fig. 12.8 SWI of a rat anesthetized with isoflurane and oxygen. Left images show precontrast and right images are 2.5 h post USPIO (P904, 300 mmol Fe/kg). Single slice SWI (top) and filtered phase images (bottom), notice the veins are much better visualized even 2.5 h post contrast

Fig. 12.9 mIP over 24 mm SW images (a) precaffeine and (b) approximately 1 h after a 200 mg caffeine tablet. The higher amounts of deoxyhemoglobin in the veins cause them to be much more visible postcaffeine

veins; this is referred to as T1–T2* coupling. The usual Gadolinium based contrast agents are paramagnetic and for a 1 mM concentration in the blood they increase the local susceptibility of the blood by 2/9ths of the usual BOLD effect. This in turn introduces an increase in the phase of the blood by 2/9ths, further improving SWI contrast. The possibility of using direct T2* shortening agents in the blood also exists with ultrasmall superparamagnetic iron oxides (USPIO) or superparamagnetic iron oxides (SPIO). While these have very large T2* shortening properties they act evenly on both arteries and veins, which causes the arteries to go dark along with the veins. While this does cause artery–vein ambiguity, the level of small vessel detail seen is quite impressive (Fig. 12.8). If this could be run in some sort of pre–post fashion to separate the arteries from veins it could find a use in imaging the small vasculature of the brain.

Role of Caffeine and Acetazolmide The cerebral vascular reserve of the brain can be tested in a number of ways. One is to stress the brain with caffeine which is a vasoconstrictive agent and a member of the methylxanthine family (which are adenosine antagonists [37]). Typically, two cups of coffee can elicit a strong vascular response [38]. The resulting reduced flow leads to an increase in the BOLD effect since the brain’s oxygen utilization remains constant. The slower flow then leads to a higher concentration of deoxyhemoglobin in the veins. This leads to an increase in the oxygen extraction fraction, OEF, to maintain the cerebral metabolic rate (CMRO2). The effect on SWI data is an improvement in visualization of the veins since deoxyhemoglobin increases, the local field increases, phase increases and local T2* effects increase [39]. The nice thing about using SWI as a high-resolution BOLD method is that the effects of caffeine can be seen throughout the entire brain (Fig. 12.9).

164

The effects of caffeine as seen with SWI have been investigated recently in a number of interesting papers [40]. Applying a region-of-interest based analysis, the authors reported an approximately exponential signal decay in venous vessels with time that was in agreement with a linear pharmacokinetic model of oral absorption of caffeine [40]. Venous response reached a maximum in the time interval of between 40 and 50 min, while allowing a significant differentiation between coffee drinkers (venous signal change: −16.5% ± 6.5%) and abstainers (venous signal change: −22.7% ± 8.3%). Only small signal changes of about −2% ± 1% were found in both gray and white matter and −1% ± 2% in the ventricles, in accordance with the earlier findings. The effects of caffeine can alter baseline cerebral blood flow and in this way modulate temporal dynamics of the BOLD response through alterations in the strength of neurovascular coupling [41–44]. Another agent that is used to study cerebral blood flow changes is the vasodilator acetazolamide (C4H6N4O3S2) also known as diamox. It has been used to study epilepsy, intracranial hypertension, glaucoma, and altitude sickness. It has been shown to increase cerebral blood flow from 30 [45] to about 50% [46, 47], and venous oxygen saturation by approximately 20% relative to the usual resting state levels [47]. As such it is used to assess hemodynamic reserve and vasomotor reactivity [48]. Acetazolamide inhibits carbonic anhydrase, which leads to the production of HCO3−. HCO3− in turn induces a local extracellular acidosis by increasing the concentrations of CO2 and H+ in the extracellular fluid in the brain, which is assumed to act as a stimulus for the increase in blood flow [49]. These increases in blood flow lead to a reduction in local oxygen saturation [50, 51].

S. Barnes and E.M. Haacke

it very useful in imaging hemorrhages from a variety of sources including trauma, stroke and aging, visualizing blood products and the vascularization of tumors, and high-resolution MR venography [9, 52–54]. It has also proven useful in other applications relating to iron such as measuring iron content in multiple sclerosis lesions, and iron accumulation in aging [55]. For a summary of clinical applications, please see the recent review by Mittal et al. [9]. Multiple sclerosis (MS) has long been thought to be an inflammatory demyelinating disease. However, both 70-yearold data [56] and recent imaging evidence [57] suggests that venous obstruction is closely related to and possibly a source of MS. Our own studies using SWI to evaluate the veins in the thalamostriate region and the iron content have shown that the iron content builds up at the confluence of the draining veins in structures such as the putamen, globus pallidus, and the caudate nucleus (Fig. 12.10). This fits with the theory that MS is a chronic cerebrospinal vascular insufficiency [57]. SWI is able to not only help visualize the small veins that appear to be at the center of the lesion [58] but also the iron content in the lesions and in the basal ganglia and thalamus.

Cerebral Microbleeds

SWI is particularly well suited for imaging venous blood as it is very sensitive to deoxyhemoglobin and other iron containing products such as ferritin and hemosiderin. This makes

Another area where SWI plays a key role is visualizing small hemorrhages thanks to the hemosiderin deposits that are eventually left behind (Figs. 12.11 and 12.12). Cerebral microbleeds (CMBs), while usually asymptomatic, are becoming an important biomarker for many other diseases such as hypertension, cerebral amyloid angiopathy, intracerebral hemorrhage, and hemorrhagic stroke [53]. SWI is more sensitive at detecting CMBs than standard T2* weighted imaging. A recent study showed that three times more CMBs could be detected with SWI compared to T2* weighted imaging [59]. SWI can also be useful in ruling out some common CMB mimics that are present in standard T2* weighted images. Calcium mimics, deposits of calcium that

Fig. 12.10 (a) SWI filtered phase showing increases in iron content directly related to the thalamostriate draining veins. (b) Minimum intensity projection processing SWI data showing the venous drainage

system for the basal ganglia and thalamus. (c) A radiographic image of the veins from a cadaver brain study (Courtesy of Georges Salamon, MD, David Geffen School of Medicine at UCLA, Los Angeles, CA)

Clinical Applications

12 Susceptibility Weighted Imaging and MR Angiography

165

much smaller than the voxel but still a source of spatially varying static fields, say for example for nanoparticles such as USPIOs, then T2¢ will be invariant at resolutions on the order of hundreds of microns or greater. However, the effects of diffusion are dependent on resolution via the gradients used to generate high-resolution images. The implications for SWI are that the phase remains invariant but T2* effects decrease at higher resolutions so SWI will maintain much of its BOLDlike contrast while T2* will become less and less sensitive. Therefore, we expect SWI to be an ideal method to for very high-resolution techniques such as detecting tiny CMBs. Fig. 12.11 Cerebral microbleeds caused by trauma in a motorcycle accident (a) SWI (b) SWI Phase

Fig. 12.12 Case with numerous cerebral microbleeds (CMBs) caused by severe CAA, imaged at 1.5 T (a) mIP over 8 mm of SW images (b) Corresponding FLAIR image

cause small hypointensities that look like CMBs, can be easily identified using the SWI phase image. Calcium will have an opposite phase shift compared to the iron in a true CMB as calcium is diamagnetic and iron is paramagnetic. If a true CMB has dark phase (as in a right-handed system) the calcium will have bright phase making it easily distinguishable. Flow voids, while not eliminated in SWI, are minimized as SWI uses a sequence that is flow compensated in all directions (read, phase, and partition). This minimizes signal loss due to flowing blood and should reduce the likelihood of confusing a CMB for a flow void.

Conclusion Importance of High-Resolution Imaging As one goes to higher and higher resolution the effects of local field variations diminishes. This is because the phase dispersion across a voxel is reduced, which reduces signal loss from dephasing. What this means is that T2¢, hence T2*, is not scale invariant. If the local fields are caused by objects

References 1. Nayler GL, Firmin DN, Longmore DB. Blood flow imaging by cine magnetic resonance. J Comput Assist Tomogr. 1986;10:715–722. 2. Nishimura DG, Macovski A, Pauly JM. Magnetic resonance angiography. IEEE Trans Med Imaging. 1986;5:140–151. 3. Haacke EM, Lenz GW. Improving MR image quality in the presence of motion by using rephasing gradients. AJR Am J Roentgenol. 1987;148:1251–1258. 4. Potchen EJ, Haacke EM, Siebert JE, Gottschalk A (eds). Magnetic Resonance Angiography: Concepts and Applications. St. Louis: Mosby; 1993:xiv, 650. 5. Reichenbach JR, Venkatesan R, Schillinger DJ, Kido DK, Haacke EM. Small vessels in the human brain: MR venography with deoxyhemoglobin as an intrinsic contrast agent. Radiology. 1997; 204:272–277. 6. Haacke EM, Xu Y, Cheng YC, Reichenbach JR. Susceptibility weighted imaging (SWI). Magn Reson Med. Sep 2004;52(3):612–618. 7. Haacke EM, Miao YW, Liu M, et al. Correlation of putative iron content as represented by changes in R2* and phase with age in deep gray matter of healthy adults. J Magn Reson Imaging. 2010;32:561–576. 8. Gelman N, Gorell JM, Barker PB, et al. MR imaging of human brain at 3.0 T: preliminary report on transverse relaxation rates and relation to estimated iron content. Radiology. 1999;210:759–767. 9. Mittal S, Wu Z, Neelavalli J, Haacke EM. Susceptibility-weighted imaging: technical aspects and clinical applications, part 2. AJNR Am J Neuroradiol. 2009;30:232–252. 10. Denk C, Rauscher A. Susceptibility weighted imaging with multiple echoes. J Magn Reson Imaging. 2010;31:185–191. 11. Reichenbach JR, Venkatesan R, Yablonskiy DA, Thompson MR, Lai S, Haacke EM. Theory and application of static field inhomogeneity effects in gradient-echo imaging. J Magn Reson Imaging. 1997;7:266–279. 12. Haacke EM, Dmitriy SL, Yablonskiy A, Lin W. In vivo validation of the bold mechanism: a review of signal changes in gradient echo functional MRI in the presence of flow. Int J Imaging Systems Technology. 1995;6:153–163. 13. Duyn JH, van Gelderen P, Li TQ, de Zwart JA, Koretsky AP, Fukunaga M. High-field MRI of brain cortical substructure based on signal phase. Proc Natl Acad Sci USA. 2007;104: 11796–11801. 14. Haacke EM, Mittal S, Wu Z, Neelavalli J, Cheng YC. Susceptibilityweighted imaging: technical aspects and clinical applications, part 1. AJNR Am J Neuroradiol. 2009;30:19–30. 15. Park SH, Moon CH, Bae KT. Compatible dual-echo arteriovenography (CODEA) using an echo-specific K-space reordering scheme. Magn Reson Med. 2009;61:767–774.

166 16. Du YP, Jin Z. Simultaneous acquisition of MR angiography and venography (MRAV). Magn Reson Med. 2008;59:954–958. 17. Xu Y, Haacke EM. An iterative reconstruction technique for geometric distortion-corrected segmented echo-planar imaging. Magn Reson Imaging. 2008;26:1406–1414. 18. Rauscher A, Barth M, Reichenbach JR, Stollberger R, Moser E. Automated unwrapping of MR phase images applied to BOLD MR-venography at 3 Tesla. J Magn Reson Imaging. 2003; 18:175–180. 19. Neelavalli J, Cheng YC, Jiang J, Haacke EM. Removing background phase variations in susceptibility-weighted imaging using a fast, forward-field calculation. J Magn Reson Imaging. 2009;29: 937–948. 20. Jin Z, Xia L, Du YP. Reduction of artifacts in susceptibilityweighted MR venography of the brain. J Magn Reson Imaging. Aug 2008;28(2):327–333. 21. Pandian DS, Ciulla C, Haacke EM, Jiang J, Ayaz M. Complex threshold method for identifying pixels that contain predominantly noise in magnetic resonance images. J Magn Reson Imaging. 2008;28:727–735. 22. McAuley G, Schrag M, Sipos P, et al. Quantification of punctate iron sources using magnetic resonance phase. Magn Reson Med. 2010;63:106–115. 23. Cheng YC, Hsieh CY, Neelavalli J, Haacke EM. Quantifying effective magnetic moments of narrow cylindrical objects in MRI. Phys Med Biol. 2009;54:7025–7044. 24. Liu T, Spincemaille P, de Rochefort L, Kressler B, Wang Y. Calculation of susceptibility through multiple orientation sampling (COSMOS): a method for conditioning the inverse problem from measured magnetic field map to susceptibility source image in MRI. Magn Reson Med. 2009;61:196–204. 25. de Rochefort L, Liu T, Kressler B, et al. Quantitative susceptibility map reconstruction from MR phase data using bayesian regularization: validation and application to brain imaging. Magn Reson Med. 2010;63:194–206. 26. Kressler B, de Rochefort L, Liu T, Spincemaille P, Jiang Q, Wang Y. Nonlinear regularization for per voxel estimation of magnetic susceptibility distributions from MRI field maps. IEEE Trans Med Imaging. 2010;29:273–281. 27. Shmueli K, de Zwart JA, van Gelderen P, Li TQ, Dodd SJ, Duyn JH. Magnetic susceptibility mapping of brain tissue in vivo using MRI phase data. Magn Reson Med. 2009;62:1510–1522. 28. Lin W, Mukherjee P, An H, et al. Improving high-resolution MR bold venographic imaging using a T1 reducing contrast agent. J Magn Reson Imaging.1999;10:118–123. 29. Deistung A, Dittrich E, Sedlacik J, Rauscher A, Reichenbach JR. ToF-SWI: simultaneous time of flight and fully flow compensated susceptibility weighted imaging. J Magn Reson Imaging. 2009; 29:1478–1484. 30. Barnes SR, Jin Z, Du YP, Deistung A. MR angiography and venography of the brain. In: Haacke EM, Reichenbach J, eds. Susceptibility Weighted Imaging in MRI: Basic Concepts and Clinical Applications: Wiley; 2011. 31. Xu Y, Haacke EM. The role of voxel aspect ratio in determining apparent vascular phase behavior in susceptibility weighted imaging. Magn Reson Imaging. 2006;24:155–160. 32. Deistung A, Rauscher A, Sedlacik J, Stadler J, Witoszynskyj S, Reichenbach JR. Susceptibility weighted imaging at ultra high magnetic field strengths: theoretical considerations and experimental results. Magn Reson Med. 2008;60:1155–1168. 33. Bottomley PA, Foster TH, Argersinger RE, Pfeifer LM. A review of normal tissue hydrogen NMR relaxation times and relaxation mechanisms from 1–100 MHz: dependence on tissue type, NMR frequency, temperature, species, excision, and age. Med Phys. 1984;11:425–448.

S. Barnes and E.M. Haacke 34. Jezzard P, Duewell S, Balaban RS. MR relaxation times in human brain: measurement at 4 T. Radiology. 1996;199:773–779. 35. Peters AM, Brookes MJ, Hoogenraad FG, et al. T2* measurements in human brain at 1.5, 3 and 7 T. Magn Reson Imaging. 2007; 25:748–753. 36. Koopmans PJ, Manniesing R, Niessen WJ, Viergever MA, Barth M. MR venography of the human brain using susceptibility weighted imaging at very high field strength. MAGMA. 2008;21: 149–158. 37. Fredholm BB, Battig K, Holmen J, Nehlig A, Zvartau EE. Actions of caffeine in the brain with special reference to factors that contribute to its widespread use. Pharmacol Rev. 1999;51:83–133. 38. Haacke EM, Filleti CL, Gattu R, et al. New algorithm for quantifying vascular changes in dynamic contrast-enhanced MRI independent of absolute T1 values. Magn Reson Med. 2007;58:463–472. 39. Haacke EM, Hu C, Parrish T, Xu Y. Whole brain stress test using caffeine: effects on fMRI and SWI at 3 T. Proc Intl Soc Mag Reson Med. 2003;11:1731. 40. Sedlacik J, Helm K, Rauscher A, Stadler J, Mentzel HJ, Reichenbach JR. Investigations on the effect of caffeine on cerebral venous vessel contrast by using susceptibility-weighted imaging (SWI) at 1.5, 3 and 7 T. Neuroimage. 2008;40:11–18. 41. Behzadi Y, Liu TT. Caffeine reduces the initial dip in the visual BOLD response at 3 T. Neuroimage. 2006;32:9–15. 42. Chen Y, Parrish TB. Caffeine’s effects on cerebrovascular reactivity and coupling between cerebral blood flow and oxygen metabolism. Neuroimage. 2009;44:647–652. 43. Liau J, Perthen JE, Liu TT. Caffeine reduces the activation extent and contrast-to-noise ratio of the functional cerebral blood flow response but not the BOLD response. Neuroimage. 2008;42: 296–305. 44. Perthen JE, Lansing AE, Liau J, Liu TT, Buxton RB. Caffeineinduced uncoupling of cerebral blood flow and oxygen metabolism: a calibrated BOLD fMRI study. Neuroimage. 2008;40:237–247. 45. Okazawa H, Yamauchi H, Sugimoto K, Toyoda H, Kishibe Y, Takahashi M. Effects of acetazolamide on cerebral blood flow, blood volume, and oxygen metabolism: a positron emission tomography study with healthy volunteers. J Cereb Blood Flow Metab. 2001;21:1472–1479. 46. Schytz HW, Wienecke T, Jensen LT, Selb J, Boas DA, Ashina M. Changes in cerebral blood flow after acetazolamide: an experimental study comparing near-infrared spectroscopy and SPECT. Eur J Neurol. 2009;16:461–467. 47. Vorstrup S, Henriksen L, Paulson OB. Effect of acetazolamide on cerebral blood flow and cerebral metabolic rate for oxygen. J Clin Invest. 1984;74:1634–1639. 48. Griffiths PD, Gaines P, Cleveland T, Beard J, Venables G, Wilkinson ID. Assessment of cerebral haemodynamics and vascular reserve in patients with symptomatic carotid artery occlusion: an integrated MR method. Neuroradiology. 2005;47:175–182. 49. Lassen NA. Is central chemoreceptor sensitive to intracellular rather than extracellular pH? Clin Physiol. 1990;10:311–319. 50. Hedera P, Lai S, Lewin JS, et al. Assessment of cerebral blood flow reserve using functional magnetic resonance imaging. J Magn Reson Imaging. 1996;6:718–725. 51. Sedlacik J, Kutschbach C, Rauscher A, Deistung A, Reichenbach JR. Investigation of the influence of carbon dioxide concentrations on cerebral physiology by susceptibility-weighted magnetic resonance imaging (SWI). Neuroimage. 2008;43:36–43. 52. Sehgal V, Delproposto Z, Haacke EM, et al. Clinical applications of neuroimaging with susceptibility-weighted imaging. J Magn Reson Imaging. 2005;22:439–450. 53. Greenberg SM, Vernooij MW, Cordonnier C, et al. Cerebral microbleeds: a guide to detection and interpretation. Lancet Neurol. 2009;8:165–174.

12 Susceptibility Weighted Imaging and MR Angiography 54. Tong KA, Ashwal S, Obenaus A, Nickerson JP, Kido D, Haacke EM. Susceptibility-weighted MR imaging: a review of clinical applications in children. AJNR Am J Neuroradiol. 2008;29:9–17. 55. Haacke EM, Cheng NY, House MJ, et al. Imaging iron stores in the brain using magnetic resonance imaging. Magn Reson Imaging. 2005;23:1–25. 56. Putnam TJ. Evidences of vascular occlusion in multiple sclerosis and “encephalomyelitis”:. Archives of Neurology and Psychiatry. 1937;37:1298–1321. 57. Zamboni P, Galeotti R, Menegatti E, et al. A prospective open-label study of endovascular treatment of chronic cerebrospinal venous insufficiency. J Vasc Surg. 2009;50:1348–1358, e1341–1343.

167 58. Tan IL, van Schijndel RA, Pouwels PJ, et al. MR venography of multiple sclerosis. AJNR Am J Neuroradiol. 2000;21: 1039–1042. 59. Nandigam RN, Viswanathan A, Delgado P, et al. MR imaging detection of cerebral microbleeds: effect of susceptibility-weighted imaging, section thickness, and field strength. AJNR Am J Neuroradiol. 2009;30:338–343. 60. Haacke EM, Tang J, Neelavalli J, Cheng YC. Susceptibility mapping as a means to visualize veins and quantify oxygen saturation. J Magn Reson Imaging. 2010;32:663–676. 61. Haacke EM, Reichenbach J, eds. Susceptibility Weighted Imaging in MRI. Hoboken, NJ: Wiley; 2011.

wwwwwwwwwwww

Non-Cartesian MR Angiography

13

Walter Block and Oliver Wieben

Introduction Magnetic resonance angiography (MRA) has several compelling features, including volumetric data acquisition and the availability of several contrast mechanisms. These can be used for imaging with and without exogenous contrast agents and to obtain anatomical as well as functional information, for example in the form of perfusion or velocity measurements. However, such MRA acquisitions require longer imaging times, which challenge scan completion without artifacts from physiological or involuntary patient motion, particularly for dynamic acquisitions, such as bolus chasing, real-time imaging, or cardiac-gated acquisitions. The inherently 3D nature of MRI allows for fine depiction of vascular territories. However, as MR samples its data in an alternative Fourier domain, only a relatively few samples can be obtained at one time. If one desires to obtain a timeresolved depiction as injected contrast enhances the vasculature, a four-dimensional space must be sampled. In one of the most demanding angiographic tasks, phase-contrast (PC) angiography can obtain quantitative velocity data that is resolved to a spatial coordinate and a cardiac cycle. This imaging task requires four-dimensional sampling (three spatial dimensions, time) of a vector quantity, the flow vector with three directional velocity components constrained by physiological and clinical implementation limits. With such demanding acquisition requirements over so many dimensions, non-Cartesian trajectories have been developed to offer increased performance in MRA. Non-Cartesian trajectories can offer increased performance in several ways, although not always simultaneously. NonCartesian methods can better utilize limited gradient hardware

W. Block, PhD () • O. Wieben, PhD Departments of Biomedical Engineering, Medical Physics, and Radiology, University of Wisconsin–Madison, Madison, WI, USA e-mail: [email protected]

speed, improve the efficiency in which k-space is covered, decrease sensitivity to motion, and improve flow properties. Some non-Cartesian methods often offer a variable sampling trajectory, where the center of k-space is sampled more often than higher spatial frequencies. These sampling patterns support time-resolved imaging in reconstruction methods that vary in performance, speed, accuracy, and complexity. In general, accuracy and performance improve when complexity and time within the reconstruction task are accounted for. Non-Cartesian methods offer possibilities to exploit the sparse nature of the vascular imaging task and the correlation between temporal frames in a time-resolved study. Vascular imaging differs from static imaging of many regions of the body in several important ways. Often, vascular images are much more sparse than nonangiographic images of other parts of the body. The sparse nature of these images can be due to the high contrast provided by injected contrast agents. The ability to subtract out static signal, as in phase-contrast imaging or through use of a precontrast mask, further increases the sparse nature of vascular images. In timeresolved imaging, significant correlation may exist between imaging frames and thus each image volume may not need to be acquired completely separately. The chapter first describes non-Cartesian acquisition and reconstruction theory, loosely classified as spiral and radial trajectories. A brief summary of methods being utilized to provide consistent performance with non-Cartesian methods is provided as these trajectories are generally less robust to several system and patient-induced imperfections than Cartesian methods. Methods that utilize non-Cartesian trajectories to improve time-resolved angiography are then discussed. Similar concepts used for accelerating time-resolved imaging can be used for quantitative resolve flow and perfusion throughout the cardiac cycle. Finally, trajectories with variable sampling densities are useful when using image estimation methods to increase performance. Here, some form of a priori information is used to constrain the reconstruction process.

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_13, © Springer Science+Business Media, LLC 2012

169

170

Non-Cartesian Trajectory Design In MR imaging, data can be sampled in k-space on any 2D or 3D trajectory that the time-varying gradients and safety regulations regarding peripheral nerve and muscle stimulation and tissue heating can support. Although the first MR imaging method proposed the acquisition of projections [1], spin-warp imaging on a Cartesian sampling grid [2] became the predominantly used trajectory. The acquisition of data on a rectilinear grid is robust against inhomogeneities in the static magnetic field resulting from imperfections the scanner. In non-Cartesian acquisitions, these inhomogeneities introduce off-resonance effects which in turn cause blurring of the PSF. Advances in scanner hardware improved the field homogeneity and alternative sampling patterns with nonuniform sampling densities revisited. Projection imaging [3] is a CT-like acquisition, where each echo represents a radial line traversing through the center of k-space. This method offers good suppression of motion artifacts and allows for imaging with very short echo times when the projections start in the center because they do not require any prewinding gradients. A disadvantage is the prolongation total imaging time because of the redundant oversampling of the central k-space region. k-space can be sampled with fewer echoes using spiral trajectories [4]. Images can be acquired with as little as 30–40 echoes, but this scheme is very sensitive to off-resonance effects. The sampling grids for these acquisitions schemes are shown in Fig. 13.1. The trajectories can also be extended or combined into 3D acquisitions, for example for truly 3D radial, 3D spiral, cone, stack of spheres, shells trajectory, cones, and spiral PR (Fig. 13.2). Hybrid 3D sampling patterns with non-Cartesian in plane encoding and traditional Fourier slice encoding have also been implemented, predominantly for the sampling of imaging volumes of shorter dimensions in the through plane direction. More complete reviews of sampling patterns can be found elsewhere in the literature [5].

W. Block and O. Wieben

General Considerations Sampling Region Sampling a cylinder of k-space saves 21.5% of the sampled space relative to a cube while sampling a sphere saves 47.6% of the required samples. While non-Cartesian trajectories can easily be tuned to cylindrical and spherical k-space regions, selection of phase-encoding and slice-encoding locations can achieve cylindrical sampling spaces. Gradient Spoiling As the readout direction is changing throughout a nonCartesian scan, some attention has to be given to the method that spoils transverse signal in gradient-recalled sequences. Winding the magnetization to the same physical k-space location after each readout is generally a good way to remove variations in the transverse steady-state signal throughout the scan. Field of View In general, non-Cartesian trajectories are designed to sample along the readout direction at k-space intervals of 1/FOV as in Cartesian trajectories, where the FOV is the largest dimension of the acquired volume. Sampling along the readout dimension is constrained by the maximum slew rate achievable, and thus k-space sampling intervals often vary, especially at the beginning of the readout. Repetitions of the spiral or radial readout are then rotated in such a way to fill k-space. To provide full k-space sampling, enough repetitions of the model readout are needed such that the space of the interleaves is less than 1/FOV in all areas of k-space. Off Resonance The extent of phase accrued by off-resonance spins during each readout is directly proportional to the readout duration. The amount of needed effort to remedy off-resonance effects, thus, increases with readout duration. The sophistication of any needed off-resonance processing depends also of course on the amount of inhomogeneity present in the vascular territory of interest. While the appearance of off resonance

Fig. 13.1 Strategies for 2D k-space sampling. Shown are the spin-warp (a), radial sampling (b), and interleaved spiral imaging (c) trajectory as examples for sampling patterns used in MR angiography

13

Non-Cartesian MR Angiography

Fig. 13.2 Strategies for 3D k-space sampling include cylindrical stack of spirals (a), spherical stack of spirals (b), a spiral–radial hybrid trajectory (c), concentric shells (d), a true 3D radial trajectory (e), and a stack of cones (f) (figure adopted from Irarrazabal and Nishimura [5]; reprinted with permission)

Fig. 13.3 Asymmetric FOV imaging with a radial trajectory with increased sampling density along the near-vertical projections. This creates a larger, horizontal FOV than vertical FOV (b) (courtesy of Steve Kecskemeti, University of Wisconsin-Madison)

varies with trajectory, off-resonance effects are generally manifested by blurring and signal dropout in non-Cartesian trajectories.

Asymmetric FOVs The largest spacing between k-space samples in the ensemble of readouts collected in the acquisition is usually designed to be no smaller than 1/FOV. The largest spacing may be designed to be 1/FOV for full sampling or greater than 1/ FOV trajectories which one intentionally oversamples in regions of sparse and high-contrast vasculature. Asymmetric FOVs that are tuned to certain vascular territory are generally easier to achieve with Cartesian trajectories than nonCartesian methods; however, non-Cartesian asymmetric FOVs are possible as shown in Fig. 13.3 [6]. Flow Sensitivity In general, trajectories whose first moment is small near the center of k-space have better flow properties [7]. Trajectories for which the first moment changes smoothly as a function of k-space radius also are more robust in MRA. In general, trajectories which originate at the center of k-space without previous slice encoding have more advantageous flow properties.

171

Sampling Density The sampling density of radial trajectories varies significantly with k-space radius, falling off as 1 / k (r ) for 2D radial trajectories and 1 / k 2 (r ) for true 3D radial trajectories. Sampling some spatial frequencies more often at the expense of others has liabilities when trying to cover all of k-space rapidly and has deleterious effects on SNR compared to flat sampling trajectories [8]. Oversampling lower spatial frequencies had shown advantages for representing time-resolved imaging [9–11], motion artifact suppression, inherent field map and coil sensitivity generation, and constrained reconstruction methods [12–14]. In general, radial trajectories in MRA have emphasized the value of variable sampling density while working to deemphasize deleterious effects from an inefficient coverage of k-space. While their acquisition is less sophisticated than spiral trajectories, the short acquisition time required for each radial line or projection significantly limits offresonance effects. Radial waveforms are generally easier to program in pulse sequences as well. Spiral waveform design initially emphasized efficient, rapid k-space coverage with flat sampling density [5, 15], where only a minimum portion of the waveform was slew rate limited. Meyers presented an analytical expression to approximate the density of spirals [4] throughout their trajectory. Very simple methods that grid spiral data points to the nearest neighbor on an oversized Cartesian matrix have also been demonstrated [16] which simplify the density compensation computation. More recently, spiral design has incorporated variable sampling density to mitigate effects from aliasing from outside the FOV and to provide some of the advantages of oversampling in representing time-resolved images volumes [17]. The advantages of spiral acquisitions grow with longer readout duration, though these increase problems with off resonance. The trajectories shown in Fig. 13.2 were first proposed by Irarrazaval and have since been utilized in numerous examples of spiral MRA, including stack of stars in the coronaries [18], cones in the peripheral vasculature [19], whole heart 3D radial imaging [20, 21], and spiral projection imaging [22]. As an example, most 3D breath-hold coronary imaging can cover only a thin plane within the breath-hold requirement. In Fig. 13.4, a variable density spiral design is used to cover the entire heart with 0.8 × 0.8 × 1.6-mm resolution in just 17 heartbeats [23].

Spiral Trajectory Design Although numerous implementations are possible, most spiral trajectories have been based on an Archimedes spiral. These trajectories follow the basis equation: k (t ) = λq (t )e - iq ( t ) .

172

W. Block and O. Wieben

Fig. 13.5 Radial sampling schemes for projections from −kr,max to +kr,max (a) and from 0 to +kr,max (b). The starting point for each readout is shown as an unfilled circle and the end point with an arrow head. Both schemes are characterized by a constant radial sampling interval Dkr and a maximum angular sampling interval Dkj,max

Projection Imaging More specific aspects of the design of various radial trajectories and their effects on PSFs are next provided. Fig. 13.4 Detail of small coronary branches. Images (a–c) were acquired with a resolution 0.8 × 0.8 × 1.6 mm3. Image (d) was acquired with a resolution of 1 × 1 × 2 mm3 and reconstructed with the iterative algorithm. Image (a) shows a detail of the conus artery. The distal RCA, including the posterior left ventricular artery (PL) and posterior descending artery (PDA), is displayed in image (b). A lesser cardiac vein (CV) is also shown. Image c shows a detail of the mid to distal RCA with an acute marginal right ventricle (RV) branch. (d) The mid LAD and diagonal branches (courtesy of J. Santos et al. [23]; reprinted with permission)

The desired gradient waveforms are given by derivative of the k-space trajectory and scaled by the inverse of the gyromagnetic ratio. Linear functions of q(t) lead to inefficient spirals with constant angular speed. The intuitive choice for efficient coverage of k-space would use a constant velocity spiral, where q (t ) = t . As this choice is not realizable in regions of the spiral where the slew rate is limited, trade-offs in the formulation of q(t) between constant angular speed and constant velocity were formulated by Bornert et al. [24]. While a more accurate solution for optimal use of gradient slew rate was formulated by King et al. [25], this solution required significant computation. A closed-form expression, which produces images which are indiscernible from the optimal solution, is provided by Glover et al. [26]. The complexity of spiral trajectory has created numerous strong publications, where computational power is often used to create shorter trajectories, more accurate sampling density functions, and more powerful off-resonance correction methods. In many cases, simpler approximations can provide adequate performance for many vascular applications. A strong review of these trade-offs is provided by Block and Frahm [27].

2D Projection Imaging In 2D projection imaging, each readout traverses through the center of k-space. The sampling trajectory can be described in polar coordinates with a radial component kr and an angle j. A total of Np repetitions are acquired with Nr samples and a sampling interval Dkr along the readout direction. As the 1D Fourier transform of each repetition provides a projection of the object, the technique is also known as projection reconstruction (PR). The ensemble of transformed projections forms a sonogram, similar to computed tomography (CT) reconstruction, which can then be reconstructed with filtered backprojection. Projection imaging, also known as radial sampling, leads to a nonuniform sampling density with emphasis on the low spatial frequencies. Let us consider the case, where each projection starts at −kr,max, traverses through the origin, and ends at +kr,max as shown in Fig. 13.5a. The radial sampling interval Dkr supports an alias-free reconstruction of distance D = 1/Dkr along the readout. The largest angular sampling distance Dkf,max occurs between adjacent spokes at the maximum sampled spatial frequency: Dkj ,max =

pN r Dk . 2N p r

(13.1)

According to the Nyquist theorem, sampling with Dkj,max = Dkr produces isotropic resolution over a radial FOV with a diameter D. This optimal sampling requires the following projections: N p,opt =

p Nr . 2

(13.2)

13

Non-Cartesian MR Angiography

173

More projections do not provide better spatial resolution or a larger FOV while fewer projections reduce the artifactfree FOV. In comparison, spin-warp imaging requires only Nr readouts (2/p = 63.7% less) for a squared FOV with identical resolution. This decrease in sampling efficiency is due to the oversampling of central k-space. It is important to note that any signal from outside the circular FOV does cause data inconsistencies in the projections and result in streak artifacts in the image. The band-pass filter applied to the received signal does limit the signal contributions along the readout direction but not perpendicular to it. While the repeated sampling of low spatial frequencies in each readout decreases the scan efficiency, radial sampling has very desirable properties for certain applications. Projection imaging is more robust to bulk motion because of the averaging effects from repeated sampling of the low spatial frequencies and more tolerable streak artifacts [3, 28]. This is also advantageous in diffusion-weighted imaging, where large gradients amplify artifacts from motion [28–30]. The trajectory can be modified so that each projection starts in the k-space origin (kr = 0) as shown in Fig. 13.5b. A free induction decay (FID) can then be acquired for the imaging of tissues with very short transverse relaxation times T2, such as in the lungs [31, 32]. Projection imaging can also be advantageous to suppress displacement artifacts in flow imaging [7]. Continuous and interleaved radial acquisitions were proposed for dynamic imaging in studies of the joints [33, 34], catheter tracking in interventional MR [35, 36], swallowing exams [37], and cardiac imaging [37]. The properties of angular undersampling for faster imaging have been explored in various studies and are discussed below.

Undersampled 2D Projection Imaging If the number of projections is decreased below Np,opt, then the angular sampling interval Dkj,max exceeds the radial interval Dkr and the high spatial frequencies are not sampled adequately. This leads to a reduced artifact-free FOV (rFOV) with a diameter d given by the inverse of the largest sampling interval: d=

1 Dkj ,max

=

2N p pDkr N r

.

(13.3)

The ratio of the diameters of the reduced FOV and the full FOV is given by Dkr d 2 Np = = . D Dkj ,max p N r

(13.4)

Figure 13.6 shows the PSFs for a fully sampled radial trajectory (Np = p/2 Nr) and with a reduced number of projections (Np << p/2 Nr).

Fig. 13.6 Radial sampling with adequate sampling (Np = p/2 Nr) results in a symmetric points spread function with an artifact-free circular FOV inside the first lobe (a) up to r = 1/Dkr. Angular undersampling causes streak artifacts outside a reduced FOV with smaller diameter as shown for an undersampling factor of 2.5 in (b) (modeled after Scheffler and Hennig [38])

As long as the imaged object does not extend outside the FOV, streak artifacts occur only outside the reduced FOV and do not interfere with the object generating the signal. In contrast to Cartesian acquisitions, where undersampling leads to coherent ghosts, undersampling creates a noise-like appearance. The property of spreading the artifact at a distance from the object is used for interventional MR with a large, static FOV and a reduced dynamic FOV [38, 39]. In another approach, Shimizu et al. [40] purposefully undersampled at very high ratios to utilize the streaks for tracking the tip of a biopsy needle. Peters et al. [41] first proposed the use of angularly undersampled projection reconstruction for large, FOV imaging in MRA using a hybrid trajectory with radial in-plane imaging and Fourier slice encoding. They demonstrated that PR provides higher spatial resolution per unit time in high-contrast environments, where the number of imaged objects is limited. This is especially true in contrast-enhanced MRA. Here, the streak artifacts from undersampling are often tolerable, leading to an increase in temporal resolution and/or spatial resolution, as shown in Fig. 13.7. The trade-off becomes the occurrence of streak artifacts and a loss in SNR due to the shortened acquisition time, similar to parallel imaging.

3D Radial or Projection Imaging In a true 3D radial sequence as shown in the right column of Fig. 13.8 and with a corresponding pulse sequence in Fig. 13.9, every projection lies in a sphere and traverses through the center of k-space in 3D radial acquisitions. Hence, this trajectory is best described in a spherical coordinate system with a radial component kr, a polar angle q, and an azimuthal angle j . A total of Np projections are acquired with Nr samples along the readout direction. The radial sampling interval Dkr determines the spatial resolution and the diameter D = 1/Dkr of the maximum achievable artifact-free spherical FOV. The Nyquist criterion is met if both angular sampling

174

W. Block and O. Wieben

Fig. 13.7 Resolution comparison between FT and PR from zoomed up regions of a phantom (a). All images were acquired with a readout length of 512 samples. The readout direction for the FT images is labeled kx. Image (b) is acquired with 128 phase encodes on a Cartesian grid. Image (c) is a PR image from 128 projections acquired in the same scan time with increased spatial resolution in the phase-encoding direction. Image

(d) is sampled on a 512 × 512 Cartesian grid with a spatial resolution of 0.3 × 0.3 mm2. The smallest dots in the image are 0.5-mm wide and spaced 0.5 mm apart. The undersampled PR image has a resolution similar to the 512 × 512 FT image, although it was acquired in one-fourth of the time. Because of the shorter acquisition time, it also has a lower SNR (adopted from Peters et al. [41]; reprinted with permission)

Fig. 13.8 Hybrid 3D PR sequences use radial imaging in the through plane direction (left) and fully sampled Fourier encoding in the slice

direction. This ensemble trajectory is often referred to as the stack of stars (middle). A truly 3D radial trajectory is shown at the right

intervals are smaller than the radial sampling interval: Dkj, Dkq £ Dkr. If we consider Np evenly spaced projections (Dkj = Dkq = Dk) that start at −kmax, traverse through the origin, and end at +kmax, then we can associate a surface area A = Dk 2 =

Fig. 13.9 Vastly undersampled Isotropic PRojection (VIPR) sequence (a) timing diagram: The Gx, Gy, and Gz gradient amplitudes define the orientation of the partial diameter in the overall (b) k-space trajectory – each excitation can cover one partial diameter as shown, a full diameter, or multiple diameters (from Barger et al. [10]; with permission)

2 4p kmax 2p 2 = k 2N p N p max

(13.5)

at kmax with each projection to cover the surface of the sampled sphere. The number of projections for optimal sampling (Dkj = Dkq = Dkr = Dk) is given by N p,opt =

p 2 Nr . 2

(13.6)

13

Non-Cartesian MR Angiography

175

This is p/2 = 57% more excitations than would be required for a cubic volume with identical spatial resolution with Fourier encoding. Because of this low sampling efficiency, Boada et al. [42] proposed a novel trajectory called twisted projection imaging (TPI) that starts as a radial projection until a certain radius when it switches to sample the surface of outward pointing cone. This scheme provides the advantages of motion robustness and FID imaging with fewer projections.

Undersampled 3D Radial Imaging By spreading the aliasing from undersampled energy from a 2D slice to a far larger 3D volume, the undersampling artifacts further diffuse into a structured background that resembles noise. This technique, termed 3DPR or Vastly undersampled Isotropic PRojection (VIPR) imaging [10, 43], provides isotropic resolution over large spherical fields of view and its pulse sequence and trajectory are shown in Fig. 13.9. Similar to the 2D case, undersampling leads to a reduced FOV, here a sphere with diameter d = 1/Dk. The ratio of the diameters for a reduced FOV and a full FOV is derived as d Dkr 2 Np = = . D Dk p Nr

(13.7)

The rFOV is proportional to the square root of the number of projections. Sparse sampling leads to artifacts from the undersampled high spatial frequencies and a reduced FOV around each object. The limiting factor in this acquisition is the decreased SNR and the artifacts introduced from undersampling.

General Reconstruction of Non-Cartesian Acquisitions In principle, there are many methods to reconstruct images from data acquired along non-Cartesian trajectories. In principle, methods that grid the acquired data onto a Cartesian grid are most utilized. With increases in computational power, parallel imaging, and model-based reconstruction, more complicated iterative algorithms are also creating increased attention. Prior to describing gridding, filtered backprojection for radial acquisitions is briefly discussed. A more thorough description of gridding is also provided in Bernstein’s Handbook [44].

Filtered Backprojection Filtered backprojection (FBP) was developed for computed tomography and is an approximate implementation

of the inverse radon transform. FBP is performed on sinogram data which are obtained by 1D Fourier transform of the radially acquired k-space data. Mathematically, the projections p(r,f) from an object distribution f(x,y) are obtained as p(r , j ) = òò f ( x, y)d ( x cos j + y sin j - r )dx dy.

(13.8)

In CT imaging, the projections are directly measured, and in MRI the projections are obtained by an inverse 1D Fourier transform of each radial line along kr: p(r , j ) = ò P (kr , j )e j2pkr r dkr .

(13.9)

The task then becomes the reconstruction of the backprojected image f¢(x,y). Since the sampling density in the center of k-space is much higher than at the edges, each projection has to be weighted with |kr| in k-space P ¢(kr , j ) = H (kr ) P (kr , j ) = kr P (kr , j ),

(13.10)

where H(kr) is the filter for the projection and H(kr) = |kr| is called the Ram–Lak filter. Unfiltered backprojection has been used in the early days of CT and results in the wellknown star artifact. The backprojected image can then be obtained as f ¢ ( x, y ) = ò

p

0

ò

R -R

p ¢(r , j )d ( x cos j + y sin j - r )dr dj . (13.11)

Different modifications to the Ram–Lak filter have been proposed for radial MR imaging. Pipe [45] discusses weighting options for the filter for noise reduction in undersampled data sets. Joseph [46] suggests to weight the central point kr = 0 by the circular area it represents, p Dkr2 / 4 , rather than setting it to zero. Backprojecting magnitude data removes errors from variable delays imparted by eddy currents as the readout direction changes. While backprojecting magnitude data removes this problem, often the data in the body is inherently complex due to B0 inhomogeneity. Methods to remove the delays before backprojection are preferable because then complex backprojection can be performed.

Reconstruction by Gridding MR data acquired along non-Cartesian trajectories are generally reconstructed with a process known as gridding [47]. If one represents M(k) as the continuous Fourier transform of the object magnetization m(x) and the k-space sample points as S(k), sampled data is represented as

176

W. Block and O. Wieben

M ^ s(k - 1) = M (k )S (k ). Gridding interpolates the sampled MR data unto a Cartesian grid using a convolution kernel, C(k), after compensating for differences in sampling density, r(k), using the expression: é M (k ) ù M c (k ) = ê s C ( k ) ú·III ( k ). ë r (k ) û Several methods exist to compute the sampling density function r(k). A simple operation is simply r (k ) = S (k ) ´ C (k ) while a more accurate iterative approach is given by Pipe [48]. The reconstructed image volume is then mc ( x ) =

1 -1 FTFT3D ( Mc (k )), c( x )

where x describes the position in the object domain. The resultant image must be divided by the Fourier transform of the convolution kernel, as convolution in one domain results in multiplication in the other. The choice of the interpolation kernel is ultimately a trade-off between precision and speed and can have great impact on the reconstruction results [49]. In general, good convolution kernels act over a quite localized region of k-space. As one tries to reduce error due to interpolation with wider kernels, the necessary time for interpolation grows rapidly. A simple and fast convolution kernel is the triangular window with a total width of two k-space samples. A popular kernel with more precise interpolation is the Kaiser–Bessel window. Dale et al. [50] introduced the use of a precalculated lookup table that allows for very rapid gridding in real-time applications. Gridding introduces some errors from the interpolation process, but has been shown to produce images with better spatial resolution than filtered backprojection [51]. As sampling in one domain connotes replication in another, the choice of interpolation kernel also determines the amount of aliasing error one suffers from adjacent image replicates in the image domain. In practice, this problem primarily affects tissue at the edge of the FOV. As vascular signal of interest is often not at the edges of the image, this problem is of less concern in vascular imaging. Intentionally, gridding the data onto a finer grid artificially creates a larger FOV in which the replicates are further apart. Known as overgridding, this process reduces aliasing error. Though initially many were overgridded by a factor of two, recent work with improved interpolation kernels demonstrates minimal error with overgridding factors as small as 1.25 [52]. As an alternative to the computation involved in interpolating data onto a Cartesian gridding, it is possible to simply use nearest-neighbor interpolation with a significantly enlarged Cartesian k-space matrix [16].

Image Degradation Due to k-Space Sampling Errors Effects of uncompensated system delays and eddy currents lead to sampling errors between the theoretical k-space sampling locations and actual k-space location. These errors manifest differently in images acquired using non-Cartesian methods relative to Cartesian trajectories. When using nonCartesian trajectories, these effects lead to blurring, particularly as one moves away from the center of the image. These errors are much more benign in Cartesian imaging, as the errors are predominantly the same along each phase-encoding acquisition. As sampling intervals become smaller with faster gradients, the size of delay that can create deleterious effects on image quality decreases. Furthermore, as these delays and eddy currents can change simply with gradient coil heating, measuring these errors quickly without a service procedure is essential for robust imaging. Correction methods to address these errors are discussed in the following sections.

Linear Eddy Current Correction Multiecho, echo-planar, and non-Cartesian trajectories place high demands on the gradient hardware, leading to increased induction of eddy currents. For conventional Cartesian acquisitions, these trajectory errors may be ignored as the resulting phase shift across the single readout direction is constant and does not appreciably affect image quality. This is not the case for multiecho, EPI, and non-Cartesian trajectories, especially those employing bipolar readouts. Methods to compensate these errors fall into two categories: system characterization and k-space measurement. System characterization methods model the gradient system as a linear system and then determine a modulation transfer function to relate the theoretical input waveforms and the actual waveforms that are created. With the modulation transfer function, the actual k-space path can be predicted for any input trajectory. While elegant, these methods require long characterization times. The characterization may no longer hold after system upgrades, system maintenance, or even changes in the gradient coil temperature. k-space measurement methods can be further broken down into methods that use numerous self-encoding pulses of different amplitudes prior to examining the readout gradient and methods that exploit localized signal. While the selfencoding methods have been proven to provide significant accuracy, they require multiple excitations and thus are rather time consuming, often requiring minutes. Due to the limitations in speed provided by these methods, methods that exploit the phase of localized signal are gaining in interest. By exciting a small-slice off isocenter and then subjecting that slice to the readout gradient under consideration, the entire test slice develops a phase which is the integral of the

13

Non-Cartesian MR Angiography

177

Fig. 13.10 k-space deviation for 1,−2,−1 gradient pulse (black waveform) is shown for each physical axis (left). Notice that much of the deviation appears similar to the gradient waveform itself, and thus the deviation can be modeled as a delay. After removing the portion of the

deviation that can be modeled as delay, much smaller deviations are remaining and primarily affect larger spatial frequencies (adopted from Brodsky et al. [54]; with permission)

applied gradient field. The integral of course is proportional to the k-space trajectory. These measurements can be repeated on each physical gradient or logical gradient if an oblique slice is being imaged. This method, suggested by Duyn et al. [53], can be performed in only a few excitations and thus can be performed easily prior to each patient scan. The difference in actual k-space trajectory and the ideal trajectory can then be provided to the reconstruction. As the difference is primarily due to short, time-constant eddy currents generated by the gradients, the actual k-space deviation is linear with the amplitude of the readout waveform selected. Thus, every waveform used during the scan need not be measured. If, for example, one readout gradient is calibrated for each axis, the k-space deviation present for scaled readout amplitudes on multiple axes is simply the scaled sum of the measured deviations on the individual gradient axes, each normalized for differences between the amplitude of the gradient using during the scan and the amplitude used during the calibration measurement. For many commercial MR systems, the uncompensated eddy currents which affect image quality can be modeled primarily as short time constant, which acts as a delay. The difference between a nominal and delayed waveform appears similar to the derivative of k-space or as the gradient waveform itself. In Fig. 13.10, a common 1,−2,1 gradient pulse used in multiple echo radial imaging is analyzed. The k-space deviations are significant, but note that majority of the deviation can be modeled as a delay. Likewise, Blatter et al. [22] studied the delays in refocusing along spiral waveforms and found changes of only fractions of a microsecond as the spiral grew in amplitude. This is logical, as there may be some

nonlinear effects occurring in the hardware as the slew requirements decrease at higher amplitudes. In practice with this method, a traditional slice-select gradient is used to select the spins at a known position, Dr, from isocenter. Note that the slice thickness should be small relative to Dr. The phase of the acquired signal is obtained and unwrapped. Choosing a slice somewhat near to isocenter greatly simplifies unwrapping the phase of the obtained signal, as this reduces the slope of the phase. Subtracting the phase accrued by simply exciting the slice and acquiring the signal with no readout gradient removes any unwanted phase due to B0 inhomogeneity. Scaling this phase measurement by 1/Dr provides the actual k-space trajectory for the gradient under test. This measurement is repeated for each of the three logical axes. Note that the magnitude of the calibration signal is actually mapping out the k-space representation of the slice profile. Thus, an RF profile that is thinner than the acquired readout resolution is helpful to remove zeros in the magnitude of the calibration signal, where the phase is indeterminate. Adding a frequency dephaser prior to the measurement can also mitigate this problem [55]. Care must also be taken in off-axis imaging, such as the breast or peripheral vasculature, to assure that slices in locations covered by the coils used during the exam are excited for the calibration.

Off-Axis Imaging Timing errors in the hardware demodulator may lead to phase errors if one uses real-time frequency demodulation to center at a location off isocenter. The easiest way to mitigate this problem, if you can increase your data acquisition rate,

178

W. Block and O. Wieben

is to avoid using the real-time frequency demodulation feature and do all demodulation during reconstruction. This solution has consequence of increasing your raw data file size also. To simplify phase unwrapping in the Duyn method when one is forced to use a slice far off isocenter to generate signal, one can acquire phase information on two slices centered a short distance on either side from the off-center image center. Real-time frequency demodulation can be performed for the image volume center during the measurement. The phase difference between these two slices is unaffected by demodulator timing errors, thus ensuring accurate k-space trajectory measurements with simple phase unwrapping. If one cannot simply increase the receiver bandwidth of an image as if you are at isocenter, then one must take care to account for these errors. The real-time frequency demodulation signal, Df (t ) = g (Gx (t )d x + Gy (t )d y + Gz (t )dz ),

(13.12)

used to center the image at a position dx, dy, dz is based on the nominal gradient waveforms and thus nominal k-space locations, marked by plus signs in Fig. 13.11. Figure 13.11

graphically describes the sources of phase errors in nonCartesian imaging. When there is a timing discrepancy between the real-time frequency demodulation hardware and the receiver, the effective demodulation can be modeled as being based on a delayed k-space trajectory along the nominal trajectory, marked by open square. However, the optimal frequency demodulation signal would be based on the actual k-space locations, marked by open circle, which are altered due to anisotropic delays and eddy currents. The delay between the actual and the nominal real-time frequency demodulation can be determined quickly [56]. Simply excite a slice and impart real-time frequency demodulation on it while playing no gradients. The difference between the actual phase accrual and the nominal accrual indicates the delay.

Time-Averaged Imaging with Radial Trajectories

Fig. 13.11 Locations of the nominal k-space sample points are marked by plus. Fast gradient characterization determines the actual locations (open circle). Due to timing errors between the frequency demodulation hardware and the receiver, the scanner creates a demodulation reference signal based on a delayed nominal trajectory (open square). Proper gridding and correction of the phase error between the phase demodulation signal required at the actual k-space points (open circle) and the phase applied by the scanner hardware (open square) provide consistent image quality (from Jung et al. [56] with permission)

Steady-state free precession (SSFP) rapidly creates high signal with T2-like contrast with bright fluid signal which can be utilized for noncontrast-enhanced angiography. However, SSFP also produces bright fat signal which may interfere with visualization of the vasculature. Methods exist to exploit the differing phase of fat and water spins in SSFP sequences, but they require tight constraints ranging from 2.4 ms at 1.5 T [57] to 3.6 ms at 1.5 and 3 T [58]. When utilized with a Cartesian trajectory, much of the TR time is spent prewinding and rewinding the spins rather than obtaining spatial encoding information. Radial sequences can be utilized to efficiently sample k-space on and out and back trajectory that eliminates the wasted time spent preparing and restoring magnetization in Cartesian trajectories. The pulse sequence, trajectory, and sample images of the peripheral vasculature obtained with 0.29-mm isotropic resolution are shown in Fig. 13.12. The inherent oversampling of the center of k-space in this radial acquisition results in signal averaging that reduces motion artifact at the expense of slightly increased blurring [44]. This principle is demonstrated in Fig. 13.13,

Fig. 13.12 Dual half-echo 3D radial pulse sequence (a) and k-space trajectory that efficiently use the available TR duration for spatial encoding. (b) Two radial lines, each one half of a diameter, are sampled

per TR. (c) Oblique MIPs of the vasculature in the knee joint acquired with 0.29-mm isotropic resolution using trajectory are shown in (b) (adapted from Jung et al. [56] with permission)

13

Non-Cartesian MR Angiography

179

Fig. 13.13 3D radial trajectory with fat/water separation is used to assess a suspected aortic aneurysm in this patient using a noncontrastenhanced steady-state acquisition with fat/water separation. Inherent

motion suppression allows a clear depiction of the ascending aorta without ECG gating

where respiratory triggering but no ECG gating is used to acquire a noncontrast-enhanced angiogram of the thoracic vasculature with a 3D radial trajectory. Similar acquisitions with a Cartesian trajectory would require ECG gating to properly depict the ascending aorta [59]. In Figs. 13.12 and 13.13, a prime benefit of the radial trajectory is effective spatial encoding. The static imaging environment in the periphery and the use of respiratory triggering in the thorax allow for longer scan times which minimize undersampling. In the next section, undersampling in sparse vascular environments using non-Cartesian trajectories is described.

window reconstruction technique with a temporal aperture that widens for k-space points acquired at larger radii is utilized [10, 11, 61–64]. The aperture widens at higher spatial frequencies to diminish the penalty from aliased, unsampled energy. These methods provide an additional order of magnitude in acceleration, though it should be noted that the acceleration factor is maximized at lower spatial frequencies and decreased for higher spatial frequencies. The k-space-weighted imaged contrast (KWIC) method was first developed to sample the center of k-space at different points in a T2 recovery curve [62]. Only a small number of radial experiments at the desired echo time were needed to characterize the center of k-space, which dominates the visualization of image contrast, especially for larger objects. As the k-space radius grew, data from an increasing aperture of acquisition points along the recovery curve is utilized. The KWIC concept can be easily extended to replace the recovery dimension with a temporal dimension for CE-MRA or perfusion imaging [11, 65]. Here, one data set can be filtered multiple times to produce a set of time-resolved images. At a desired time point, data from only a limited number of radial projections is used at the center of k-space. Larger annuli of k-space require data taken over a larger time interval to fill the space. The KWIC method has been utilized for dynamic imaging of the breast, lung, and liver [11, 66] in the assessment of perfusion when characterizing lesions. The process of creating temporal weighting with KWIC can be viewed as a density compensation problem in which different density compensation functions are generated for each reconstructed time point. If one views the different compensation functions as time progresses, the filter has an appearance like that of a moving tornado and is thus sometimes referred to as a tornado filter. A general, deterministic algorithm to generate temporal filters for variable density k-space acquisitions is provided by Liu et al. [63]. The method allows more flexibility in selecting the temporal weighting of data at the center of k-space and properly calculates the

Time-Resolved MRA Achieving a time-resolved image series by simply acquiring the same set of k-space samples at each time point is the simplest method for 4D imaging, but it also fails to exploit the significant correlation of the image volumes between time frames. Only a relatively small percentage of an imaging volume contains vascular signal while the larger amounts of static tissue are intentionally suppressed in MRA. Radial and interleaved k-space trajectories provide a basis to exploit these characteristics to improve performance of timeresolved MRA.

Temporal Processing The oversampling of the center of k-space with radial methods can be exploited in time-resolved MRA to depict temporal enhancement during a contrast injection or resolve flow throughout the cardiac cycle [60]. The projection acquisition order can be subdivided into interleaved sets so that spatial frequency orientations throughout k-space are sampled on an interval less than or equal to the desired frame rate. A sliding

180

Fig. 13.14 Time-resolved CE-MRA using 3D radial acquisitions and temporal filtering is utilized for head and neck (left) and abdominal study (right), where a postprocessing algorithm renders early enhancing vessels red and later enhancing vessels blue

density compensation function for more complicated trajectories, such as those using ramp sampling. The use of these types of tornado filters to depict the enhancement patterns throughout the body is depicted in Fig. 13.14.

Quantitative Velocity Imaging Using 3D Radial Trajectories MRA can not only provide morphological information such as the patency of vessels, but also quantitative flow measurements with velocity-sensitive encoding techniques. Phasecontrast MRI is commonly used in clinical applications, such as the evaluation of valve disease. However, the extensive scan time for PC MR acquisitions has limited their use to one-directional velocity mapping of a single slice to be completed in a single breath hold. Only recent advances in MR hardware design, particularly more powerful gradient systems and the use of parallel imaging techniques, have permitted the use of three-directional PC MR with volumetric coverage in broader human studies because of the extensive scan time associated with multidirectional velocity encoding, respiratory and cardiac gating, and volumetric acquisitions. These studies have mainly focused on the largest vessel in the human body, the aorta, because of the necessary compromises in spatial resolution to limit the scan time to below 15–20 min [67]. PC MRA with alternative trajectories, such as 3D radial sampling with PC VIPR [68], allows for significant data undersampling because of the data sparsity from the inherent subtraction process in the reconstruction, thereby providing high signal from the vessels and only small contributions from the background signal. In addition, view sharing like techniques within the cardiac cycle can be applied for additional significant savings in scan time. The obtained high spatial resolution in clinically feasible scan times and advantageous properties in respect to motion suppression make PC VIPR a viable alternative for high-resolution MRA in

W. Block and O. Wieben

Fig. 13.15 The dual-echo PC VIPR pulse sequence (a) and k-space trajectory (b). Velocity encoding is achieved with the use of standard bipolar gradients superimposed onto the prewinders along the direction of velocity encoding. Acquisitions with reversed bipolar gradients are used for the removal of phase contributions that are not contributed to motion

patients in which the standard contrast-enhanced (CE) MRA is contraindicated due to risk associated with nephrogenic systemic fibrosis (NSF) [69] and improves the ability to quantify velocity and flow parameters as well as hemodynamic derivatives, such as relative pressures [60], pulse wave velocities, and wall shear stress. To reliably achieve high-quality images, several correction schemes are applied to account for the effects of T1 saturation, trajectory errors, motion, and aliasing associated with undersampling. A dual-echo acquisition as shown in Fig. 13.15 can be used to derive a field map and reduce off-resonance effects with a computationally more demanding reconstruction process [70]. Trajectory errors are compensated for the gridding process by independently mapping out trajectory derivations for the radial readout gradients as well as the bipolar gradients in x, y, and z. This is accomplished with a short calibration routine using the Duyn method [53] and using the principle of superposition for the individual combinations of projection angles and flow-encoding gradients. Ultimately, the phase-contrast data are reconstructed as magnitude images, velocity vector fields, and angiograms calculated similar to complex difference images, and additional postprocessing is applied to derive hemodynamic parameters from the velocity fields. Advanced visualization software can be used to interactively display the volumetric, cine velocity vector fields, possibly in combination with superimposed vascular anatomy as shown in Fig. 13.16. However, these displays require significant postprocessing since no such software packages dedicated to medical imaging currently exist.

Advanced Temporal Processing Newer reconstruction methods aim to speed time-resolved imaging by undersampling while using other information to limit undersampling artifacts. In general, these image

13

Non-Cartesian MR Angiography

181

Fig. 13.16 PC VIPR exam of a 18-month-old male with pulmonary venolobar syndrome consisting of a hypoplastic right pulmonary artery (RPA), partial anomalous pulmonary venous return (PAPVR) to superior vena cava (SVC) and inferior vena cava (IVC), and an anomalous systemic pulmonary artery from the abdominal aorta to the right lower lobe. LA left atrium. The noncontrast-enhanced MR angiogram (a) shows a posterior view from volume rendered PC VIPR data. Flow-time curves were

measured at different locations using advanced visualization software, here in the ascending aorta (b) and anomalous systemic artery (c). With such 4D MR flow imaging, measurements can be made in any arbitrary orientation after the images have been acquired. (d, e) Hemodynamic analysis for a patient with aortic coarctation. (d) Flow velocity profiles showing the highest velocity immediately distal to the coarctation. (e) Pressure difference map showing the drop over the coarctation

estimation methods require sparsity in some domain. These methods include a family of methods that use a high SNR image with little temporal resolution to constrain an image reconstruction using a limited amount of temporal data, known as HighlY constrained back Projection (HYPR) [71] and HYPR Local Reconstruction (HYPR LR) [13]. These single-pass methods are generally rapid, but their accuracy can vary with the sparsity of the object. Improved accuracy can be achieved with iterative reconstruction methods, such as IHYPR [72] or conjugate gradient HYPR [73]. The relation of these methods to compressed sensing is developed in Lustig et al. [14]. Here, the requirements for sparsity are more general. For example, sparsity can be present in the actual image itself, the derivative of the image, or the wavelet representation of the image.

8. Tsai CM, Nishimura DG. Reduced aliasing artifacts using variabledensity k-space sampling trajectories. Magn Reson Med. 2000;43:452–458. 9. Korosec FR, Frayne R, Grist TM, Mistretta CA. Time-resolved contrast-enhanced 3D MR angiography. Magn Reson Med. 1996; 36:345–351. 10. Barger AV, Block WF, Toropov Y, Grist TM, Mistretta CA. Timeresolved contrast-enhanced imaging with isotropic resolution and broad coverage using an undersampled 3D projection trajectory. Magn Reson Med. 2002;48:297–305. 11. Song HK, Dougherty L. Dynamic MRI with projection reconstruction and KWIC processing for simultaneous high spatial and temporal resolution. Magn Reson Med. 2004;52:815–824. 12. Mistretta CA, Wieben O, Velikina J, Block W, Perry J, Wu Y, Johnson K. Highly constrained backprojection for time-resolved MRI. Magn Reson Med. 2006;55:30–40. 13. Johnson KM, Velikina J, Wu Y, Kecskemeti S, Wieben O, Mistretta CA. Improved waveform fidelity using local HYPR reconstruction (HYPR LR). Magn Reson Med. 2008;59:456–462. 14. Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magn Reson Med. 2007;58:1182–1195. 15. Meyer CH, Hu BS, Nishimura DG, Macovski A. Fast spiral coronary artery imaging. Magn Reson Med. 1992;28:202–213. 16. Oesterle C, Markl M, Strecker R, Kraemer FM, Hennig J. Spiral reconstruction by regridding to a large rectilinear matrix: a practical solution for routine systems. J Magn Reson Imaging. 1999;10: 84–92. 17. Tsai CM, Nishimura DG. Reduced aliasing artifacts using variabledensity k-space sampling trajectories. Magn Reson Med. 2000;43:452–458. 18. Thedens DR, Irarrazaval P, Sachs TS, Meyer CH, Nishimura DG. Fast magnetic resonance coronary angiography with a threedimensional stack of spirals trajectory. Magn Reson Med. 1999;41: 1170–1179. 19. Gurney PT, Hargreaves BA, Nishimura DG. Design and analysis of a practical 3D cones trajectory. Magn Reson Med. 2006;55: 575–582. 20. Stehning C, Bornert P, Nehrke K, Eggers H, Stuber M. Freebreathing whole-heart coronary MRA with 3D radial SSFP and self-navigated image reconstruction. Magn Reson Med. 2005;54: 476–480. 21. Holmqvist MH, Cao J, Hernandez-Pineda R, Jacobson MD, Carroll KI, Sung MA, Betty M, Ge P, Gilbride KJ, Brown ME, Jurman ME,

References 1. Lauterbur PC. Image formations by induced local interactions: examples employing nuclear magnetic resonance. Nature. 1973;242:190–191. 2. Edelstein WA, Hutchison JM, Johnson G, Redpath T. Spin warp NMR imaging and applications to human whole-body imaging. Phys Med Biol. 1980;25:751–756. 3. Glover GH, Pauly JM. Projection reconstruction techniques for reduction of motion effects in MRI. Magn Reson Med. 1992;28: 275–289. 4. Meyer CH, Hu BS, Nishimura DG, Macovski A. Fast spiral coronary artery imaging. Magn Reson Med. 1992;28:202–213. 5. Irarrazabal P, Nishimura DG. Fast three dimensional magnetic resonance imaging. Magn Reson Med. 1995;33:656–662. 6. Larson PE, Nishimura D. Variable field-of-view three-dimensional projection-reconstruction imaging. Proceedings of the International Society of Magnetic Resonance in Medicine; Berlin, Germany. May 19–25, 2007; p. 417. 7. Nishimura DG, Jackson JI, Pauly JM. On the nature and reduction of the displacement artifact in flow images. Magn Reson Med. 1991;22:481–492.

182

22.

23.

24. 25. 26. 27. 28.

29.

30.

31. 32.

33.

34.

35.

36.

37.

38. 39. 40.

41.

42.

43.

W. Block and O. Wieben Lawson D, Silos-Santiago I, Xie Y, Covarrubias M, Rhodes KJ, Distefano PS, An WF. Elimination of fast inactivation in Kv4 A-type potassium channels by an auxiliary subunit domain. Proc Natl Acad Sci USA. 2002;99:1035–1040. Blatter DD, Parker DL, Robison RO. Cerebral MR angiography with multiple overlapping thin slab acquisition. Part I. Quantitative analysis of vessel visibility. Radiology. 1991;179:805–811. Santos JM, Cunningham CH, Lustig M, et al. Single breath-hold whole-heart MRA using variable-density spirals at 3T. Magn Reson Med. 2006;55:371–379. Bornert P, Schomberg H, Aldefeld B, Groen J. Improvements in spiral MR imaging. Magma. 1999;9:29–41. King KF, Foo TK, Crawford CR. Optimized gradient waveforms for spiral scanning. Magn Reson Med. 1995;34:156–160. Glover GH. Simple analytic spiral K-space algorithm. Magn Reson Med. 1999;42:412–415. Block KT, Frahm J. Spiral imaging: a critical appraisal. J Magn Reson Imaging. 2005;21:657–668. Gmitro AF, Alexander AL. Use of a projection reconstruction method to decrease motion sensitivity in diffusion-weighted MRI. Magn Reson Med. 1993;29:835–838. Jung KJ, Cho ZH. Reduction of flow artifacts in NMR diffusion imaging using view-angle tilted line-integral projection reconstruction. Magn Reson Med. 1991;19:349–360. Trouard TP, Theilmann RJ, Altbach MI, Gmitro AF. High-resolution diffusion imaging with DIFRAD-FSE (diffusion-weighted radial acquisition of data with fast spin-echo) MRI. Magn Reson Med. 1999;42:11–18. Bergin CJ, Pauly JM, Macovski A. Lung parenchyma: projection reconstruction MR imaging. Radiology. 1991;179:777–781. Gewalt SL, Glover GH, Hedlund LW, Cofer GP, MacFall JR, Johnson GA. MR microscopy of the rat lung using projection reconstruction. Magn Reson Med. 1993;29:99–106. Rasche V, de Boer RW, Holz D, Proksa R. Continuous radial data acquisition for dynamic MRI. Magn Reson Med. 1995;34: 754–761. Rasche V, Holz D, Proksa R. MR fluoroscopy using projection reconstruction multi-gradient-echo (prMGE) MRI. Magn Reson Med. 1999;42:324–334. Rasche V, Holz D, Kohler J, Proksa R, Roschmann P. Catheter tracking using continuous radial MRI. Magn Reson Med. 1997;37: 963–968. Shankaranarayanan A, Wendt M, Aschoff AJ, Lewin JS, Duerk JL. Radial keyhole sequences for low field projection reconstruction interventional MRI. J Magn Reson Imaging. 2001;13:142–151. Rasche V, Proksa R, Sinkus R, Bornert P, Eggers H. Resampling of data between arbitrary grids using convolution interpolation. IEEE Trans Med Imaging. 1999;18:385–392. Scheffler K, Hennig J. Reduced circular field-of-view imaging. Magn Reson Med. 1998;40:474–480. Weiss S, Rasche V. Projection-reconstruction reduces FOV imaging. Magn Reson Imaging. 1999;17:517–525. Shimizu K, Mulkern RV, Oshio K, Panych LP, Yoo SS, Kikinis R, Jolesz FA. Rapid tip tracking with MRI by a limited projection reconstruction technique. J Magn Reson Imaging. 1998;8: 262–264. Peters DC, Korosec FR, Grist TM, Block WF, Holden JE, Vigen KK, Mistretta CA. Undersampled projection reconstruction applied to MR angiography. Magn Reson Med. 2000;43:91–101. Boada FE, Gillen JS, Shen GX, Chang SY, Thulborn KR. Fast three dimensional sodium imaging. Magn Reson Med. 1997;37: 706–715. Glover GH, Pauly JM. Projection reconstruction techniques for reduction of motion effects in MRI. Magn Reson Med.1992;28: 275–289.

44. Bernstein MA, King KF, Zhou ZJ. Handbook of MRI Pulse Sequences. Amsterdam; Boston: Academic Press; 2004. xxii,1017. 45. Pipe JG. Reconstructing MR images from undersampled data: dataweighting considerations. Magn Reson Med. 2000;43:867–875. 46. Joseph PM. Sampling errors in projection reconstruction MRI. Magn Reson Med.1998;40:460–466. 47. O’Sullivan J. A fast sinc function gridding algorithm for Fourier inversion in computer tomography. IEEE Trans Med Imaging. 1985;M1:200. 48. Pipe JG, Menon P. Sampling density compensation in MRI: rationale and an iterative numerical solution. Magn Reson Med. 1999;41:179–186. 49. Jackson J, Meyer C, Nishimura D. Selection of a convolution function for Fourier inversion using gridding. IEEE Trans Med Imaging. 1991;10:473–478. 50. Dale B, Wendt M, Duerk JL. A rapid look-up table method for reconstructing MR images from arbitrary K-space trajectories. IEEE Trans Med Imaging. 2001;20:207–217. 51. Lauzon ML, Rutt BK. Polar sampling in k-space: reconstruction effects. Magn Reson Med. 1998;40:769–782. 52. Beatty PJ, Nishimura DG, Pauly JM. Rapid gridding reconstruction with a minimal oversampling ratio. IEEE Trans Med Imaging. 2005;24:799–808. 53. Duyn JH, Yang Y, Frank JA, van der Veen JW. Simple correction method for k-space trajectory deviations in MRI. J Magn Reson. 1998;132:150–153. 54. Brodsky EK, Samsonov AA, Block WF. Charaterizing and correcting gradient errors in non-cartesian imaging: are gradient errors linear time-invariant (LTI)? Magn Reson Med. 2009;62: 1466–1476. 55. Beaumont M, Lamalle L, Segebarth C, Barbier EL. Improved k-space trajectory measurement with signal shifting. Magn Reson Med. 2007;58:200–205. 56. Jung Y, Jashnani Y, Kijowski R, Block WF. Consistent non-cartesian off-axis MRI quality: calibrating and removing multiple sources of demodulation phase errors. Magn Reson Med. 2007;57:206–212. 57. Vasanawala SS, Pauly JM, Nishimura DG. Linear combination steady-state free precession MRI. Magn Reson Med. 2000;43: 82–90. 58. Leupold J, Hennig J, Scheffler K. Alternating repetition time balanced steady state free precession. Magn Reson Med. 2006;55: 557–565. 59. Groves EM, Bireley W, Dill K, Carroll TJ, Carr JC. Quantitative analysis of ECG-gated high-resolution contrast-enhanced MR angiography of the thoracic aorta. AJR Am J Roentgenol. 2007; 188:522–528. 60. Lum DP, Johnson KM, Paul RK, Turk AS, Consigny DW, Grinde JR, Mistretta CA, Grist TM. Transstenotic pressure gradients: measurement in swine--retrospectively ECG-gated 3D phase-contrast MR angiography versus endovascular pressure-sensing guidewires. Radiology. 2007;245:751–760. 61. Pipe JG. Reconstructing MR images from undersampled data: dataweighting considerations. Magn Reson Med. 2000;43:867–875. 62. Song HK, Dougherty L. k-space weighted image contrast (KWIC) for contrast manipulation in projection reconstruction MRI. Magn Reson Med. 2000;44:825–832. 63. Liu J, Redmond MJ, Brodsky EK, Alexander AL, Lu A, Thornton FJ, Schulte MJ, Grist TM, Pipe JG, Block WF. Generation and visualization of four-dimensional MR angiography data using an undersampled 3-D projection trajectory. IEEE Trans Med Imaging. 2006;25:148–157. 64. Lai P, Huang F, Li Y, Nielles-Vallespin S, Bi X, Jerecic R, Li D. Contrast-kinetics-resolved whole-heart coronary MRA using 3DPR. Magn Reson Med. 2010;63:970–978.

13

Non-Cartesian MR Angiography

65. Song HK, Dougherty L, Schnall MD. Simultaneous acquisition of multiple resolution images for dynamic contrast enhanced imaging of the breast. Magn Reson Med. 2001;46:503–509. 66. Lin W, Guo J, Rosen MA, Song HK. Respiratory motion-compensated radial dynamic contrast-enhanced (DCE)-MRI of chest and abdominal lesions. Magn Reson Med. 2008;60:1135–1146. 67. Wigstrom L, Sjoqvist L, Wranne B. Temporally resolved 3D phasecontrast imaging. Magn Reson Med. 1996;36:800–803. 68. Gu T, Korosec FR, Block WF, Fain SB, Turk Q, Lum D, Zhou Y, Grist TM, Haughton V, Mistretta CA. PC VIPR: a high-speed 3D phase-contrast method for flow quantification and highresolution angiography. AJNR Am J Neuroradiol. 2005;26: 743–749. 69. Thomsen HS, Morcos SK, Dawson P. Is there a causal relation between the administration of gadolinium based contrast media and

183

70.

71.

72.

73.

the development of nephrogenic systemic fibrosis (NSF)? Clin Radiol. 2006;61:905–906. Johnson KM, Lum DP, Turski PA, Block WF, Mistretta CA, Wieben O. Improved 3D phase contrast MRI with off-resonance corrected dual echo VIPR. Magn Reson Med. 2008;60:1329–1336. Mistretta CA, Wieben O, Velikina J, Block W, Perry J, Wu Y, Johnson K, Wu Y. Highly constrained backprojection for timeresolved MRI. Magn Reson Med. 2006;55:30–40. O’Halloran RL, Wen Z, Holmes JH, Fain SB. Iterative projection reconstruction of time-resolved images using highly-constrained back-projection (HYPR). Magn Reson Med. 2008;59:132–139. Griswold MA. More optimal HYPR reconstruction using a combination of HYPR and conjugate-gradient minimization. Proceedings of the International Society of Magnetic Resonance in Medicine; Berlin, Germany. May 19–25, 2007; p. 188.

wwwwwwwwwwww

Parallel Imaging in Angiography Nicole Seiberlich and Mark Griswold

Introduction One of the main goals in magnetic resonance angiography (MRA) is to acquire data quickly, both to reduce longer scan times to avoid artifacts due to patient motion (for time-offlight and phase-contrast imaging) and to capture relevant dynamic information (contrast-enhanced MRA). However, the amount of time it takes to generate an MRI image depends directly on the desired spatial resolution. Images with higher spatial resolution require more k-space data, and thus a longer time is needed to acquire these data. In order to meet both goals of high spatial resolution and high temporal resolution simultaneously, one must look to advanced image acquisition and reconstruction techniques, such as parallel imaging. In parallel imaging, the amount of k-space data acquired is reduced, thereby reducing the amount of time needed to acquire this data. To understand how this data undersampling affects the final image, one must consider the relationship between k-space data and the image acquired. The spacing between phase encoding lines in k-space determines the field-of-view (FoV) of the image, and the extent of k-space coverage determines the resolution. One option to reduce the scan time would be to reduce the extent of k-space coverage. While the scan would be shorter, the resolution would decrease, and fine details in the image would be lost. If a high resolution is still required but the scan time must be shortened, another option is to increase the spacing between phase encoding lines. This will reduce the FoV. However, if the object to be imaged extends beyond this smaller FoV, aliasing or fold-over artifact will occur. This happens because multiple frequencies are indistinguishable from one another when one reduces the sampling rate in one direction of k-space, leading to the aliasing present in Fig. 14.1. At this

N. Seiberlich, PhD () • M. Griswold, PhD Department of Radiology, University Hospitals of Cleveland/ Case Western Reserve University, Cleveland, OH, USA e-mail: [email protected]

14

point, the goal of shorter imaging time has been achieved, although the resulting image cannot be used because features cannot be easily distinguished due to the artifacts. In such a case, a parallel imaging reconstruction algorithm, which uses spatial information inherent in an array of receiver coils to complement gradient encoding, can be used to generate an unaliased image. In such an array of coils, each coil is more sensitive to certain parts of the object than others. An example of independent coils which acquire information from the same object with differing sensitivities is shown in Fig. 14.2. Here, each of the eight coils positioned around the head can be used to acquire a different view of the object. The variations in coil sensitivities can be used to localize the origin of the signal received and is the basis of parallel imaging. The concept of using spatial information from coil arrays to replace some or all gradient encoding was introduced in the late 1980s and early 1990s [1–3]. However, due to the lack of availability of either receiver array coils or multichannel MR systems, the acceleration of in vivo images could not be accomplished robustly with these early algorithms. This changed in 1997 when Sodickson et al. [4] introduced a method known as SMASH. This reconstruction was easy to calculate and provided the first realistic in vivo parallel imaging results. Part of the reason for SMASH’s early success was the simplicity of the concept: the idea is to use the variations in the sensitivity profiles of the individual coils in the array to exactly mimic the phase modulations normally generated by the phase encoding gradients. The other reason for its success was that multichannel receiver coils and systems were also becoming more commonplace and robust, which allowed this new parallel imaging method to be applied reliably in vivo. While SMASH was the first realistic parallel imaging method, it was shown relatively quickly to be limited in terms of real clinical application. However, the arrival of SMASH and these array coils started a revolution that has generated many other more robust and specialized parallel imaging methods, such as AUTO-SMASH [5], SENSE [6], PILS [7], and GRAPPA [8].

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_14, © Springer Science+Business Media, LLC 2012

185

186

N. Seiberlich and M. Griswold

Fig. 14.1 Top left: A schematic description of R = 4 undersampled k-space, where the solid lines have been acquired, and the dotted lines have been skipped to shorten scan time. Top right: Performing a Fourier transform leads to an aliased image. Bottom left: Parallel imaging in the form of GRAPPA can be used to reconstruct the missing lines in k-space, which leads to an unaliased image after Fourier transform. Bottom right: A SENSE reconstruction performed in the image domain can unfold the aliased image, yielding the reconstruction shown here

Fig. 14.2 An example of an eight-coil receiver array surrounding the object to be imaged, in this case, a transverse slice of the brain. Each of the eight coils is more sensitive to the portion of the brain closest to the coil, and this sensitivity falls off as one moves away from the coil. These variations in coil sensitivity are used for signal localization in parallel imaging. Note that the data from each coil is collected simultaneously and independently, as depicted with the arrows

As parallel imaging has developed, these methods have been demonstrated to be especially useful for MR angiography (MRA). For instance, many thoracic and abdominal scans are limited by the amount of time a patient can hold their breath. Employing parallel imaging allows one to increase the spatial resolution or total coverage of a scan during that breathhold. The ability to obtain a higher temporal resolution while maintaining or even improving the spatial resolution in contrast-enhanced angiography can yield additional diagnostic information for the clinician. Additionally, the timing of the contrast bolus becomes less crucial as multiple phases can be captured efficiently. The acceleration of phase-contrast and time-of-flight scans is also possible, thereby reducing artifacts associated with patient motion. It is not an exaggeration to say that parallel imaging has been employed in all aspects of MRA because of these potential benefits. In general, there are two classes of methods for reconstructing the unaliased image. The first class operates in the image domain, and employs coil sensitivity maps to “unfold” the aliased image. Although the simplest image domain method is PILS, which relies on having localized and independent receiver coils, the most commonly used algorithm in this class is SENSE. The second class of algorithms seeks to

reconstruct the missing k-space data using coil sensitivity variations in place of gradient encoding. Once the data in k-space have been reconstructed, a Fourier Transform can be performed on the original and reconstructed data to arrive at the unaliased image. The GRAPPA method, which is based on SMASH, is one of the most flexible and often employed k-space reconstruction algorithms. In addition to these standard parallel imaging methods which one can use when sampling on a Cartesian grid, one can also apply parallel imaging to non-Cartesian trajectories, such as the radial trajectory often used in MRA. The two most commonly used non-Cartesian parallel imaging techniques are analogous to the Cartesian techniques mentioned above, namely, CG SENSE [9] and radial GRAPPA [10]. CG SENSE is an iterative method that works primarily in the image domain, like SENSE, while radial GRAPPA reconstructs the missing radial projections in k-space, like GRAPPA. There are also several parallel imaging techniques which can be applied specifically to dynamic acquisitions. These methods, TSENSE [11]/TGRAPPA [12], rely on interleaved data acquisition techniques for the collection of robust calibration data. In this chapter, each of these different techniques is explained, and the applications of these methods to MRA are described.

14

Parallel Imaging in Angiography

General Remarks on Parallel Imaging Before discussing specific reconstruction algorithms, a few general remarks about parallel imaging must be made (see [13, 14] for more detailed comments on parallel imaging). The following points are important to keep in mind: • Parallel imaging algorithms are not new imaging sequences, but merely techniques for the reconstruction of undersampled data • Parallel imaging requires data acquired simultaneously and independently with multiple receiver coils, with each coil having a differing spatial sensitivity profile • Parallel imaging generally does not change the contrast behavior of the underlying imaging sequence • Parallel imaging can be used to decrease the amount of time needed to perform an imaging sequence, increase the resolution given a specific measurement time, or perform a combination of the two • The SNR in the reconstructed image is generally reduced by a factor greater than or equal to the square root of the reduction factor, as R-times fewer data points are acquired [see (14.1)] As a convention, the amount of data acquired as compared to the amount needed for a fully sampled image is described as the acceleration factor, R. For example, if 256 k-space lines are required for a given FoV and resolution, and only 64 are actually acquired, the acceleration factor of the dataset is said to be R = 4, which is depicted schematically in Fig. 14.1. This acceleration occurs in the phase encoding direction, as the total time of an MR experiment is dependent on the number of phase encoding steps used. Thus, for 2D Cartesian imaging, undersampling is performed in a single direction; for 3D Cartesian imaging, two acceleration directions are available. Acceleration factors of between 2 and 4 are currently used in a clinical setting, although factors between 9 and 12 are possible for 3D imaging in high SNR applications using appropriate receiver arrays. The acceleration factor is limited by the sensitivity variations inherent in the receiver coil used to make the measurement. As a rough limit, the acceleration factor cannot be larger than the number of coils employed, although most array configurations cannot accomplish such high acceleration factors due to the distribution of the sensitivity variations in two (or even three) spatial directions. However, the trend toward greater numbers of elements in coil arrays is influenced by the fact that more independent coils lead to higher possible parallel imaging acceleration factors. As stated in the bullet points above, parallel imaging generally does not change the contrast in the accelerated image. This does not hold true in a number of special cases, especially for multiecho or single shot acquisitions. In these

187

cases, the use of parallel imaging can improve the SNR and decrease susceptibility artifacts, as well as change the contrast, due to shorter echo train lengths [15, 16].

SNR Losses in Parallel Imaging The acceleration factor describes not only the time savings, but is also an indication of the minimum SNR loss for an image. Because R-times fewer points are acquired in an accelerated scan, the SNR of the reconstructed image will be at least R times lower than that of the unaccelerated image. An additional factor which describes the encoding efficiency of the receiver coil also influences the SNR. This parameter, known as the g-factor, is essentially a measure of how well the coil array spatially encodes the image. Given completely independent coil sensitivity profiles (i.e., perfect encoding ability), the g-factor is equal to one; if the sensitivity of each coil is identical, the g-factor is equal to infinity, since the array provides no spatial encoding, and thus the parallel imaging reconstruction cannot be solved in this case. Thus when using the SENSE reconstruction, the SNR for a reconstructed image can be described by the following equation: SNR pMRI =

SNR g× R

(14.1)

A similar effect occurs for GRAPPA reconstructions. Owing to the inclusion of this g-factor, the actual SNR loss is dependent on the coil array used to acquire the data. Arrays with more variations in coil sensitivity allow for higher acceleration factors at better SNR. In general, as the acceleration factor increases, either the SNR in the resulting image will decrease, and/or residual aliasing artifacts remain. Thus, parallel imaging can only be successfully implemented when the resulting loss in SNR does not lead to unacceptably degraded image quality. However, especially when performing contrast-enhanced examinations, MRA images generally exhibit a high SNR, making the impact of these SNR losses potentially less significant than in some other applications.

Cartesian Parallel Imaging Methods Parallel imaging was first developed for Cartesian data acquisitions, and these Cartesian methods make up the vast majority of parallel imaging reconstructions in the clinical setting. This section serves to outline the basics of several parallel imaging methods and to explain their strengths and weaknesses in comparison to one another. In addition, the application of each of these methods to MRA is discussed.

188

Fig. 14.3 Top: An example of an object imaged using a highly localized coil. When the single coil image is undersampled by a factor of R = 4, no aliasing artifacts occur because of the limited range of the coil sensitivities. Bottom: If four independent and localized coils which

N. Seiberlich and M. Griswold

completely cover the object but do not overlap are used for the measurement, each undersampled image shows a specific portion of the object with no aliasing. These single coil images can simply be combined together using SoS to form the final image

Parallel Imaging with Localized Sensitivities The most basic parallel imaging method is PILS (Parallel Imaging with Localized Sensitivities [7]). Imagine that the sensitivity profiles of each of the receiver coils used in an acquisition are highly localized and homogeneous, as in Fig. 14.3. In this case, each coil is only sensitive to a unique portion of the image. Thus the effective FoV for each single coil can be reduced without any fold-over of any signal. If we assume that the rest of the FoV is covered by coils such as this, the entire process requires only cropping of the individual coil images followed by combination of the resulting single coil images using standard methods. This process is described schematically in Fig. 14.3. In order for PILS to yield artifact-free images, localized and homogeneous coil sensitivities are required, as shown in the top center of Fig. 14.3. However, few clinical receiver coils have such properties; usually the coils will be more sensitive to one side of the object than another (for an example of such behavior, see Fig. 14.2). This inhomogeneous and nonlocal sensitivity profile leads to several problems in the PILS reconstruction. Firstly, the reduced FoV images may show shading, where areas of the object closer to the coils have a higher signal than those further away from the coils. Secondly, if the sensitivity is not limited to a small area, aliasing artifacts resulting from the extended coil sensitivity may appear in the folded images, leading to artifacts in the final image. Thus, although PILS is the simplest incarnation of parallel imaging, it cannot in general be employed when clinical coils with complex coil sensitivities are used for data collection. However, it has found limited use for the acceleration of MRA, for instance using the spiral trajectory in coronary artery angiography [17].

Fig. 14.4 When working with a coil that does not have localized sensitivities, one must consider the effects of the coil sensitivity on each of the aliased pixels. If the image on the left is undersampled by a factor of R = 4, the four pixels denoted as r1 through r4 will alias on top of one another. However, each of these pixels will be weighted by the coil sensitivity at the appropriate location. Using aliased images from a number of receiver coils and a knowledge of the coil sensitivity, the SENSE reconstruction can be employed to generate unaliased images

Sensitivity Encoding SENSE (Sensitivity Encoding [6]) is the most common image-based parallel imaging method. It functions by using explicit knowledge of the coil sensitivity profiles to separate pixels which are aliased in the reduced FoV image. In Fig. 14.4, the relationship between the object to be imaged, the coil sensitivity, and the undersampled single coil image is shown for one receiver coil. Here, using an acceleration factor of R = 4, four different pixels fold on top of one another. Each pixel is weighted by the sensitivity of the receiver coil at its source location in space. When using multiple coils, the following equation can be written to describe the relationship between the actual image and the aliased image (for simplification, the noise characteristics of each coil have been neglected):

14

Parallel Imaging in Angiography

é I A ù C A1 C A 2 êI ú C ê B ú = B1 C B 2 ê I C ú CC1 CC 2 ê ú ë I D û C D1 C D 2

C A3 CB3 CC 3 CD3

189

CA4 CB 4 CC 4 CD 4

ér1 ù êr ú 2 ×ê ú êr ú ê 3ú ëê r 4 ûú

(14.2)

This equation can be written more efficiently using a matrix formulation:   I = Cˆ × r (14.3)  In both of these equations, the values in I represent the aliased voxels from the coils A through D, r are the actual voxel values at locations 1 through 4, and the matrix Cˆ contains the coil sensitivity values for the four different coils at the appropriate locations. If the  coil sensitivity matrix is known, then the aliased pixels I can be used to determine the actual values of the underlying pixels in the proper locations  r by taking the inverse of the matrix Cˆ :    Cˆ -1 × I = Cˆ -1 × Cˆ × r = r (14.4) If the number of coils is greater than the acceleration factor, which is often the case, the sensitivity matrix Cˆ is not square, and the Moore-Penrose pseudoinverse (denoted as pinv) must be employed in place of the inverse: ˆ = (A ˆ H A) ˆ -1 A ˆH pinv[A]

(14.5)

ˆ H is the conjugate transpose (otherwise known as the where A ˆ. Hermetian conjugate) of the matrix A In order to employ SENSE, one must first obtain a coil sensitivity map. In areas of the body with little motion, such a coil sensitivity map can be obtained by acquiring one image with the receiver array to be used during accelerated acquisition and one image with a homogeneous single coil (such as a body coil). Then the images generated from each channel of the array are divided by the image from the body coil, resulting in images which show the sensitivity profiles of the single channels in the array without contamination from the object to be imaged. Similarly, instead of acquiring a separate image with the body coil, each image from the array can be divided by the Sum-of-Squares (SoS) combination of the multichannel images [18]. While this method again removes any influence from the object itself, the final SENSE images will have a potentially nonuniform SoS weighting. Another option is to use an adaptive combine method [19], which uses statistical estimation methods to determine the coil sensitivity patterns. For all of these methods, the coil sensitivity maps images do not have to have the same resolution as the final images, as the coil sensitivities generally vary smoothly over space. Thus, lower resolution images can also be employed to find the coil sensitivity maps. When using a different resolution, however, filtering should be used to assure that there are no artifacts due to Gibb’s ringing etc., which could also lead to artifacts in the SENSE reconstruction.

SENSE has been shown to deliver the best possible image reconstruction with optimal SNR given an accurate sensitivity map [14]. Because the sensitivity profiles depend on the loading of the coil and the placement of the object in the array, the sensitivity measurement must be performed separately for each patient. If the patient moves between the acquisition of the sensitivity profile information and the SENSE reconstruction, the coil sensitivity information will no longer match between the two datasets, and there will be residual aliasing in the final reconstructed image. In certain imaging applications, i.e., lung imaging, dynamic imaging, or single-shot methods, the acquisition of a sensitivity map is either time consuming or difficult due to SNR limitations. SENSE has been applied robustly to the acceleration of many different types of MRA examinations, and is often used in combination with other image reconstruction techniques, especially view-sharing or keyhole-type methods. For example, imaging of the coronary arteries has been accelerated using SENSE [20–26], improving the spatial resolution and shortening the length the patient must hold their breath. This method has also been applied to the assessment of AVMs in the brain [27–29] and other abnormalities in the vasculature of the head and neck [30, 31] using dynamic imaging. Imaging of the renal arteries [32–34] and peripheral vasculature [35– 37] has been improved, resulting in less venous contamination and higher spatial and temporal resolutions, and multistation exams can be performed using SENSE or SENSE-like techniques [38]. Nondynamic methods can also be accelerated using SENSE [39], and signal increases due to flexibility in imaging parameters have even been reported [40, 41]. By combining additional reconstruction techniques such as partial Fourier and view-sharing with the SENSE reconstruction, the temporal footprint in MRA exams of cranial and peripheral vasculature can be dramatically decreased [42, 43].

Generalized Autocalibrating Partially Parallel Acquisitions One of the most commonly used k-space based parallel imaging algorithm is GRAPPA [8]. The basic idea of GRAPPA is that coil sensitivity variations can be used to generate missing spatial harmonics in the undersampled k-space data. This can be understood by examining the 1D Fourier Transform: S ( k y ) = ò r( y) × e

ik y y

× dy

(14.6)

In standard imaging experiments, the spin density r ( y ) is ik y modulated by the appropriate spatial harmonic, e y , through the application of an encoding gradient. The sum of these products yields the signal S (ky ) at that particular location in k-space. Including the influence of the coil sensitivity leads to the following equation for the signal from a coil L: SL ( k y ) = ò r ( y) × CL ( y) × e

iky y

× dy

(14.7)

190

N. Seiberlich and M. Griswold

Fig. 14.5 A schematic description of GRAPPA. Top left: The undersampled R = 4 multichannel k-space data, where the black points have been acquired, and the white points have not. The autocalibration signal (ACS) can be seen as the extra acquired data in the center of k-space. Bottom left: The first step in GRAPPA is to determine the weight sets for a given kernel using the ACS data. A 2 × 3 kernel is shown here, where six source points from each coil are used to fit a target point in a single coil. This kernel appears multiple times in the ACS data, and all

In parallel imaging, when the data are undersampled to reduce scan time, some of these k-space lines are skipped. By examining the encoding equations for two lines separated by a distance m in k-space, one can see that the only difference between the two is the spatial harmonic applied, i.e., for the signal at a k-space location ky + mDky, the signal can be written as: SL (k y + mDk y ) = ò ρ ( y)·CL ( y)·e

i ( k y + mDk y ) y

·dy

(14.8)

Thus, if one can recreate the additional spatial harmonic i ( mDk y ) y of the missing k-space line, or e , the missing lines can be reconstructed. In GRAPPA, this reconstruction involves using a linear combination of the coil sensitivity variations of all the coils in the array to mimic the gradient encoding: CL ( y) × e

i ( mDky ) y

NC

» å nK , L , m × C K ( y )

(14.9)

K =1

where both K and L run from 1 to the number of coils in the array, or NC, and for a complete reconstruction of the missing k-space lines, m runs from 1 to R − 1. By substituting (14.9) into (14.8), the signal from an acquired line of k-space

 repetitions are used to determine the GRAPPA weights w for the kernel. Top center: The same kernel appears in the undersampled portion of k-space, and when the weights are applied to the source points, the missing target point can be reconstructed. Bottom right: The same weight set can be used to reconstruct all of the missing points in the first coil with the same relationship to the source points. In order to reconstruct the entire missing k-space points, weight sets for each missing spatial harmonic and coil must be calculated, using the same process shown here

can be used to generate an approximation of the signal from a missing line: NC

SL (ky + mDky ) » å nK ,L ,m × SK (ky)

(14.10)

K =1

or, in matrix form:   S (ky + mDky ) » wˆ m × S (ky )

(14.11)

Equation (14.10) is simply the matrix version of (14.11), and both state that a linear combination of acquired signals from all receiver coils yields an approximation of the missing signal for a single coil. The elements of the matrix wˆ , referred to as the GRAPPA weight set, are the values of n for each of the coil combinations, and the subscript m remains to denote that a different weight set is required for each missing spatial harmonic, i.e., each of the missing k-space lines. Before GRAPPA can be employed for a reconstruction, the weight sets must be determined. This can be accomplished by acquiring a few additional lines in the center of k-space (where the signal level is the highest), as shown on the upper left-hand side of Fig. 14.5. These autocalibration

14

Parallel Imaging in Angiography

191

signal lines, or ACS, can be used in conjunction with (14.11) to generate the values of the weight set:   SACS (ky + mDky ) » wˆ m × SACS (ky )

(14.12)

This GRAPPA kernel (and therefore the harmonic relationship) appears many times in the autocalibration signal, although the weight set for each equation is the same throughout the entire k-space. Thus, the many appropriate signal and target vectors can be written together in the form of a matrix. The weight sets can then be determined with the help of the pseudoinverse, which allows the nonsquare matrix made up of the source signals to be inverted and results in the least-squares solution to this linear problem:   SACS (ky + mDky ) × pinv[ SACS (ky )] » wˆ m

using combined methods have been reported for imaging of cerebral AVMs [70], carotid arteries [71], and lower extremity vasculature [72].

(14.13)

In this equation, the pseudoinverse is denoted as “pinv.” Once the weight sets have been determined, they can be used to reconstruct the portions of the dataset where data are missing to arrive at the final image. When using GRAPPA to reconstruct undersampled data, the additional ACS lines can be acquired before, after, or during the accelerated acquisition. Because the calibration and reconstruction are both performed in k-space, no explicit knowledge of the coil sensitivity map is necessary, making GRAPPA more robust in situations where it is difficult to obtain an accurate estimate of the sensitivity profiles. Like SENSE, the GRAPPA technique has been used in many aspects of MRA to accelerate the imaging process. The total acquisition time for noncontrast techniques such as time-of-flight [44, 45] and phase-contrast angiography [46, 47] can be shortened using GRAPPA, making the images less prone to motion artifacts. However, most of the focus has been on decreasing the temporal footprint for dynamic contrast-enhanced imaging. In the brain, this has enabled the dynamic investigation of cranial and cerebral arteries and veins with high spatial resolutions [48–51]. Renal and abdominal angiography has also been performed using GRAPPA and similar parallel imaging techniques to increase the spatial and temporal resolution [52–56]. The application of GRAPPA to peripheral [57–60] and whole body angiography [61–64] has also been explored. Because reconstructing images of the lungs can be difficult using SENSE due to problems acquiring coil sensitivity maps, the acceleration of pulmonary imaging is performed mostly using GRAPPA [65–67]. Additionally, coronary angiography in threedimensions has benefitted from scan time reduction with GRAPPA, allowing free-breathing acquisitions to become feasible [68, 69]. Additionally, by combining parallel imaging with view-sharing, higher temporal resolutions can be obtained than when using either acceleration method alone. Dynamic imaging with high temporal and spatial resolution

Three Dimensional Parallel Imaging Although the formulations discussed in the previous sections are for two-dimensional parallel imaging algorithms, these methods can also be used in three dimensions. In the 3D case, the undersampling can occur in either the phase or partition direction, or both. By dividing the total undersampling between the two encoding directions, higher acceleration factors can often be achieved with reduced residual artifacts. This is due to the fact that the receiver coil sensitivity profiles tend to vary in several directions which can both be exploited for parallel imaging acceleration. For example, it is generally better to perform a total acceleration of R = 4 by applying R = 2 in the phase encoding direction and R = 2 in the partition encoding direction instead of R = 4 in the phase direction and R = 1 in the partition direction (shown in Fig. 14.6). Another option in three dimensional encoding is to shift the acquired read-out lines to generate more benign aliasing artifacts. The CAIPIRINHA method [73] allows the acquisition pattern to be tailored to fit the anatomy and coil sensitivity distributions. Several examples of CAIPIRINHA patterns for a total acceleration factor of R = 4 are also shown in the bottom row Fig. 14.6. When performing 3D reconstructions, both the SENSE and GRAPPA methods work basically as described above. In order to perform the 3D SENSE reconstruction, a three dimensional map of the coil sensitivities is required, but the actual implementation of the algorithm is the same. Similarly, while Fig. 14.5 shows how GRAPPA is used to reconstruct a two-dimensional dataset using a 2D GRAPPA kernel, the same principles can be applied to three-dimensional data using a 3D kernel [74], where the source points are taken from the read, phase, and partition directions.

Non-Cartesian Parallel Imaging Methods In the previous section, Cartesian k-space based parallel imaging methods were explained and discussed. However, these methods can only be applied when the k-space undersampling is on a grid or where there are regular patterns in the undersampled k-space data. Such regular undersampling leads to well-defined aliasing characteristics in the image domain. Thus, given an acceleration factor R in a single direction, R voxels must be separated from one another; that this is the case can be seen in (14.2). However, when non-Cartesian k-space data are undersampled, aliasing artifacts appear in all directions, and each voxel in the image domain can

192

N. Seiberlich and M. Griswold

Fig. 14.6 Top: Standard SENSE-like acquisition patterns for a total acceleration factor of R = 4, where the phase encoding direction is top to bottom, the partition encoding direction left to right, and the read encoding direction goes into the page. The data can be undersampled in the phase encoding direction alone (Ry = 4, Rz = 1, left), the partition

encoding direction (Ry = 1, Rz = 4, center), or both (Ry = 2, Rz = 2, right). Bottom: CAIPIRINHA-like acquisition patterns, where the total acceleration factor is still R = 4, but the arrangement of acquired points has an additional sheer in comparison to the SENSE-type patterns

potentially alias with all of the other voxels [75, 76]. Thus, non-Cartesian parallel imaging is considerably more complicated and requires different reconstruction algorithms.

is the number of k-space points acquired for each channel.  The vector v has a length N 2, where N is the image matrix size. The encoding matrix, which necessarily has a size of NC × nk ´ N 2, can be written as:   EL ,k ,r = e i×k ×r × CL (r )

Conjugate Gradient SENSE Conjugate Gradient SENSE (CG SENSE [9], similar to [77]) was one of the first parallel imaging methods to be described for non-Cartesian parallel imaging reconstruction. Although it shares a name with Cartesian SENSE, the principles of CG SENSE are different than its Cartesian counterpart. As stated above, pixels from all of the FoV can be aliased in an undersampled non-Cartesian image. The relationship between the image and the acquired non-Cartesian k-space data can be written as a matrix equation (again ignoring the noise properties of the system):   Eˆ × v = m

(14.14)

 where m is a vector containing the acquired k-space points  for each coil, v is a vector containing the unknown image voxel values, and Eˆ is a matrix which represents a combina tion of coil and gradient encoding. The vector m has a length of NC × nk, where NC is the number of receiver coils, and nk

(14.15)

 where L runs from 1 to NC, r is the pixel position in the image domain, and k is the k-space sampling point. Thus, if NC·nk is greater than N 2, it should be possible to reconstruct the missing pixel values by inverting the encoding matrix:    ( Eˆ H Eˆ ) × v = Eˆ H × m = v

(14.16)

given that one has acquired the k-space data and has a knowledge of the k-space trajectory and coil sensitivity maps. Again, the influence of noise correlation between the different receiver channels has been neglected. Solving (14.16) directly would require immense amounts memory and computation time due to the large sizes of the matrices and vectors involved. However, this formulation is  ˆ ·r = b where the values in the a linear equation of the form A  matrix Aˆ and vector b are known, and the values in the vector  r are unknown. This means that although it would be computationally challenging to directly solve (14.16), other

14

Parallel Imaging in Angiography

193

Fig. 14.7 A schematic of the CG SENSE algorithm, as described in the text

solutions for this basic type of linear system can be used. In CG SENSE, Pruessmann et al. [9] chose to solve the equation using the well-known iterative Conjugate Gradient method, which is often employed when solving large systems of linear equations. Decades of experience with the CG algorithm outside of MRI have shown that the algorithm converges predictably after a small number of iterations, which is important in for fast and robust image reconstructions. One of the important features of the CG SENSE method is that the encoding matrix depicted in (14.15) is made up of Fourier terms and the coil sensitivities, and one can use a Fourier transform and  coil sensitivity multiplication instead of calculating the E and E H matrices explicitly. The implementation of the CG SENSE method is depicted schematically in Fig. 14.7. First the original undersampled non-Cartesian data are gridded, transformed to the image domain, and multiplied by the complex conjugate of the coil sensitivities to yield a first estimate of the image. This estimate is fed into the Conjugate Gradient algorithm, and the residual is converted back to the non-Cartesian data space by multiplying by the coil sensitivities, Fourier transform, and degridding. This operation continues until the residual from the CG step reaches a predetermined level, resulting in the final image. The CG SENSE method is important for parallel imaging because it allows one to reconstruct arbitrary undersampled trajectories. However, like SENSE, CG SENSE requires a coil sensitivity map, which means that the two algorithms have similar limitations. Although the coil map can sometimes be extracted from the undersampled non-Cartesian data, errors in this map lead to reconstruction artifacts which cannot be removed. Finally, when performing dynamic imaging, CG SENSE must be performed separately for each time

frame, which can make the reconstruction process much more time consuming than k-space based methods, which typically require only one weight set for all time frames. Because of the complexity of the reconstruction algorithm as compared to its Cartesian counterpart, CG SENSE is currently employed less often for MRA than SENSE. However, as trajectories such as radial can be advantageous for dynamic imaging due to their oversampling of the center of k-space, the use of non-Cartesian parallel imaging is becoming increasingly important. For instance, it has been shown that CG SENSE can be used for rapid time-resolved volumetric imaging of the coronary arteries [78]. Additionally, CG SENSE may prove to be instrumental in combining parallel imaging ideas with the emerging field of compressed sensing [79], as these methods both use iterative processes for reconstruction.

Radial GRAPPA As can be seen in the basic GRAPPA equation, (14.11), as well as the schematic shown in Fig. 14.5, GRAPPA can be applied efficiently only when the undersampling in k-space leaves regular patterns of missing data points. If this is not the case, a separate GRAPPA weight set is required for each missing point in k-space, a time-consuming and computationally intensive task. A schematic of the irregular undersampling found in non-Cartesian trajectories, in this case radial, can be seen on the far left hand side of Fig. 14.8. It is clear that if the GRAPPA weight set was determined for the specific pattern highlighted in gray with the solid outline, the application of the weight set to other source points would lead to errors in the reconstructed points. The source points

194

N. Seiberlich and M. Griswold

Fig. 14.8 Left: A schematic of an undersampled radial acquisition, where the acquired points are shown in gray and the missing points in white. No regular kernel patterns exist in this data, as described in the text. Center: A zoom of a radial segment, showing that while each potential kernel for GRAPPA reconstruction has a different configuration,

these configurations are similar to one another. Right: The central approximation of radial GRAPPA is that within a segment, similar patterns exist. Using this assumption, Cartesian GRAPPA is applied to calibrate and reconstruct within each segment

highlighted in gray with the dotted outline have a similar degree of undersampling, but the direction of undersampling is different. Similarly, the source points highlighted in black have a similar direction of undersampling, but the degree of undersampling is different. Thus, the regular undersampling patterns in k-space that are crucial for a successful GRAPPA reconstruction are not present in undersampled non-Cartesian datasets, and standard Cartesian GRAPPA cannot be applied to such datasets. However, one can see that in cases where the non-Cartesian trajectory is highly symmetric, similar patterns do exist. Non-Cartesian GRAPPA takes advantage of these similar patterns to reconstruct the missing k-space points. In 2003, it was proposed that one could assume that the undersampling pattern is similar enough in some portions of k-space such that that one could use of the same GRAPPA weight set for reconstruction for all missing points in a local region [10]. For instance, the large outlined area in Fig. 14.8 shows such a segment; the data points are not parallel, but the weight set determined using these data will given an approximate fit for the entire segment. Thus, although the source points used with the weight set do not have exactly the proper relationship in k-space with the target point, the approximation allows for the reconstruction of the missing radial rays. In order to perform radial GRAPPA, a fully sampled radial calibration dataset must be acquired. The data must be fully sampled to allow the calculation of GRAPPA weights for each segment; a low resolution dataset would not provide information about the outer portions of k-space (unlike in the Cartesian case), because the geometry in these areas is different in radial acquisitions. Both the fully sampled and undersampled data are divided into segments. For each

segment, GRAPPA weights are calculated using the fully sampled radial data and then the weights are applied to the same segment in the undersampled data, just as in Cartesian GRAPPA. Once this process has been repeated for each of the segments, the reconstructed data are gridded and converted to the image domain using the Fourier Transform. While this method does not require a coil sensitivity map and is not iterative, it has a number of drawbacks. The first is that radial GRAPPA is only a rough approximation to the actual solution, and thus the angular weight set which is used in a given segment may not accurately reconstruct the spatial harmonics which are missing. This approximation stems from the fact that the distances between the source and target points change by a small amount even within a single segment. The second drawback is that a fully sampled radial dataset is required to determine the weight sets for each angular segment, as the acceleration factor and direction, and therefore the weight sets, are different for each segment. Thus, the use of radial GRAPPA is generally limited to dynamic studies where a single fully sampled dataset can be acquired and used for the reconstruction of subsequent undersampled datasets. Another potential problem is determining the optimal segment size. If large segments are chosen, the approximation of similar geometries within the segment does not hold, and errors result in the reconstruction. If the segments are chosen to be small, there may not be enough information to allow for a proper calibration of the GRAPPA weights. This segmentation choice is especially crucial at high acceleration factors. One option to address this problem is to use multiple fully sampled datasets for the calibration. Then, very small segments can be used in k-space, because additional calibration information can be obtained from the multiple

14

Parallel Imaging in Angiography

Fig. 14.9 Left: An example of a Stack-of-Stars trajectory, which is undersampled in both the radial and Cartesian directions by a factor of R = 2 (total acceleration factor of R = 4). Center: The undersampling in the radial direction is shown (solid lines are acquired, dotted lines are not acquired). Radial GRAPPA can be used to reconstruct undersampling in this direction. Right: The undersampling in the Cartesian direction can be reconstructed using standard Cartesian GRAPPA applied separately for each acquired stack of projections

repetitions through time. This method, known as throughtime radial GRAPPA [80], is especially advantageous when a high frame rate and high acceleration factor are required for dynamic imaging. Again, due to the more complex nature of radial GRAPPA in comparison to standard GRAPPA and the fact that k-space based non-Cartesian parallel imaging methods are not implemented on clinical MRI scanners, radial GRAPPA is not commonly used for MRA image reconstruction. However, a three-dimensional version of the radial trajectory is advantageous for angiography, namely, the Stack-of-Stars trajectory (shown in Fig. 14.9). This trajectory is radial in the kx–ky plane, and Cartesian in the kz direction. It is possible to undersample the data in both the radial and Cartesian directions; the figure shows an undersampling factor of 2 in the radial direction and a factor of 2 in the Cartesian direction, leading to a total acceleration factor of 4. The reconstruction of either or both of these undersampling directions can be performed using GRAPPA. When undersampling in only the Cartesian direction (similar to what is depicted at the far right side of Fig. 14.9), standard Cartesian GRAPPA can be used, as shown in [81]. If the data are undersampled in both the radial and Cartesian directions, a 3D formulation of radial GRAPPA can be employed [82].

Dynamic Parallel Imaging Methods: TSENSE and TGRAPPA In MRI and especially MRA, the ability to obtain dynamic information is becoming more and more important. To meet this need, parallel imaging methods which specifically exploit the temporal nature of such data have been developed. As above, there are primarily two approaches to this

195

problem: one based on images domain processing similar to SENSE, which is referred to as Temporal SENSE or TSENSE [11], while the other involves k-space process similar to GRAPPA and is referred to as Temporal GRAPPA or TGRAPPA [12]. Both of these are based on the same basic observation: if the coil sensitivity maps or calibration data do not reflect the underlying sensitivity profiles of the coils due to motion or other issues, no parallel imaging algorithm will be able to accurately reconstruct unaliased images, and residual artifacts will remain. One could imagine acquiring a separate coil sensitivity maps or calibration data for each undersampled frame, but this method would decrease the overall acceleration factor, leading to a lower temporal resolution. This is obviously problematic in areas of the body where motion occurs (i.e., the heart, abdomen) Thus, techniques that are more tolerant of mismatches between the aliased images and the calibration data in dynamic imaging have been sought, and it is to this end that both temporal SENSE (TSENSE) and temporal GRAPPA (TGRAPPA) have been developed. Owing to the nature of these dynamic reconstructions, temporally resolved data is required. Thus in MRA this technique can be best applied to dynamic contrast-enhanced MRA acquisitions [83–85]. TSENSE works by combining an interleaved acquisition scheme with the SENSE reconstruction. In an interleaved acquisition scheme, the acquired k-space lines are shifted by one line from frame to frame. Thus is one wishes to acquire a dataset accelerated by a factor of R, one could obtain a fully sampled dataset every R frames. Thus, the fundamental idea of TSENSE is to assemble at least R frames together and to calculate the coil sensitivity maps directly from this acquired data. The individual frames are then reconstructed using the SENSE method based on these maps. Since the coil sensitivity maps are acquired essentially simultaneously with the undersampled data, and since they have the full image resolution, TSENSE is clearly more robust in a dynamic imaging situation. It has further been shown that the addition of an UNFOLD-like [86] temporal filter on these SENSEreconstructed images will further reduce any remaining aliasing resulting from any inaccuracies in the coil sensitivity maps. Thus, TSENSE results in improved image quality over SENSE or UNFOLD alone. TGRAPPA functions in a similar way; a total of R frames of an R-times accelerated dataset can be combined to make a fully sampled dataset. This fully sampled dataset is used to calibrate the GRAPPA weights, which are then applied to each of the undersampled datasets. In this way, no additional calibration data must be acquired, and the GRAPPA weights more accurately reflect the underlying coil sensitivities. In practice, a sliding window approach over more than R frames can used to generate improved GRAPPA weights for each accelerated frame.

196

N. Seiberlich and M. Griswold

References 22. 1. Kelton J, Magin RM, Wright SM. An algorithm for rapid image acquisition using multiple receiver coils. Proc Intl Soc Magn Reson Med. 1989;8:1172. 2. Ra JB, Rim CY. Fast imaging using subencoding data sets from multiple detectors. Magn Reson Med. 1993;30:142–145. 3. Carlson JW, Minemura T. Imaging time reduction through multiple receiver coil data acquisition and image reconstruction. Magn Reson Med. 1993;29:681–687. 4. Sodickson DK, Manning WJ. Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med. 1997;38:591–603. 5. Jakob PM, Griswold MA, Edelman RR, Sodickson DK. AUTOSMASH: a self-calibrating technique for SMASH imaging. SiMultaneous Acquisition of Spatial Harmonics. MAGMA. 1998;7: 42–54. 6. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999;42: 952–962. 7. Griswold MA, Jakob PM, Nittka M, Goldfarb JW, Haase A. Partially parallel imaging with localized sensitivities (PILS). Magn Reson Med. 2000;44:602–609. 8. Griswold MA, Jakob PM, Heidemann RM, Nittka M, Jellus V, Wang J, Kiefer B, Haase A. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med. 2002;47: 1202–1210. 9. Pruessmann KP, Weiger M, Börnert P, Boesiger P. Advances in sensitivity encoding with arbitrary k-space trajectories. Magn Reson Med. 2001;46:638–651. 10. Griswold MA, Heidemann RM, Jakob PM. Direct parallel imaging reconstruction of radially sampled data using GRAPPA with relative shifts. Proc Intl Soc Mag Reson Med. 2003;11:2349. 11. Kellman P, Epstein FH, McVeigh ER. Adaptive sensitivity encoding incorporating temporal filtering (TSENSE). Magn Reson Med. 2001;45:846–852. 12. Breuer FA, Kellman P, Griswold MA, Jakob PM. Dynamic autocalibrated parallel imaging using temporal GRAPPA (TGRAPPA). Magn Reson Med. 2005;53:981–985. 13. Heidemann RM, Seiberlich N, Griswold MA, Wohlfarth K, Krueger G, Jakob PM Perspectives and limitations of parallel MR imaging at high field strengths. Neuroimaging Clin N Am. 2006; 16:311– 320, xi. 14. Blaimer M, Breuer F, Mueller M, Heidemann RM, Griswold MA, Jakob PM. SMASH, SENSE, PILS, GRAPPA: how to choose the optimal method. Top Magn Reson Imaging. 2004;15:223–236. 15. Griswold MA, Jakob PM, Chen Q, Goldfarb JW, Manning WJ, Edelman RR, Sodickson DK. Resolution enhancement in singleshot imaging using simultaneous acquisition of spatial harmonics (SMASH). Magn Reson Med. 1999;41:1236–1245. 16. Weiger M, Boesiger P, Hilfiker PR, Weishaupt D, Pruessmann KP., Sensitivity encoding as a means of enhancing the SNR efficiency in steady-state MRI. Magn Reson Med. 2005;53:177–185. 17. Kressler B, Spincemaille P, Nguyen TD, Cheng L, Xi Hai Z, Prince MR, Wang Y. Three-dimensional cine imaging using variable-density spiral trajectories and SSFP with application to coronary artery angiography. Magn Reson Med. 2007;58:535–543. 18. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased array. Magn Reson Med. 1990;16:192–225. 19. Walsh DO, Gmitro AF, Marcellin MW. Adaptive reconstruction of phased array MR imagery. Magn Reson Med. 2000;43:682–690. 20. Huber ME, Kozerke S, Pruessmann KP, Smink J, Boesiger P. Sensitivity-encoded coronary MRA at 3T. Magn Reson Med. 2004; 52:221–227. 21. Jahnke C, Paetsch I, Schnackenburg B, Bornstedt A, Gebker R, Fleck E, Nagel E. Coronary MR angiography with steady-state free

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

precession: individually adapted breath-hold technique versus freebreathing technique. Radiology. 2004;232:669–676. Huber ME, Kozerke S, Boesiger P. Improved artery delineation in dual-stack coronary magnetic resonance angiography using parallel imaging at 3 T. J Magn Reson Imaging. 2005;21:443–448. Niendorf T, Saranathan M, Lingamneni A, Pedrosa I, Spencer M, Cline H, Foo TK, Rofsky NM. Short breath-hold, volumetric coronary MR angiography employing steady-state free precession in conjunction with parallel imaging. Magn Reson Med. 2005;53: 885–894. Plein S, Radjenovic A, Ridgway JP, Barmby D, Greenwood JP, Ball SG, Sivananthan MU. Coronary artery disease: myocardial perfusion MR imaging with sensitivity encoding versus conventional angiography. Radiology. 2005;235:423–430. Niendorf T, Hardy CJ, Giaquinto RO, Gross P, Cline HE, Zhu Y, Kenwood G, Cohen S, Grant AK, Joshi S, Rofsky NM, Sodickson DK. Toward single breath-hold whole-heart coverage coronary MRA using highly accelerated parallel imaging with a 32-channel MR system. Magn Reson Med. 2006;56:167–176. Gharib AM, Herzka DA, Ustun AO, Desai MY, Locklin J, Pettigrew RI, Stuber M. Coronary MR angiography at 3T during diastole and systole. J Magn Reson Imaging. 2007;26:921–926. Summers PE, Kollias SS, Valavanis A. Resolution improvement in thick-slab magnetic resonance digital subtraction angiography using SENSE at 3T. J Magn Reson Imaging. 2004;20:662–673. Taschner CA, Gieseke J, Le Thuc V, Rachdi H, Reyns N, Gauvrit JY, Leclerc X. Intracranial arteriovenous malformation: timeresolved contrast-enhanced MR angiography with combination of parallel imaging, keyhole acquisition, and k-space sampling techniques at 1.5 T. Radiology. 2008;246:871–879. Petkova M, Gauvrit JY, Trystram D, Nataf F, Godon-Hardy S, Munier T, Oppenheim C, Meder JF. Three-dimensional dynamic time-resolved contrast-enhanced MRA using parallel imaging and a variable rate k-space sampling strategy in intracranial arteriovenous malformations. J Magn Reson Imaging. 2009;29:7–12. Parmar H, Ivancevic MK, Dudek N, Gandhi D, Mukherji SK. Dynamic MRA with four-dimensional time-resolved angiography using keyhole at 3 tesla in head and neck vascular lesions. J Neuroophthalmol. 2009;29:119–127. Willinek WA, Hadizadeh DR, von Falkenhausen M, Urbach H, Hoogeveen R, Schild HH, Gieseke J. 4D time-resolved MR angiography with keyhole (4D-TRAK): more than 60 times accelerated MRA using a combination of CENTRA, keyhole, and SENSE at 3.0T. J Magn Reson Imaging. 2008;27:1455–1460. Bicakci K, Soker G, Binokay F, Akgul E, Aksungur E, Sertdemir Y. Estimation of the ratio of renal artery stenosis with magnetic resonance angiography using parallel imaging technique in suspected renovascular hypertension. Nephron Clin Pract. 2006;104: c169-c175. Gufler H, Weimer W, Neu K, Wagner S, Rau WS. Contrast enhanced MR angiography with parallel imaging in the early period after renal transplantation. J Magn Reson Imaging. 2009;29:909–916. Muthupillai R, Douglas E, Huber S, Lambert B, Pereyra M, Wilson GJ, Flamm SD. Direct comparison of sensitivity encoding (SENSE) accelerated and conventional 3D contrast enhanced magnetic resonance angiography (CE-MRA) of renal arteries: effect of increasing spatial resolution. J Magn Reson Imaging. 2010;31:149–159. Maki JH, Wilson GJ, Eubank WB, Hoogeveen RM. Utilizing SENSE to achieve lower station sub-millimeter isotropic resolution and minimal venous enhancement in peripheral MR angiography. J Magn Reson Imaging. 2002;15:484–491. Wang MS, Haynor DR, Wilson GJ, Leiner T, Maki JH. Maximizing contrast-to-noise ratio in ultra-high resolution peripheral MR angiography using a blood pool agent and parallel imaging. J Magn Reson Imaging. 2007;26:580–588. Ruhl KM, Katoh M, Langer S, Mommertz G, Guenther RW, Niendorf T, Spuentrup E. Time-resolved 3D MR angiography of

14

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

Parallel Imaging in Angiography the foot at 3 T in patients with peripheral arterial disease. AJR Am J Roentgenol. 2008;190:W360-W364. de Vries M, Nijenhuis RJ, Hoogeveen RM, de Haan MW, van Engelshoven JM, Leiner T. Contrast-enhanced peripheral MR angiography using SENSE in multiple stations: feasibility study. J Magn Reson Imaging. 2005;21:37–45. Sumi T, Sumi M, Van Cauteren M, Kimura Y, Nakamura T. Parallel imaging technique for the external carotid artery and its branches: comparison of balanced turbo field echo, phase contrast, and timeof-flight sequences. J Magn Reson Imaging. 2007;25:1028–1034. Riederer SJ, Hu HH, Kruger DG, Haider CR, Campeau NG, Huston J 3rd. Intrinsic signal amplification in the application of 2D SENSE parallel imaging to 3D contrast-enhanced elliptical centric MRA and MRV. Magn Reson Med. 2007;58:855–864. Yu J, Schär M, Vonken EJ, Kelle S, Stuber M. Improved SNR efficiency in gradient echo coronary MRA with high temporal resolution using parallel imaging. Magn Reson Med. 2009;62:1211–1220. Haider CR, Hu HH, Campeau NG, Huston J 3rd, Riederer SJ. 3D high temporal and spatial resolution contrast-enhanced MR angiography of the whole brain. Magn Reson Med. 2008;60:749–760. Haider CR, Glockner JF, Stanson AW, Riederer SJ. Peripheral vasculature: high-temporal- and high-spatial-resolution threedimensional contrast-enhanced MR angiography. Radiology. 2009;253:831–843. Ozsarlak O, Van Goethem JW, Parizel PM. 3D time-of-flight MR angiography of the intracranial vessels: optimization of the technique with water excitation, parallel acquisition, eight-channel phased-array head coil and low-dose contrast administration. Eur Radiol. 2004;14:2067–2071. Gaa J, Weidauer S, Requardt M, Kiefer B, Lanfermann H, Zanella FE. Comparison of intracranial 3D-ToF-MRA with and without parallel acquisition techniques at 1.5T and 3.0T: preliminary results. Acta Radiol. 2004;45:327–332. Hope MD, Purcell DD, Hope TA, von Morze C, Vigneron DB, Alley MT, Dillon WP. Complete intracranial arterial and venous blood flow evaluation with 4D flow MR imaging. AJNR Am J Neuroradiol. 2009;30:362–366. Hope TA, Hope MD, Purcell DD, von Morze C, Vigneron DB, Alley MT, Dillon WP. Evaluation of intracranial stenoses and aneurysms with accelerated 4D flow. Magn Reson Imaging. 2010; 28:41–46. Meckel S, Mekle R, Taschner C, Haller S, Scheffler K, Radue EW, Wetzel SG. Time-resolved 3D contrast-enhanced MRA with GRAPPA on a 1.5-T system for imaging of craniocervical vascular disease: initial experience. Neuroradiology. 2006;48:291–299. Markl M, Uhl M, Wieben O, Ness T, Langer M, Hennig J, Bley TA. High resolution 3T MRI for the assessment of cervical and superficial cranial arteries in giant cell arteritis. J Magn Reson Imaging. 2006;24:423–427. von Morze C, Purcell DD, Banerjee S, Xu D, Mukherjee P, Kelley DA, Majumdar S, Vigneron DB. High-resolution intracranial MRA at 7T using autocalibrating parallel imaging: initial experience in vascular disease patients. Magn Reson Imaging. 2008;26: 1329–1333. Lettau M, Sartor K, Heiland S, Hähnel S. 3T high-spatial-resolution contrast-enhanced MR angiography of the intracranial venous system with parallel imaging. AJNR Am J Neuroradiol. 2009;30: 185–187. Schoenberg SO, Rieger J, Weber CH, Michaely HJ, Waggershauser T, Ittrich C, Dietrich O, Reiser MF. High-spatial-resolution MR angiography of renal arteries with integrated parallel acquisitions: comparison with digital subtraction angiography and US. Radiology. 2005;235:687–698. Michaely HJ, Herrmann KA, Kramer H, Dietrich O, Laub G, Reiser MF, Schoenberg SO. High-resolution renal MRA: comparison of image quality and vessel depiction with different parallel imaging acceleration factors. J Magn Reson Imaging. 2006;24:95–100.

197 54. Nael K, Saleh R, Lee M, McNamara T, Godinez SR, Laub G, Finn JP, Ruehm SG. High-spatial-resolution contrast-enhanced MR angiography of abdominal arteries with parallel acquisition at 3.0 T: initial experience in 32 patients. AJR Am J Roentgenol. 2006; 187:W77–W85. 55. Fenchel M, Nael K, Deshpande VS, Finn JP, Kramer U, Miller S, Ruehm S, Laub G. Renal magnetic resonance angiography at 3.0 Tesla using a 32-element phased-array coil system and parallel imaging in 2 directions. Invest Radiol. 2006;41:697–703. 56. Lum DP, Busse RF, Francois CJ, Brau AC, Beatty PJ, Huff J, Brittain JH, Reeder SB. Increased volume of coverage for abdominal contrast-enhanced MR angiography with two-dimensional autocalibrating parallel imaging: initial experience at 3.0 Tesla. J Magn Reson Imaging. 2009;30:1093–1100. 57. Tongdee R, Narra VR, McNeal G, Hildebolt CF, El-Merhi F, Foster G, Brown JJ. Hybrid peripheral 3D contrast-enhanced MR angiography of calf and foot vasculature. AJR Am J Roentgenol. 2006; 186:1746–1753. 58. Kramer H, Michaely HJ, Matschl V, Schmitt P, Reiser MF, Schoenberg SO. High-resolution magnetic resonance angiography of the lower extremities with a dedicated 36-element matrix coil at 3 Tesla. Invest Radiol. 2007;42:477–483. 59. Potthast S, Bongartz GM, Huegli R, Schulte AC, Schwarz JG, Aschwanden M, Bilecen D. Intraarterial contrast-enhanced MR aortography with and without parallel acquisition technique in patients with peripheral arterial occlusive disease. AJR Am J Roentgenol. 2007;188:823–829. 60. Zenge MO, Vogt FM, Brauck K, Jökel M, Barkhausen J, Kannengiesser S, Ladd ME, Quick HH. High-resolution continuously acquired peripheral MR angiography featuring partial parallel imaging GRAPPA. Magn Reson Med. 2006;56:859–865. 61. Quick HH, Vogt FM, Maderwald S, Herborn CU, Bosk S, Göhde S, Debatin JF, Ladd ME. High spatial resolution whole-body MR angiography featuring parallel imaging: initial experience. Rofo. 2004;176:163–169. 62. Nikolaou K, Kramer H, Grosse C, Clevert D, Dietrich O, Hartmann M, Chamberlin P, Assmann S, Reiser MF, Schoenberg SO. Highspatial-resolution multistation MR angiography with parallel imaging and blood pool contrast agent: initial experience. Radiology. 2006;241:861–872. 63. Nael K, Fenchel M, Krishnam M, Laub G, Finn JP, Ruehm SG. High-spatial-resolution whole-body MR angiography with highacceleration parallel acquisition and 32-channel 3.0-T unit: initial experience. Radiology. 2007;242:865–872. 64. Fenchel M, Doering J, Seeger A, Kramer U, Rittig K, Klumpp B, Claussen CD, Miller S. Ultrafast whole-body MR angiography with two-dimensional parallel imaging at 3.0 T: feasibility study. Radiology. 2009;250:254–263. 65. Nikolaou K, Schoenberg SO, Attenberger U, Scheidler J, Dietrich O, Kuehn B, Rosa F, Huber A, Leuchte H, Baumgartner R, Behr J, Reiser MF. Pulmonary arterial hypertension: diagnosis with fast perfusion MR imaging and high-spatial-resolution MR angiography – preliminary experience. Radiology. 2005;236:694–703. 66. Nael K, Fenchel M, Krishnam M, Finn JP, Laub G, Ruehm SG. 3.0 Tesla high spatial resolution contrast-enhanced magnetic resonance angiography (CE-MRA) of the pulmonary circulation: initial experience with a 32-channel phased array coil using a high relaxivity contrast agent. Invest Radiol. 2007;42:392–398. 67. Attenberger UI, Ingrisch M, Dietrich O, Herrmann K, Nikolaou K, Reiser MF, Schönberg SO, Fink C. Time-resolved 3D pulmonary perfusion MRI: comparison of different k-space acquisition strategies at 1.5 and 3 T. Invest Radiol. 2009;44:525–531. 68. Park J, McCarthy R, Li D. Feasibility and performance of breathhold 3D true-FISP coronary MRA using self-calibrating parallel acquisition. Magn Reson Med. 2004;52:7–13. 69. Jin H, Zeng MS, Ge MY, Yang S, Chen CZ, Shen JZ, Li RC. A study of in vitro and in vivo MR of free-breathing whole-heart

198

70.

71.

72.

73.

74.

75. 76. 77.

N. Seiberlich and M. Griswold 3D coronary angiography using parallel imaging. Int J Cardiovasc Imaging. 2009;25(Suppl 1):121–129. Oleaga L, Dalal SS, Weigele JB, Hurst RW, Lee J, Voorhees A, Melhem ER. The role of time-resolved 3D contrast-enhanced MR angiography in the assessment and grading of cerebral arteriovenous malformations. Eur J Radiol. 2010;73:e117-e121. Lim RP, Shapiro M, Wang EY, Law M, Babb JS, Rueff LE, Jacob JS, Kim S, Carson RH, Mulholland TP, Laub G, Hecht EM. 3D time-resolved MR angiography (MRA) of the carotid arteries with time-resolved imaging with stochastic trajectories: comparison with 3D contrast-enhanced Bolus-Chase MRA and 3D time-offlight MRA. AJNR Am J Neuroradiol. 2008;29:1847–1854. Lim RP, Jacob JS, Hecht EM, Kim DC, Huffman SD, Kim S, Babb JS, Laub G, Adelman MA, Lee VS. Time-resolved lower extremity MRA with temporal interpolation and stochastic spiral trajectories: preliminary clinical experience. J Magn Reson Imaging. 2010;31:663–672. Breuer FA, Blaimer M, Heidemann RM, Mueller MF, Griswold MA, Jakob PM. Controlled aliasing in parallel imaging results in higher acceleration (CAIPIRINHA) for multi-slice imaging. Magn Reson Med. 2005;53:684–691. Breuer FA, Blaimer M, Seiberlich N, Griswold MA, Jakob PM. A 3D GRAPPA algorithm for volumetric parallel imaging. Proc Intl Soc Mag Reson Med. 2006;14:286. Lauzon ML, Rutt BK. Effects of polar sampling in k-space. Magn Reson Med. 1996;36:940–949. Scheffler K, Hennig J. Reduced circular field-of-view imaging. Magn Reson Med. 1998;40:474–480. Kannengiesser SAR and Noll TG. Towards a practical generalized image reconstruction method for MRI. In: Proc Intl Soc Mag Reson Med. 2002;10:155.

78. Bi X, Park J, Larson AC, Zhang Q, Simonetti O, Li D. Contrastenhanced 4D radial coronary artery imaging at 3.0 T within a single breath-hold. Magn Reson Med. 2005;54:470–475. 79. Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magn Reson Med. 2007;58:1182–1195. 80. Seiberlich N, Ehses P, Duerk JL, Griswold MA. Through-Time Radial GRAPPA Calibration: Application to Real-Time Cardiac Imaging. In: Proc 3rd International Workshop on Parallel Imaging (2009), San Diego CA. 81. Jeong HJ, Eddleman CS, Shah S, Seiberlich N, Griswold MA, Batjer HH, Carr JC, Carroll TJ. Accelerating time-resolved MRA with multiecho acquisition. Magn Reson Med. 2010;63: 1520–1528. 82. Seiberlich N, Breuer FA, Blaimer M, Speier P, Griswold MA, Jakob PM. 3D Cylindrical GRAPPA. Proc Intl Soc Mag Reson Med. 2006;14:7. 83. Kellman P, Derbyshire JA, Agyeman KO, McVeigh ER, Arai AE. Extended coverage first-pass perfusion imaging using slice-interleaved TSENSE. Magn Reson Med. 2004;51:200–204. 84. Han M, Daniel BL, Hargreaves BA. Accelerated bilateral dynamic contrast-enhanced 3D spiral breast MRI using TSENSE. J Magn Reson Imaging. 2008;28:1425–1434. 85. Lai P, Huang F, Larson AC, Li D. Fast four-dimensional coronary MR angiography with k-t GRAPPA. J Magn Reson Imaging. 2008;27:659–665. 86. Madore B, Glover GH, Pelc NJ. Unaliasing by fourier-encoding the overlaps using the temporal dimension (UNFOLD), applied to cardiac imaging and fMRI. Magn Reson Med. 1999;42: 813–828.

Targeted Agents for Wall Imaging Emily A. Waters and Thomas J. Meade

Introduction MRI is an increasingly important diagnostic tool for cardiovascular pathology, with the ability to assess ventricular wall motion, myocardial perfusion and viability, coronary artery disease, and the function of the cardiac valves. However, MR imaging is primarily used for diagnosis of functional abnormalities, which do not occur until relatively late in disease processes (often after irreversible tissue damage has occurred). Molecular imaging allows observation of disease processes as they occur, before functional abnormalities become apparent [1]. Not only does it hold great promise for early and more accurate diagnoses but it also enables direct observation of therapeutic efficacy in individual patients, allowing optimization of the treatment regimen for each person. Furthermore, by allowing direct observation of physiological processes at the molecular level, molecular imaging enables longitudinal study of disease biochemistry. Current methods of diagnosing atherosclerosis focus on measuring lumenal stenosis. However, the limitations of this approach are evident in the coronary arteries. In this case, the lesions causing a stenosis that is considered clinically significant is frequently not the lesions responsible for major myocardial infarctions. Large stenoses that are clinically significant by angiography are frequently responsible for stable angina. Acute coronary syndromes are more often caused by the so-called “vulnerable plaque” that may begin as a 40–50% stenosis which then ruptures. Inflammatory and thrombogenic contents are spilled into the bloodstream, resulting in a large clot that rapidly blocks the coronary artery. The same is true in the carotid arteries, where strokes frequently result from the rupture of a plaque not considered clinically signifi-

E.A. Waters, PhD () • T.J. Meade, PhD Department of Chemistry, Northwestern University, Evanston, IL, USA e-mail: [email protected]

15

cant under the current standards. Consequently, there is a great deal of interest in differentiating vulnerable plaques from stable plaques, and it is in this context where molecular imaging techniques may play a major role. In postmortem studies, ruptured plaques have been characterized by a large lipid-rich necrotic core and a thin fibrous cap [2]. These features have been associated with extensive macrophage infiltration, reduced numbers of smooth muscle cells, and angiogenesis induced by hypoxia at the center of the plaque. The fragile and leaky vessels that form as a result of angiogenesis can result in intraplaque hemorrhage, further destabilizing plaques. Macrophage infiltration and hypoxia are both components of a more generalized inflammatory response. As part of an effort to identify vulnerable plaques, techniques are being developed to directly image the components of the plaque in the vessel wall. Dynamic contrast enhanced imaging has been used to assess the relative vascularity of atherosclerotic plaques. However, targeted molecular imaging of various inflammatory components may ultimately provide a more sensitive and specific diagnosis [3]. Targets of interest include macrophages, lipid-rich areas, fibrin (for detection of small thrombi), integrins and adhesion molecules (to measure inflammation), and matrix metalloproteinases (to measure degradation of extracellular matrix) (Fig. 15.1). Invasive angiography via cardiac catheterization is the most commonly used imaging technique for the diagnosis of atherosclerotic plaques. However, this technique only detects plaques which have grown large enough to impinge on the artery lumen. Because of the extensive positive remodeling which occurs during the process of atherosclerosis, many of the plaques that are most likely to rupture do not appear clinically significant on a coronary angiogram. While positron emission tomography (PET) has been used for molecular imaging due to its extremely high sensitivity, its limited resolution may interfere with coronary vessel wall imaging. Intravascular ultrasound (IVUS) is capable of distinguishing between plaques that are more fibrous and heavily calcified and plaques that have higher lipid content. However, this

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_15, © Springer Science+Business Media, LLC 2012

199

200

E.A. Waters and T.J. Meade

Fig. 15.1 A schematic showing the progression of an atherosclerotic lesion, developing from a normal blood vessel (far left) to a vessel with an atherosclerotic plaque and superimposed thrombus (far right), highlighting potential targets for molecular imaging at each stage. AHA

American Heart Association, ICAM1 intercellular adhesion molecule 1, LDL low-density lipoprotein, MMP matrix metalloproteinase, VCAM1 vascular cell-adhesion molecule 1. (Figure reproduced with permission from Sanz J et al. [3])

technique is invasive and therefore unsuitable for longitudinal or repeat studies. Computed tomography (CT) has been used for coronary artery imaging, but it does not have sufficient soft tissue contrast to differentiate components of the vessel wall. Additional disadvantages of CT include the need to administer drugs to lower patients’ heart rate, and the high radiation dose which renders the technique unsuitable for screening and repeated follow-up.

a large concentration of gadolinium and targeting ligands (including HDL mimicking nanoparticles, micelles, liposomes, and nanoemulsions) (Table 15.1 [4–12]). Iron oxide nanoparticles are superparamagnetic and are most often used as T2 shortening agents, producing a signal void on an image. Since the particles produce signal voids with a diameter far greater than their actual diameter, iron oxide nanoparticles can be detected at extremely low concentration, and imaging of single cells has been demonstrated. The differently sized particles of iron oxide (ranging from ultrasmall iron oxide nanoparticles (USPIOs)), which have a diameter of only a few nanometers, to micron-sized particles of iron oxide (MPIOs) have different pharmacokinetics and biodistribution, and have been used for a range of applications [13]. USPIOs have a long circulating blood half-life and are taken up by macrophages in atherosclerotic plaques both in rabbits and in humans even without further surface modification [13].

Classes of MRI Contrast Agents The three major classes of contrast agent platforms that have been developed for molecular imaging of atherosclerosis are: iron oxide nanoparticles (ranging in size from a few nanometers to micrometer size), small molecule gadolinium chelates conjugated to a targeting ligand, and nanoparticles that carry

15

Targeted Agents for Wall Imaging

201

Table 15.1 Types of nanoparticles that have been used for MRI molecular imaging of atherosclerosis Class of agent Iron oxide nanoparticle (superparamagnetic)

Lipid-based nanostructure (paramagnetic)

Nanoparticle type USPIO CLIO SPIO MPIO HDL mimic Micelle

Diameter ~20 nm ~40 nm ~100 nm ~5 mm ~15 nm ~20 nm

Liposome PFC emulsion

80–500 nm ~200 nm

Targeting moiety None Peptide None Antibody None Peptide, antibody, hydrophobic tyrosine residues

Target Macrophage [5] VCAM (activated endothelium) [4] Macrophage Platelets [6] Areas of HDL trafficking [7] Oxidation [8], macrophages [9] lipid-rich areas [10]

Peptide, antibody

avb3 integrin [11] fibrin [12]

Monocrystalline iron oxide nanoparticles with surface cross-linking to further increase circulation time and allow for efficient surface functionalization are referred to as CLIO (cross-linked iron oxides). These particles have been successfully targeted to Vascular Cell Adhesion Molecule for imaging of plaque inflammation, and have additionally been modified with a near infrared dye for optical colocalization [4]. Intermediate sized particles (SPIOs, or superparamagnetic iron oxide nanoparticles) are rapidly cleared by the reticuloendothelial system (RES), so they have been less popular platforms for molecular imaging, but have still been shown to accumulate in macrophages in atherosclerotic plaques [13]. MPIOs have an advantage for imaging of intravascular targets because the iron payload they carry is orders of magnitude larger than smaller particles (e.g., USPIOs), resulting in a large contrast effect for single particles. Each 4.5 mm diameter particle creates a signal void on the image of approximately 100 mm [14]. Small-molecule contrast agents consisting of a gadolinium chelate conjugated to a targeting ligand represent an improvement in specificity over the existing gadolinium based agents, which remain extracellular, although they do diffuse out of the vasculature into interstitial spaces. Sensitivity is a major challenge with these agents, however, since there is typically only one gadolinium chelate per targeting molecule. As a result, these agents are generally suitable for imaging epitopes found in high concentrations, such as fibrin [15], but less so for very sparse epitopes such as integrins. To image sparse epitopes, it is necessary to amplify contrast agent signal. This can be done by creating a larger structure such as a nanoparticle, which both increases the relaxivity of individual gadolinium chelates by increasing their rotational correlation time, and increases the local concentration of gadolinium by delivering many gadolinium atoms per targeting ligand [16]. A widely used strategy for this has been the development of lipid based particles. Micelles are comprised of aggregates of amphiphilic molecules (often lipid based) which assemble naturally in aqueous solution [9]. Liposomes are comprised of a water-filled phospholipid

bilayer, and are similar to cell membranes [16]. Perfluorocarbon emulsions are comprised of a droplet of liquid perfluorocarbon surrounded by a phospholipid monolayer [17]. HDL mimicking particles are similarly comprised of a hydrophobic apolipoprotein core surrounded by a phospholipid monolayer [7]. The advantage of the outer lipid layer in all these systems is that amphipathic targeting ligands, gadolinium chelates, optical dyes, or other moieties of interest can be incorporated.

Molecular Imaging of Macrophages Macrophages and macrophage derived foam cells are associated with the growth and destabilization of atherosclerotic plaques. Untargeted dextran coated USPIOs were originally used as contrast agents for lymph node imaging [13]. As a result of their small size, these particles are not immediately taken up by the liver and spleen, and hence their circulation half-life is prolonged. They can escape the circulatory system via capillary pores and interendothelial junctions, at which point they are phagocytosed by cells of the Mononuclear Phagocytic System [13]. Ruehm et al. [5] showed that these particles could be observed in macrophages located in aortic atherosclerotic plaques of Watanabe Heritable Hyperlipidemic rabbits approximately 5 days after administration. Promising results were obtained in a study of human subjects scheduled to go carotid endarterectomy for symptomatic carotid atherosclerosis, although imaging artifacts complicated interpretation of some of the data [2]. The primary disadvantage of iron oxide nanoparticles is that they create negative contrast in the form of signal voids that cannot be easily distinguished from other sources of susceptibility artifacts. These artifacts, such as air or calcium deposits, are frequently present in regions of atherosclerosis. In addition, long delays are required between agent administration and imaging to allow sufficient time for macrophages to take up the contrast agent [13]. Macrophage targeted paramagnetic micelles create a signal enhancement (positive contrast) in a shorter period of time.

202

Mulder et al. developed a system of paramagnetic micelles coated with a hydrophilic PEG layer that reduces the interaction of micelles with the immune system and the reticuloendothelial system, increasing their blood halflife [9]. The 15 nm diameter is sufficiently small, and the micelles are sufficiently flexible, that they diffuse into tissue with increased permeability (such as plaques). This is in contrast to larger nanoparticles that remain intravascular. To specifically target the micelles to macrophages, the investigators covalently conjugated an antibody to the macrophage scavenger receptor (MSR) via a maleimide linkage. The micelles were intravenously administered at a dose of 0.075 mmol Gd/kg to atherosclerotic 13 month old Apolipoprotein E (ApoE) −/− mice that had been maintained on a Western diet for 20 weeks. At 24 h after injection, the aortas of mice that had been treated with MSR-targeted micelles had significantly higher contrast-to-noise ratios between aorta and spinal muscle than the aortas of mice treated with untargeted micelles [9].

Molecular Imaging of Lipid-Rich Plaque Regions and Lipoproteins Since a large lipid core is a prerequisite for plaque rupture [2], one strategy for targeted plaque imaging is been to image lipid-rich regions directly. For this purpose, Beilvert et al. [10] developed a contrast agent based on the human apolipoprotein A-I mimetic amphiphatic peptide D-4F, that has four hydrophobic phenylalanine residues on its lipophilic side. The investigators attached hydrophobic tyrosine groups to pegylated paramagnetic micelles as a simple lipid-targeting method. One year old ApoE −/− mice that had been maintained on a Western diet for at least 24 weeks were injected intravenously with immunomicelles at a dose of 60 mmol Gd/kg [10]. Significant enhancement of MRI signal in the aortic wall was observed at 6 and 24 h postinjection; the signal enhancement had subsided by 48 h postinjection. Significant enhancement was not observed at any time point with control micelles injected into ApoE −/− mice or tyrosine micelles injected into wild-type mice. Fluorescence microscopy showed that tyrosine micelles did not colocalize with macrophages, foam cells, or smooth muscle cells, suggesting that the micelles remained in the extracellular space in the atherosclerotic plaque. Micelles did colocalize well with apolipoprotein B and somewhat with the proteoglycan decorin (which is associated with the retention of atherogenic lipoproteins in the vessel wall). The authors hypothesized that the lipophilic properties of the micelles combined with long circulation times and increased permeability of the atherosclerotic plaque lead to accumulation of micelles in the lipid core of plaques [10]. The strategy of targeting lipid-rich areas by introducing hydrophobic amino acid residues is a much simpler approach than other targeted contrast agents that have used antibodies,

E.A. Waters and T.J. Meade

peptides, peptidomimetics, and small molecules as targeting moieties [10]. Another amphiphilic micellar paramagnetic contrast agent (Gadofluorine M) has been shown to accumulate in the extracellular matrix of plaques without interacting with lipid-rich areas [18]. These agents could be used in conjunction with one another to show both the size of the lipid core and the thickness of the fibrous cap, both important features in plaque stability and likelihood of rupture [2]. Lipoprotein mimicking nanoparticles have the potential to incorporate targeting ligands to specific plaque components, carry many Gd(III) chelates per particle, and increase the relaxivity of individual Gd(III) chelates. HDL particles consist of a hydrophobic core surrounded by phospholipids, cholesterol, and apolipoproteins (mostly apoA-I) [19]. They are promising as an MR imaging agent platform because they are large enough to attach multiple Gd(III) chelates, but small enough to be readily trafficked into plaques. The surface monolayer is easily modified, and the proteins on the surface are endogenous and not immunogenic. Unlike LDL particles, HDL particles are not atherogenic [19]. Frias et al. [7] modified established protocols for HDL reconstitution to develop discoidal HDL-mimicking nanoparticle contrast agents with lipophilic Gd(III) chelates incorporated into the surface. Fluorescently labeled lipids were also incorporated, and the particles had a hydrodynamic size range of 8–13 nm. Particles were injected into wild-type and 13-month-old ApoE mice at a dose of 200 mL of 2 mM Gd(III) solution. Uptake patterns corresponded to the plaque composition, with earlier uptake (peak at 24 h) in more macrophage-laden plaques, and slower uptake (peak at 72 h) in advanced plaques with fewer macrophages. Confocal microscopy of matched histological samples showed that nanoparticles were located inside intimal foam cells. Foam cells are macrophages that have endocytosed large quantities of lipoprotein and comprise a major component of the plaque lipidrich core. Oxidized low-density lipoprotein (OxLDL) promotes the progression of atherosclerosis in animal models and has been implicated in the destabilization of vulnerable plaques in humans [19]. It can promote macrophage apoptosis and hinder phagocytic clearance of apoptotic cells within lesions. Conversely, OxLDL is greatly diminished in plaques in the early stages of regression and stabilization. Because OxLDL is immunogenic, antibodies can be isolated from animals and humans with atherosclerosis [8]. Tracking OxLDL with MRI may help not only to identify vulnerable plaques but also to assess plaque response to therapies. Briley-Saebo et al. [8 ] attached antibodies to oxidation-specific epitopes to paramagnetic immunomicelles. An S-acetythioglycolic acid N-hydroxysuccinimide ester (SATA) linker was attached to the antibodies, which were then conjugated to maleimide groups on the surface of the micelles. Three targeted micelle formulations were prepared, bearing antibodies to MDA2, IK17, and E06,

15

Targeted Agents for Wall Imaging

203

respectively; control formulations were untargeted or conjugated to IgG antibodies. Average micelle diameters ranged from 14 to 22 nm. The micelles were administered at a dose of 0.075 mmol Gd/ kg to cholesterol-fed 1-year-old ApoE −/− mice. In wild-type mice, the half-life of all micelles was ~1.5 h. By contrast, in ApoE −/− mice, the oxidation-targeted micelles had much longer half-lives of 14–19 h, compared to ~1.5 h for the IgGtargeted or unlabeled micelles. The highest MRI contrast enhancement between the arterial wall and adjacent muscle occurred at 72 h after administration of targeted micelles. MDA2, IK17, and E06 micelles were associated with 125, 137, and 231% enhancement, respectively; no enhancement was seen for control micelles in ApoE −/− mice or for targeted micelles in wild-type mice. A competitive inhibition study performed for the MDA2 micelles by pretreating mice with 6 mg free MDA2 resulted in a sixfold reduction in signal enhancement at 72 h. Localization of targeted micelles within the aortic wall was confirmed by confocal microscopy; control micelles were not seen in the aortic wall [8].

Molecular Imaging of Plaque-Associated Thrombus Thromboembolic events frequently occur as a consequence of plaque rupture, and may result in heart attack, stroke, or unstable angina [2, 19]. However, due to their complex composition, thrombi can be difficult to distinguish from plaque on an MR image. Fibrin is a promising target for MR imaging, because it is highly abundant in arterial thrombi such as those associated with plaque rupture, unlike sparse molecular epitopes. An example fibrin-targeted contrast agent is EP-1873, developed by Botnar et al. [15] EP-1873 is a small peptide with two attached Gd-DTPA chelates, which binds fibrin, but does not bind to circulating fibrinogen in the blood. EP-1873 was studied in a rabbit model of plaque rupture, where an aortic balloon injury was followed by 8 weeks of 1% cholesterol diet to induce atherogenesis. This was followed by induction of plaque rupture by intravenous injection of Russell’s Viper Venom and histamine. MR images of rabbits with both acute and subacute thrombus were acquired on a clinical 1.5T imaging system [15]. Enhancement was detected throughout the thrombus in both acute and subacute cases, suggesting permeation of the contrast agent throughout the whole thrombus (Fig. 15.2). In acute thrombi, EP-1873 was administered 30 min after pharmacological thrombus induction. Each thrombus was detectable by MR imaging 30 min after administration of contrast, with progressive increase in the area and intensity of signal over the next 6 h (consistent with ongoing thrombus formation). In subacute thrombi, EP-1873 was administered 24 h after pharmacological thrombus induction. Again,

Fig. 15.2 (a) Reformatted view of a coronal 2D dataset shows subrenal aorta ~20 h after EP-1873 administration. Three well-delineated mural thrombi (arrows) can be observed, with good contrast between thrombus (numbered), arterial blood (dotted arrow), and vessel wall (dashed arrow). The in-plane view of the aorta allows simultaneous display of all thrombi, showing head, tail, length, and relative location. (b–d) Corresponding cross-sectional views show good agreement with histopathology (e–g). (Reprinted from Botnar R et al. [15])

thrombus was detectable within 30 min, but in this case, the thrombus-blood CNR peaked 20 h after injection. In both cases, enhancement was sufficient to create maximum intensity projections (MIPs) showing the distribution of thrombi throughout the aorta [15]. While fibrin is a key component of large arterial clots, microdeposits of fibrin are found on the surface of unstable plaques. However, further signal amplification is needed to detect these. Morawski et al. [12] developed a lipid-perflurocarbon nanoemulsion with gadolinium chelates anchored in the lipid membrane. The resulting nanoparticles have an average diameter of 200 nm; their larger size keeps them primarily intravascular, which is advantageous for vascular targets. The formulation used in this study was targeted via a fibrin-specific monoclonal antibody which was anchored in the lipid membrane; other formulations have included targeting ligands to avb3 and tissue factor targeting. The investigators incubated fibrin clots formed in vitro from combining human plasma with calcium citrate and thrombin with the contrast agent, as well as a postsurgical human carotid endarterectomy specimen (Fig. 15.3). Samples were imaged on both the proton and fluorine frequencies. The concentration of nanoparticles present in the sample was

204

Fig. 15.3 (a) Optical image of a 5-mm cross-section of a human carotid endarterectomy sample. This section showed moderate luminal narrowing as well as several atherosclerotic lesions. (b) A 19F projection image acquired at 4.7T through the entire carotid artery sample

quantified by comparing the 19F spectral signal with the signal from a reference standard. The investigators were able to quantitatively measure the proportions present of two different nanoparticles formulated with different perfluorocarbons (and hence having discrete spectral signals), with the future goal of being able to simultaneously quantify particles targeted to two different epitopes [12]. Because these nanoparticles deliver tens of thousands of gadolinium chelates per bound targeting moiety, they can be imaged at dramatically lower concentrations than agents with only a few chelates per targeting moiety. The nanoparticles can be imaged either by traditional proton MRI or by 19F “hot spot” imaging, which takes advantage of the lack of naturally occurring 19F in the body to create an image of the nanoparticles with no background signal. This can then be coregistered with an anatomical proton image for increased specificity [12]. Activated platelets are another promising target for the early detection of small thrombi [3]. Von zur Muhlen et al. [6] conjugated a single-chain antibody that selectively binds to ligandinduced binding sites (LIBS) of the activated platelet integrin GPIIb/IIIa onto micron-sized particles of iron oxide (MPIO). They developed a mouse model of stable semiocclusive carotid thrombosis modified from a model of fully occlusive carotid thrombosis induced with ferric chloride in wild-type mice. Using the LIBS-MPIO, the investigators were able to image wall-adherent carotid thrombi in vivo in mice, and were further able to dynamically image the dissolution of the thrombus after administration of the thrombolytic drug urokinase. The investigators performed ex vivo imaging of LIBSMPIO bound to human carotid endarterectomy specimens. MPIO-induced signal voids were observed on the surface of endarterectomy specimens surgically removed from patients with symptomatic carotid artery atherosclerosis and incubated with LIBS-MPIO (Fig. 15.4). Immunohistochemistry confirmed binding of the LIBS-MPIO to activated platelets on the surface of the atherosclerotic plaques. Binding did not occur in specimens incubated with control MPIO [6].

E.A. Waters and T.J. Meade

shows high signal along the lumen due to nanoparticles bound to fibrin. (c) Concentration map of bound nanoparticles in the carotid sample. (Reprinted by permission from Morawski AM et al. [12])

Fig. 15.4 MRI and histology of symptomatic human carotid plaques. (a) Transversal MRI sections show the eversion specimen before and after incubation with LIBS MPIOs. Black arrows depict areas of contrast agent binding within the vessel lumen on the surface of the plaque. (b) No luminal binding can be observed in a plaque incubated with control MPIOs before and after contrast agent incubation. (c) Immunohistochemistry for platelets demonstrates the area depicted with black arrows in A. Platelets appear red; MPIOs appear as yellow round structures (typical appearance of MPIOs in paraffin-processed tissue in contrast to their appearance in frozen-section histology), confirming binding of LIBS MPIOs to areas of platelet adhesion/aggregation. (Reprinted from Von zur Muhlen C et al. [6])

15

Targeted Agents for Wall Imaging

Molecular Imaging of Cytokines, Integrins, and Adhesion Molecules A number of inflammatory processes are associated with atherosclerosis and may contribute to plaque destabilization [19]. Cellular adhesion molecules such as VCAM-1 are associated with the recruitment of leukocytes from the vasculature into plaques, where they become activated and contribute to the inflammatory response [3]. Matrix metalloproteinases (MMPs) are secreted by activated macrophages, degrading extracellular matrix and possibly helping to erode the plaque’s fibrous cap [20]. Integrins, such as the avb3 integrin, are expressed on the endothelium of angiogenic vessels, which are associated with destabilizing intraplaque hemorrhage and with further recruitment of inflammatory cells to the plaque. All of these markers of bioactivity are desirable targets for molecular imaging. However, they are present in nanomolar concentrations. As a result, significant signal amplification is required to detect these epitopes by MR imaging.

Matrix Metalloproteinases MMPs degrade the extracellular matrix in plaque areas subject to high mechanical stresses, eroding and thinning the fibrous cap, and promoting plaque rupture [20]. Lancelot et al. [21] developed the agent P947 by coupling an MMP inhibitor to the gadolinium chelate Gd-DOTA. The agent was found to bind to the soluble MMPs (−1, −2, −3, −8, −9, and −13) with micromolar affinity and to the membranebound MMP-14 with 100 times lower affinity. P947 was injected into ApoE −/− mice intravenously via the tail vein at a dose of 100 mmol/kg. The clearance profile of P947 was similar to that of Gd-DOTA, but with higher uptake in the artery walls. Significant increases in CNR between the aortic wall and skeletal muscle were noted beginning at 1 h and lasting up to 22 h after injection. Plaques were clearly delineated from surrounding tissue and the enhancement pattern showed higher signal intensities at the fibrous cap and shoulder regions of plaques. This result is consistent with the known distribution of MMPs within plaques. Little enhancement was observed in lipid cores. Additional testing was performed in excised human carotid endarterectomy specimens that were incubated with P947, washed, and imaged [21]. P947 accumulated in plaques at micromolar concentrations and enabled investigators to distinguish between MMP-rich plaques and MMP-poor plaques. However, the investigators concluded that it would be necessary to target a broad spectrum of MMPs to reach a level of agent accumulation that would enable clinically relevant MR imaging. A contrast agent with greater capacity for signal amplification would be required to target specific MMPs.

205

P-Selectin and Vascular Cell Adhesion Molecule P-selectin is expressed on activated endothelial cells at early stages of inflammation, and it is implicated in the recruitment of monocytes and other inflammatory cells to atherosclerotic plaque [19, 20]. Because it is expressed on activated platelets, it could be used as an early indicator of thrombus. The macromolecular blood pool contrast agent, P717 (CMD-A2-Gd-DOTA, research substance, Guerbet, France) is comprised of a 20-kDa dextran with multiple Gd-DOTA groups attached to the side chains. Alsaid et al. modified P717 with additional carboxylate and sulfate groups to mimic the natural binding ligand of p-selectin [22]. The resulting agent (F-P717) was shown to bind preferentially to activated human platelets, but not to resting platelets or blood cells including erythrocytes, leukocytes, or neutrophils [23]. To test the agent’s ability to target early-stage inflamed atherosclerotic plaques, in vivo studies were performed with rhodamine isothiocyanate-labeled F-P717 in Apolipoprotein E −/− mice [22]. MR images were acquired on a 2T horizontal bore MR imaging system (Oxford Instruments). Angiographic studies showed that FP-717 was renally cleared from the blood pool more rapidly than P-717. Black-blood images of the abdominal aorta showed a rapid and sustained enhancement of the aortic wall, beginning 10 min after injection of F-P717 and lasting for up to 6 h. Immunofluorescence showed colocalization of the rhodamine-labeled F-P717 with P-selectin, which was imaged with a fluorolabeled antibody. No MR signal enhancement or P-selectin expression was observed in the aortas of wild-type mice. Vascular Cell Adhesion Molecule 1 (VCAM-1) is a cell adhesion molecule implicated in a variety of inflammatory processes. It is upregulated on vascular endothelial cells, and contributes to monocyte and leukocyte recruitment to atheromas. Since VCAM-1 is upregulated prior to clinical manifestation of disease it is a promising target for molecular imaging and targeted therapies [19]. VCAM-1 has the additional advantage of being an intravascular target. Therefore, extravasation of contrast agent is not required for imaging. Kelly et al. [4] used phage display methods to screen for peptide sequences to target VCAM-expressing endothelial cells. They determined that the sequence VHSPNKK (which has homology with a known ligand to VCAM-1) could bind VCAM-1 and block leukocyte–endothelial cell interactions. They conjugated these peptides to cross-linked particles of iron oxide (CLIO). Control particles were formulated with a scrambled peptide, and fluorophores were conjugated to both targeted and control particles for optical corroboration of results. The in vivo binding of these particles was tested in the ears of wild-type mice injected with TNF-alpha to induce

206

E.A. Waters and T.J. Meade

Fig. 15.5 MRI of atherosclerotic lesions with VNP. (a) Cholesterolfed apoE −/− mice were imaged via MRI using Gd-PGC to delineate vascular lumen and structural aortic abnormalities such as narrowing (arrows). Axial images (at the level of the green line) were then obtained before (b) and 24 h after (c) administration of the magnetic VNP. Note the decrease in signal intensity of the eccentrically thickened aortic wall between the two arrows. (d) Ex vivo MRI confirms extensive low signal changes in the aortic wall induced by VNP accumulation (arrows),

further corroborated by macroscopic epifluorescence imaging of Cy5.5 in VNP (e). (f–h) Histological validation of MR findings. (f) Hematoxylin/eosin section of the aorta image in (b, c) confirms eccentric wall thickening (x2). (g, h) Comparative immunofluorescence of VCAM-1 expression (g: green) and VNP accumulation (h: red). Nuclei in (g, h) are counterstained with DAPI (blue). Note that there is extensive colocalization of VCAM-1 expression and VNP. Bars = 10 mm. (Reprinted from Kelly K et al. [4])

inflammation, after which the authors transitioned to an atherosclerosis model. The mice used in this study were 1-year-old ApoE −/− mice maintained for 3 months on a high-cholesterol diet, corresponding to a relatively early stage of atherosclerosis. The mice were injected intravenously with targeted or control nanoparticles and imaged 24 h after injection. In mice injected with targeted nanoparticles, signal reduction consistent with T2 shortening from iron oxide nanoparticles was observed, especially around the aortic root. The presence of nanoparticles was confirmed by high resolution ex vivo MR imaging of the excised aortic root and by fluorescent imaging of the aorta. The nanoparticle distribution colocalized well with VCAM-1 expression (Fig. 15.5). Leukocyte recruitment is an important feature of inflammation in early atherosclerosis [19, 20]. The dynamics of leukocyte–endothelial cell interaction are complex and involve multiple receptor-ligand interactions. Initial rolling along the endothelium is mediated by selectins, whereas adhesion and infiltration is mediated by cell adhesion molecules such as

VCAM and ICAM-1. A dual targeted, micron-sized particle could mimic leukocyte behavior and may be more likely to bind in a vessel where it is exposed to high shear stresses. McAteer et al. prepared micron-sized particles of iron oxide with targeting ligands to both P-selectin and VCAM [14]. ApoE −/− mice were maintained on a Western diet (21% milk fat, 0.15% cholesterol) for 25 weeks, beginning at age 3 weeks. Mice were administered dual-targeted MPIO, MPIO singly targeted to VCAM or P-selectin, or control MPIO targeted to IgG via the left ventricle of the heart. Mice were sacrificed 5 min after injection and the aortas excised and imaged with MRI ex vivo. The number of VCAM-MPIO or P-selectin-MPIO bound per aortic root section was similar; dual targeted MPIO showed significantly higher binding efficiency (5.5-fold and 6.9-fold increase, respectively, over IgG-MPIO) (Fig. 15.6a–b). A second group of mice were intravenously administered either dual-targeted or IgG targeted MPIO, sacrificed 30 min after injection, and the aortas excised and imaged with MRI ex vivo. In these mice, the number of dual-targeted MPIO

15

Targeted Agents for Wall Imaging

207

Fig. 15.6 (a) Aortic roots after left ventricular injection of targeted MPIO. VCAM-1-MPIO and P-selectin-MPIO showed similar binding to plaque endothelium. Dual ligand MPIO recognizing VCAM-1 and P-selectin showed enhanced binding. (b) Dense dual-targeted MPIO binding to endothelium overlying atherosclerotic plaque. Scale bar 20 mm (***P < 0.001, **P < 0.01, *P < 0.05). (c) Ex vivo MRI of aortic

roots 30 min after IV injection of MPIO. Retention of dual-targeted MPIO was 3.5-fold greater than IgG-1 MPIO (P < 0.01). (d) MPIO appeared as distinct circular low-signal areas adherent to endothelium overlying atherosclerotic plaque. Scale bar 500 mm. (e) 3D reconstruction of segmented images. (Reprinted from McAteer M et al. [14])

was 3.5-fold higher than IgG-MPIO. MPIO binding was localized to the endothelium of atherosclerotic plaque throughout the aorta; no binding was observed in atherosclerosis free regions (Fig. 15.6c–e).

study of rabbits pretreated with targeted nonparamagnetic (i.e., MR-invisible) nanoparticles. The presence of angiogenesis in the aortic wall was correlated to the MR images with immunohistochemistry. No MR enhancement was noted in rabbits fed a control diet, corresponding to the lack of angiogenesis noted by immunohistochemistry. This approach to detecting atherosclerosis was significantly more sensitive to early stages of disease than delayed contrast enhancement MRI (DCE-MRI); no difference was observed in aortic enhancement between rabbits fed a cholesterol-supplemented diet and rabbits fed a control diet. There is significant interest in using imaging agents as drug delivery vehicles. The ability to directly image drug delivery would be beneficial not only to observe the initial drug distribution, but also to serially monitor therapeutic effect over time. Finally, targeted therapies create the potential to treat diseases using very small concentrations of otherwise highly toxic drugs. Combination therapies using small a few doses of these targeted highly toxic drugs could greatly improve the utility of traditional therapies such as statins. Winter et al. [25] dissolved the lipophilic antiangiogenic drug fumagillin in the lipid membrane of a paramagnetic lipid-perfluorocarbon nanoemulsion. Fumagillin is a highly potent antiangiogenic drug derived from the fungus Aspergillus fumigatus, which was tested as an anticancer drug, but was deemed too toxic for systemic administration in humans. However, delivery via a targeted nanoparticle system allows administration of miniscule, and consequently

Integrins and Angiogenesis Angiogenesis is stimulated by hypoxia and is implicated in a variety of inflammatory processes, including cancer, diabetic retinopathy, and atherosclerosis [24]. In the case of atherosclerosis, the resulting “leaky” vasculature can result in microhemorrhages which destabilize plaque and increase the risk of rupture. Angiogenesis can serve as a vehicle for the recruitment and delivery of additional inflammatory cells and cytokines [20]. Winter et al. [11] targeted a paramagnetic lipid-perfluorocarbon nanoemulsion bearing Gd chelates anchored in the lipid membrane to the avb3 integrin via an RGD peptide motif. The investigators maintained New Zealand White rabbits on a 1% cholesterol diet for 80 days to induce atherosclerotic plaques in the abdominal aorta. The rabbits were intravenously administered either avb3 targeted or control (untargeted) nanoparticles and imaged 2 h later on a 1.5T clinical MR scanner. Significantly greater enhancement of the aortic wall was noted in rabbits treated with the angiogenesis-targeted nanoparticles; this enhancement was inhibited in a competition

208

E.A. Waters and T.J. Meade

Fig. 15.7 (a) Segmentation of the aortic wall and color-coded signal enhancement before and after targeted fumagillin treatment (Top) Black blood image of the thoracic aorta (arrow) and segmentation of the vessel wall (outlined in yellow) is shown for the week 0 image. The color-coded overlay of signal enhancement (%) shows patchy areas of high angiogenesis. On the week 1 image, the signal enhancement has clearly decreased due to the antiangiogenic effect of targeted fumagillin treatment. (Bottom) The level of signal enhancement gradually increases at weeks 2 and 3 after fumagillin treatment, until week 4, when the level of enhancement is practically identical to the week 0 image.

(b–e) avb3-targeted nanoparticles bind to the advential vasculature. (b) Hematoxylin and eosin staining of aorta shows small plaque, media, and adventia (×10). (b) High-power fluorescent image demonstrates colocalization of rhodamine-labeled avb3-targeted nanoparticles (c, ×60) with FITC-labeled lectin, a vascular endothelial marker, (d, ×60), indicating that the nanoparticles are vascularly constrained to the neovessels of the adventitia. No rhodamine-labeled avb3-targeted nanoparticles were detected in the lumen plaque (not shown). (Reprinted with permission from Winter P et al. [25])

far less toxic, doses than would be required for systemic administration. New Zealand White rabbits were maintained on a 0.25% cholesterol diet, after which they were treated with 0, 1, or 2 doses of fumagillin bearing targeted nanoparticles, and atorvastatin was, or was not, added to their diets. Rabbits treated with statins alone did not experience a major reduction in angiogenesis. Rabbits treated with one or two doses of fumagillin, but not statins, experienced transient large decreases in angiogenesis which returned to baseline within 4 weeks (Fig. 15.7). Rabbits treated with two doses of fumagillin nanoparticles and statins experienced a sustained reduction in angiogenesis which was substantially greater than the additive effects of statins and fumagillin bearing targeted nanoparticles. These results demonstrate (1) imaging of biomarker response to therapy can help to measure the efficacy of the therapy and (2) novel targeted theranostic approaches may greatly boost the efficacy of longer term traditional therapies.

Myeloperoxidase is expressed by activated macrophages and foam cells in vulnerable atherosclerotic plaques. It consumes hydrogen peroxide to create reactive oxygen species which further destabilize plaque and contribute to plaque progression and rupture. MPO(Gd) is a myeloperoxidase sensitive contrast agent which is oligomerized in the presence of myeloperoxidase, resulting in an increase in molecular weight and a consequent increase in relaxivity. The oligomer can then bind to proteins, further increasing its relaxivity and promoting local retention of the agent [27]. New Zealand White rabbits were maintained on a 0.125– 0.25% cholesterol diet for approximately 28 months to induce atherogenesis [28]. Rabbits were serially scanned for 2 h after the administration of either MPO(Gd) or Gd-DTPA. At 10 min after administration, the enhancement in the aorta was similar between the two agents, but at 2 h after administration, there was significantly greater contrast in the rabbits receiving MPO(Gd). This contrast persisted for at least 4 h after injection. To confirm that MPO(Gd) enhancement was due to myeloperoxidase rather than nonspecific accumulation, an analog was synthesized that was a substrate for other peroxidases but not myeloperoxidase. In rabbits treated with this analog (bis-tyr-DTPA(Gd)), minimal aortic enhancement was observed. In rabbits receiving MPO(Gd), a significant linear correlation was found between the number of enhanced pixels by MRI and the number of pixels staining positive for myeloperoxidase by immunohistochemistry (Fig. 15.8).

Myeloperoxidase One challenge of targeted MR molecular imaging is that it can be difficult to distinguish signal from bound, and unbound, targeted MRI contrast agents. Activatable agents address this problem by remaining quiescent until they reach the target of interest, at which time they are activated and become MRI-visible [26].

15

Targeted Agents for Wall Imaging

Fig. 15.8 Comparison of imaging with MPO(Gd) and a nonactivatable analog. ΔCNR (a) and enhancement ratio (ER) analysis (b) of diseased walls in cholesterol-fed rabbits (n = 3 rabbits, 11 sections analyzed) imaged after administration of DTPA(Gd), MPO(Gd), and a nonactivatable analog of MPO(Gd) [bis-tyr-DTPA(Gd)], with each agent administered at least 3 days apart. Significantly increased ΔCNR and ER was found with MPO(Gd) compared with both control agents, which supports the concept that the increased enhancement observed is the result of myeloperoxidase activation of MPO(Gd). No significant

Future Directions In vivo molecular imaging of atherosclerotic plaques, while highly desirable, must overcome significant hurdles to be realized in practice. The coronary arteries have a narrow diameter and are subject to both cardiac and respiratory motion. Differentiating plaque components from each other, even with the use of contrast agents, will require high resolution images and sophisticated motion compensation techniques. However, existing cardiac imaging techniques are continually being improved on and molecular imaging will certainly benefit from these advances. Signal amplification is another major hurdle, as existing MRI contrast agents typically must be present in micromolar concentrations to be observable. Strategies for signal amplification include the use of macromolecules with several gadolinium chelates conjugated to a rigid backbone; nanoparticles that can deliver a great deal of contrast with a single particle;

209

(NS) differences were observed between the analog and DTPA(Gd). (c) Representative 2-h delayed images demonstrated that only the activatable MPO(Gd) was able to show focal increased enhancement to confirm myeloperoxidase activity. The window/level settings were determined on the precontrast images to ensure that the paraspinal muscles appeared similar between images for each agent, and then these settings were applied to the postcontrast images. (d) Myeloperoxidase immunohistochemistry (IHC) confirmed MPO(Gd) imaging findings. (Reprinted from Ronald J et al. [28])

and enzyme activatable contrast agents, where a single enzyme can activate multiple contrast agent molecules, thus increasing local concentrations [26]. While promising preliminary data has been acquired for a number of targeting ligands and contrast agent platforms, more detailed studies of clearance, biodistribution, toxicity, and immunogenicity are required. Finally, new targeting ligands are continually being developed. Several strategies employed by individual investigators could be more widely applied to expand the repertoire of targeting ligands available for targeted MRI molecular imaging of atherosclerosis. For example, phage display can be used to discover short peptide sequences that can be used as an alternative to monoclonal antibodies, which are generally bulky and can be synthetically challenging to work with [4]. Alternately, designing contrast agents that mimic the natural ligand of the target of interest, or that mimic endogenous substances, can result in increased binding specificity.

210

E.A. Waters and T.J. Meade

References 15. 1. Sinusas AJ, Bengel F, Nahrendorf M, et al. Multimodality cardiovascular molecular imaging, part I. Circ Cardiovasc Imaging. 2008 2008;1:244–256. 2. Thim T, Hagensen M, Bentzon J, Falk E. From vulnerable plaque to atherothrombosis. J. Intern. Med. 2008;263:506–516. 3. Sanz J, Fayad ZA. Imaging of atherosclerotic cardiovascular disease. Nature. 2008;451:953–957. 4. Kelly K, Allport J, Tsourkas A, Shinde-Patil V, Josephson L, Weissleder R. Detection of vascular adhesion molecule-1 expression using a novel multimodal nanoparticle. Circ Res. 2005;96:327–336. 5. Ruehm S, Corot C, Vogt P, Kolb S, Debatin J. Magnetic resonance imaging of atherosclerotic plaque with ultrasmall superparamagnetic particles of iron oxide in hyperlipidemic rabbits. Circulation. 2001;103:415–422. 6. Von zur Muhlen C, Von Elverfeldt D, Moeller J, et al. Magnetic resonance imaging contrast agent targeted toward activated platelets allows in vivo detection of thrombosis and monitoring of thrombolysis. Circulation. 2008;118(3):258–267. 7. Frias J, Ma Y, Williams K, Fayad Z, Fisher E. Properties of a versatile nanoparticle platform contrast agent to image and characterize atherosclerotic plaques by magnetic resonance imaging. Nano Lett. 2006;6:2220–2224. 8. Briley-Saebo KC, Shaw PX, Mulder WJ, et al. Targeted molecular probes for imaging atherosclerotic lesions with magnetic resonance using antibodies that recognize oxidation-specific epitopes. Circulation. 2008;117:3206–3215. 9. Mulder W, Strijkers G, Briley-Saboe K, et al. Molecular imaging of macrophages in atherosclerotic plaques using bimodal PEGmicelles. Magn Reson Med. 2007;58:1164–1170. 10. Beilvert A, Cormode D, Chaubet F, et al. Tyrosine polyethylene glycol (PEG)-micelle magnetic resonance contrast agent for the detection of lipid rich areas in atherosclerotic plaque. Magn Reson Med. 2009;62:1195–1201. 11. Winter P, Morawski A, Caruthers S, et al. Molecular imaging of angiogenesis in early-stage atherosclerosis with alpha(v)beta3-integrin-targeted nanoparticles. Circulation. 2003;108:2270–2274. 12. Morawski AM, Winter PM, Yu X, et al. Quantitative “magnetic resonance immunohistochemistry” with ligand-targeted (19)F nanoparticles. Magn Reson Med. 2004;52:1255–1262. 13. Tang TY, Muller KH, Graves MJ, et al. Iron oxide particles for atheroma imaging. Arterioscler Thromb Vasc Biol. 2009;29:1001–1008. 14. McAteer M, Schneider J, Ali Z, et al. Magnetic resonance imaging of endothelial adhesion molecules in mouse atherosclerosis using

16.

17.

18.

19. 20.

21.

22.

23.

24.

25.

26. 27.

28.

dual-targeted microparticles of iron oxide. Arterioscler Thromb Vasc Biol. 2008;28:77–83. Botnar R, Perez A, Witte S, et al. In vivo molecular imaging of acute and subacute thrombosis using a fibrin-binding magnetic resonance imaging contrast agent. Circulation. 2004;109:2023–2029. Aime S, Castelli DD, Crich SG, Gianolio E, Terreno E. Pushing the sensitivity envelope of lanthanide-based magnetic resonance imaging (MRI) contrast agents for molecular imaging applications. Acc Chem Res. 2009;42:822–831. Winter PM, Caruthers SD, Yu X, et al. Improved molecular imaging contrast agent for detection of human thrombus. Magn Reson Med. 2003;50:411–416. Meding J, Urich M, Licha K, et al. Magnetic resonance imaging of atherosclerosis by targeting extracellular matrix deposition with Gadofluorine M. NMR Biomed. 2007;2:120–129. Lusis AJ. Atherosclerosis. Nature. 2000;407:233–241. Nahrendorf M, Sosnovik DE, French BA, et al. Multimodality cardiovascular molecular imaging, part II. Circ Cardiovasc Imaging. 2009;2:56–70. Lancelot E, Amirbekian V, Brigger I, et al. Evaluation of matrix metalloproteinases in atherosclerosis using a novel noninvasive imaging approach. Arterioscler Thromb Vasc Biol. 2008;28:425–432. Alsaid H, De Souza G, Bourdillon M, et al. Biomimetic MRI Contrast agent for imaging of inflammation in atherosclerotic plaque of ApoE−/− mice: a pilot study. Invest Radiol. 2009;44:151–158. Chaubet F, Bertholon I, Serfaty J, et al. A new macromolecular paramagnetic MR contrast agent binds to activated human platelets. Contrast Media Mol Imaging. 2007;2:178–188. Slevin M, Kumar P, Wang Q, et al. New VEGF antagonists as possible therapeutic agents in vascular disease. Expert Opin Invest Drugs. 2008;17:1301–1314. Winter P, Caruthers S, Zhang H, Williams T, Wickline S, Lanza G. Antiangiogenic synergism of integrin-targeted fumagillin nanoparticles and atorvastatin in atherosclerosis. JACC Cardiovasc Imaging. 2008;1:624–634. Major JL, Meade TJ. Bioresponsive, cell-penetrating, and multimeric MR contrast agents. Acc Chem Res. 2009;42:893–903. Rodriguez E, Nilges M, Weissleder R, Chen J. Activatable magnetic resonance imaging agents for myeloperoxidase sensing: mechanism of activation, stability, and toxicity. J Am Chem Soc. 2009;132:168–177. Ronald J, Chen J, Chen Y, et al. Enzyme-sensitive magnetic resonance imaging targeting myeloperoxidase identifies active inflammation in experimental rabbit atherosclerotic plaques. Circulation. 2009;120:592–599.

Part II Clinical Applications

wwwwwwwwwwww

Intracranial Arterial and Venous Disease

16

Dariusch R. Hadizadeh, Horst Urbach, and Winfried A. Willinek

Technical Considerations MRA Techniques Time-of-Flight MR-Angiography Inflow or time-of-flight MR angiography (TOF-MRA) sequences were the first sequences that were exclusively developed to image vascular structures in various regions of the body and have been used since the early 1980s [1]. Using gradient echo sequences with very short repetition times (TR < T1) and partial flip angle techniques, the background tissue is effectively saturated and only blood that enters the imaging volume during the repetition time (and has therefore not been saturated before) is visualized. Background suppression is further enhanced by magnetization transfer contrast (MTC) [2]. However, this nonsubtractive imaging technique is highly dependent on the flow direction and the coverage of anatomic volume is limited. This limitation has partly been overcome by the application of multiple overlapping thin slab acquisition (MOTSA) [3], which is now routinely implemented in TOF-MRA sequences and allows for better coverage when imaging intracranial arteries. Short echo times (TE) reduce intravoxel dephasing artifacts and tilted optimized nonsaturating excitation (TONE) [4] is used to counteract effects of linearly decreasing signal across the imaging slab by accordingly increasing flip angles within the imaging volume. Technical developments and the introduction of parallel imaging have made 3-dimensional TOF-MRA (3D TOFMRA) a robust noncontrast-enhanced imaging technique for D.R. Hadizadeh, MD () Department of Neuroradiology, University of Bonn, Bonn, Germany e-mail: [email protected] H. Urbach, MD Department of Neuroradiology, University of Bonn, Bonn, Germany Department of Radiology, University of Bonn, Bonn, Germany W.A. Willinek, MD Department of Radiology, University of Bonn, Bonn, Germany

the visualization of intracranial arteries with short acquisition time, effective background presaturation and high spatial resolution. Limitations still apply including thrombi that may simulate flow, in-plane and in-volume saturation effects, and low sensitivity particularly to slow flow.

Phase-Contrast MR-Angiography A few years after the introduction of time-of-flight MR angiography another technique was developed that allowed the visualization of flowing blood based on phase shifts in a magnetic field that is established by bipolar gradients [5]. In this setting, the change in precession frequency results in a phase shift that is measured in three flow-directions while a forth measurement is added with flow-compensation to allow for extraction of the pure flow-information by subtraction of the flow-encoded and flow-compensated data sets. The maximum phase shift is defined as the expected maximum flow velocity and information about flow up to the defined maximum velocity is visualized. Therefore, this method allows for venous-only imaging by defining the flow velocity above the venous and at the same time below the arterial flow velocity level. However, arterial-only imaging by PC-MRA is limited. Advantages of this technique include a better differentiation of flow and thrombus, variable flow sensitivity and the possibility to exactly quantify flow velocity [6]. Disadvantages are (a) long acquisition times that are inherent to this method because of the necessity of four separate measurements for the extraction of flow information and (b) dephasing effects if flow is turbulent. Contrast-Enhanced MR-Angiography Since its introduction in the early 1990s, contrast-enhanced MR angiography (CE-MRA) has become the method of choice for MR vascular imaging in almost all regions of the body [7]. This method is based on the fact that the center of k-space encodes the contrast-resolution while the periphery of k-space encodes the spatial resolution. As long as the center of k-space data is acquired during the arterial bolus passage of a T1-shortening contrast agent, arterial-only imaging

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_16, © Springer Science+Business Media, LLC 2012

213

214

is realized, even though acquisition of the periphery of k-space takes place when both arteries and veins are filled with the contrast agent. However, the short arterial window for acquisition of central k-space information limits acquisition time and hence maximum spatial resolution. Many attempts were made to modify central k-space acquisition during the arterial phase of bolus passage to further accelerate imaging or increase spatial resolution including 3D TRICKS [8], elliptical centric view ordering [9] and randomly segmented central k-space ordering [CENTRA] [10]. Finally, the application of parallel imaging dramatically reduced imaging time of the gradient-echo T1-weighted sequences that are used in CE-MRA and allowed for high-resolution 3D data set acquisition during a single breath hold which is especially useful in supraaortic imaging. However, exact timing of data acquisition during the arterial bolus passage remains crucial for CE-MRA. Practical applications for exact bolus timing have been realized by low-dose test bolus acquisition, fluoroscopic triggering, and automated signal-dependent bolus triggering.

D.R. Hadizadeh et al. Table 16.1 Recommendations for the clinical application of MRA-techniques Indication Stroke Vasculitis Aneurysm Untreated Treated Clipped Coiled Venous thrombosis AVM, AVF

Recommended MRA-sequence TOF-MRA, CE-MRA TOF-MRA, CE-MRA TOF-MRA, CE-MRA TOF-MRA CE-MRA TOF-MRA ± CE PC-MRA, time-resolved CE-MRA, CE-MRA Time-resolved CE-MRA, TOF-MRA

PC-MRA in visualizing sinus thrombosis due to the fact that the latter method may suffer from artifacts that mimic or falsely rule out sinus thrombosis [13]. Recommendations on the use of different MRA-techniques for various diseases of intracranial vessels are summarized in Table 16.1.

Clinical Recommendations While CE-MRA has replaced noncontrast-enhanced MRA techniques in most body regions, 3D TOF-MRA and PC-MRA are still routinely applied when imaging intracranial vessels due to some specific advantanges for this region of the body. For instance, the spatial resolution that has been achieved with 3D TOF-MRA for the visualization of the circle-ofWillis and its branches at high field strength is still unequaled by CE-MRA and therefore this method is of specific value for noninvasive screening for intracranial stenoses and aneurysms [11]. 3D PCA-MRA, on the contrary, allows for the clear, selective visualization of the intracranial venous system without the application of Gadolinium-based contrast agents and is routinely used for exclusion of sinus thrombosis in many centers. Limitations of either method due to slow flow and differentiation of thrombus vs. flow effects have been overcome to a large extent by sequence modifications as they are readily available in modern standard whole-body MR scanners with field-strengths of 1.0–3.0 T. CE-MRA has gained an important role in high-spatial resolution imaging of the extracranial supraaortic arteries and is readily acquired in less than 2 min [12]. In clinical protocols, the application of the contrast-agent for CE-MRA can be used for vascular imaging in the first step and visualization of contrast-uptake by intracranial lesions due to a breakdown of the blood–brain barrier in the second step, such that altogether no additional contrast agent is applied by the addition of the CE-MRA sequence. Limitations include a possible overinterpretation of stenosis grade, lack of visualization of the origin of the vertebral arteries at the subclavian arteries partly due to vessel movement, and trade-off between temporal and spatial resolution. Some investigators favor the use of CE-MRA over

Arterial Disease Stenoocclusive Disease/Stroke Atherosclerotic Disease With an age-adjusted annual incidence of 5.1/1,000 patient years, stroke is among the leading causes of invalidity in the western hemisphere [14]. While both computed tomography angiography (CTA) and CE-MRA have been shown to achieve high sensitivity and specificity in intracranial arterial stenosis visualization, CE-MRA techniques have the advantage that they can easily be combined with both diffusion-weighted (DWI) and perfusion-weighted imaging (PWI) techniques that allow for early detection of infarcted brain areas and identification of tissue-at-risk. TOF-MRA can be applied for high-resolution imaging of intracranial artery occlusion, while CE-MRA may be added to diagnose extracranial supraaortic artery disease that can have a major impact on the practicability of therapeutic intravascular options (Fig. 16.1). Borderline infarctions at the marginal areas of vascular territories, and territorial, basal ganglia, and brainstem infarctions are reliably diagnosed by DWI. Macro- and microangiopathic insults depending on the affected vessel size, hemorrhagic transformation, and subcortical arteriosclerotic encephalopathy with the appearance of lacunar infarctions are accurately identified by MRI [15]. The combination of MRA, structural imaging, DWI, and PWI facilitate clinical decision-making for local intraarterial thrombolysis or mechanical recanalization. Table 16.2 provides recommendations for a comprehensive MRI protocol to diagnose ischemic brain disease including differential diagnoses.

16

Intracranial Arterial and Venous Disease

215

Fig. 16.1 A 85-year-old male patient with high-grade tandem stenoses (arrows) of the right middle cerebral artery; (a) Supraaortic CE-MRA, (b) TOF-MRA of the circle-of-Willis

Table 16.2 Recommended MRI protocol in differential diagnosis of ischemic brain disease MR-sequence DWI PWI T2* GRE/ SWI T2 FLAIR

Plane tra, cor, (sag) tra tra tra

T1 ± CE T1 fatsat TOF CE-MRA

tra tra2 CB SA

Best for visualization of… Infarcted area Tissue-at-risk Microbleeds/calcifications Chronic ischemic events/ acute edema Blood–brain barrier disruption Intramural hematoma in dissection Intracranial stenoocclusive disease Supraaortic stenoocclusive disease

tra transversal, cor coronal, sag sagittal, tra2 transversal (artery of interest), CB circle-of-Willis and basilar artery, SA supraaortic and intracranial arteries

In case of subclavian steal-syndrome, clinically presenting with blood-pressure differences between both upper extremities and clinical symptoms while exercising with the ipsilateral arm, the combination of TOF-MRA and CE-MRA can display the subclavian stenosis and may also suggest retrograde flow in the ipsilateral vertebral artery [16]. Only recently, time-resolved CE-MRA has been introduced and may serve as one alternative for visualization of dynamic and morphologic changes similar to DSA.

Vasculitis Intracerebral vasculitis is a rare inflammatory disease (<4/100,000) that is commonly overestimated by neurological referrals [17]. Primary and secondary forms are distinguished according to the underlying diseases including periarteriitis nodosa, Behçet disease, viral (HBV, HCV, HIV, Herpes zoster, CMV), bacterial (tuberculosis, lues, borreliosis), and parasite (cysticercosis) infections, systemic lupus erythomatodes, giant cell arteriitis, radiation, malignant diseases, drug abuse, and organ transplantation. Primary forms of vasculitis are classified by the size of the affected arteries and detection of antibodies directed against neutrophil cytoplasmatic antigens (antineutrophil cytoplasmatic antibodies, ANCA): the sensitivity of MRA is highest for the diagnosis of “large vessel arteriitis” including giant cell arteriitis, Takayasu disease, and primary CNS angiitis.

Fig. 16.2 A 66-year-old female patient with vasculitis (giant cell arteriitis) with multiple acute and subacute infarctions (hollow arrows) and high-grade stenoses of both distal internal cerebral arteries (segments C5, C6; arrows) and the distal right vertebral artery (segments V3, V4; arrowheads) and occlusion of the left vertebral artery; (a) transversal diffusion-weighted image and transversal FLAIR, (b) TOFMRA of the circle-of-Willis, (c) Supraaortic CE-MRA (d) selective DSA of the left internal cerebral artery and right vertebral artery show stenoses accordingly to MRA findings

When MRI or laboratory findings indicate vasculitis, digital subtraction angiography (DSA) is considered the method of choice for further evaluation. However, even DSA cannot sufficiently visualize small vessels (<200 mm) and definitive diagnosis may only be possible by biopsy (biopsy must include dura, leptomeninx, cortex, and white matter) [18]. Typical findings include perivascular inflammation, segmental vascular stenoses and dilatation with formation of pseudoaneurysms (“string-of-pearls” sign) [19]. Signs in structural imaging are unspecific and include a mixed pattern of microangiopathic, hemodynamic, and territorial infarctions in multiple territories, parenchymal hemorrhage, and T2-hyperintense lesions in the cortex, basal ganglia, and white matter (Fig. 16.2). However, though MR findings may be rather unspecific, the absence of any abnormalities in structural imaging, on the contrary, may be used to exclude the diagnosis of vasculitis. Complications of vasculitis include intracranial hemorrhage in the case of vascular

216

D.R. Hadizadeh et al.

Table 16.3 Morikawi classification of Takayasu disease

Table 16.4 Suzuki stages for staging of Moyamoya disease

Morikawi type I IIa IIb

Suzuki stage I II

III IV

Affected vessels Aortic arch Ascending aorta, aortic arch and its branches Ascending and descending thoracic aorta, aortic arch and its branches Descending thoracic aorta, abdominal arteries, optionally renal arteries Findings of both types IIb and III

rupture and ischemic infarctions due to stenosis or occlusion of affected vessels and are best visualized in standard structural imaging of the brain [19]. Overall, MRI offers high sensitivity, but at the same time low specificity in the diagnosis of vasculitis and may particularly be useful in active disease [20]. On the contrary, some findings in MRA may be rather specific: for instance, Wegener granulomatosis allows for a rather specific diagnosis by the simultaneous visualization of vascular inflammation and a thickened, enhancing, granulomatous Dura [21]. Takayasu disease, on the contrary, leads to bilateral stenoses at the origin of the subclavian arteries, common carotid arteries, and brachiocephalic trunk (Morikawi classification [22], see Table 16.3) and is best visualized with supraaortic CE-MRA that may allow the visualization of vascular abnormalities at an early stage of the disease. In TOF-MRA at high field strengths, vasculitis-associated changes are more pronounced and may be more readily visualized due to an overestimation effect and the high spatial resolution that can be achieved with this method.

Moyamoya Disease Moyamoya disease (jap. moyamoya “puff of smoke”; the name refers to the “cloudy” angiographic appearance of collateral vessels in DSA) is a progressive stenoocclusive disease of the basal cerebral arteries that commonly bilaterally affects the supraclinoid internal carotid arteries and leads to the formation of tiny basal collaterals from the lenticulostriate, choroidal, and anterior and posterior perforating arteries, as well as ethmoidal branches of ophthalmic arteries [23]. The primary disease has an unknown cause, is very rare (incidence: <1:1 Mio.), and mainly affects Japanese women under the age of 20 years that initially present with transients ischemic attacks and eventually ischemic hemiparesis. The secondary, symptomatic form of the disease, Moyamoya syndrome, affects adults and occurs after vasculitis, trauma, radiation, tuberculous meningitis or is related to neurofibromatosis, sickle cell anemia, atherosclerosis, Down syndrome, or tumor compression. In contrast to the primary form, these patients commonly present with subarachnoid hemorrhage. Several disease stages (Suzuki stages, Table 16.4) are discriminated [24].

III IV

V VI

Description Stenosis of the intracranial carotid fork Dilation of main cerebral arteries and emergence of a fine pseudo rete mirabile Further expansion of network and perfusion deficit in anterior and middle cerebral artery territories Occlusion of internal carotid artery or stenosis of all three cerebral arteries, decrease in collateral circulation, defection of posterior cerebral artery with appearance of ethmoidal collaterals Occlusion of main cerebral arteries and formation of leptomeningeal anastomoses The intracranial circulation is only fed by the external cerebral artery

Fig. 16.3 A 44-year-old male patient with Moyamoya syndrome and low caliber of the internal carotid arteries of both sides (arrows in b) and defects after prior infarctions in the territories of both middle cerebral arteries without evidence of acute infarctions (arrows in c); (a) TOF-MRA of the circle-of-Willis, (b) Supraaortic CE-MRA, (c) Transversal FLAIR, and transversal diffusion-weighted image

CE- and TOF-MRA allow for the visualization of larger collateral vessels and structural imaging at the same time reveals ischemic lesions (lacunar, territorial, hemodynamic infarctions; Fig. 16.3) [25]. Therefore MRI and MRA are considered best for follow-up of Moyamoya disease after initial diagnosis by DSA.

Dissection Twenty percent of strokes in young and middle-aged patients are estimated to be caused by arterial dissections [26]. The clinical history often reveals hyperextension trauma of the neck or chiropractic manipulations. Other risk factors are

16

Intracranial Arterial and Venous Disease

arterial hypertension, fibromuscular dysplasia, migraine, cystic media necrosis, drug abuse, oral contraceptives, infection, Marfan syndrome, and Ehlers–Danlos syndrome. Up to 60% of dissections are termed “spontaneous” [27]. Symptoms of arterial dissection include headache, neck pain, cranial nerve deficits, and the sudden appearance of a Horner syndrome [28]. Complications are ischemic infarctions and the formation of pseudoaneurysms with a usually benign clinical course. The vast majority of arterial dissections occur in extracranial arteries and the site of hemorrhage in these cases is the tunica media (middle coat) that is markedly thickened. The predilection site for dissections of the extracranial vertebral artery is the atlas loop; dissections of the internal cerebral artery usually start at the distal portion of the extracranial internal carotid artery and end before entering the petrous portion of the temporal bone. Dissections of the distal portions of the internal cerebral artery (incidence: 2.5– 3/100,000) and vertebral artery (incidence: 1–1.5/100,000) are common causes of ischemic events in young adults [29]. Dissections of intracranial arteries are considered to be rare. In these patients, hemorrhage into the vascular wall is secondary to a tear in the tunica intima (inner coat) and results in either stenosis/occlusion of the affected artery over a variable distance with subsequent ischemia of the affected vessel territory or subarachnoid hemorrhage. The vessel wall composition of intra- and extracranial arteries differs: up to 1 cm before entering the dura mater, the wall of the vertebral artery is composed of a thick tunica adventitia (outer coat), thick external elastic membrane, and a middle coat with a high content of muscular and elastic fibers. Within the next 2 cm, the content of collagen fibers of the outer coat of the vertebral artery is reduced and its thickness decreases to 2/3 of its original size, while the external elastic membrane is only composed of spare fibers and the middle coat is thinned out and contains less elastic fibers. In intracranial arteries, hemorrhage occurs either between the inner and middle coat, which results in vessel occlusion or between the middle and outer coat which manifests with subarachnoid hemorrhage [30]. In addition, arterioarterial emboli may occur due to the thrombogenic environment caused by a rupture in the vessel wall. CE-MRA and fat-suppressed transversal or coronal T1-weighted sequences are best for direct visualization of the stenotic vessel segment and the intramural hematoma (Fig. 16.4). Additional functional information and follow-up is readily available by ultrasonography and helps make the decision as to whether it is necessary to perform invasive DSA in these predominantly young patients. Inhomogeneous flow may be shown by TOF- and PC-MRA techniques and the possibility to quantify flow and define flow directions may be of specific value when using PC-MRA. The value of timeresolved CE-MRA for the evaluation of this disease entity remains to be determined although initial results on abdominal dissecting aortic aneurysms are promising [31]. Noninvasive

217

Fig. 16.4 A 27-year-old female patient with acute dissection of the left internal cerebral artery with string sign (arrows in a) and intramural hematoma (arrow in b). The right internal carotid artery and intracranial arteries are regularly visualized (arrow in c); (a) Supraaortic CE-MRA, (b) fat-suppressed T1-weighted transversal section at the level of the dissection of the left internal carotid artery, (c) TOF-MRA of the circle-of-Willis

diagnosis of this disease is crucial to start early anticoagulation or antiplatelet therapy as the treatments of choice, which are considered to improve clinical outcome even though large randomized controlled trials are lacking [32].

Aneurysms While in the extracranial circulation an aneurysm is defined by the dilatation of the diameter of the artery to more than 1.5 times of its regular diameter, in intracranial arteries every outpouching of an artery is referred to as an aneurysm. Vascular bifurcations are predilection sites for aneurysm formation and the risk for aneurysm formation is increased in smokers, alcohol abuse, trauma, and postinfection (mycotic). Other risk factors include autosomal dominant polycystic kidney disease, aortic coarctation, fibromuscular hyperplasia, age, female gender, hypertension, and having one or more affected relatives with subarachnoid hemorrhage (SAH) [33]. The age peak of aneurysm detection is 40–60 years, which is probably due to the fact that intracranial aneurysms are usually asymptomatic until ruptured. The general incidence of unruptured aneurysms is estimated with 0.5%, whereas the incidence of aneurysmal hemorrhage is estimated at ten cases per 100,000. However, 10% of patients with SAH due to aneurysm bleeding have a proven history of familial occurrence of aneurysmal SAH with two or more patients affected within the same family, suggesting some genetic predisposition for this disease. The overall frequency of multiple aneurysms has been reported

218

D.R. Hadizadeh et al.

to be 20% in women and 12% in men [34–36]. The average annual rupture incidence is given with 1.4% and cumulative rates of bleeding from the aneurysm range from 10% at 10 years to 32% at 30 years after the diagnosis [37]. The potential for rupture and subsequent hemorrhage depends on the size and shape of the aneurysm. The primary symptom of aneurysm rupture is known as the “worst headache of my life,” usually occurs while the person is active, and is often accompanied with a transient loss of consciousness and neck stiffness [38].

Table 16.5 Hunt and Hess classification of patients with intracranial aneurysm Grade I II III IV V

Criteria Asymptomatic to minimal headache, slight nuchal rigidity Moderate or severe headache, nuchal rigidity, no neurological deficit other than cranial nerve palsy Drowsiness, confusion, or mild focal deficit Stupor, moderate to severe hemiparesis, possibly early decerebrate rigidity and vegetative disturbances Deep coma, decerebrate rigidity, moribund appearance

Fig. 16.5 A 49-year-old female patient with a coiling of a broad-based aneurysm at the basilar tip with a diameter of 5 mm and no signs of residual perfusion in postinterventional studies (arrows); (a) Preinterventional CTA shows the broad-based basilar tip aneurysm, (b) Postinterventional DSA shows positioning of the coils and excludes residual perfusion of the aneurysm, (c) Postinterventional TOF-MRA of the circle-of-Willis (wholevolume MIP and single slices at the region of the aneurysm’s base; arrowhead: basilar artery), (d) Postinterventional Supraaortic CE-MRA

Up to 80% of patients who are diagnosed with subarachnoid hemorrhage in the emergency department suffer from ruptured aneurysms. The prognosis correlates with the level of consciousness and neurologic deficits on presentation (Hunt and Hess Grades I-V [39], Table 16.5, survival rates: 46% (I/II), 4% (IV/V)) [40]. Therapeutic options (clipping vs. coiling) depend on the site of the aneurysm, its shape and relationship to other arteries including those that originate from the aneurysm itself. Therefore, high spatial resolution vascular imaging is crucial for clinical decision-making in the workup of both incidental and symptomatic intracranial aneurysms. In this context, high field strength offers important advantages over lower field strength due to an increase in available signal that can be used to apply high parallel imaging factors and can be invested into shorter acquisition times or increased spatial resolution. In addition, T1 shortening effects result in improved vessel-totissue contrast and improved sensitivity to Gd-chelates (Fig. 16.5). The follow-up of patients with intracranial stents, and aneurysms after coiling or clipping remains challenging, but first results that compare TOF-MRA at 3.0 T to DSA have provided promising results and suggest that MRA may not only be an adjunctive tool, but ready to replace DSA in the

16

Intracranial Arterial and Venous Disease

219

follow-up of patients with previously coiled intracranial aneurysms and additional DSA may only be necessary in selected cases [41].

Venous Disease Thrombosis Sinus thrombosis is a rare, but serious complication of prothrombotic stages that is found in both noninfectious and septic conditions [42]. Infectious forms occur following purulent otomastoid, orbit, and central face cutaneous infections, sinusitis, meningitis, or septic embolization. Noninfectious causes are more common and occur postnatally, in dehydrated and malnutritioned children, after hormone administration, in patients with coagulopathies, posttraumatically including cerebral birth traumas, paraneoplastic, in pregnancy, surgery, and in patients with arteriovenous malformations, extracorporal membrane oxygenation (ECMO), and jugular thrombosis after central venous catheterization; intracranial tumors are the most common causes, but in 30% of cases the underlying etiology cannot be identified [43]. Women are more commonly affected than men mainly due to prothrombotic factors including third-generation oral contraceptives with a higher incidence in third-trimester pregnancy [44]. In fact, sinus thrombosis is most often diagnosed in premature birth, pregnancy and the third decade of life. Venous congestion with surrounding edema and hemorrhage is often found and leads to symptoms of intracranial hypertension including headache (varying from a subclinical course to severe headache), vomiting, nausea, altered vigilance, confusion, psychosis, focal seizures, impaired vision, and hemiparesis. Owing to its significant mortality and striking benefit of treatment of this disease (mortality: treated below 15% vs. untreated up to 70%), an aggressive diagnostic approach is recommended although data on therapeutic approaches is limited [45, 46]. Inflammatory parameters in serum and cerebral spinal fluid are often increased, but rather unspecific [47]. Imaging plays a key role in the diagnosis of sinus thrombosis and important signs include brain edema, direct thrombus visualization (“empty delta or triangle sign”: the fresh unorganized thrombus does not enhance in the confluence of sinusus) [48]. Other signs include bilateral frontal/ parietal hemorrhagic infarction, and local hyperemia. An accurate knowledge of venous drainage areas is of utmost importance for the correct diagnosis. Differential diagnoses include sinus hypoplasia, arachnoid granulations (Pacchioni’s granulations), and pseudotumor cerebri [49]. Structural MRI and CE-MRA are the primary diagnostic tools for the evaluation of suspected sinus thrombosis and allow for direct visualization of the thrombus and parenchymal changes, slow and complex flow, segmental recanalization, formation

Fig. 16.6 A 29-year-old female patient with thrombosis of the superior sagittal sinus (arrows); (a) coronal T1-weighted nonenhanced and (b) coronal T2-weighted images show signal abnormalities in the superior sagittal sinus resembling thrombotic tissue, (c) PC-MRA of intracranial venous structures

of collaterals, and sinus hypoplasia (Fig. 16.6). Pitfalls include the lack of visualization of hypoplastic sinuses, which most commonly present as an asymmetric visualization of the transversal sinuses and a hypoplastic anterior third of the superior sagittal sinus. In these cases, structural imaging, especially sagittal and coronal T2-weighted spin echo sequences, allows for differentiation of hypoplastic from normal-sized thrombotically occluded sinuses [49]. The exclusive visualization of venous blood-flow by PC-MRA with low flow-velocity-settings is of secondary importance in the diagnosis of sinus thrombosis.

Combined Arteriovenous Disease Cerebral Arteriovenous Malformations In cerebral arteriovenous malformations (cAVM), altered smooth muscle maturation proteins (significantly decreased expression of “smoothelin” reflects disappearance of contractile property) lead to a lack of capillary beds and abnormal

220

D.R. Hadizadeh et al.

angioarchitecture in a defined area of the brain (nidus) [50]. This may cause an apoplectic bleeding under high-flow conditions, and therefore, patients with cAVM carry a lifetime risk of focal neurological deficits, seizures and hemorrhagic stroke with possibly life-threatening or disabling outcome [51]. Cerebral AVMs have been found as incidental findings in 0.2% in a population of 2,536 healthy young men [52, 53]. The perioperative risk is assessed by the Spetzler-Martin classification [54] and a total score >2 indicates an increased operative risk (see Table 16.6). For this reason, vascular imaging is very important prior to clinical decision-making to estimate the risks and benefits of invasive therapy. Very short mean transit times of feeding arteries and draining veins in cAVM of typically 1.0–2.1 s (high flow AVM: 200 ms) and the small size of the involved vessels (circle-of-Willis: 2–5 mm; perforating arteries 0.3–0.7 mm) make these vascular abnormalities a challenge for MR imaging. Therefore, DSA has remained the standard procedure for diagnostic imaging of cAVM [55]. However, techniques

Table 16.6 Spetzler-Martin malformations Nidus size

Eloquence Venous drainage

classification

<3 cm 3–6 cm >6 cm Noneloquent Eloquent Superficial Deep

of

that allow for both high temporal and spatial resolution are highly desirable for the visualization of cAVM and many attempts were made in the near past to overcome limitations of standard imaging procedures with the help of complex k-space acquisitions schemes. For instance, a combination of fast imaging strategies (10) including keyhole [56, 57] and parallel imaging [58] has achieved a temporal resolution of roughly 600 ms and at the same time a spatial resolution of 1.1 mm³ [59] and has been shown to allow for reliable Spetzler-Martin-Classification of cAVM [60] (Fig. 16.7). Nidus size, eloquence of adjacent brain tissue and venous drainage are precisely determined on MR imaging. The combination of high-resolution T2-weighted sequences and MRA allows for unique comprehensive information about the clinical risk for rupture and the history of previous hemorrhage that is not provided by any other single imaging modality [61]. However, limitations still exist especially regarding the clear definition of all small feeding arteries and therefore further developments will be needed to address all issues related to noninvasive cAVM visualization to replace diagnostic DSA by MRA.

arteriovenous Score 1 2 3 0 1 0 1

Fig. 16.7 Maximum intensity projections of consecutive frames of dynamic time-resolved CE-MRA of a right temporal cerebral arteriovenous malformation in a 20-year-old female patient allowing for visualization of contrast bolus passage: arterial enhancement (arrows),

Dural AV Fistula While traumatic fistulas are often found in extracranial vessels as for example between the internal carotid artery and the cavernous sinus, the etiology of dural arteriovenous fistula (dAVF) often remains unclear. The incidence of dAVF has been shown to be increased after surgery, infection, and pregnancy and a frequent association with sinus thrombosis

enhancement of the nidus (hollow arrows), and venous drainage (arrowheads) of the cAVM are displayed in separate time frames due to high temporal resolution

16

Intracranial Arterial and Venous Disease

was found [62]. Therefore, it has been discussed that these acquired lesions in adults may be the result of an enlargement of physiologic arteriovenous shunts as an attempt of recanalization after thrombosis. Dural AVFs comprise 10–15% of intracranial arteriovenous lesions and are located between pachymeningeal arteries and dural sinuses, dural veins, or leptomeningeal veins [63]. Common sites are the cavernous sinus, sigmoid sinus, and occipital regions of the brain. Depending on the site of the dAVF, symptoms may include headache, pulsatile tinnitus, pulsatile exophthalmus, and cranial nerve palsies. The classification of dAVFs is based on venous drainage patterns (Table 16.7) and Djindjian types III and IV are associated with a high risk of hemorrhage (50–66%) [64]. MRA allows for noninvasive visualization of large dAVF lesion and identification of feeding arteries and draining veins (Fig. 16.8). Smaller lesions are best visualized with high spatial-resolution TOF- or CE-MRA protocols. PC-MRA may offer valuable information about venous drainage patterns, but suffers from limitations in

Table 16.7 Djindjian classification of AVFs Djindjian type Description I Antegrade, extracranial flow direction, limited to affected dural sinus, ipsilateral or bilateral tinnitus IIa Insufficient antegrade venous drainage and subsequent reflux of arterial blood into sinus IIb Insufficient antegrade venous drainage and subsequent reflux of arterial blood into bridging vein III Direct cortical drainage into nonectatic cortical vein IV Direct cortical drainage into ectatic cortical vein at site of fistula V Drainage into spinal perimedullary vein (ascending cord paralysis)

Fig. 16.8 A 58-year-old male patient with arteriovenous fistula of the left marginal sinus in the hypoglossus channel with early filling of the left internal cerebral vein (arrow in a) and enlarged arteries in the region of interest (arrows in b); (a) Supraaortic CE-MRA, (b) TOF-MRA at

221

the visualization of feeding arteries. Time-resolved CE-MRA techniques may offer valuable functional information that may help in the identification of dAVFs, but this has not yet been verified by systematic studies. MRA has potential to noninvasively provide valuable information about dAVF but it may miss particularly smaller dural dAVFs or those without dilated venous drainage and therefore up to date DSA still remains the reference standard and should be performed especially in equivocal cases [63].

Differential Diagnosis of “Atypical” Intracerebral Hemorrhage Differential diagnostic considerations of spontaneous intracranial hemorrhage primarily depend on the site of hemorrhage. A hemorrhage of the putamen–claustrum region that is usually secondary to hypertension is referred to as a “typical” intracerebral hemorrhage. Another common location for hypertensive hemorrhages is the cerebellum. All other sites of hemorrhage are considered “atypical” and may be related to aneurysmal rupture, trauma, hemorrhagic transformation of infarction, anticoagulation, cerebral amyloid angiopathy, neoplasm, vasculitis, Moyamoya disease, cavernoma, arteriovenous malformations, and cortical vein or sinus thromboses [65]. Differential diagnoses of the cause for hemorrhage include a variety of both clinical and imaging parameters. High spatial resolution TOF-MRA allows for the direct visualization of most ruptured intracranial aneurysms and hemorrhage is most commonly found in the subarachnoid space and is readily detected on T2*-weighted imaging. A history of anticoagulative therapy and the identification of fluid/fluid levels are characteristic for coagulopathic hemorrhage. Foci of prior subcortical hemorrhage are typically identified in cases of cerebral amyloid angiopathy. Intraparenchymal

the region of interest. (c) Selective DSA of the ascending pharyngeal artery displays pathologic arteriovenous shunt vessels (arrows) and early filling of the draining veins (arrowhead)

222

neoplasms are in the majority of cases surrounded by marked vasogenic edema due to the mass effect of the lesion. Arteriovenous malformations may be difficult to diagnose when small and are identified by the blooming effect of hemosiderin deposits and a “bag of worms” appearance in T2-weighted spin echo sequences due to flow voids in highflow nidus vessels. Signs of cerebral vein and sinus thrombosis have been extensively discussed in the preceding paragraph. Isolated cortical vein thrombosis is found in 1% of patients suspected of having stroke, and is characterized by hemorrhage due to venous occlusion and subsequent increase of venous pressure [66]. In MRI, hemosiderin-sensitive sequences with T2*-weighting allow for visualization of the thrombotic clot with the corresponding blooming artifact, and FLAIR-sequences visualize the corresponding edema that is often the most prominent sign in imaging. In T1-weighted sequences, the thrombus is first isointense and later hyperintense. CE- and PC-MRA may allow for a direct visualization of the missing draining veins [67]. In case of atypical intracerebral hemorrhage, MRA should be performed to detect vascular abnormalities as an adjunct to structural MRI [68]. In this case, TOF- and CE-MRA sequences are the best choice to identify the underlying source of hemorrhage. If there is any doubt about the underlying disease, DSA is still required as the imaging modality of choice to exclude vascular disease.

References 1. Nishimura DG. Time-of-flight MR angiography. Magn Reson Med. 1990;14:194–201. 2. Balaban RS, Ceckler TL. Magnetization transfer contrast in magnetic resonance imaging. Magn Reson Q. 1992;8:116–137. 3. Parker DL, Yuan C, Blatter DD. MR angiography by multiple thin slab 3D acquisition. Magn Reson Med. 1991;17:434–451. 4. Atkinson D, Brant-Zawadzki M, Gillan G, Purdy D, Laub G. Improved MR angiography: magnetization transfer suppression with variable flip angle excitation and increased resolution. Radiology. 1994;190:890–894. 5. Dumoulin CL, Yucel EK, Vock P, et al. Two- and three-dimensional phase contrast MR angiography of the abdomen. J Comput Assist Tomogr. 1990;14:779–784. 6. Dumoulin CL. Phase contrast MR angiography techniques. Magn Reson Imaging Clin N Am. 1995;3:399–411. 7. Prince MR, Yucel EK, Kaufman JA, Harrison DC, Geller SC. Dynamic gadolinium-enhanced three-dimensional abdominal MR arteriography. J Magn Reson Imaging. 1993;3:877–881. 8. Korosec FR, Frayne R, Grist TM, Mistretta CA. Time-resolved contrast-enhanced 3D MR angiography. Magn Reson Med. 1996;36:345–351. 9. Huston J, III, Fain SB, Riederer SJ, Wilman AH, Bernstein MA, Busse RF. Carotid arteries: maximizing arterial to venous contrast in fluoroscopically triggered contrast-enhanced MR angiography with elliptic centric view ordering. Radiology. 1999;211:265–273. 10. Willinek WA, Gieseke J, Conrad R, et al. Randomly segmented central k-space ordering in high-spatial-resolution contrastenhanced MR angiography of the supraaortic arteries: initial experience. Radiology. 2002;225:583–588.

D.R. Hadizadeh et al. 11. Willinek WA, Born M, Simon B, et al. Time-of-flight MR angiography: comparison of 3.0-T imaging and 1.5-T imaging--initial experience. Radiology. 2003;229:913–920. 12. Nael K, Fenchel M, Salamon N, et al. Three-dimensional cerebral contrast-enhanced magnetic resonance venography at 3.0 Tesla: initial results using highly accelerated parallel acquisition. Invest Radiol. 2006;41:763–768. 13. Tsuruda JS, Shimakawa A, Pelc NJ, Saloner D. Dural sinus occlusion: evaluation with phase-sensitive gradient-echo MR imaging. AJNR Am J Neuroradiol. 1991;12:481–488. 14. Carandang R, Seshadri S, Beiser A, et al. Trends in incidence, lifetime risk, severity, and 30-day mortality of stroke over the past 50 years. JAMA. 2006;296:2939–2946. 15. Gold G, Kovari E, Herrmann FR, et al. Cognitive consequences of thalamic, basal ganglia, and deep white matter lacunes in brain aging and dementia. Stroke. 2005;36:1184–1188. 16. Bitar R, Gladstone D, Sahlas D, Moody A. MR angiography of subclavian steal syndrome: pitfalls and solutions. AJR Am J Roentgenol. 2004;183:1840–1841. 17. Scott DG, Watts RA. Systemic vasculitis: epidemiology, classification and environmental factors. Ann Rheum Dis. 2000;59:161–163. 18. Harris KG, Tran DD, Sickels WJ, Cornell SH, Yuh WT. Diagnosing intracranial vasculitis: the roles of MR and angiography. AJNR Am J Neuroradiol. 1994;15:317–330. 19. Greenan TJ, Grossman RI, Goldberg HI. Cerebral vasculitis: MR imaging and angiographic correlation. Radiology. 1992;182:65–72. 20. Blockmans D, Bley T, Schmidt W. Imaging for large-vessel vasculitis. Curr Opin Rheumatol. 2009;21:19–28. 21. Murphy JM, Gomez-Anson B, Gillard JH, et al. Wegener granulomatosis: MR imaging findings in brain and meninges. Radiology. 1999;213:794–799. 22. Moriwaki R, Noda M, Yajima M, Sharma BK, Numano F. Clinical manifestations of Takayasu arteritis in India and Japan – new classification of angiographic findings. Angiology. 1997;48:369–379. 23. Satoh S, Shibuya H, Matsushima Y, Suzuki S. Analysis of the angiographic findings in cases of childhood moyamoya disease. Neuroradiology. 1988;30:111–119. 24. Suzuki J, Takaku A. Cerebrovascular “moyamoya” disease. Disease showing abnormal net-like vessels in base of brain. Arch Neurol. 1969;20:288–299. 25. Scott RM, Smith ER. Moyamoya disease and moyamoya syndrome. N Engl J Med. 2009;360:1226–1237. 26. Schievink WI. Spontaneous dissection of the carotid and vertebral arteries. N Engl J Med. 2001;344:898–906. 27. Schwartz NE, Vertinsky AT, Hirsch KG, Albers GW. Clinical and radiographic natural history of cervical artery dissections. J Stroke Cerebrovasc Dis. 2009;18:416–423. 28. Ozdoba C, Sturzenegger M, Schroth G. Internal carotid artery dissection: MR imaging features and clinical-radiologic correlation. Radiology. 1996;199:191–198. 29. Cimini N, D’Andrea P, Gentile M, et al. Cervical artery dissection: a 5-year prospective study in the Belluno district. Eur Neurol. 2004;52:207–210. 30. Wilkinson IM. The vertebral artery. Extracranial and intracranial structure. Arch Neurol. 1972;27:392–396. 31. Schoenberg SO, Wunsch C, Knopp MV, et al. Abdominal aortic aneurysm. Detection of multilevel vascular pathology by timeresolved multiphase 3D gadolinium MR angiography: initial report. Invest Radiol. 1999;34:648–659. 32. Goyal MS, Derdeyn CP. The diagnosis and management of supraaortic arterial dissections. Curr Opin Neurol. 2009;22:80–89. 33. Clarke M. Systematic review of reviews of risk factors for intracranial aneurysms. Neuroradiology. 2008;50:653–664. 34. Rosenorn J, Eskesen V, Schmidt K. Unruptured intracranial aneurysms: an assessment of the annual risk of rupture based on epidemiological and clinical data. Br J Neurosurg. 1988;2:369–377.

16

Intracranial Arterial and Venous Disease

35. Weir B. Unruptured intracranial aneurysms: a review. J Neurosurg. 2002;96:3–42. 36. Kaminogo M, Yonekura M, Shibata S. Incidence and outcome of multiple intracranial aneurysms in a defined population. Stroke. 2003;34:16–21. 37. Juvela S, Porras M, Heiskanen O. Natural history of unruptured intracranial aneurysms: a long-term follow-up study. J Neurosurg. 1993;79:174–182. 38. Brisman JL, Song JK, Newell DW. Cerebral aneurysms. N Engl J Med. 2006;355:928–939. 39. Hunt WE, Hess RM. Surgical risk as related to time of intervention in the repair of intracranial aneurysms. J Neurosurg. 1968;28:14–20. 40. Bracard S, Lebedinsky A, Anxionnat R, et al. Endovascular treatment of Hunt and Hess grade IV and V aneuryms. AJNR Am J Neuroradiol. 2002;23:953–957. 41. Urbach H, Dorenbeck U, von Falkenhausen M, et al. Threedimensional time-of-flight MR angiography at 3 T compared to digital subtraction angiography in the follow-up of ruptured and coiled intracranial aneurysms: a prospective study. Neuroradiology. 2008;50:383–389. 42. Filippidis A, Kapsalaki E, Patramani G, Fountas KN. Cerebral venous sinus thrombosis: review of the demographics, pathophysiology, current diagnosis, and treatment. Neurosurg Focus. 2009;27:E3. 43. Saadatnia M, Fatehi F, Basiri K, Mousavi SA, Mehr GK. Cerebral venous sinus thrombosis risk factors. Int J Stroke. 2009;4:111–123. 44. de Bruijn SF, Stam J, Vandenbroucke JP. Increased risk of cerebral venous sinus thrombosis with third-generation oral contraceptives. Cerebral Venous Sinus Thrombosis Study Group. Lancet. 1998;351:1404. 45. Einhaupl KM, Villringer A, Meister W, et al. Heparin treatment in sinus venous thrombosis. Lancet. 1991;338:597–600. 46. Medel R, Monteith SJ, Crowley RW, Dumont AS. A review of therapeutic strategies for the management of cerebral venous sinus thrombosis. Neurosurg Focus. 2009; 27:E6. 47. Allroggen H, Abbott RJ. Cerebral venous sinus thrombosis. Postgrad Med J. 2000;76:12–15. 48. Lee EJ. The empty delta sign. Radiology. 2002;224:788–789. 49. Leach JL, Fortuna RB, Jones BV, Gaskill-Shipley MF. Imaging of cerebral venous thrombosis: current techniques, spectrum of findings, and diagnostic pitfalls. Radiographics. 2006;26(Suppl 1):S19-S41. 50. Moftakhar P, Hauptman JS, Malkasian D, Martin NA. Cerebral arteriovenous malformations. Part 1: cellular and molecular biology. Neurosurg Focus. 2009;26:E10. 51. Al Shahi R, Warlow C. A systematic review of the frequency and prognosis of arteriovenous malformations of the brain in adults. Brain. 2001;124:1900–1926.

223 52. Weber F, Knopf H. Incidental findings in magnetic resonance imaging of the brains of healthy young men. J Neurol Sci. 2006;240:81–84. 53. Kondziolka D, McLaughlin MR, Kestle JR. Simple risk predictions for arteriovenous malformation hemorrhage. Neurosurgery. 1995;37:851–855. 54. Spetzler RF, Martin NA. A proposed grading system for arteriovenous malformations. J Neurosurg. 1986;65:476–483. 55. Friedlander RM. Clinical practice. Arteriovenous malformations of the brain. N Engl J Med. 2007;356:2704–2712. 56. Jones RA, Haraldseth O, Muller TB, Rinck PA, Oksendal AN. K-space substitution: a novel dynamic imaging technique. Magn Reson Med. 1993;29:830–834. 57. van Vaals JJ, Brummer ME, Dixon WT, et al. “Keyhole” method for accelerating imaging of contrast agent uptake. J Magn Reson Imaging. 1993;3:671–675. 58. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999;42: 952–962. 59. Willinek WA, Hadizadeh DR, von Falkenhausen M, et al. 4D timeresolved MR angiography with keyhole (4D-TRAK): more than 60 times accelerated MRA using a combination of CENTRA, keyhole, and SENSE at 3.0T. J Magn Reson Imaging. 2008;27:1455–1460. 60. Hadizadeh DR, von Falkenhausen M, Gieseke J, et al. Cerebral arteriovenous malformation: Spetzler-Martin classification at subsecond-temporal-resolution four-dimensional MR angiography compared with that at DSA. Radiology. 2008;246:205–213. 61. Nussel F, Wegmuller H, Huber P. Comparison of magnetic resonance angiography, magnetic resonance imaging and conventional angiography in cerebral arteriovenous malformation. Neuroradiology. 1991;33:56–61. 62. Morales H, Jones BV, Leach JL, Abruzzo TA. Documented development of a dural arteriovenous fistula in an infant subsequent to sinus thrombosis: case report and review of the literature. Neuroradiology. 2010;52:225–229. 63. Gupta A, Periakaruppan A. Intracranial dural arteriovenous fistulas: A Review. Indian J Radiol Imaging. 2009;19:43–48. 64. Cognard C, Gobin YP, Pierot L, et al. Cerebral dural arteriovenous fistulas: clinical and angiographic correlation with a revised classification of venous drainage. Radiology. 1995;194:671–680. 65. Smith EE, Rosand J, Greenberg SM. Imaging of hemorrhagic stroke. Magn Reson Imaging Clin N Am. 2006;14:127–40, v. 66. Chang R, Friedman DP. Isolated cortical venous thrombosis presenting as subarachnoid hemorrhage: a report of three cases. AJNR Am J Neuroradiol. 2004;25:1676–1679. 67. Kimber J. Cerebral venous sinus thrombosis. QJM. 2002;95:137–142. 68. Dainer HM, Smirniotopoulos JG. Neuroimaging of hemorrhage and vascular malformations. Semin Neurol. 2008;28:533–547.

wwwwwwwwwwww

Carotid and Vertebral Circulation: Clinical Applications

17

Sugoto Mukherjee and Max Wintermark

Introduction Magnetic resonance angiography (MRA) is a broad term referring to a plethora of techniques, all of which can be used to directly image flow in arteries and veins in the extra- and intracranial circulation. These techniques include 2D and 3D time of flight (TOF), 2D and 3D phase contrast (PC), newer techniques such as 3D TOF with multiple overlapping thin slab acquisition (MOTSA), as well as contrast-enhanced MRA (CEMRA). Although intra- and extracranial MRA is a relatively young technique, many of the above-mentioned techniques are mature enough to be a staple part of routine, clinical MR protocols. Hence, their clinical applications, the technical background and the various pitfalls require familiarity from the neuroradiologists. This chapter summarizes the various established as well as emerging clinical uses of these techniques. Understanding the inherent complementary nature of many of the techniques is essential to maximize the diagnostic information for every individual patient. As the physics and the technical background of these techniques have been discussed elsewhere in this book, we stress on their clinical applications.

Extracranial Carotid Atherosclerosis Stroke is the third leading cause of death in the USA, behind heart disease and cancer, and the second leading cause of death worldwide. The effects of stroke are devastating, with approximately 20% mortality and significant morbidity among the 80% survivors. Carotid artery atherosclerosis is a significant cause of many of these strokes and cerebrovascular disease. Hence, carotid artery imaging is of vital impor-

S. Mukherjee, MD () • M. Wintermark, MD Neuroradiology Division, Department of Radiology, University of Virginia, Charlottesville, VA, USA e-mail: [email protected]

tance in detection and quantification of disease, as well as in presurgical mapping and posttreatment surveillance. Although it is beyond the scope of this chapter to detail the entire pathophysiology of cerebrovascular ischemic events, many of these are secondary to vascular occlusion from embolic events, propagation of a thrombus, or local perfusion failure. Current understanding of the significance of carotid artery stenosis has been largely shaped by landmark clinical trials, such as the North American Symptomatic Carotid Endarterectomy Trial (NASCET), the European Carotid Surgery Trial (ECST), and the Veteran’s Administration Symptomatic Stenosis Trial (VASST). Although there are controversies associated with each of the above studies, all of these have concluded the beneficial effects of carotid endarterectomy in symptomatic patients with ipsilateral high-grade carotid stenosis (70–99%). This has prompted a search for easy to perform, reproducible and accurate, noninvasive techniques to image the degree of carotid stenosis, given the known risks with invasive catheter angiography [1, 2]. An overview of these landmark studies is essential to understand the background, the controversies and predict some of the future directions in carotid imaging. The primary aim in the NASCET study was to determine whether carotid endarterectomy in combination with the best medical therapy was superior to the best medical treatment alone in patients with carotid stenoses and transient cerebral ischemia or partial stroke. Unlike multiple previous attempts, the study goal was accomplished by a large, randomized, multicenter trial with strict inclusion and exclusion criteria. One important exclusion criterion was that of a tandem stenosis, a finding in 2% of this population. This study demonstrated a clear benefit of carotid endarterectomy in symptomatic patients with 70–90% stenosis, prompting stopping the NASCET study in this subset of patients. Additionally, the benefits were directly related to the severity of the stenosis and were greater for higher-grade stenosis than lower ones. The NASCET study concluded that there was a 17% ± 3.5% risk reduction in any ipsilateral stroke in patients treated with carotid endarterectomy and optimal medical care.

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_17, © Springer Science+Business Media, LLC 2012

225

226

S. Mukherjee and M. Wintermark

the distal segment. The ECST method requires an assumption of the true lumen, a potential source of error [1, 2]. Role of carotid endarterectomy in asymptomatic patients is unclear. The CASANOVA (Carotid Artery Stenosis with Asymptomatic Narrowing: Operation Versus Aspirin) concluded that surgical treatment for asymptomatic patients with less than 90% stenosis was not recommended. However, the American Heart Association Multidisciplinary Consensus Group has suggested that in asymptomatic patients with greater than 75% stenosis, it may be acceptable to perform CEA. The Asymptomatic Carotid Surgery Trial (ACAS) further contributed to the controversy by showing that stenoses greater than or equal to 60% of luminal diameter benefit from endarterectomy, with a 6% absolute risk reduction over 5 years. In recent years, percutaneous transluminal angioplasty (PTA) and/or carotid stenting have emerged as viable alternatives in patients, especially those who are not surgical candidates [1, 4].

Fig. 17.1 Graphic depicting the NASCET calculation with a comparative catheter angiogram demonstrating the measurements

For 50–69% stenosis, carotid endarterectomy leads to moderate reduction in stroke risk, with questionable benefit in patients with less than 50% stenosis. These results were confirmed in the other trials [1–3]. Different methods have been utilized in measuring carotid stenosis, with variations in technique and accuracy. The more common methods include the NASCET method, which, measures the residual lumen diameter at the most stenotic portion of the vessel and compares this to the lumen diameter of the normal internal carotid artery distal to the stenosis. The ECST method measures the lumen diameter at the most stenotic portion of the vessel and compares this to the estimated probable original diameter at the site of maximum stenosis. The Common Carotid (CC) method measures the residual lumen diameter at the most stenotic portion of the vessel and compares this to the lumen diameter in the proximal common carotid artery. Despite the differences, the results of all three methods have a nearly linear relationship to each other and provide data of similar prognostic value. The NASCET method has become the standard, being adopted by surgical community. Even though the NASCET method was devised for conventional angiography, it can be used for noninvasive methods such as MR and CT angiography. The NASCET trial uses this formula: [1 − (stenosis diameter/normal distal lumen diameter) × 100], to determine the percentage stenosis (Fig. 17.1). Note should be made that the same degree of stenosis is quantified as a higher percentage stenosis when measured by the ECST or CC methods than when measured by the NASCET method as the maximal stenosis is at the carotid bulb, which is always wider than

Overview of MRA Techniques Previously, angiography had been the well-defined practice standard for carotid disease evaluation, also forming the basis of the NASCET and ECST trials. The inherent risks of catheter angiography when evaluating patients for endarterectomy have led efforts to develop and use noninvasive techniques to screen potential surgical candidates, as well as to establish an entirely noninvasive approach to the preoperative imaging of patients with symptomatic carotid arterial disease. Although MRA has made rapid strides, there’s still a role for catheter angiography and CT angiography, especially in patients with greater than 70% stenosis. In this segment, we review the MRA techniques, their relative advantages and limitations, and their applications to clinical practice. MR angiography techniques used for extracranial carotid imaging can be broadly categorized into contrast-enhanced and noncontrasted time-of-flight and phase-contrast techniques. Both TOF and PC techniques may use 2D or 3D Fourier transformation for image reconstruction to get 2D and 3D images. Variations of these techniques include inversion recovery techniques, MOTSA, and adiabatic fast scanning. Quantitative 2D advances PC techniques have been used to provide velocity information in carotid and vertebrobasilar disease. CEMRA is, however, the most frequently used MRA technique for imaging the extracranial carotid.

Time-of-Flight MRA Techniques Time-of-flight MRA generates images from the difference in signal due to the wash in of blood with respect to the stationary tissue. The stationary tissue magnetization and signal is

17 Carotid and Vertebral Circulation: Clinical Applications

Fig. 17.2 Thin section 3D TOF MRA (MOTSA) of the extracranial neck demonstrates excellent flow related enhancement and uniform signal intensity within the bilateral common carotid bifurcations (arrows), almost comparable to contrast-enhanced MRA of the neck

further decreased due to the repeated application of section sensitive gradient – refocusing pulses with short repetition time (TR). Modern 2D TOF MRA uses an additional venous presaturation pulse, with the images being acquired in thinner 1.5 mm sections and displayed using maximum intensity projection (MIP) techniques. When compared to the 2D TOF MRA described above, the 3D TOF MRA involves additional excitation of a 30–60 mm thick slab of tissue and then separating it out in thin slices (1 mm or less), by phase encoding in the same direction. The final images are reconstructed using a 3D Fourier transformation. As with 2D TOF MRA, a 90-degree venous presaturation pulse is applied to reduce venous flow related artifacts. The flow in 3D TOF MRA is maximized by orienting the imaging slab perpendicular to the direction of flowing blood (axial in the neck), optimizing the TR and the flip angle and using a transmit/receive head coil to minimize saturation of flowing spins outside the imaging volume [3]. Recent advances in TOF MRA techniques have tried to address the issues of flow saturation effect and coverage. A new technique, known as MOTSA, combines elements of 2D and 3D methods, for improved flow related enhancement while acquiring thinner sections (Fig. 17.2). In this technique, a series of overlapping 3D gradient-recalled echo image sets are acquired in the axial plane in a sequential or interleaved fashion. Signal loss due to intravoxel dephasing is avoided by using smaller gradient moments and shorter TE’s. Another technique, targeted at reducing saturation related signal loss is tilted optimized nonsaturating excitation (TONE). TONE uses ramped radiofrequency (RF) pulses so that the flip angle applied to the moving spins increases over the imaged volume. This technique can be combined with MOTSA to reduce scanning times [3, 5].

227

Fig. 17.3 3D TOF MRA images demonstrate the easily recognizable ill-defined hypo intensity (filling defects) consistent with recirculation artifacts in the bilateral carotid bulbs

Limitations of TOF MRA Techniques Cervical TOF MRA techniques are plagued with saturation related artifacts, inherent to their technique. Flowing blood saturation may occur due to vessel shape, orientation, course, and slow and turbulent flow as well as slice thickness. 3D TOF MRA is even more prone to saturation effects, as it excites a thicker slab of tissue. It is less sensitive to slow flow. Saturation effects are of a lesser concern in 2D TOF MRA due to thinner slices. However, 2D TOF MRA is more prone to T2* effects (from air and bone interfaces adjacent to the skull base) and complex flow, classically seen at carotid bifurcations, stenoses, and vascular bends. Overestimation of carotid stenosis is a major problem with both TOF MRA techniques. 2D TOF MRA suffers to a greater degree because of smaller voxel size, longer echo times (TEs), and a greater slice select gradient strength/ duration causing increased intravoxel dephasing. 3D TOF MRA, on the contrary, is prone to overestimation due to the saturating effects of slowly flowing blood. The difficulty in distinguishing between a significant stenosis and a complete occlusion, both appearing as “flow void”/“flow gap” or apparent vascular interruption, is a known dilemma with TOF MRA imaging. Although this is more often seen with 2D TOF MRA, such apparent vascular interruption has also been noted with high-grade stenosis on 3D TOF MRA. This flow void artifact has been confirmed to represent a highgrade stenosis, usually above 70% severity on digital subtraction angiography (DSA) [3]. Recirculation artifacts at the carotid bulb due to flow separations, reversal, and secondary flows represent a common source of artifacts (Fig. 17.3). These artifacts appear as ill-

228

defined areas of signal loss in the proximal internal carotid artery. This artifact can be identified by its poorly defined margins, preservation of posterior wall contour, a less dark defect and lack of any caudal or cephalic extension. A unique and easily recognizable artifact seen with MOTSA imaging is the so-called “venetian blind” artifact due to variation in signal intensity at the slab interfaces. Sufficient overlapping of the volumes, by reducing the slice thickness, increasing the position in the imaging volume occupied by the excitation pulse, and decreasing the field of view can avoid this artifact [6]. As the source images for TOF MRA are T1 weighted, any substance with a T1 shortening effect such as lipid, blood products and gadolinium may be mistaken for flow related enhancement. Other artifacts include those due to local magnetic field inhomogeneities induced by air, bone, or metal interfaces, leading to dephasing and loss of signal, as well as motion-related artifacts.

Efficacy of TOF MRA Techniques in Carotid Atherosclerosis The studies that examined the efficacy of TOF MRA techniques have reported various results, in part due to the variability in the assessed imaging techniques. A number of these studies have compared 2D TOF with Doppler ultrasound. Many of these have suggested the combined usage of 2D TOF MRA and Doppler sonography as a viable alternative to DSA, with the latter being used for cases of where MRA and Doppler ultrasound were discordant. Doppler studies have better sensitivity and specificity as compared to 2D TOF MRA in cases where 50% stenosis was used as the cutoff. Others studies have found a greater sensitivity for TOF MRA for greater than 70% stenosis. 100% correlation with DSA exists when both ultrasound and TOF MRA are in agreement. Disease distal to the bifurcation as well as tandem stenosis has not been addressed in these studies. Combining the 2D and 3D techniques, with 2D better for slow flow and identifying string signs, and 3D offering better spatial resolution for wall irregularity, can address many of the major issues raised by TOF MRA [7, 8].

Contrast-Enhanced 3D Fourier Transform MRA Techniques Contrast-enhanced 3D Fourier transform MRA is the workhorse for stroke imaging work up and evaluation for carotid bifurcation disease. This has been made possible by both software and hardware upgrades, allowing for rapid dynamic acquisition of complete spatial frequency information, permitting very short sequential 3D acquisitions.

S. Mukherjee and M. Wintermark

Contrast-enhanced MRA uses the T1 shortening effect of paramagnetic contrast, thereby overcoming the various flow related saturation and spin dephasing shortcomings of time-of-flight MRA techniques. This also makes it possible to cover a longer segment of the vessel using a smaller number of sections as it avoids the need to image the vessel perpendicular to the flow direction, further decreasing the time for acquisition. The basic premise of this technique is the acquiring of data from the center of the k-space in the region of interest post infusion of a paramagnetic contrast material, using fast 3D gradient echo pulse sequence with short repetition. Timing such an acquisition is critical as there is a very short optimal time window of between 5 and 10 s between the arterial and venous enhancement phases. Acquiring the center of the k-space before the arterial peak is reached reduces the SNR and produces artifacts. Conversely, delayed acquisition, in addition to decreased SNR, results in venous enhancement, which complicates image interpretation. To achieve maximum signal to noise ratio, peak arterial enhancement must coincide with acquisition of the center of the k-space. Several techniques have been devised to ensure this. The main focus of these techniques is to capture the first pass of gadolinium by coordinating the initiation of the MRA sequence. To capture the first-pass arterial enhancement, the transit time of contrast agent from the injection site to the neck and from the arterial phase to the venous phase should be known. A “best guess” approach, although acceptable in many cases requires infusion of a larger volume of contrast dose over a longer time period to compensate for potential timing errors. An alternative method is to perform real-time fluoroscopic triggering using a gradientrecalled echo fluoroscopic sequence. When coupled especially with elliptic centric view ordering, this can provide images with high spatial resolution and venous suppression. This technique provides a quicker trajectory through the contrast portions of 3D k-space and therefore provides better separation between the contrast enhancement within the arteries and the veins compared with standard centric ordering [3, 8, 9]. A newer technique involves the use of a “time-resolved” 2D or 3D sequence to eliminate the need for exact timing of contrast-agent bolus. This involves rapid and repeated acquiring of images during the passage of contrast bolus with retrospective evaluation of k-spaces. Extensive postprocessing is then used to separate the signal of the arteries from the background. Advantages include obtaining real-time flowrelated information, making it a possible alternative to digital subtraction angiography. Spatial resolution is a casualty of this technique. Modifications of these include a combination of parallel imaging and time resolved imaging with stochastic trajectories. An important aspect in acquiring contrast-enhanced MRA images is the dosing and manner of contrast injections. A power injector ensures rapid and constant delivery of

17 Carotid and Vertebral Circulation: Clinical Applications

contrast. With the development of better bolus timing techniques and phase ordering schemes, it is possible to obtain good-quality images with 0.1–0.2 mmol/kg of contrast material. A saline flush following the injection of contrast agent allows to achieve a more concentrated bolus. The risk of nephrogenic systemic fibrosis in patients with renal impairment has become a critical issue and has increased our awareness of contrast dosage, type and accumulated doses over a lifetime. Breath holding techniques are especially important in elliptical-centric phase-encoding techniques. Coil selection is another very important parameter that affects the SNR and anatomic coverage. Correct placement of the coil is desirable to cover the aortic arch, origins of the great vessels, and carotid siphons in addition to carotid bifurcations. The known signal drop-off at the aortic arch level with routine head and neck coils can be avoided by using a body coil at the price of decreased spatial resolution for carotid bifurcations. 3 T scanners represent an opportunity for better MRA images. 3 T scanners almost double the SNR and contrast to noise ratio when compared to 1.5 T scanners. High spatial resolution MRA can be performed on 3 T scanners with doses as low as 0.047 mmol/kg without compromising image quality, acquisition, or resolution. Better neurovascular coils can allow the entire course of the carotid arteries to be imaged during a single scan. As expected, the problems encountered with 3 T scanners include signal loss due to T2* dephasing, which can be corrected by decreasing TE, voxel sizes, and the injection rates so as to decrease the gadolinium concentration [10]. Further MRA advancements are forthcoming with the increased availability of 3 T scanners and parallel imaging [3, 11].

229

Fig. 17.4 Short segment irregular high grade (>70% stenosis) is nicely demonstrated on contrast-enhanced MRA, CTA, and catheter angiography [arrows in (a), (b), and (c) respectively]. Both the focal stenosis and the immediately proximal ulceration on the MRA match very closely to the similar lesions seen on the catheter angiogram and the CTA

angiography, enhanced MR angiography, and DSA for degree and length of stenosis with CTA. An extensive metaanalysis of the published data on MRA and carotid artery disease till November 2006 was performed by Debrey et al. [11], which concluded that TOF MRA and contrast-enhanced MRA showed high accuracy for the detection of high-grade carotid stenosis, with contrast-enhanced MRA having the edge over TOF MRA but only poor (TOF MRA) to fair (contrast-enhanced MRA) sensitivity for moderately severe stenosis. It was recommended that a second noninvasive imaging study (Doppler ultrasound, CTA) be obtained to confirm the degree of stenosis in such circumstances.

Efficacy of Contrast-Enhanced MRA in Carotid Atherosclerosis Limitations of Contrast-Enhanced MRA Contrast-enhanced MRA has been found to be comparable and even superior to TOF MRA, CTA and DSA in several studies. Contrast-enhanced MRA is definitely superior to 3D TOF MRA with 100% and 90% sensitivity and specificity when compared to 95.5% and 87.2%, respectively with 3D TOF MRA, using rotational angiography as the gold standard (Fig. 17.4). Rotational angiography represents the current gold standard, scoring over DSA, due to increased number of projections. Contrast-enhanced MRA offers several advantages over traditional TOF techniques with higher quality images that are less prone to artifacts. Contrastenhanced MRA produces a reproducible three-dimensional image of the carotid bifurcation with good sensitivity for detecting high-grade carotid stenosis. CT angiography is a fairly robust technique with comparable results to MRA. There is good correlation between CT

Although contrast-enhanced MRA scores over carotid duplex ultrasound in many studies, it does have its own limitations. Some of the other inherent disadvantages of MR scanning pertain to MRA scanning such as difficulty in performing the scans when the patient is critically ill, unable to lie supine, or has claustrophobia, a pacemaker, or ferromagnetic implants.

Vertebral Artery Atherosclerosis Posterior circulation stroke accounts for a fifth of all strokes. One fourth of patients with posterior circulation ischemic stroke are thought to have vertebral artery stenosis with artery-to-artery embolism being the likely mechanism for their stroke. More importantly, recent studies have shown

230

that patients with acute posterior circulation stroke and transient ischemic attack (TIA) have a similar, or perhaps even higher, rate of recurrence compared with patients with anterior circulation stroke, highlighting the need for effective secondary prevention. However, optimal management of vertebral artery stenosis remains uncertain. Angioplasty and stenting are preferred over surgical revascularization for vertebral artery stenosis because of difficult surgical access. Given the invasive nature of catheter angiography, more so in the vertebrobasilar circulation, accurate noninvasive diagnosis of vertebral stenosis is essential. As with carotid disease, the choices are between MR angiography, CT angiography, and Doppler ultrasound. Contrast MRA scores over TOF MRA in most of the studies for evaluation of vertebral artery stenosis. Doppler ultrasound has the least sensitivity with higher specificity due to limited visualization of the vertebral arteries. However, Doppler has been found to be comparable and even better to the other techniques for vertebral artery origin stenosis. Both CT angiography and Doppler show a higher sensitivity for vertebral artery origin stenosis, scoring over MRA. For the rest of the extracranial vertebral artery, both contrast-enhanced MRA and contrast CT angiography (CTA) appear to provide better vessel visualization. Controversy exists about the cutoff of 70–99% stenosis being used for vertebral arteries, analogous to that of carotid artery stenosis. This is a relevant point given that the vertebral artery lumen is much smaller (3–5 mm) and there are no landmark studies such as NASCET for the vertebral arteries, which have validated a similar cut off criteria for vertebral artery stenosis. Some argue the presence or absence of stenosis greater than 50% is important both in identifying vertebral stenosis as a cause of stroke and in identifying potential stenosis for further intervention [12, 13]. This has typically led to many studies using a 50–99% stenosis as the cutoff point. On a practical note, making exact stenosis measurements in vertebral arteries are more difficult as compared to carotid arteries. A recent review of the relevant literature attempted to summarize the imaging findings with the above thoughts in mind evaluating the effectiveness of the various noninvasive imaging modalities, in occluded vessels, 50–99%, 50–70%, and 70–99% stenosis. For occluded vessels, all the different modalities were highly sensitive and specific across the board. Although there were far fewer studies comparing 70–99% and 50–79% stenosis, both contrast-enhanced MRA and CTA score well over Doppler USG with CTA being slightly more sensitive than MRA on this Meta analysis. However, contrast-enhanced MRA was comparable and with better sensitivity when compared to CTA when comparing 50–99% stenosis, which also had the added advantage of increased number of studies. A recent study in the journal Stroke, came to similar conclusions. They graded vertebral artery stenosis into four categories (<50%, 50–69%, 70–99%,

S. Mukherjee and M. Wintermark

and 100%) and concluded that contrast-enhanced MRA had a higher sensitivity and specificity for detecting vertebral stenoses; CTA also has a high sensitivity and specificity, although perhaps slightly less than contrast-enhanced MRA. By contrast, duplex ultrasound has a low sensitivity although a high specificity. The technical factors involved in contrast-enhanced MRA are similar to those in carotid MRA, which has been discussed in detail above. A big advantage of MRA is that it can be combined with MRI of the brain, which has much greater sensitivity for detecting vertebrobasilar distribution infarcts and can be performed as part of a stroke work up.

Carotid and Vertebral Artery Dissection Craniocervical (carotid and vertebral artery) dissections represent a common cause of strokes in younger and middle aged patients. Many of these are being increasingly recognized and detected due to advances in noninvasive imaging as well as an understanding and awareness of the fact that many of these may be spontaneous in nature. An important factor to remember in extracranial dissections is the difference in prognostic and therapeutic implications as compared to intracranial dissections as well as more common extracranial atherosclerotic disease. Carotid and vertebral artery dissections are either posttraumatic or spontaneous in the majority of cases. Spontaneous cases of dissection either have no precipitating history or may have preceding history of trivial trauma, such as contact sports, cervical manipulation or even coughing and vomiting. Posttraumatic dissections are usually seen in the setting of severe blunt head and neck trauma. The association of intrinsic arteriopathies with dissection, including fibromuscular dysplasia, Marfan syndrome, Ehler-Danlos (IV) syndrome, and cystic medial necrosis, is well known. Fibromuscular dysplasia can be found in up to 15% of patients with cervical artery dissection. The association of spontaneous dissections with intracranial aneurysm, a widened aortic root, arterial redundancies, and migraine also suggests an indirect evidence of an arteriopathy. Dissection results from hemorrhage into the vessel wall usually within the media. Noninvasive angiographic techniques such as MR and magnetic resonance angiography (MRA) as well as computed tomography angiography (CTA) have shown accurate or even superior results compared with DSA. The main advantage of these techniques is the direct visualization of the vessel wall confirming the intramural hematoma on the T1- and T2-weighted images. Approximately 60% of spontaneous internal carotid artery dissections are solely extracranial, 20% are both intracranial and extracranial, and 20% are primarily intracranial. Subarachnoid hemorrhages typically occur in the setting of

17 Carotid and Vertebral Circulation: Clinical Applications

Fig. 17.5 Axial CT angiography images show the severely narrowed lumen of the right distal cervical ICA [arrow in (a)]. See contralateral ICA for comparison. The fat suppressed T1 weighted axial MR reveals hyperintense crescent within the wall of the right ICA, pathognomonic of intramural hematoma following a dissection [arrow in (b)]. Near occlusive luminal narrowing is evident on the contrast-enhanced MRA within the mid and distal cervical right ICA [arrow in (c)]

intracranial extension of dissection. Spontaneous carotid dissections are typically located a few centimeters above the carotid bifurcation whereas spontaneous vertebral artery dissections involve the V3 segment, adjacent to the lateral masses of the C1 and C2 vertebral bodies after the artery has exited the transverse foramina and before it has entered the foramen magnum. Approximately 10% of vertebral artery dissections are solely intracranial and another 10% likely originate from an extracranial segment of the artery and extend to the intracranial portion of the vertebral artery. Most dissections, however, end at the skull base. Recurrent spontaneous dissections should raise the flag of an underlying arteriopathy. The imaging findings in craniocervical dissection are quite varied and also depend on the modality used. Catheter angiography demonstrates the changes of luminal dilatation or narrowing, pseudoaneurysmal formation, intimal flap and double lumen appearance. Dissection represents an entity in which MR imaging represents a definite improvement over catheter angiography as it allows depiction of the wall hematoma in addition to the lumen. Cross-sectional MR sequences allow accurate visualization of the intramural hematoma, seen as a crescentic rim of hyperintense signal on T1 and T2 weighted sequences (Fig. 17.5). Typically, these are most obvious a few days after the onset of dissection due to the blood products being in the intracellular methemoglobin stage, and persist for weeks to months. In the acute and

231

hyperacute stages, the hematoma may be isointense, making it difficult to diagnose. Another advantage of MRI and MRA over catheter angiography includes the findings of increased diameter of the affected vessel, due to the presence of the intramural hematoma. This finding is not well apparent on catheter angiography as they image solely the arterial lumen, and not the external diameter. MRA also scores over catheter angiography in the depiction of luminal narrowing or occlusion, as its shows that the arterial narrowing is not simply due to normal variation or atherosclerotic disease by also depicting the intramural hematoma. MRI along with MRA readily identifies the intimal flap and double lumen, which is often seen on the MRA axial source and on the contrastenhanced spoiled gradient recalled acquisition images. A specific sign on dissection on catheter angiography, known as “pearl and string” sign, consisting of a long segment of arterial narrowing in conjunction with regions of arterial dilation, can also be seen on MRA. All these above findings have been validated on multiple prospective studies comparing MRI and 3DFT TOF MRA to conventional angiography with MR imaging and MRA demonstrated an excellent sensitivity and specificity, approaching and even crossing 90%. Many of these studies have concluded that MR imaging/MRA represents the most efficacious method of evaluation of a patient with a clinical suspicion of dissection, and is reliable for following the vascular response to treatment [14, 15]. As with all MR techniques, there have been a few welldescribed pitfalls and limitations of MR imaging and MRA in the evaluation of craniocervical dissection, few of which are discussed below. The perivascular venous plexus surrounding the vertebral arteries can sometimes mimic the changes of vertebral dissection. Addition of a caudal presaturation pulse helps in distinguishing this from a true extracranial vertebral artery dissection. Another common cause of false positive study includes mistaking the periarterial fat for the intramural thrombus. This is, however, easily avoidable by using axial T1 fat saturation techniques. Sometimes, turbulent flow with artifacts can mimic an intimal flap, frequently seen with proximal high-grade stenosis. This can be avoided by remembering to image the entire volume of the specific vessel. The tortuous course of the vertebral artery, such as the horizontal V3 segment of the vertebral artery can make diagnosis difficult. Other usual limitations of TOF MRA such as in plane saturation and flow related artifacts should always be kept in mind. Some of these pitfalls specific to TOF techniques can be avoided by using contrastenhanced MR angiography. Contrast-enhanced MRA can be rapidly performed using elliptical centric phase encoding, offering more reliable assessment of vascular stenosis due to lack of signal loss within the vessel lumen. False negative studies include missing the abnormal arterial segment in the imaged volume, a confounding coexisting

232

S. Mukherjee and M. Wintermark

hyperintense intraluminal thrombus as missing the intramural hematoma in the very early stages (first couple of days) when it may be hypointense on both T1 and T2 weighted sequences. Few other caveats need to be mentioned here. These include the possibility of missing the intimal flaps on contrast-enhanced MRA, because of the high signal of the flowing blood. In cases of suspected severe luminal narrowing, a 2D TOF may be more helpful. Phase-contrast techniques may be useful in cases of severely limited flow, because they do not demonstrate the high signal intensity of methemoglobin and allow selection of the velocity of blood to be imaged. MRA sequences obtained for suspected arterial dissection should always be read in conjunction with their source images as well as an additional axial T1 fat sat sequence obtained as part of a dedicated dissection protocol. Typically, the signal intensity of the flow-related enhancement of blood in the residual lumen is mildly brighter than that of the intramural hematoma, which allows the two entities to be distinguished [14, 15].

Fibromuscular Dysplasia Fibromuscular dysplasia is a nonatherosclerotic, noninflammatory vascular disease that most commonly affects the renal and internal carotid arteries but has been described in almost every arterial bed in the body. It consists of smooth muscle hyperplasia or thinning, proliferation of fibrous tissue, and elastic fiber destruction. The pathological classification scheme for fibromuscular lesions is based on the arterial layer – intima, media, or adventitia – in which the lesion predominates. Macroaneurysms and dissections are the usual complications of fibromuscular dysplasia. Multiple angiographic studies indicate that the cervical internal carotid artery is diseased in approximately 95% of cases of cephalic fibromuscular dysplasia, with bilateral involvement in 60–85%. When the carotid arteries are involved, the cervical segment C1-C2 most often is affected. Intracranial disease is extremely rare and is almost always associated with cervical carotid disease; it involves the supraclinoid segment of the internal carotid arteries and the proximal middle cerebral arteries. Catheter angiography has been the mainstay for diagnosing fibromuscular dysplasia. Consequently, the three main patterns on carotid FMD involvement are based on their angiographic appearance. These are as follows: • Type 1: This is the “classical” most common form, seen in almost 80–85% of patients with FMD. Angiography reveals the typical string-of-beads appearance, with alternating segments of stricture and dilation. This type usually is a result of medial fibroplasia of the arterial wall (Fig. 17.6).

Fig. 17.6 Contrast-enhanced MRA of the neck depicts alternate segments of narrowing and dilatation (arrows) involving mid cervical segments of bilateral internal carotid arteries, in this patient with known renal artery stenosis and fibromuscular dysplasia, strongly suggestive of type I fibromuscular dysplasia lesion

• Type 2: Typically seen with the intimal form, in 6–12% of patients. The imaging appearance is that of a long segment tubular stenosis. • Type 3: This uncommon form seen in 4–6% of patients is characterized by involvement of only one side of an artery, causing small multiple diverticulas of the arterial wall. Isolated reports of web-like septum have also been described, and probably represent an atypical presentation of FMD. Although more and more FMD cases are diagnosed on 2D and 3D TOF neck MRA, these are difficult to diagnose, more so in patients with classic alternating areas of stenosis and dilation. This has been perceived to be due to turbulent flow through the disease segment, with dephasing artifacts. Also, the stenotic bands within the involved segments may be mistaken for a common slice misregistration artifact due to motion seen on TOF MRA. Although 3D TOF has greater spatial resolution than the 2D TOF technique, the disturbed flow is also likely to be problematic with this technique. Given these limitations, the sensitivity or specificity of MRA for this arteriopathy is expected to be lower than conventional angiography and even CT angiography. When evident, the imaging findings on MRA do demonstrate the string of beads appearance, as well as long-segment tubular stenosis or ovoid-shaped outpunching. A big plus however, is the beneficial effect of MRA in detecting associated intracranial

17 Carotid and Vertebral Circulation: Clinical Applications

233

Fig. 17.7 Left-sided subclavian steal syndrome, with a proximal left subclavian artery stenosis [arrows in (a)], on CE-MRA image. Sequential frames from time resolved MRA elegantly displays the reversal of flow in the left vertebral artery [arrows in (b, c)]. Additional

TOF source images from the upper and mid neck shows lack of flow related enhancement in the left vertebral artery [arrows in (d, e)] due to the superiorly placed saturation band, suppressing flow from both the venous system, as well as the retrograde flow in the left vertebral artery

aneurysms, findings seen in almost 21–51% of patients with fibromuscular dysplasia [16, 17].

encoding. An added advantage of cine phase-contrast MRA is quantification of the reversed flow, by measuring the blood flow changes in both the vertebral arteries and the basilar artery. Other approaches to image subclavian steal syndrome have included a combination of a 2D TOF “localizer” sequence and a contrast-enhanced MRA, as well as using dynamic time-resolved echo-shared angiographic technique with parallel imaging, which provides anatomic and dynamic flow pattern information, and may obviate the need for catheter angiography altogether [18, 19] (Fig. 17.7).

Subclavian Steal Syndrome Subclavian steal syndrome refers to subclavian artery stenoocclusive disease proximal to the origin of the vertebral artery, resulting in flow reversal in the vertebral artery from the contralateral vertebral artery across the vertebral basilar junction and in a retrograde direction ipsilateral to the stenosis or occlusion. This condition can cause episodic brainstem ischemia or stroke, with occasional arm claudication, typically secondary to arm exercise. The imaging diagnosis of subclavian steal syndrome hinges on demonstration of a proximal subclavian artery stenosis or occlusion along with reversal of flow in the ipsilateral vertebral artery, which can be elegantly demonstrated on Doppler US and conventional angiography. Although the subclavian stenosis can be imaged on conventional MRA sequences, the flow reversal phenomena require additional manipulation of the TOF or PC MRA techniques. Imaging subclavian steal syndrome on 2D TOF requires removing or repositioning the superior saturation pulse (placed to suppress signal from the venous system), allowing the reversed flow in the vertebral artery to be visualized. A further refinement of this technique involves using two saturation pulses superior and inferior to the image acquisition, which will just show the reversed flow in the vertebral artery. Phase-contrast 2D MRA demonstrates the reversed flow as high signal intensity, when acquired with superior to inferior flow

Intracranial Carotid Artery Atherosclerosis Atherosclerotic stenoocclusive disease of the intracranial arteries is a major cause of ischemic stroke, with atherosclerotic occlusive disease involving the carotid siphon being second in frequency only to that of the carotid bifurcation in the neck. Numerous previous studies have shown that hypertension, and diabetes mellitus are associated with more extensive intracranial atherosclerosis. Compared with extracranial stenosis, intracranial stenoses do not correlate as well with the typical atherosclerotic risk factors for peripheral and coronary vascular disease (i.e., male sex and hypercholesterolemia). Also, patients with intracranial disease are usually significantly younger. MRA of the intracranial internal carotid artery represents a challenge given the extreme tortuous course and altered flow dynamics in addition to susceptibility related artifacts from the adjacent sphenoid sinus. Potential acute changes in flow direction and velocity can cause severe phase

234

S. Mukherjee and M. Wintermark

Fig. 17.8 3D TOF MRA of the head demonstrates extensive lenticulostriate and leptomeningeal collaterals [black arrows in (a, b)] along with severe occlusive narrowing of the supraclinoid internal carotid artery extending to the proximal middle cerebral arteries, [white arrows

in (a, b)] suggestive for a moyamoya disease. The black arrow in (b) refers to the classic “puff of smoke” appearance of the thalamoperforator and posterior choroidal vessels as described in the literature

dispersion artifacts, some of which can be decreased by using shorter gradient echo times, small voxel size, flow compensation measures, and partial echo sampling. Many of these are used routinely nowadays. 3D TOF MRA is currently the standard pulse sequence to evaluate the intracranial carotid circulation. Other technical advances in TOF imaging of the intracranial circulation include imaging at higher 3 T strengths with better signal to noise ratio. This along with parallel imaging technique results in substantial improvement in spatial resolution in 3D TOF MRA imaging [3, 20].

arteries, which subsequently extends into the proximal segments of the middle and anterior cerebral arteries with secondary collateralization. This collateral flow is through the adjacent perforators in the basal ganglia as well as leptomeningeal vessels. Suzuki and Kodama have classified the angiographic pattern into six stages, with Stage I representing narrowing of the internal carotid bifurcation, and stage VI having the classic Moyamoya pattern with extensive pial collaterals through the external carotid. The clinical presentation tends to be related to the age when the disease develops. In children, the first symptom of Moyamoya disease is often recurrent transient ischemic attacks (TIA, commonly referred to as “ministrokes”), or stroke, frequently accompanied by muscular weakness or paralysis affecting one side of the body, or seizures, progressing to a severe vegetative state. In the adult population, patients often present with hemorrhagic strokes due to recurrent blood clots with intraventricular and subarachnoid hemorrhages. The imaging appearance is fairly characteristic, regardless of the etiology. On MR and catheter angiography, the imaging appearance depends on the severity of the disease. Initially, there is narrowing of the supraclinoid internal carotid artery and the proximal vessels of the circle of Willis followed in the later stages by development of lenticulostriate and thalamoperforator collaterals (Fig. 17.8). Findings of transdural and transosseous external carotid to internal carotid leptomeningeal collaterals are seen in the longstanding cases. MRA, although having certain limitations has been found to accurately demonstrate the supraclinoid and proximal anterior and middle cerebral artery narrowing. Initial studies by Yamada and colleagues comparing 3D TOF MRA have proved this with overestimation of stenosis in a few cases.

Moyamoya Disease and Moyamoya Syndrome Moyamoya disease represents a rare arteriopathy, characterized by progressive occlusive changes of the vessels of the Circle of Willis. Patients with the characteristic moyamoya pattern who also have well-recognized associated conditions (such as neurofibromatosis I, sickle cell disease, postradiation, Down’s syndrome, renal artery stenosis, etc.) are categorized as having the moyamoya syndrome, whereas patients with no known associated risk factors are said to have moyamoya disease. By definition, the pathognomonic arteriographic findings are bilateral in moyamoya disease, although the severity can differ between sides. Patients with unilateral findings have the moyamoya syndrome, even if they have no other associated risk factors. However, contralateral disease eventually develops in up to 40% of patients initially presenting with unilateral findings. The “moyamoya” term in Japanese refers to the “puff or spiral of smoke” seen best on catheter angiography secondary to the prominent cloud-like lenticulostriate and thalamostriate collaterals. This condition is characterized by severe occlusive narrowing of the distal/supraclinoid internal carotid

17 Carotid and Vertebral Circulation: Clinical Applications

235

Fig. 17.9 A 12-year-old with sickle cell disease and history of multiple recent and remote strokes. 3D TOF of the head reveals severe stenosis of the bilateral supraclinoid ICA’s [arrows in (a)] extending in to the proximal middle and anterior cerebral arteries with prominent lenticulostriate

branches (secondary moyamoya syndrome). CE-MRA demonstrates the prominent enhancing lenticulostriate collaterals [arrow in (b)]. Multiple old infarcts are well seen on the axial T2 images, involving the right basal ganglia and the right parietooccipital lobe [arrow in (c)]

One limitation of MRI is in identifying the smaller leptomeningeal collaterals. Some of these may be corrected using contrast-enhanced MR angiography to increase the intravascular contrast to counteract the saturation effects, or by using a 2D TOF technique in the later stage of disease to pick up the slower flow and conversely using a 3D TOF technique in the earlier stages of the disease to avoid overestimation of the severity of the stenosis. Another study by Yamada et al. [21] in 2001 utilized high-resolution turbo MR angiography with zero-filling interpolation (ZFI) technique, to overcome some of the above shortcomings. His studies showed high accuracy (98%) of the high-resolution turbo MR angiography technique in the assessment of both stenoocclusive lesions and collateral vessels in moyamoya disease with significantly decreased scan times. Furthermore, a recent study by Fushimi et al., compared 3- and 1.5-T 3D time-of-flight (TOF) magnetic resonance (MR) angiography in patients with moyamoya disease, with special emphasis on the visualization of abnormal net-like vessels (moyamoya vessels) and concluded that moyamoya vessels were better visualized on MIPs obtained with 3 T imaging than on MIPs obtained with 1.5-T imaging [22].

aneurysm formation. Infarctions occur most extensively in the territory supplied by the internal carotid artery, specifically in the region supplied by the distal watershed territories of the anterior and middle cerebral arteries. Focal narrowing of the distal internal carotid artery and the adjacent middle and anterior cerebral arteries sometimes leads to a secondary “moyamoya” pattern, seen in almost 20–40% of patients with sickle cell disease and stroke. MRA findings have been seen to mirror the changes seen on catheter angiography, with several papers on 3D TOF demonstrating vascular occlusions and stenosis, comparable to conventional angiography. Findings commonly seen on MRA include stenosis of the distal internal carotid artery and vessels of the proximal circle of Willis as well as the findings of moyamoya pattern (Fig. 17.9). An important caveat in MRA in these patients is the turbulent dephasing due to anemia, and/or rapid flow can mimic stenosis. The lowest possible TE should be used in these cases if stenosis is suspected. An additional source of information is the finding of multiple dots in the basal ganglia on the MRA source images due to moyamoya. TOF MRA without contrast being noninvasive and carrying decreased risk of precipitating strokes should be the preferred approach in these young patients. An MRA combined with parenchymal MRI appears to be well suited for the neurovascular evaluation of these patients. Other findings in this subset of patients include fusiform aneurysms, and primitive carotid basilar arterial communications [23].

Sickle Cell Vasculopathy Sickle cell disease is a cause of thrombotic stroke secondary to abnormal adherence of sickled red blood cells, causing capillary occlusions and resulting in ischemia and infarctions. It is a primary cause of stroke in African-American children, with stroke incidence being dependent on the Hemoglobin S. There is a 1% annual risk of stroke in patients with sickle cell disease with 11% of these patients suffering an episode of acute cerebral infarction by the age of 20 years. Although, the initial sequelae are at the micro vascular level, subsequent injury to the large vessel walls with intimal proliferation results in a large vessel vasculopathy and

Vascular Compression of the Facial or Trigeminal Nerve Vascular loop compression of the trigeminal and the facial nerve root zone is a common cause of trigeminal neuralgia and hemifacial spasm respectively. The vessels commonly involved are the superior cerebellar artery, posterior inferior

236

cerebellar artery and the basilar artery in trigeminal neuralgia, and the anterior inferior cerebellar artery, posterior inferior cerebellar artery and the vertebral artery in hemifacial spasm. Other causes of trigeminal neuralgia and hemifacial spasm include extraaxial mass lesions in the cerebellopontine angle and intraparenchymal lesions, including multiple sclerosis. Although there is controversy regarding the association of neurovascular compression of the fifth and seventh cranial nerves, there is a definite role of imaging in excluding surgical pathologies and localizing possible treatment targets. Also, even though neurovascular compression has been demon-

Fig. 17.10 A 53-year-old male with left-sided hemifacial spasm. Thin section axial T2 reveals an extremely tortuous and ectatic left distal vertebral artery, deforming the nerve root exit zone of the left facial nerve at the level of the internal auditory canal [arrow in (a)]. The source 3D TOF MRA confirms the tortuous course of the offending vessel, [arrow in (b)] which is the left vertebral artery, corresponding to the T2 image. The MRA image should always be read in conjunction with conventional sequences to evaluate the adjacent brain parenchyma

Fig. 17.11 Axial T2 fat saturated sequence shows a left carotid body paraganglioma, in between the left internal and external carotid arteries [arrow in (a)]. The intensely vascular lesion is also seen in the contrast MRA source study [arrow in (b)]. The neck CE-MRA maximum inten-

S. Mukherjee and M. Wintermark

strated in multiple MRI and MRA based studies, a confounding factor is the similar imaging appearance in asymptomatic patients. The current thinking is that a vessel merely being in contact with root exit zone of the facial nerve is not sufficient to cause symptoms, but rather that the root exit zone must be compressed or deformed by the vessel [24]. MRA plays a complementary role in the evaluation of neurovascular compression. The MRA should be centered in the posterior fossa. Various investigators have used a variation of the 3D TOF MRA technique, known as MR tomographic angiography to better evaluate this condition. This basically consists of using the source data of the conventional 3D TOF MRA technique and reforming it in submillimeter coronal, sagittal, and oblique sections with the window and levels adjusted to allow visualization of both vascular structures and the adjacent brainstem parenchyma as well as nerve-root exit zones. Thin section high resolution T2 weighted sequences serve as important adjuncts (Fig. 17.10). Coronal reformations from both these sequences appear to be the most reliable for demonstrating the nerve-root exit zone of the cranial nerves. Gadolinium-enhanced MR has not been found to be of any additional value [25].

Head and Neck Neoplasms Extracranial MRA plays a limited role in the evaluation of neoplastic lesions in the head and neck. They have their primary role as an adjunct to cross-sectional MR imaging.

sity projection demonstrates the intensely enhancing left carotid body paraganglioma, nestled in between the left external and internal carotid arteries, corresponding to the preoperative catheter angiogram [arrows in (c) and (d)]

17 Carotid and Vertebral Circulation: Clinical Applications

In certain patients with head and neck lesions, MRA in combination with cross-sectional imaging is a valuable diagnostic tool for the detection of vascular involvement, either invasion or encasement or just displacement of the vessels. They can sometimes help in distinguishing skull base and head and neck vascular lesions, in defining their vascularity. MR Angiography does have a role in evaluation of head neck and skull base paragangliomas. Advanced contrastenhanced MRA techniques in combination with conventional MR sequences is superior to conventional MR imaging alone in assess paragangliomas in the head and neck, helping in narrowing the differential diagnosis, as well as obviating the need for catheter angiography in many of these cases. These are also helpful detecting additional paragangliomas, in patients with familial paragangliomas who are prone to present with multiple lesions. The combination of contrastenhanced MRA and conventional MRI have the added advantages of demonstrating the lesion better as well as demonstrating the enlarged feeding arteries in paragangliomas, as well as having lesser image degradation due to pulsation artifacts [25] (Fig. 17.11).

237

6.

7.

8.

9.

10.

11.

12.

13.

Summary Given the rapid advances in MRA techniques over the past decade, it is probably safe to assume that newer techniques are on the horizon. Current research is being based on software/postprocessing advances as well as the increasing availability of higher gradients and higher field scanners. The preference for MRA due to its noninvasiveness and its use as part of a comprehensive stroke workup along with conventional MRI sequences will make it even more widely acceptable. As with all evolving techniques, further studies are needed for validating their accuracy and reliability.

References 1. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. North American Symptomatic Carotid Endarterectomy Trial Collaborators. N Engl J Med. 1991;325:445–453. 2. MRC European Carotid Surgery Trial: interim results for symptomatic patients with severe (70–99%) or with mild (0–29%) carotid stenosis. European Carotid Surgery Trialists’ Collaborative Group. Lancet. 1991;337:1235–1243. 3. Heiserman JE, Masaryk TJ, Aygun N. MR angiography: techniques and clinical applications. In: Atlas SW, ed. Magnetic Resonance Imaging of the Brain and Spine. Philadelphia, PA: Lippincott Williams & Wilkins; 2009:841–881. 4. Carotid surgery versus medical therapy in asymptomatic carotid stenosis. The CASANOVA Study Group. Stroke. 1991;22: 1229–1235. 5. De Marco JK, Schonfeld S, Keller I, Bernstein MA. Contrastenhanced carotid MR angiography with commercially available

14.

15.

16. 17. 18.

19.

20. 21.

22.

23.

24.

25.

triggering mechanisms and elliptic centric phase encoding. AJR Am J Roentgenol. 2001;176:221–227. Ahn KJ, You WJ, Lee JH, et al. Re-circulation artefact at the carotid bulb can be differentiated from true stenosis. Br J Radiol. 2004;77:551–556. DeMarco JK, Huston J 3rd, Bernstein MA. Evaluation of classic 2D time-of-flight MR angiography in the depiction of severe carotid stenosis. AJR Am J Roentgenol. 2004;183:787–793. Anzalone N, Scomazzoni F, Castellano R, et al. Carotid artery stenosis: intraindividual correlations of 3D time-of-flight MR angiography, contrast-enhanced MR angiography, conventional DSA, and rotational angiography for detection and grading. Radiology. 2005;236:204–213. Lim R, Shapiro M, Wang E, et al. 3D time-resolved MR angiography (MRA) of the carotid arteries with time-resolved imaging with stochastic trajectories: comparison with 3D contrast-enhanced bolus-chase MRA and 3D time-of-flight MRA. AJNR Am J Neuroradiol. 2008;29:1847–1854. Bernstein MA, Huston J 3rd, Lin C, Gibbs GF, Felmlee JP. Highresolution intracranial and cervical MRA at 3.0 T: technical considerations and initial experience. Magn Reson Med. 2001;46:955–962. Debrey SM, Yu H, Lynch JK, et al. Diagnostic accuracy of magnetic resonance angiography for internal carotid artery disease: a systematic review and meta-analysis. Stroke. 2008;39:2237–2248. Khan S, Cloud GC, Kerry S, Markus HS. Imaging of vertebral artery stenosis: a systematic review. J Neurol Neurosurg Psychiatry. 2007;78:1218–1225. Khan S, Rich P, Clifton A, Markus HS. Noninvasive detection of vertebral artery stenosis: a comparison of contrast-enhanced MR angiography, CT angiography, and ultrasound. Stroke. 2009;40: 3499–3503. Provenzale JM. MRI and MRA for evaluation of dissection of craniocerebral arteries: lessons from the medical literature. Emerg Radiol. 2009;16:185–193. Provenzale JM, Sarikaya B. Comparison of test performance characteristics of MRI, MR angiography, and CT angiography in the diagnosis of carotid and vertebral artery dissection: a review of the medical literature. AJR Am J Roentgenol. 2009;193:1167–1174. Osborn AG, Anderson RE. Angiographic spectrum of cervical and intracranial fibromuscular dysplasia. Stroke. 1977;8:617–626. Steiger HJ, Turowski B. Fibromuscular dysplasia. N Engl J Med. 2004;351:509–510; author reply 509–510. Virmani R, Carroll TJ, Hung J, Hopkins J, Diniz L, Carr J. Diagnosis of subclavian steal syndrome using dynamic time-resolved magnetic resonance angiography: a technical note. Magn Reson Imaging. 2008;26:287–292. Turjman F, Tournut P, Baldy-Porcher C, Laharotte JC, Duquesnel J, Froment JC. Demonstration of subclavian steal by MR angiography. J Comput Assist Tomogr. 1992;16:756–759. Laub GA, Kaiser WA. MR angiography with gradient motion refocusing. J Comput Assist Tomogr. 1988;12:377–382. Yamada I, Nakagawa T, Matsushima Y, Shibuya H. High-resolution turbo magnetic resonance angiography for diagnosis of moyamoya disease. Stroke. 2001;32:1825–1831. Fushimi Y, Miki Y, Kikuta K, et al. Comparison of 3.0- and 1.5-T three-dimensional time-of-flight MR angiography in moyamoya disease: preliminary experience. Radiology. 2006;239:232–237. Moritani T, Numaguchi Y, Lemer NB, et al. Sickle cell cerebrovascular disease: usual and unusual findings on MR imaging and MR angiography. Clin Imaging. 2004;28:173–186. Tien RD, Wilkins RH. MRA delineation of the vertebral-basilar system in patients with hemifacial spasm and trigeminal neuralgia. AJNR Am J Neuroradiol. 1993;14:34–36. Neves F, Huwart L, Jourdan G, et al. Head and neck paragangliomas: value of contrast-enhanced 3D MR angiography. AJNR Am J Neuroradiol. 2008;29:883–889.

wwwwwwwwwwww

Thoracic Aorta Emily Ward and James C. Carr

Introduction Diseases of the thoracic aorta are a significant cause of morbidity and mortality and can result in potentially catastrophic consequences, if the condition remains undiagnosed. Conventional digital subtraction angiography (DSA) has been the gold standard for imaging the thoracic aorta for many years; however, this is associated with well-recognized complications due to its invasive nature [1] and is essentially a projectional two-dimensional (2D) technique, thus providing only limited information about vessel morphology. DSA also uses ionizing radiation and potentially nephrotoxiciodinated contrast. DSA is now primarily used for guiding interventional procedures, such as stent graft placement, and is occasionally employed as a diagnostic tool in the setting of trauma [2]. Computed tomography (CT) is now the most frequently utilized modality for evaluating the thoracic aorta and has high diagnostic accuracy for detection of aortic pathology, particularly with the advent of multidetector scanners [3, 4]. CT has the advantage of being quick and readily available in most hospital settings; however, it too employs ionizing radiation [5] and potentially nephrotoxic contrast agent. Transesophageal echocardiography (TEE) can also be used to assess the thoracic aorta, particularly in the diagnosis of aortic dissection; however, it is relatively invasive and provides limited coverage of the entire vessel [6]. Magnetic resonance imaging (MRI) is increasingly becoming the first-line investigation for evaluating diseases in the thoracic aorta [7, 8]. MRI possesses the capability for multiplanar imaging, uses a well-tolerated contrast agent and does not involve ionizing radiation. With recent advances in E. Ward, MB, BCH, BAO () Department of Radiology, Northwestern Memorial Hospital, Chicago, IL, USA e-mail: [email protected] J.C. Carr, MD Feinberg School of Medicine, Northwestern University, Chicago, IL, USA

18

gradient hardware, much shorter repetition times (TR) are now achievable resulting in significant increases in acquisition speed. This has prompted the development of new pulse sequences and ultrafast magnetic resonance angiography (MRA) techniques. The principal challenge to universal implementation of MRI as a first-line imaging tool for the thoracic aorta remains its less widespread availability compared to CT.

MR Imaging Techniques MRI strategies for evaluating the thoracic aorta are based on contrast-enhanced MRA (CE-MRA) combined with other techniques to assess vessel lumen and wall morphology. Time-resolved MRA (TR-MRA) is now feasible with most modern MRI systems and can be used as an adjunct to conventional CE-MRA or, in some instances, as a replacement. Balanced steady-state free precession techniques (bSSFP) now form an integral part of any thoracic aorta protocol, either as a method for assessing relevant adjacent structures, such as the aortic valve or as a replacement for CE-MRA, where Gadolinium contrast cannot be used. Phase contrast MRI (PC-MRI) is also increasingly employed as a valuable tool to characterize flow disturbance, such as those occur with bicuspid aortic valve (BAV) disease or aortic coarctation. MRI protocols for imaging the thoracic aorta are described in Table 18.1.

Steady-State Free Precession Balanced SSFP is a gradient echo technique that is widely used for cine imaging of the heart [9–14]. bSSFP is T2*weighted and produces high signal from blood without the need for a contrast agent. Contrast to noise depends on T2/T1 differences, which, at short TR, are high for blood and soft tissues. In order to produce artifact-free images, bSSFP must be implemented at short TR and, as a result, is most

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_18, © Springer Science+Business Media, LLC 2012

239

240

E. Ward and J.C. Carr

Table 18.1 MRI protocols for imaging the thoracic aorta Protocol for contrast enhanced MR angiography (CE-MRA) of the thoracic aorta Technique

Comment

1 2D single shot SSFP (axial and coronal) 2 T1 GRE FS (axial and coronal) 3 Cine SSFP (AV, LVOT, 3 chamber) 4 PC-MRA (AV, LVOT, 3 chamber) 5 Dark blood TSE (axial/LAO) Inject contrast 6 TR-MRA (LAO/Coronal)

Free breathing; ECG gated to diastole Pre- and postcontrast If aortic valve disease suspected, e.g., BAV If aortic valve disease/coarct present, e.g., BAV; can be carried out post contrast Optional: to evaluate vessel wall

7

ECG-gated CE-MRA (LAO)

4 cc GAD @ 4 cc/s; this can be used for timing; alternatively, this can be repeated several times at different temporal resolution, if required Breath-hold; MPR postprocessing to obtain orthogonal measurements

Protocol for non contrast MR angiography (NC-MRA) of the thoracic aorta Technique

Comment

1 2 3 4 5 6

Free breathing; ECG gated to diastole Noncontrast If aortic valve disease suspected, e.g., BAV If aortic valve disease/coarct present, e.g., BAV; can be carried out postcontrast Optional: to evaluate vessel wall Free breathing; respiratory gated with navigator

2D single shot SSFP (axial and coronal) T1 GRE FS (axial and coronal) Cine SSFP (AV, LVOT, 3 chamber) PC-MRA (AV, LVOT, 3 chamber) Dark blood TSE (axial/LAO) 3D SSFP (axial/LAO)

successfully used on scanners with high-performance gradients. The typical TR is 3.2 ms with an echo time (TE) of 1.6 ms. Signal intensity is maximal at a flip angle of 60–70°. A 256 imaging matrix is used resulting in in-plane resolution of approximately 2.0 × 1.5 mm2. bSSFP can be implemented in single-shot, cine, or three-dimensional (3D) modes for imaging the thoracic aorta [15].

Single-Shot Two-Dimensional SSFP The single-shot strategy is an electrocardiographically (ECG) triggered 2D acquisition. A trigger delay (~200–400 ms) can be used to push the acquisition further into diastole, depending on the R-R interval. The acquisition time per image is of the order of 200–400 ms depending on imaging parameters, resulting in 1 image per heartbeat. Parallel imaging is commonly utilized at twofold acceleration, to further shorten the acquisition speed. Because of the rapid acquisition, the technique is essentially independent of respiratory motion artifact and can be successfully carried out without breath holding, which is particularly advantageous in critically ill patients. The aorta is typically covered in an interleaved manner in axial, coronal, and sagittal orientations at the beginning of the study. Frequently, the pathologic process is evident on these initial images, as is any incidental pathology such as pulmonary or breast disease. This technique is particularly advantageous in patients with suspected acute dissection, where CTA is contraindicated due to renal impairment or iodinated contrast allergy.

Cine SSFP Cine SSFP is a breath-hold 2D ECG-triggered segmented k-space acquisition, similar to what is used for cine MRI of the heart. The imaging time is approximately 4–6 s per slice with parallel imaging [16–19]. Cine images are typically acquired at multiple selected anatomic levels and orientations. Cine SSFP of the aortic valve is an essential component of any protocol to assess ascending thoracic aortic aneurysms (TAAs), where BAV disease needs to be excluded. In cases of ascending aortic dissection, coronal and long axis images through the aortic valve are used to exclude significant aortic insufficiency as a complication of dissection. A sagittal oblique “candy-cane” image through the upper chest is particularly useful for demonstrating the aortic arch. Real-time SSFP, which does not require breath-holding or ECG-triggering, is a useful alternative cine technique in patients who cannot hold their breath [20, 21]. Despite the lower spatial and temporal resolution with this technique, diagnostic images of the thoracic aorta and aortic valve can usually be obtained. Three-Dimensional SSFP Recently, 3D SSFP techniques have been developed to image the coronary arteries and thoracic aorta [22]. The technique employs a nonselective radiofrequency pulse with segmented acquisition to obtain isotropic 3D data with very high spatial resolution. Imaging times are long; therefore the acquisition requires respiratory gating using a motion adaptive

18

Thoracic Aorta

respiratory navigator. One drawback is that image quality may be compromised in patients with arrhythmia or irregular respiratory cycles. Thus, the technique has become a popular noncontrast alternative to CE-MRA in “at risk” patients.

Black-Blood Techniques “Black blood” techniques demonstrate the wall of the thoracic aorta and can be useful for evaluating a variety of pathologies, including acute aortic syndromes (intramural hematoma (IMH), penetrating ulcerative plaque, dissection), atherosclerosis, vasculitis, or neoplasms. A number of different strategies can be used. ECGtriggered breath-hold “black blood” turbo spin echo (TSE) [23], utilizes a double inversion technique to null the blood signal. The first 180° inversion recovery pulse is nonslice selective and occurs at the R wave trigger. It inverts the magnetization in the entire tissue volume, including the blood signal. This is followed immediately by a slice-selective “reversion” 180° pulse, which regenerates the magnetization in the slice to be imaged. At a specific inversion time (TI), the blood signal is completely nulled. Blood flowing into the slice is nulled from the first 180° inversion pulse resulting in a black-blood appearance. Imaging occurs during diastole. Approximately 12–16 lines of k-space are acquired in each RR interval and the center of k-space is acquired at the TI. The entire image is acquired in a segmented manner over several heartbeats. This technique is particularly useful when targeted to a specific abnormality, such as wall thickening in vasculitis or a penetrating atherosclerotic ulcer (PAU). ECG-triggered “black blood” HASTE utilizes a blackblood preparation to null signal from blood [24]. This is a single-shot technique with the entire image being acquired in a single heartbeat. This is a useful alternative in patients who cannot hold their breath. More recently, a fast diffusion-prepared (DP) balanced SSFP-based MR technique that allows for 3D dark-blood imaging has been described [25]. Because the 3D DP-SSFP technique relies on blood motion (rather than inflow) to suppress MR signal from blood, it can be of use for time-efficient dark blood MR imaging of the entire thoracic aorta when set to be acquired in a sagittal–oblique plane. In addition, because this is a 3D technique, the images offer improved slice resolution and more intuitive visualization of the thoracic aorta relative to 2D methods.

T1-Weighted Gradient Echo Fat-Saturated Imaging T1-weighted gradient echo fat-saturated (GRE-FS) imaging pre- and postcontrast injection is routinely used in evaluating

241

pathology in the chest and also provides an overview of the entire thoracic aorta and surrounding structures [26]. It is particularly helpful in demonstrating intramural abnormalities, such as IMH, and may replace black-blood techniques in many situations. A 2D breath-hold gradient echo pulse sequence with fat saturation is used to cover the thorax in an interleaved manner. The entire thoracic aorta is covered in axial and either coronal or sagittal orientations pre- and postcontrast injection. The postcontrast set of images is usually obtained following the MRA.

Contrast-Enhanced MRA Time-Resolved MRA Advances in gradient strength have resulted in much shorter TRs than were previously achievable. Much shorter acquisition times are now possible allowing CE-MRA to be implemented with subsecond temporal resolution (Fig. 18.1) [9, 27]. This is particularly advantageous in the thoracic aorta, where high temporal resolution is useful for evaluating high-flow vascular lesions, such as shunts and dissections. TR-MRA is also accurate in detecting other pathology in the thoracic aorta and may completely replace conventional CE-MRA in many situations. The basic pulse sequence is a 3D gradient echo acquisition, similar to conventional CE-MRA. Asymmetric k-space scanning is utilized in all three axes to shorten the acquisition time. A TR of 1.6 ms and a TE of 0.8 ms are used and the flip angle is typically 20–25°. Parallel imaging can be used both in the x–y direction and the z direction to further improve the imaging time. Echo sharing can also be incorporated using the TRICKS or TWIST [28] to reduce the time per frame. The temporal resolution can be dialed up or down by varying the in-plane spatial resolution (i.e., matrix size and number of phase encoding steps) and through plane spatial resolution (i.e., slice thickness and number of slices). This approach is not dissimilar to DSA, where the frame time is tailored to the anatomic region and specific pathology. For example, a time resolution of 1 s per frame may be needed to image a rapidly filling shunt or dissection, whereas 3–5 s temporal resolution is sufficient for aneurysmal disease or vasculitis. Gadolinium contrast is injected rapidly (i.e., 4–5 cc/s) via an intravenous cannula placed in an antecubital vein. Since acquisition times per frame are short, small doses of contrast (i.e., 5–10 cc) are sufficient, allowing for repeat studies in different orientations, if needed. The contrast injection and MR acquisition are started simultaneously, avoiding the need for a timing run and simplifying the study for the operator. Patients are asked to breath-hold for as long as they can. Approximately 24 3D volumes are typically acquired in a single breath-hold. The first 3D set serves as a mask and subtraction occurs in-line. Maximum intensity projection (MIP) images are produced

242

E. Ward and J.C. Carr

Fig. 18.1 Images from a dynamic CE-MRA. Five measurements are performed after contrast injection. Temporal resolution must be in the order of 2–4 s to get a good differentiation between the arterial and venous phase

automatically. The entire series can be viewed as a cine loop with a frame time of 900 ms or less. In order to reduce the acquisition time per 3D volume to a minimum, spatial resolution may need to be sacrificed resulting in near-projectional MRA. The main objective is to depict high-flow vascular abnormalities in the plane of the scan with as high temporal resolution as possible and, as a result, through-plane information is less important. In-plane resolution is only moderately reduced and is sufficient for diagnosing most thoracic aortic pathology. Because of the rapid frame rates, there does not appear to be an appreciable reduction in image quality in patients who cannot hold their breath and therefore, this technique is particularly advantageous in critically ill individuals. Above all, because of the small contrast load, TR-MRA can be repeated a number of times in combination with conventional CE-MRA in order to provide a comprehensive assessment of vascular abnormalities. TR-MRA can also serve as the timing bolus acquisition for conventional-timed CE-MRA.

Conventional CE-MRA Conventional CE-MRA [29–33] is carried out for comprehensive evaluation of the thoracic aorta; however, most abnormalities are already evident from the TR-MRA images. Conventional CE-MRA provides better spatial resolution, particularly in the z direction, and may produce better depiction of abnormalities, such as PAUs. Conventional CE-MRA can be performed without or with ECG triggering, although the latter is preferred. CE-MRA performed without ECG triggering requires less time to acquire and is useful in patients who have difficulty holding their breath. However, non-ECG triggered CE-MRA is less useful for assessing the

ascending thoracic aorta, which is commonly degraded by motion artifact from cardiac pulsation or obscured by overlapping vascular structures. In patients that require highresolution images of the aortic root and ascending aorta, ECG triggering is employed and is the preferred method when available [29, 32, 34, 35]. Image data is acquired only during the same phase of the cardiac cycle with each heartbeat, usually at end-diastole. As a result, the images are not degraded by cardiac motion, but do require a longer breath hold. The basic pulse sequence for conventional CE-MRA is a standard 3D gradient echo acquisition. The contrast transit time is calculated from the subsecond TR-MRA. A 512 matrix size or higher is used yielding a typical voxel size of 1.3 × 0.8 × 1.3 mm3. 0.1–0.2 mmol/kg of Gadolinium is injected at 2.5 ml/s via an 18G cannula placed in an antecubital vein. Images are acquired during breath-holding. Subtracted 3D sets are calculated from the raw data and these are subjected to an MIP postprocessing algorithm. A key component of CE-MRA of the thoracic aorta is image postprocessing, which can be performed at the scanner or on a separate workstation in a dedicated 3D postprocessing lab. The unsubtracted postcontrast 3D data set must be analyzed using multiplanar formatting to generate orthogonal images of the thoracic aorta at multiple anatomic landmarks. Using ECG-gating, which allows for improved quality images at the aortic root and sinuses, facilitates these measurements. These images are used to generate orthogonal dimensions at the following points in the thoracic aorta: annulus, sinuses of valsalva, sinotubular junction, mid-ascending thoracic aorta, proximal aortic arch, distal aortic arch, and descending thoracic aorta at level of diaphragm (Fig. 18.2a–c). The sinus of valsalva measurement

18

Thoracic Aorta

243

is maximal sinus-to-sinus measurement. The proximal arch measurement is an important determinant about what type of surgery needs to be performed. It is critical that the anatomic landmarks are chosen accurately, particularly in follow up

studies, where comparison of measurements is being made to assess aneurysm progression. Assessment of aneurysm volume may be a more reliable and consistent measurement method by removing operator subjectivity [36, 37].

Fig. 18.2 (a–d). Images of the thoracic aorta. (a) Diagrammatic representation; (b) ECG-gated MRA showing an aneurysm of the aortic root. Figures (a) and (b) show the levels for measurement of the thoracic aortic diameter. Measurements are takes at the aortic annulus, the sinuses of valsalva, the sino-tubular junction, the mid-ascending aorta, the proximal aortic arch, the distal aortic arch, the mid-descending aorta, and the descending aorta at the level of the diaphragm. (Panel

(a) is reprinted with permission from Evangelista et al. [68].). (c) Navigator 3D SSFP post gadofosveset. The advantage of this technique is high spatial resolution by imaging in the steady-state with a blood pool agent (gadofosveset); (d) Magnitude and phase imaging through the aortic valve in this patient showed aortic stenosis, highlighting the importance of imaging the aortic valve in all patients with a suspected aortic aneurysm

244

E. Ward and J.C. Carr

Fig. 18.3 (a and b). MRA and 4D flow image from a patient with a bicuspid aortic valve with a growing aneurysm. Figure (a) is a morphologic image of the thoracic aorta; Figure (b) shows helical flow in the thoracic aneurysm using 4D technique. (Courtesy of Alex J. Barker, Ph.D, University Medical Center Freiburg.)

Phase Contrast MRA Phase contrast MRA (PC-MRA) is most useful when CE-MRA shows a stenosis in the thoracic aorta or when an aortic valve stenosis is suspected [38, 39]. PC-MRA utilizes velocity differences or phase shifts in moving spins to produce image contrast in vessels. Phase shifts for moving spins are generated by applying magnetic field gradients of opposite polarity. The velocity encoding (VENC) value controls the amplitude of the magnetic gradient and is typically 150–200 cm/s in the aorta. If the VENC is too low, aliasing will occur resulting in mismapping or dropout of signal from the center of the lumen. PC-MRA results in two sets of data, i.e., phase images and magnitude images. The sequence can be either ECG-triggered or retrospectively gated using pulse recording. It is important that the region of interest is positioned in the center of the magnetic field to avoid artifact from eddy currents, although this is less important nowadays due to improved magnetic field homogeneity with modern scanners. Typically, the 2D PC-MRA slice is orientated perpendicular to the vessel of interest and the velocity is encoded through the plane of the slice. Alternatively, the velocity may be encoded within the plane of the slice to demonstrate a stenotic or regurgitant jet in profile. This is particularly useful, where jets are eccentrically orientated in order to facilitate positioning of axial through plane slices orthogonal to the jet. This allows for more accurate measurement of peak antegrade velocity and regurgitant fraction. Time-flow and time-velocity curves can be generated from the PC-MRA data and peak flow and peak velocity values are calculated. Analysis of curve shape and slope helps decide whether stenoses are significant or not. These mea-

surements can be particularly useful for routine follow-up of stenoses. PC-MRA is an important component of TAA protocols, where BAV disease needs to be excluded. Similarly, PC-MRA may be helpful in characterizing and following progression of aortic stenoses, such as coarctations, by measuring peak velocity through the area of narrowing. Conventional PC-MRA encodes velocity in a single direction. Velocity can also be encoded in the x, y, and z directions simultaneously allowing for depiction of turbulence and vortical flow within the plane of the image [40]. This may be an important marker for evaluating aneurysm progression, although this needs to be established. Tri-directional VENC has also been combined with a 3D spatial acquisition potentially allowing true functional assessment of vascular hemodynamics with MRI [41, 42]. Parameters, such as wall shear stress and pulse wave velocity, can be calculated from these “7D” images thus facilitating characterization of vascular physiology (Fig. 18.3) [43].

Non-Contrast MRA With the recent concerns about NSF and its association with Gadolinium contrast administration, there has been a resurgence of interest in non-contrast MRA (NC-MRA) techniques for imaging the vasculature throughout the body. The success of NC-MRA in different parts of the body depends largely on anatomic region, the size of the vessels being imaged and vascular flow. NC-MRA is particularly suited to imaging the thoracic aorta due to its large size and rapid flow, although it is affected by breathing motion, therefore respiratory gating is required. NC-MRA of the thoracic aorta

18

Thoracic Aorta

depends mostly on two techniques: bSSFP and fast spin echo (FSE).

3D FSE NC-MRA using FSE relies on an ECG-gated 3D partialFourier FSE sequence, which is triggered to systole and diastole [44]. The technique depends on the loss of arterial signal during fast flow in systole, whereas arterial signal is bright during the slower flow in diastole. Venous signal is bright during systole and diastole. Subtracting the systolic data set from the diastolic data set produces the NC-MRA image. Each partition of the 3D imaging set is acquired as a single shot image at end diastole. It typically takes about 90 s to 3 min for each 3D acquisition resulting in a total imaging time of 3–6 min. The thoracic aorta produces a consistent flow void in systole due to rapid flow resulting in excellent image quality on the subtracted set. In order to shorten overall acquisition time, it may be sufficient to only acquire the diastolic image since venous overlay does not significantly impact the thoracic aorta. Balanced Steady-State Free Precession Techniques bSSFP is ideally suited to NC-MRA due to the nonflowdependent bright blood signal since image contrast is determined by T2/T1 ratios [27, 45]. An SSFP requires implementation with short TR resulting in shortened acquisition times for a 3D acquisition; however, the sequence necessitates high performance gradient hardware to maintain the short TR and avoid troublesome banding artifacts. SSFP imaging of the thoracic aorta does not require fat or venous suppression and therefore is ideally suited to NC-MRA without any arterial spin labeling in this vascular distribution. The technique involves a 3D acquisition gated to diastole with respiratory triggering resulting in an overall acquisition time of 5–10 min. A nonselective RF pulse can be used to shorten the TR and reduce sensitivity to field in homogeneities. Isotropic voxel sizes can be achieved, especially when parallel imaging is used, allowing image reconstruction in multiple orientations.

Abnormalities of the Thoracic Aorta Acute Aortic Syndromes Aortic Dissection Aortic dissection results from the passage of blood out of the true lumen through a defect in the vessel wall into the tunica media, resulting in separation of the layers of the aorta, typically separating media from intima [46–48]. This results in the creation of a true and false lumen separated by an intimal flap. The false lumen may thrombose over time. The true

245

incidence of acute aortic dissection is difficult to define although population-based studies suggest that the incidence ranges from 2 to 3.5 cases per 100,000 which correlates to 6,000–10,000 cases annually in the USA [49–53]. Anatomically, aortic dissection is classified according to the origin of the intimal tear or whether the dissection involves the ascending aorta. Accurate classification is essential as it drives decisions about surgical versus nonsurgical management. The two most commonly used classification systems are the DeBakey and Stanford systems (Fig. 18.4). The Stanford classification system divides dissections into two categories, those that involve the ascending aorta and those that do not. Note that involvement of the aortic arch without involvement of the ascending aorta is classified as type B. Type A dissections are surgical emergencies, whereas type B dissections are treated medically. Occasionally, when the false lumen is thrombosed on imaging despite classic symptoms, the dissection is termed “non-communicating dissection” or “chronic aortic dissection.” This entity is also indistinguishable from IMH, which likely represents a spectrum of the same pathology. There are a number of risk factors for aortic dissection. These include conditions associated with increased aortic wall stress, such as uncontrolled hypertension and coarctation of the aorta, conditions associated with aortic media abnormalities such as Marfan syndrome or the inflammatory vasculitides and other conditions such as pregnancy [54]. A family history of TAA is an important risk factor. Patients with acute aortic syndrome usually present the same way regardless of whether due to dissection, IMH, or penetrating ulcer. The most common presenting symptom is pain. Although classically described as tearing or ripping, it may also be sharp or stabbing, anterior or posterior. The side effects result from either proximal or distal migration of the dissection to involve and occlude branch vessels of the aorta. When the dissection migrates proximally to involve the coronary arteries and aortic valve, the consequences are catastrophic. Complications due to proximal migration include myocardial infarction, aortic regurgitation, heart failure/shock, pericardial effusion/tamponade, and neurologic complications, such as stroke. Complications due to distal migration include mesenteric ischemia/infarction, renal failure, and limb ischemia. The main objectives of imaging are to identify the presence of a dissection, to correctly classify it as type A or type B and to detect complications. Urgent and definitive imaging of the thoracic aorta using echocardiography, CTA or MRA is recommended in all patients with suspected dissection. Echo, either transthoracic or tranesophageal, has the advantage of being easily available; however, it does not produce images of the entire aorta. CT angiography is usually the first-line test in patients with acute aortic syndrome due to the presence of CT scanners in most

246

E. Ward and J.C. Carr

Fig. 18.4 Aortic dissection classification: DeBakey and Stanford Classifications. (Reprinted with permission from the Cleveland Clinic Foundation.)

emergency departments and its rapid scan capabilities. MRI is usually reserved for patients with a contraindication to CT. MRI has high sensitivity and specificity for detection of dissection and is now regarded as the gold standard for assessment of this abnormality [3, 26, 32]. Single-shot SSFP has been shown to be highly accurate for diagnosis and classification of aortic dissection [15]. Because of the inherent high signal to noise from blood, both true and false lumen and dissection flap are clearly visible. In fact, the diagnosis is made on the SSFP localizers in many instances. Moreover, the technique can be implemented without breath holding in a total imaging time of less than 2 min. This is ideally suited to critically ill patients who cannot hold their breath. Breath-hold cine SSFP can be used to evaluate aortic insufficiency and hemopericardium in type A dissections; however, these patients are rarely fit enough to hold their breath. In these situations, non-ECG-triggered realtime SSFP, although of lower spatial resolution, can be a useful alternative [20, 21]. T1 GRE-FS imaging pre- and postcontrast is essential for detecting IMH or thrombosed false lumen associated with a dissection. Recent intramural hemorrhage will appear hyperintense on noncontrast images. It may however be more difficult to detect older hemorrhage, which may appear isointense or hypointense to surrounding structures. It may be very difficult to distinguish IMH from a thrombosed false lumen of an aortic dissection. CE-MRA is also accurate in diagnosing aortic dissections; however, the diagnosis is usually made already on the SSFP images. CE-MRA is useful for assessing the

proximal and distal extent of the dissection and its involvement of branch vessels. A second CE-MRA acquisition of the abdomen may be required to assess distal dissections extending into the abdominal aorta, particularly to evaluate involvement of the renal and mesenteric vasculature. This can usually be achieved as a stepping table MRA to cover both thorax and abdomen in the same study. TR-MRA can demonstrate sequential filling of the true and false lumen and may help identify the entry and exit points of the dissection. In addition, TR-MRA can clearly demonstrate pseudoaneurysms or IMHs.

Intramural Hematoma There is considerable clinical and imaging overlap between IMH, thrombosed false lumen in a localized dissection and PAU [52, 55–59]. Approximately 10–20% of patients with a clinical presentation of dissection present with imaging findings of IMH. IMH may arise either from hemorrhage from vas vasorum within the vessel wall or from microscopic tears within the aortic intima. The hematoma may propagate in an antegrade or retrograde direction similar to a dissection. IMH may arise from either a thrombosed false lumen or a penetrating ulcer. Extensive IMH is identical to noncommunicating dissection with a thrombosed false lumen. Imaging depends on demonstrating free blood within the aortic wall. MRI is ideally suited to making this diagnosis due to its inherent sensitivity to depiction of blood products. In all cases, the intramural blood is detected on precontrast T1 GRE-FS or black-blood techniques.

18

Thoracic Aorta

Penetrating Atherosclerotic Ulcer PAU represents an atherosclerotic lesion that penetrates the internal elastic lamina and allows hematoma formation within the media of the aortic wall. PAU is considered a prelude to IMH and aortic dissection and represents part of the spectrum of acute aortic syndromes. PAUs usually occur in areas of atherosclerosis and therefore are most common within the descending thoracic aorta. PAUs are well depicted on postcontrast T1 GRE-FS images or the source images from a conventional CE-MRA [60].

247 Table 18.2 Normal adult thoracic aorta diameters Thoracic aorta

Range of reported mean (cm)

Root (female) Root (male) Ascending (female, male) Mid-descending (female) Mid-descending (male) Diaphragmatic (female) Diaphragmatic (male)

3.50–3.72 3.63–3.91 2.86 2.45–2.64 2.39–2.98 2.40–2.44 2.43–2.69

(Adapted from Hiratzka et al. Circulation 2010 [54])

Pseudoaneurysms of the Thoracic Aorta Pseudoaneurysms refer to a localized contained rupture of the thoracic aorta and can result from blunt trauma or following aortic surgery or catheter-based interventions. Pseudoaneurysms have a narrow neck and are eccentrically located within the vessel. The condition can be diagnosed on conventional CE-MRA, although TR-MRA may have a valuable role in demonstrating the point of communication and the rate of filling of the aneurysm sac.

Thoracic Aortic Aneurysm Aneurysms of the thoracic aorta are relatively common and are important because of their potentially reversible lethal consequences [61]. TAAs are usually caused by degenerative disease resulting in dilatation of the aorta. The incidence of TAAs is estimated to be around 10.4 cases per 100,000 person-years. There are multiple risk factors and causes for the development of TAA. The commonest cause is atherosclerosis, where aneurysms most commonly affect the descending aorta. Less common causes include trauma, congenital abnormalities, connective tissue disorders, such as Erhlers-Danlos and Marfan syndrome, vasculitides such as Takayasu’s arteritis or giant cell arteritis (GCA) and rare infections such as syphilis. TAA is also associated with BAV disease. TAAs can be classified according to the anatomic segment of the thoracic aorta that is involved. Aneurysms of the aortic root and ascending aorta are the most common followed by the descending aorta and aortic arch. The location of the aneurysm largely determines the type of treatment, surgical versus endovascular. TAAs can be further classified into true or false aneurysms and fusiform or saccular aneurysms. Most TAAs are fusiform in shape, whereas saccular aneurysms may suggest pseudoaneurysm formation or mycotic aneurysm. TAAs are commonly asymptomatic, being detected incidentally on a chest X-ray or cross-sectional imaging, such as CT or MRI. When symptoms occur, they may be due to compression, causing complaints, such as hoarseness or stridor, or aortic valve involvement, causing aortic regurgitation and

heart failure. The most catastrophic complications of TAA are dissection and rupture. Rupture usually occurs when TAAs reach more than 5.5 cm in size. The average rate of expansion of TAAs is estimated to be 0.10–0.42 cm/year; therefore, regular surveillance with cross-sectional imaging is necessary. Normal thoracic aortic diameters vary depending on the segment of the thoracic aorta and generally decrease as one progresses distally within the thoracic aorta (Table 18.2). Values vary between male and female and increase with age. In smaller patients, it may be more helpful and accurate to calculate values normalized to body surface area. MRI is ideally suited to evaluation of aneurysms because of its ability for multiplanar imaging [7, 62]. CE-MRA is most useful for depicting location, extent, shape, and exact diameter of TAAs. The preferred technique is ECG-gated CE-MRA, where the acquisition is gated to diastole, as described above in technique section. This produces highquality sharp images of the aortic root and ascending aorta, thereby facilitating measurements in these locations. It is essential that orthogonal measurements be obtained from all of the relevant anatomic landmarks in the thoracic aorta (Fig. 18.2a, b). These dimensions are included systematically as part of a structured report for thoracic MRA. It is important to remember that MIP images represent a cast of the lumen; therefore measurements should be obtained from source images where the vessel wall is visible. MRI is frequently utilized as a follow-up tool for monitoring the progression of disease and therefore, in order to produce consistent results, vessel dimensions should be measured at the same anatomic locations each time. Another important advantage of MRI is the ability to use multiple pulse sequences allowing more comprehensive evaluation of pathology. Where aneurysms involve the ascending aorta or sinuses of valsalva, concomitant aortic valve disease, such as BAV, should be evaluated using cine imaging of the heart. PC-MRA can then be used to characterize the degree of stenosis or regurgitation (Fig. 18.2d). In patients who cannot receive Gadolinium contrast due to low GFR or allergy, NC-MRA using bSSFP serves as a

248

E. Ward and J.C. Carr

Fig. 18.5 (a–d). Coarctation of the aorta. Time-resolved MRA images from a patient with coarctation of the aorta and numerous enlarged collateral arteries. Figure (a) shows the obstruction distal to the origin of the left subclavian artery (arrow). Figures (b–d) shows progressive filling of significant neck and chest wall collateral vessels secondary to the obstruction (arrows). (Courtesy of Andrada Popescu, MD Northwestern Memorial Hospital, Chicago, IL)

suitable alternative to CE-MRA. Image processing and measurement should occur in a similar fashion to CE-MRA.

disease progression over time; however, it is important to obtain measurements in similar anatomic locations in order to produce accurate and consistent results.

Aortic Stenosis Congenital Abnormalities Stenoses in the thoracic aorta are commonly caused by atherosclerosis but are usually multifocal and rarely flow limiting. Coarctation of the aorta causes a more severe stenosis at the junction of the aortic arch and descending aorta and may be focal (usually incidental finding in adults) or diffuse (usually symptomatic in infants) (Fig. 18.5). It is usually associated with multiple chest wall collaterals. Residual coarctation can present later in life following a previous coarctation repair. Pseudocoarctation resembles a true coarctation but is caused by aortic kinking just distal to the origin of the left subclavian artery. It is not hemodynamically flow-limiting and therefore is not associated with multiple collaterals. Less common causes of aortic stenosis include extrinsic compression from tumor or rare inflammatory conditions, such as Takayasu’s arteritis. CE-MRA is the most useful technique for evaluating stenoses in the thoracic aorta. TR-MRA will accurately depict aortic stenoses but is particularly useful in hemodynamically significant lesions, such as coarctations where it demonstrates gradual filling of chest wall collaterals. The addition of SSFP or T1 GRE-FS will exclude an adjacent mass causing extrinsic compression. PC-MRA can be used to measure velocity and flow both proximal and distal to a stenosis and helps assess the significance of a stenosis (Fig. 18.2d) [63]. It may be more useful in monitoring

There are a number of congenital abnormalities, which affect the thoracic aorta, including right-sided aortic arch, aberrant subclavian artery, aortic coarctation, patent ductus arteriosis, double aortic arch, and aortic hypoplasia. All of these conditions are best visualized using CE-MRA [64, 65]. With congenital abnormalities, 3D postprocessing techniques can be used to great effect allowing the abnormality to be viewed from different orientations. Cine imaging of the heart should always accompany an assessment for congenital abnormalities of the aorta in order to detect accompanying lesions in the heart. Aortic coarctations and PDAs are usually evaluated using a combination of CE-MRA, balanced SSFP, and PC techniques. Because of the high spatial resolution of 3D CE-MRA, this technique is used for accurate morphological assessment of these abnormalities. Using a computer workstation, accurate orthogonal measurements of the aorta can be made at multiple locations proximal and distal to the coarctation or across the PDA. PC-MRA is used to quantify flow rates and flow velocities through the abnormality, as well. In patients with aortic coarctation, the minimal cross-sectional area and the mean deceleration of flow in the descending aorta have been found to be highly predictive of a moderate or severe coarctation gradient (³20 mmHg), when compared to

18

Thoracic Aorta

249

Fig. 18.6 Predominant range of vascular involvement by vasculitides. Large artery refers to the aorta and large branches to the extremities as well as the head and neck. Medium-sized artery refers to the main visceral arteries. Small artery refers to the distal intraparenchymal arteries. (Adapted with permission from Jennette et al. [69].)

Fig. 18.7 System for classifying Takayasu arteritis according to the site of involvement. (Reprinted with permission from Nastri [70].)

pressure measurements made with catheter angiography. These measurements can then be used to follow patients prior to and after treatment.

Vasculitis Arteries, including the thoracic aorta, can be affected by a number of vasculitides (Fig. 18.6). Because of the high spatial and contrast resolution offered by newer MRI techniques, which permit assessment of the aortic wall, rheumatologists routinely include MR imaging in the work-up of patients with vasculitides, particularly in patients with vasculitides affecting larger vessels – GCA and Takayasu’s arteritis [66, 67].

Takayasu’s Arteritis Takayasu’s arteritis, also known as pulseless disease, is an idiopathic vasculitis of the elastic arteries, involving the aorta and its branches. The disease affects all ethnic and racial groups but is more common in Asian populations. Women are affected ten times more than men and the disease usually begins in the second to third decades of life. The clinical presentation usually occurs in two phases: acute and chronic. The acute phase is characterized by systemic symptoms, such as weight loss and fatigue, whereas the chronic phase is the burnt-out stage causing fixed arterial occlusions and stenoses, similar to atherosclerosis. The disease can also cause TAA. The disease is classified based on the extent of involvement of the aorta and its branches (Fig. 18.7).

250

A hallmark of the disease is great vessel involvement, typically bilateral subclavian artery lesions. Conventional CE-MRA is the mainstay for assessing patients with Takayasu’s arteritis. The images are acquired in a coronal orientation. Both thorax and abdomen should be imaged in two separate stations using a moving table approach, since thoracic aortic and abdominal aortic diseases frequently coexist. In addition to demonstrating the degree of luminal narrowing, CE-MRA demonstrates segmental thickening of the vessel wall, mural thrombus, inflammatory changes in the periaortic fat, vascular dilatation, and vascular stenoses. TR-MRA may also be employed in the thorax to give an overview of the thoracic aorta and, in particular, the branches of the aortic arch. The protocol should include dark blood sequences to visualize the vessel wall in addition to CE-MRA sequences. Recently, delayed CE-MRI techniques, similar to those used to assess myocardial viability in the heart, have been used to characterize the degree of inflammation in the aortic wall of patients with Takayasu’s arteritis [66]. MRI with cine imaging of the heart can also be used to evaluate the affect of the disease on cardiovascular function. In addition to imaging at the time of presentation, MRA can be used to assess patient response to therapy.

Giant Cell Arteritis GCA, also known as temporal arteritis, is a vasculitis of the elastic arteries and their branches. In contradistinction to Takayasu’s arteritis, GCA that occurs in patients over 50 years of age has a female to male ratio of 3:2 and occurs more commonly in persons of northern European ancestry. Affected patients present with nonspecific constitutional symptoms but the majority complain of headache and scalp tenderness. TAA or dissection can occur in up to 20% of patients. The MRI protocol is identical to that used for Takayasu’s arteritis and relies on demonstration of vascular occlusive disease or aneurysm, extent of involvement and aortic wall thickening. Other Inflammatory Conditions Behcet’s disease is rare and is characterized by uveitis, aphthous stomatitis, and genital ulcers. Large and small vessel vasculitis can coexist in multiple sites resulting in stenotic lesions and aneurysms of the thoracic aorta and aortic arch vessels. Behcet’s disease is one of the few conditions that can affect both arteries and veins causing superficial thrombophlebitis and deep venous thrombosis in the venous system. Ankylosing spondylitis is one of the spondyloarthropathies, which is associated with HLA B27 and seronegativity for rheumatoid factor. Typical features include male preponderance, sacroilitis, and arthritis and there is an association with aortitis.

E. Ward and J.C. Carr

Infective TAAs are caused by infection with bacterial, fungal, viral, or tuberculous organisms. Aneurysms are usually saccular although fusiform aneurysms have also been described. Infection can be due to contiguous spread from adjacent structures, such as mediastinitis, septic emboli in infective endocarditis, and hematogenous spread in intravenous drug abuse. Syphilitic aortitis is a specific infection caused by treponema pallidum, which characteristically causes aneurysms in the ascending thoracic aorta, usually 10–15 years after the initial infection.

Conclusion There are numerous pulse sequences available now for evaluating the diverse pathology, which affects the thoracic aorta. Preliminary imaging using SSFP and pre- and postcontrast T1 GRE-FS is usually required to assess morphology of the aorta and adjacent structures. CE-MRA is the mainstay in the investigative approach. The addition of time-resolved CE-MRA is particularly useful for assessing high-flow vascular lesions such as shunts, while at the same time not adding much to the overall contrast load. PC-MRA may help further characterize stenotic lesions and can be useful for monitoring progression of disease.

References 1. Waugh JR, Sacharias N. Arteriographic complications in the DSA era. Radiology.1992;182:243–246. 2. Mirvis SE, Pais SO, Gens DR. Thoracic aortic rupture: advantages of intraarterial digital subtraction angiography. AJR Am J Roentgenol. 1986;146:987–991. 3. Hartnell GG. Imaging of aortic aneurysms and dissection: CT and MRI. J Thorac Imaging. 2001;16:35–46. 4. Rubin GD. Helical CT angiography of the thoracic aorta. J Thorac Imaging. 1997;12:128–149. 5. Hidajat N, Maurer J, Schroder RJ, Wolf M, Vogl T, Felix R. Radiation exposure in spiral computed tomography. Dose distribution and dose reduction. Invest Radiol. 1999;34:51–57. 6. Scott CH, Keane MG, Ferrari VA. Echocardiographic evaluation of the thoracic aorta. Semin Roentgenol. 2001;36:325–333. 7. Roberts DA. Magnetic resonance imaging of thoracic aortic aneurysm and dissection. Semin Roentgenol. 2001;36:295–308. 8. Fattori R, Nienaber CA. MRI of acute and chronic aortic pathology: pre-operative and postoperative evaluation. J Magn Resonan Imaging. 1999;10:741–750. 9. Carr JC, Simonetti O, Bundy J, Li D, Pereles S, Finn JP. Cine MR angiography of the heart with segmented true fast imaging with steady-state precession. Radiology. 2001;219:828–834. 10. Barkhausen J, Ruehm SG, Goyen M, Buck T, Laub G, Debatin JF. MR evaluation of ventricular function: true fast imaging with steady-state precession versus fast low-angle shot cine MR imaging: feasibility study. Radiology. 2001;219:264–269. 11. Oppelt A, Graumann R, Barfuss H. FISP-a new fast MRI sequence. Electromedica. 1986;54:15–18. 12. Fang W, Pereles FS, Bundy J et al. Evaluating left ventricular function using real-time TrueFISP: a comparison with conventional

18

13.

14.

15.

16.

17.

18.

19.

20.

21.

22. 23.

24.

25.

26.

27.

28.

29. 30.

31.

Thoracic Aorta techniques. Proceedings of the Eighth Meeting of the International Society for Magnetic Resonance in Medicine. Berkeley, CA: International Society for Magnetic Resonance in Medicine, 2000; 8:308. Deimling M, Heid O. Magnetization prepared True FISP imaging [abstract]. Proceedings of the Second Meeting of the Society of Magnetic Resonance. Berkeley, CA: Society of Magnetic Resonance. 1994:495. Bundy J, Simonetti O, Laub G, Finn JP. Segmented TrueFISP imaging of the heart. Proceedings of the Seventh Meeting of the International Society for Magnetic Resonance in Medicine. Philadelphia, PA, 1999;2:1282. Pereles FS, McCarthy RM, Baskaran V, et al. Thoracic aortic dissection and aneurysm: evaluation with nonenhanced true FISP MR angiography in less than 4 minutes. Radiology. 2002;223: 270–274. Sodickson DK, Manning WJ. Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med. 1997;38:591–603. Sodickson DK, McKenzie CA, Li W, Wolff S, Manning WJ, Edelmann RR. Contrast-enhanced 3D MR angiography with simultaneous acquisition of spatial harmonics: a pilot study. Radiology. 2000;217:284–289. Weiger M, Pruessmann KP, Boesiger P. Cardiac real-time imaging using SENSE. SENSitivity Encoding scheme. Magn Reson Med. 2000;43:177–184. Weiger M, Pruessmann KP, Kassner A, Roditi G, Lawton T, Reid A, Boesiger P. Contrast-enhanced 3D MRA using SENSE. J Magn Reson Imaging. 2000;12:671–677. Barkhausen J, Goyen M, Ruhm SG, Eggebrecht H, Debatin JF, Ladd ME. Assessment of ventricular function with single breathhold real-time steady-state free precession cine MR imaging. AJR Am J Roentgenol. 2002;178:731–735. Lee VS, Resnick D, Bundy JM, Simonetti O, Lee P, Weinreb JC. Cardiac function: MR evaluation in one breath hold with real-time true fast imaging with steady-state precession. Radiology. 2002;222:835–842. Francois CJ, Carr JC. MRI of the thoracic aorta. Cardiol Clin. 2007;25:171–184, vii. Simonetti OP, Finn JP, White RD, Laub G, Henry DA. “Black blood” T2-weighted inversion-recovery MR imaging of the heart. Radiology.1996;199:49–57. Stehling MK, Holzknecht NG, Laub G, Bohm D, von Smekal A, Reiser A. Single-shot T1- and T2-weighted magnetic resonance imaging of the heart with black blood: preliminary experience. Magma. 1996;4(3–4):231–240. Koktzoglou I, Kirpalani A, Carroll TJ, Li D, Carr JC. Dark-blood MRI of the thoracic aorta with 3D diffusion-prepared steady-state free precession: initial clinical evaluation. AJR Am J Roentgenol. 2007;189:966–972. Summers RM, Sostman HD, Spritzer CE, Fidler JL. Fast spoiled gradient-recalled MR imaging of thoracic aortic dissection: preliminary clinical experience at 1.5 T. Magn Reson Imaging. 1996;14:1–9. Finn JP, Baskaran V, Carr JC, et al. Thorax: low-dose contrastenhanced three-dimensional MR angiography with subsecond temporal resolution – initial results. Radiology. 2002;224:896–904. Korosec FR, Frayne R, Grist TM, Mistretta CA. Time-resolved contrast-enhanced 3D MR angiography. Magn Resn Med. 1996;36: 345–351. Prince MR. Gadolinium-enhanced MR aortography. Radiology. 1994;191:155–164. Princ MR, Narasimham DL, Stanley JC, et al. Breath-hold gadolinium-enhanced MR angiography of the abdominal aorta and its major branches. Radiology. 1995;197:785–792. Krinsky GA, Rofsky NM, DeCorato DR, et al. Thoracic aorta: comparison of gadolinium-enhanced three-dimensional MR

251

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44. 45.

46. 47.

48.

49.

50.

51.

angiography with conventional MR imaging. Radiology. 1997; 202:183–193. Prince MR, Narasimham DL, Jacoby WT, et al. Three-dimensional gadolinium-enhanced MR angiography of the thoracic aorta. AJR Am J Roentgenol. 1996;166:1387–1397. Krinsky GA, Reuss PM, Lee VS, Carbognin G, Rofsky NM. Thoracic aorta: comparison of single-dose breath-hold and doubledose non-breath-hold gadolinium-enhanced three-dimensional MR angiography. AJR Am J Roentgenol. 1999;173:145–150. Bireley WR 2nd, Diniz LO, Groves EM, Dill K, Carroll TJ, Carr JC. Orthogonal measurement of thoracic aorta luminal diameter using ECG-gated high-resolution contrast-enhanced MR angiography. J Magn Reson Imaging. 2007;26:1480–1485. Groves EM, Bireley W, Dill K, Carroll TJ, Carr JC. Quantitative analysis of ECG-gated high-resolution contrast-enhanced MR angiography of the thoracic aorta. AJR Am J Roentgenol. 2007;188: 522–528. Czermak BV, Mallouhi A, Perkmann R, et al. Serial CT volume and thrombus length measurements after endovascular repair of Stanford type B aortic dissection. J Endovasc Ther. 2004;11:1–12. Kritpracha B, Beebe HG, Comerota AJ. Aortic diameter is an insensitive measurement of early aneurysm expansion after endografting. J Endovasc Ther. 2004;11:184–190. Dumoulin CL, Yucel EK, Vock P, et al. Two- and three-dimensional phase contrast MR angiography of the abdomen. J Comput Assist Tomogr.1990;14:779–784. Lundin B, Cooper TG, Meyer RA, Potchen EJ. Measurement of total and unilateral renal blood flow by oblique-angle velocityencoded 2D-cine magnetic resonance angiography. Magn Reson Imaging. 1993;11:51–59. Liu X, Zhao X, Huang J, et al. Comparison of 3D free-breathing coronary MR angiography and 64-MDCT angiography for detection of coronary stenosis in patients with high calcium scores. AJR Am J Roentgenol. 2007;189:1326–1332. Bock J, Frydrychowicz A, Lorenz R, et al. In vivo noninvasive 4D pressure difference mapping in the human aorta: phantom comparison and application in healthy volunteers and patients. Magn Reson Med. 2011. Bock J, Frydrychowicz A, Stalder AF, et al. 4D phase contrast MRI at 3 T: effect of standard and blood-pool contrast agents on SNR, PC-MRA, and blood flow visualization. Magn Reson Med. 2010;63:330–338. Barker AJ, Markl M. Editorial: the role of hemodynamics in bicuspid aortic valve disease. Eur J Cardiothorac Surg. 2011;39: 805–806. Miyazaki M, Lee VS. Nonenhanced MR angiography. Radiology. 2008;248:20–43. Francois CJ, Lum DP, Johnson KM, et al. Renal arteries: isotropic, high-spatial-resolution, unenhanced MR angiography with threedimensional radial phase contrast. Radiology. 2011;258:254–260. DeSanctis RW, Doroghazi RM, Austen WG, Buckley MJ. Aortic dissection. N Engl J Med.1987;317:1060–1067. Debakey ME, Henley WS, Cooley DA, Morris GC Jr, Crawford ES, Beall AC Jr. Surgical management of dissecting aneurysms of the aorta. J Thorac Cardiovasc Surg. 1965;49:130–149. Nienaber CA, von Kodolitsch Y, Nicolas V, et al. The diagnosis of thoracic aortic dissection by noninvasive imaging procedures. N Engl J Med.1993;328:1–9. Svensson LG, Crawford ES, Coselli JS, Safi HJ, Hess KR. Impact of cardiovascular operation on survival in the Marfan patient. Circulation. 1989;80(3 Pt 1):I233-I242. Svensson LG,Labib SB, Eisenhauer AC, Butterly JR. Intimal tear without hematoma: an important variant of aortic dissection that can elude current imaging techniques. Circulation. 1999;99: 1331–1336. Meszaros I, Morocz J, Szlavi J, et al. Epidemiology and clinicopathology of aortic dissection. Chest. 2000;117:1271–1278.

252 52. Bickerstaff LK, Pairolero PC, Hollier LH, et al. Thoracic aortic aneurysms: a population-based study. Surgery. 1982;92: 1103–1108. 53. Clouse WD, Hallett JW Jr, Schaff HV, et al. Acute aortic dissection: population-based incidence compared with degenerative aortic aneurysm rupture. Mayo Clinic Proc. 2004;79:176–180. 54. Hiratzka LF, Bakris GL, Beckman JA, et al. 2010 ACCF/AHA/ AATS/ACR/ASA/SCA/SCAI/SIR/STS/SVM guidelines for the diagnosis and management of patients with Thoracic Aortic Disease: a report of the American College of Cardiology Foundation/ American Heart Association Task Force on Practice Guidelines, American Association for Thoracic Surgery, American College of Radiology, American Stroke Association, Society of Cardiovascular Anesthesiologists, Society for Cardiovascular Angiography and Interventions, Society of Interventional Radiology, Society of Thoracic Surgeons, and Society for Vascular Medicine. Circulation. 2010;121:e266-e369. 55. Wolff KA, Herold CJ, Tempany CM, Parravano JG, Zerhouni EA. Aortic dissection: atypical patterns seen at MR imaging. Radiology. 1991;181:489–495. 56. Harris JA, Bis KG, Glover JL, Bendick PJ, Shetty A, Brown OW. Penetrating atherosclerotic ulcers of the aorta. J Vasc Surg. 1994;19:90–98; discussion 98–99. 57. Cooke JP, Kazmier FJ, Orszulak TA. The penetrating aortic ulcer: pathologic manifestations, diagnosis, and management. Mayo Clinic Proc. 1988;63:718–725. 58. Yucel EK, Steinberg FL, Egglin TK, Geller SC, Waltman AC, Athanasoulis CA. Penetrating aortic ulcers: diagnosis with MR imaging. Radiology. 1990;177:779–781. 59. Pressler V, McNamara JJ. Aneurysm of the thoracic aorta. Review of 260 cases. J Thorac Cardiovasc Surg. 1985;89:50–54.

E. Ward and J.C. Carr 60. Mohiaddin RH, McCrohon J, Francis JM, Barbir M, Pennell DJ. Contrast-enhanced magnetic resonance angiogram of penetrating aortic ulcer. Circulation. 2001;103:E18-E19. 61. Pitt MP, Bonser RS. The natural history of thoracic aortic aneurysm disease: an overview. J Cardiac Surg. 1997;12(2 Suppl):270–278. 62. Schmidta M, Theissen P, Klempt G, et al., Long-term follow-up of 82 patients with chronic disease of the thoracic aorta using spinecho and cine gradient magnetic resonance imaging. Magn Reson Imaging.2000;18:795–806. 63. Mohiaddin RH, Kilner PJ, Rees S, Longmore DB. Magnetic resonance volume flow and jet velocity mapping in aortic coarctation. J Am Coll Cardiol. 1993;22:1515–1521. 64. Roche KJ, Krinsky G, Lee VS, Rosky N, Genieser NB. Interrupted aortic arch: diagnosis with gadolinium-enhanced 3D MRA. J Comput Assist Tomogr. 1999;23:197–202. 65. Soler R, Rodriguez E, Reguejo I, Fernandez R, Raposo I. Magnetic resonance imaging of congenital abnormalities of the thoracic aorta. Eur Radiol. 1998;8:540–546. 66. Desai MY, Stone JH, Foo TK, Hellmann DB, Lima JA, Bluemke DA. Delayed contrast-enhanced MRI of the aortic wall in Takayasu’s arteritis: initial experience. AJR Am J Roentgenol. 2005;184:1427–1431. 67. Gotway MB, Araoz PA, Macedo TA, et al. Imaging findings in Takayasu’s arteritis. AJR Am J Roentgenol. 2005;184:1945–1950. 68. Evangelista A, Flachskampf FA, Erbel R, et al. Echocardiography in aortic diseases: EAE recommendations for clinical practice. Eur J Echocardiogr. 2010;11: 645–658. 69. Jennette JC, Falk RJ. The pathology of vasculitis involving the kidney. Am J Kidney Dis. 1994; 24:130–141. 70. Nastri MV, Baptista LP, Baroni RH, et al. Gadolinium-enhanced three-dimensional MR angiography of Takayasu arteritis. Radiographics. 2004;24:773–786

Pulmonary MRA James F.M. Meaney and Peter Beddy

Magnetic resonance angiography (MRA) is an effective tool for evaluation of the pulmonary vasculature [1]. The most common indication for pulmonary vascular imaging is suspected pulmonary embolism (PE), nonetheless virtually all disorders of the pulmonary vasculature can be assessed [1]. In addition, the ability to evaluate pulmonary ventilation and perfusion, and to assess right ventricular function offers significant benefit over CT. Non-contrast approaches although currently limited compared to contrast-enhanced techniques potentially offer a truly non-invasive approach, and bloodpool agents may allow imaging at high resolution, even in patients with limited breath-hold capability.

Applied Pulmonary Artery Anatomy The main pulmonary arteries, which arise from the bifurcation of the pulmonary trunk, are relatively large arteries and are well visualised on CT and MR because of their large size (Fig. 19.1). The branching pattern of the main pulmonary arteries mirrors that of the segmental bronchi [1]. On the right, the first branch from the main pulmonary artery, the upper lobe artery gives rise to apico-posterior, anterior, and posterior segmental arteries. Beyond the takeoff of the right upper lobe pulmonary artery, the right main pulmonary artery continues as the inter-lobar artery which continues as the right lower lobe artery after the take-off of the middle lobe artery. The middle lobe artery is a short trunk (usually less than 1 cm) that passes antero-laterally and horizontally and immediately divides into medial and lateral segmental arteries. There are five segmental arteries to the right lower lobe (apical, anterior, posterior, medial, and lateral). J.F.M. Meaney, MD () Trinity College Dublin, Dublin, Ireland Department of Radiology, St. James’s Hospital, Dublin 8, Ireland e-mail: [email protected] P. Beddy, MD Department of Radiology, St. James’s Hospital, Dublin 8, Ireland

19

The arrangement on the left is similar apart from the following – the lingular arises directly from the left upper lobe pulmonary artery and divides into superior and inferior branches, the apical and posterior arteries to the left upper lobe, like their accompanying bronchi, are shared (the apicoposterior segmental artery) as are the left medial and left posterior segmental arteries (the mesio-basal segmental artery). Therefore, there are ten segmental arteries on the right and eight on the left. The correct identification of the pulmonary arteries is important, as correlation of findings at DSA, a projectional technique, with cross-sectional techniques such as MRA and CTA depends on accurate anatomical mapping of the individual vessels. Although the importance of embolism to the small sub-segmental arteries is unknown, initial CTA and MRA studies largely ignored them as a result of poor spatial resolution (now historical) and absence of a robust gold standard for these vessels. It is clear fact that embolism to the segmental (and larger) arteries, which can easily be diagnosed with cross-sectional techniques, is significant, but it remains unclear as to whether embolism to sub-segmental arteries, which is more difficult to diagnose with cross-sectional techniques, is unknown. Therefore, nomenclature of the pulmonary arteries respecting the distance from the pulmonary valve is given by order, and it is accepted that embolism to the first–fourth order branches (segmental and above) is not only significant but also diagnosable with MRA, whilst embolism to fifth order and greater (sub-segmental) is less reliable but of uncertain significance.

Background to Imaging of Pulmonary Embolism Clinical Considerations, Conventional Catheter Angiography, Spiral CT, and MRA Pulmonary embolism is a major cause of death and disability in western society. Unsuspected PE is the cause of death in 5% of unselected autopsies and a major contributor to the

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_19, © Springer Science+Business Media, LLC 2012

253

254

J.F.M. Meaney and P. Beddy

conditions is identical. However, when non-invasive testing does not inform therapeutic decision making, imaging of the pulmonary arteries becomes essential [28].

Catheter Angiography for Diagnosis of PE

Fig. 19.1 Normal pulmonary MRA. (a) Whole-volume MIP demonstrates excellent depiction of the central, lobar, segmental, and some subsegmental pulmonary arteries in the coronal plane. (b) Individual partitions (2 mm) from coronally acquired acquisition with excellent breath-holding and clear demonstration of central, lobar, segmental, and some sub-segmental arteries. (c) Sub-volume coronal oblique reformat show excellent depiction of the right-sided pulmonary arteries to segmental level. (d) Coronal oblique reformat along the left main and descending pulmonary artery demonstrates main, lobar, and segmental arteries

cause of death in 10% [2–14]. PE is said to be the cause of death overall in 15% of patients dying in hospital [4, 5, 7, 8]. It had been estimated that 21% would have had a favourable prognosis had the diagnosis been established, whilst the remainder were split evenly between those in whom establishing a diagnosis would not have made a difference due to severe associated co-morbid conditions and those who would have a favourable prognosis in any case [7, 9]. The importance of correctly diagnosing PE lies first in the tenfold difference in mortality between untreated PE (30%) and treated PE (3%) and second, in the risk of potentially fatal hemorrhagic complications in patients undergoing anticoagulation, thus mandating the patients without PE is not anti-coagulated [15, 16]. Because catheter angiography, the reference standard for the diagnosis of PE is time-consuming, not widely available and has a rare but fatal complication risk [17–21], diagnostic strategies for PE are complex and focus on pre-selecting patients with higher risk of PE who would benefit most from imaging of the pulmonary arteries [22–28]. These strategies include blood D-dimer measurement [22, 23] and/or a search for lower extremity DVT with duplex, compression and colour-flow sonography [24–28], which obviates the need for pulmonary artery imaging if positive as treatment of both

Catheter angiography is widely accepted as the technique most likely to rule-in or rule-out pulmonary embolism, given its exquisite resolution and the fast that small sub-segmental arteries can be visualised [1, 28–32]. Likewise, given the resolution constraints of cross-sectional methods for the diagnosis of PE, it is acknowledged that diagnosis of subsegmental embolism with cross-sectional techniques is imperfect. However, because of the many published studies lauding the accuracy of CTA, catheter angiography is no longer widely used. Therefore, if we accept that DSA is the gold standard, how do we determine a possible role for new or improved techniques with higher resolution? The intermittent use of pulmonary arteriography for clinical trials solely to determine accuracy of a new technique test is unacceptable, considering its radiation dose, risk of local complications, and nephrotoxicity [33] in addition to the small but significant mortality rate [30, 31]. However, before accepting that DSA is the only appropriate reference standard for validation of competing modalities, the question of accuracy of DSA must be addressed (Table 19.1).

How Accurate Is Catheter Angiography and Does It Stand Up as a Gold Standard? Despite DSAs’ unrivalled in-plane resolution, its throughplane resolution is minimal and careful correlation between DSA and other tests reveals several flaws. For example, review of the original PIOPED data (VQ versus catheter angiography) demonstrated that although there was excellent inter-observer agreement for PE between experienced pairs of reviewers, the agreement for blinder evaluation of the subsegmental arteries was extremely poor [14, 18, 30]. Wittram et al. [31] in a detailed comparison of discordant findings between CTA and catheter arteriography in 20 patients from the follow-up PIOPED II study found 13 patients with positive findings at CTA and were negative at DSA. The largest thrombi missed at DSA were sub-segmental in eight patients, segmental in two and lobar in three. Likewise, Qanadli [32] reported discordance between CTA and DSA in 6 of 157 cases, with three positive and three negative for PE at CTA. Consensus evaluation by experienced reviewers reported that two of the discordant readings were true positive at CTA and false negative at angiography, one was false negative at CTA, two were false negative, and one false positive at DSA [33]. Of note was the fact that significantly more emboli were

19

Pulmonary MRA

255

Table 19.1 Limitations of DSA Limitation There is exquisite in-plane resolution, but no through-plane resolution May not be possible in patients due with renal impairment May not be possible in patients due with hypersensitivity Optimal technique not universally agreed, and selective injection of each pulmonary artery not always performed Balloon occlusion with injection into the isolated arterial segment gives the greatest likelihood of demonstrating sub-segmental emboli, but is almost never used High inter-observer error for small arteries Accuracy has been questioned in carefully controlled comparisons with MRA and CTA in cases of discordance

revealed by CTA compared to DSA (92 versus 56). A further confounding observation in the correlation of findings at cross-sectional imaging and DSA relates to the well-known fact that thrombi can form on catheters during angiography, thus inducing the very pathology being sought by the test, and also the fact that thrombi frequently dissolve rapidly, giving yet another reason for discordance between DSA and non-invasive angiographic techniques [31–34]. Therefore, catheter angiography is an imperfect standard for the diagnosis of PE, and its abandonment for routine diagnosis in the light of potential side effects appears warranted [17–21, 33]. A pragmatic recent approach has been to use a composite reference standard composed of Wells clinical criteria, CTA, and VQ scan results in association with clinical likelihood to validate the accuracy of newer tests.

Spatial Resolution, Anatomic Coverage, Sub-segmental Emboli, and the Potential for Non-invasive Cross-Sectional Imaging of PE Concerns regarding the accuracy of cross-sectional imaging techniques for the detection of small PE [35–41] may be overcome by multidetector-row (64–320 slice) spiral CT (isotropic spatial resolution of 0.6 mm) and highly accelerated parallel imaging techniques for MR which now allows visualisation of pulmonary vessels down to sixth order branches (first two divisions beyond the segmental arteries) but whether this improved visualisation of peripheral branches leads to improved detection rate, and whether this in turn leads to improved outcome is uncertain. The improved inter-observer correlation of sub-segmental emboli detection with high-resolution multidetector-row spiral CT which exceeds that of pulmonary angiography suggests that small peripheral emboli detected at CT might indeed be a true finding [31, 32]. But, outcome data from numerous follow-up

Comment Can be partially negated by performing multiple views Can be addressed by pre- and post-hydration, use of N-acetyl cysteine and LOCM Can pre-treat with steroids

CT studies (and DSA) where excellent outcome has been reported in patients with a negative scan for PE using older (single slice) technology which has lower resolution (and which might miss small PE), raises questions as to whether patients with such minimal thrombo-embolic disease should be exposed to the inherent risks of anticoagulation therapy [42–45] (Table 19.2). It is appropriate to look to the mechanism by which emboli reach the lung and what size of artery is occluded. As an embolus traverses the right cardiac chambers churning motion of the heart breaks up the clot into many fragments, which then reach the pulmonary circulation coming to lodge in the smallest artery accessible. Clearly the emboli vary in size, and, for example, a large fragment can present as a central type saddle embolus, whereas a similar sized embolus in another patient could fragment into multiple smaller fragments which would shower into the pulmonary circulation. However, as treatment decisions are based on a simple dichotomy (if PE is present, then anticoagulation warranted; if PE is not present, then anticoagulation withheld), determining the presence of any embolus allows appropriate decision making, a fact that is aided by the observation that isolated sub-segmental embolism is reasonably unusual. However, this point is also disputed and recent studies put the incidence of sub-segmental PE at between 6 and 30% [46–48] (Table 19.3). Therefore, if we accept that sub-segmental embolism is not uncommon, a non-invasive test that did not have sufficient resolution to confirm sub-segmental PE would only be acceptable if (isolated) sub-segmental PE is insignificant. However, it has been established in followup of pooled data from several thousand patients with negative spiral CT that patient outcome was not adversely affected, if anticoagulation was withheld based on a negative spiral CT test with earlier CT approaches (mostly single slice) [45]. As it is highly likely that many of these patients had sub-segmental embolism, it would appear logical that

256

J.F.M. Meaney and P. Beddy

Table 19.2 Arguments supporting and against importance of sub-segmental PE Evidence cited in support of sub-segmental embolism being unimportant The high inter-observer difference on DSA for sub-segmental PE implies that many cases are missed. However, despite this, patients with negative DSA (some of whom must have PE) have a low incidence of PE or sudden death at early follow-up There was only one death in 20 PIOPED patients with isolated sub-segmental PE missed on initial evaluation of angiograms but picked up at subsequent review

Multiple PEs unrelated to the cause of death are frequently found at autopsy

Evidence cited in support of sub-segmental embolism being important Logic dictates that multiple sub-segmental emboli, obstructing the same vascular territory as a larger central embolus, would have an identical effect

Logic dictates that an embolus broken up into multiple tiny fragments that lodge within sub-segmental arteries only because of their small size, probably has the same clinical significance as the “same” embolus, broken into fewer but larger fragments that lodge within segmental and larger arteries Autopsy studies in patients dying from PE often reveal smaller recent PE’s raising the possibility that a small PE may “herald” a larger potentially fatal one

Pooled data from 3,500 patients with negative spiral CT showed that patient outcome was not adversely affected, if anticoagulation was withheld based on a negative spiral CT test

Table 19.3 Incidence of isolated sub-segmental PE PIOPED Oser Stein Diffin Oudkerk de Monteye Oudkerk

References [54] [48] [55] [44] [56]

Incidence of isolated sub-segmental PE (%) 30 6 7 18 22 3

diagnosis of PE to segmental level only was acceptable, and that diagnosis of smaller emboli, although preferable, might not affect the outcome [45].

MRA for PE Initial non-contrast techniques for evaluation of the pulmonary arteries were unreliable [49–54] and were rapidly supplanted by contrast-enhanced techniques which became the first reliable MRA technique for pulmonary embolism diagnosis (Figs. 19.2 and 19.3) [55–59]. Scans were performed in the coronal plane and excluded sub-segmental arteries within the periphery of the lung both anterior and posterior to the coronally oriented 3D imaging volume [55, 56]. This situation was broadly analogous to early CTA approaches which evaluated the pulmonary circulation from approximately the level of the aortic arch to the level of the diaphragm. Because of spatial resolution constraints of MRA, even those subsegmental arteries included within the FOV (within the lateral parts of both lungs) were beyond the resolution capability. This situation was rapidly improved upon, by using a sagittal double-slab, dual injection technique [57]. Each lung was

Fig. 19.2 Patient with saddle pulmonary embolus. (a) Coronal wholevolume MIP from a CE-MRA demonstrating a linear filling defect straddling both left and right main pulmonary arteries consistent with a saddle embolus. (b) Coronal sub-volume reformat confirms the filling defect straddling both left and right main pulmonary arteries consistent with a saddle embolus. (c) Axial GRE acquisition demonstrates a linear filling defect within the right main pulmonary artery, although image quality is impaired compared to the breath-hold CE-MRA. (d) Axial balanced image demonstrating the pulmonary embolus in addition to a small right pleural effusion

evaluated separately at higher spatial resolution and inclusion of the entire pulmonary vasculature could be achieved. More recently, the development of time-resolved and faster scanning with parallel imaging techniques that allow improved spatial and temporal resolution has resulted in an MRA technique now closely matches the resolution of CTA [60–66]. In addition, the ability to determine pulmonary

19

Pulmonary MRA

257

The frequently cited advantages of superior mediastinal and parenchymal evaluation by CT is less certain nowadays as MR offers similar capability, without the attendant risks of radiation exposure. Although the risks of contrast-induced nephropathy in CT [33] must be weighed against the risks of NSF [83–85], standard gadolinium contrast dose and careful selection of a low risk (for NSF) agent can eliminate entirely the risk of NSF, whereas the risk of CIN is much higher and harder to eliminate completely [33, 84]. Two potentially significant but largely unexplored advantages of MR over CT for the diagnosis of PE are the use of blood-pool contrast agents which allow data acquisition during breath-holding [86–89] and the use of non-contrast techniques [90–96]. Because of severe motion artefact from cardiac and diaphragmatic motion coupled with long scan times, non-contrast techniques have not made a significant impact on clinical practice. Nonetheless, with resurgence in interest in non-contrast MRA in patients with impaired renal function at risk of NSF, the promise offered by a completely non-invasive technique is appealing [90].

MR Technique for Diagnosing PE Comparison of CE-MRA with Catheter Angiography

Fig. 19.3 (a) Axial subtracted MRA demonstrates contrast surrounding embolic material within both the right and left main pulmonary arteries (arrows). (b) Axial and (c) coronal real-time MR images in the same patient show bilateral pulmonary emboli [arrow (b), broken arrow (c)] (Courtesy of Emily Ward, MD, Northwestern University Feinberg School of Medicine, Chicago, IL.)

perfusion with MR (usually performed with a dynamic coronal approach during breath-holding following injection of a small volume of contrast agent), which may improve confidence in diagnosing small PE, promises additional benefit [67–73]. Although not widely available at present, MR assessment of ventilation may add to the approach, harking back to an earlier era of VQ scanning, albeit with MR with the additional benefit of angiographic imaging at high resolution [74–79]. Yet another advantage of MR is the ability to more comprehensively evaluate RV function in acute PE; although some information can also be obtained with CT, this may be of benefit primarily in chronic thrombo-embolic pulmonary embolism [80–82] (Table 19.4).

The initial two studies, published in 1997 (23 patients) and 1999 (46 patients), examined both lungs in the coronal plane and addressed the arteries to segmental level only. They, respectively, reported sensitivity of 85–87% and specificity of 95–96% [55, 56]. Oudkerk’s more comprehensive study of 141 patients in whom comparison of MRA and DSA was available in 115 patients was performed at much higher resolution by using the double sagittal approach [57]. These authors reported sensitivity and specificity of 77% but missed PE in three of five patients with sub-segmental PE only. Of note was the fact that MRA detected PE in two patients with normal DSA, for a specificity of 98%. These three studies benefitted from the higher accuracy typically reported from single centres which benefit from a high concentration of local expertise, however, a multicentre trial of MRA versus other modalities, PIOPED III, was recently reported (Table 19.5).

MRA Versus CTA for Diagnosis of PE Ohno et al. [63] reported clear superiority of MRA (4-s timeresolved technique) over MDCT (4-slice) and VQ scanning in patients with suspected PE using what was at that time state-of-the-art technology for both CT and MR. They also reported excellent results for MRA compared to catheter pulmonary angiography (Table 19.6).

258

J.F.M. Meaney and P. Beddy

Table 19.4 Comparison of different techniques for pulmonary MRA Technique First pass MRA Time-resolved imaging

Non-contrast MRA

Advantages Well understood and widely practised technique Excellent for patients with limited breath-holding capability Can “isolate” the arterial phase, therefore, no problems with venous overlay etc. Higher-resolution imaging possible. Images can be acquired with navigators during free breathing Better resolution than isotope techniques Can focus attention on the arteries supplying an area with poor perfusion Role uncertain but in combination with perfusion gives information analogous to isotope VQ scanning Eliminates risks associated with gadolinium use

Cardiac (RV) assessment

Particularly useful in patients with impaired renal function due to risk of NSF Most useful for the assessment of RV function in patients with chronic thromboembolism

Blood-pool agent imaging Perfusion

Ventilation scanning

Disadvantages Artefacts, primarily from respiratory motion Sacrifices spatial resolution in favour of temporal resolution

Poor contrast between flowing blood and embolus in some cases Long scan times requiring multiple breath-holds or triggering which introduces artefacts Significantly adds to examination time

Table 19.5 Accuracy of MRA for the diagnosis of PE Author Meaney Gupta Oudkerk Pleszwski PIOPED III

Year 1997 1999 2002 2006 2010

Number of patients 23 46 141 48 371

Arbiter DSA DSA DSA Composite Composite

Sensitivity (%) 87 85 77 82 78

Specificity (%) 95 96 98 100 90

Table 19.6 MRA and CTA verses angiography: diagnostic capability of data-sets per vascular zone Vascular zone All zones MRA CTA Central vessels MRA CTA Peripheral vessels MRA CTA

Sensitivity (%)

Specificity (%)

PPV (%)

83 75

97 97

64 64

99 98

96 96

100 100

99 99

92 97

100 100

99 99

68 54

96 96

47 42

98 98

95 94

PIOPED III The PIOPED III study was a multicentre trial, published in 2010, designed to determine the sensitivity and specificity of contrast-enhanced MRA alone and in conjunction with contrast-enhanced MRV of the lower extremity veins [59]. Acknowledging the fact that DSA has all but disappeared from clinical practice, PIOPED III used a composite reference

NPV (%)

Accuracy (%)

standard composed of CTA, Wells clinical criteria, D-Dimer assay, sonovenography, and (very occasionally) VQ scanning. Patients were recruited from seven acute hospital sites within the USA. Three hundred and seventy-one of 818 patients had reliable data that could be evaluated, of whom 104 had PE whilst 267 did not. Criteria for the diagnosis of PE were definite positive findings at CTA/CTV in 98% of cases and high probability VQ scan result in the remaining 2% of patients with Wells criteria indicating high likelihood of PE. The majority of patients

19

Pulmonary MRA

259

were scanned at 1.5 T, although one centre recruited exclusively at 3 T. In accordance with evolving clinical concerns regarding MR contrast agent safety, half way through the study a doubledose regime was replaced with single dose. PE was excluded by negative CTA/CTV, normal D-dimers, negative sonovenography, and exceptionally negative VQ scan result. Disappointingly 25% of patients had a nondiagnostic scan, although this varied between 8 and 52% for the different studies. Of patients with technically adequate studies, the sensitivity was 78% (CI 67–86%) and specificity was 99%. Amongst patients with technically adequate studies, sensitivity and specificity were 92% (83–97%) and 96% (91–99%), respectively. With prevalence of PE between 11 and 50% across the recruiting centres, they calculated a positive predictive value of 74–98% and a negative predictive value of 92–99%. Inter-centre comparison revealed sensitivity from 45 to 100% and specificity from 95 to 100%. Only one patient in the study had sub-segmental PE only. Several animal studies have also addressed accuracy of CTA versus MRA under ideal conditions.

Additional MR Techniques Time-Resolved MRA Extremely rapid imaging of the pulmonary vasculature in an effort to capture an arterial phase is possible, albeit at the expense of spatial resolution [64–66]. Thus, a “clean” arterial phase is acquired in favour of a higher-resolution scan that can depict sub-segmental arteries. Nonetheless, in a study comparing time-resolved MRA, MDCT, combined CE-MRA plus MDCT, VQ scanning alone, and VQ scanning plus MDCT (Tables 19.8 and 19.9), Ohno et al. reported that time-resolved CE-MRA with a 4-s acquisition was superior to all other non-invasive modalities including MDCT, particularly for the sub-segmental arteries [63]. This approach has also been validated in patients with contra-indications to iodinated contrast material, in whom neither CTA nor arteriography were appropriate. Ersoy et al. reported excellent results for fast (3.3 s) validated by available reference standards and follow-up [102].

Accuracy of MRA in Animal Studies Because of ethical and safety issues, validation of technique from animal studies provides an additional comfort level for new cross-sectional imaging modalities for PE [97–101]. However, animal studies are carried out in suspended respiration during general anaesthesia and do not reflect the situation in humans where respiratory motion artefact is frequently problematic. Haage reported superiority of MRA over CTA and superiority of real-time MRA over “static” contrast-enhanced MRA [97]. Seo also reported increased accuracy of MRA over CTA [101], whereas in an earlier study by Hurst [98] CTA had slightly better accuracy. A more recent study reported slightly better diagnostic accuracy of (dual energy) CTA versus unenhanced MRA, enhanced MRA, and pulmonary perfusion at 3 T although the difference was not statistically significant [100] (Table 19.7).

Perfusion Imaging Perfusion imaging with isotope scintigraphy is limited for the diagnosis of PE, but MRI perfusion with its superior spatial and temporal resolution offers potential benefit [67–73]. Although studies comparing accuracy of scintigraphic and MR pulmonary perfusion are similar, reproducibility is greater with MR. It is postulated that MR pulmonary perfusion may lead to increased accuracy, as perfusion defects would focus attention on the arteries supplying this area, and may lead to repeat imaging of the relevant area at higher resolution, particularly with blood-pool agents. MRPP is performed with a low dose (2–4 cc) of gadolinium contrast agent prior to the 3D CE-MRA and does not interfere with the subsequent angiographic examination.

Table 19.7 Animal studies of CTA versus MRA for diagnosis of PE Author Hurst Haage Seo` Zhang

Year 1999 2003 2003 2010

Number of species 7 Dogs 9 Pigs 5 Pigs 18 Rabbits

CTA Sensitivity (%) 64–76 71 61 94

Specificity NS 96 97

MRA Sensitivity (%) 48–52 80–97 90 89–94

Specificity NS 87 89

Size of emboli 3.7 mm 4 mm 3 mm 2 mm

Gold standard DSA + autopsy DSA Autopsy Autopsy

260

J.F.M. Meaney and P. Beddy Table 19.8 Accuracy of CE-MRA, MDCT, and combined CE-MRA + MDCT on a per vascular zone basis Modality Overall CE-MRA MDCT CE-MRA + MDCT Central CE-MRA MDCT CE-MRA + MDCT Peripheral CE-MRA MDCT CE-MRA + MDCT

Sensitivity (%)

Specificity (%)

PPV (%)

NPV (%)

Accuracy (%)

83 75 83

97 97 97

64 64 65

99 98 99

96 96 96

100 100 100

99 99 99

92 97 95

100 100 100

99 99 99

68 54 68

96 96 96

47 42 47

98 98 98

95 94 95

Overall results (all vascular segments) and results for central and peripheral segments are presented (Ohno)

Table 19.9 Accuracy of CE-MRA, MDCT, combined CE-MRA + MDCT, VQ scanning, and combined VQ + MDCT on a per patient basis Modality CE-MRA MDCT CE-MRA + MDCT VQ scanning MDCT + VQ

Sensitivity (%) 92 83 92 67 92

Specificity (%) 94 94 94 78 94

Blood-Pool Imaging Despite many agents being proposed [86–89], only a single blood-pool contrast agent (Ablavar) has been released commercially to date [87]. No prospective comparative study of blood-pool agent compared to competing modalities currently exists, nonetheless, because of the long vascular persistence time, repeated breath-hold studies targeted to a potentially abnormal area can be performed at high resolution after an initial screening examination performed at lower resolution. Alternatively, high-resolution breath-hold images of the entire pulmonary vasculature can be performed in multiple acquisitions tailored to the breath-hold capability of the patient. Another possibility, especially promising in patients with severe respiratory compromise is the use of navigator-echo techniques to eliminate the need for breathholding. Blood-pool agents offer superior ability to detect clots within the lower extremities in patients with suspected PE, because of their long intravascular half-life.

Non-contrast MRA Kluge et al. evaluated the diagnostic value of real-time magnetic resonance imaging for the diagnosis of acute pulmonary embolism compared to contrast-enhanced MRA in 39

PPV (%) 85 83 85 50 85

NPV (%) 97 94 97 88 97

Accuracy (%) 94 92 94 75 94

consecutive patients with suspected PE using real-time true fast imaging with steady-state precession (TrueFISP) at 1.5-T [90]. Scan time of 0.4 s per image allowed visualisation of the pulmonary vasculature in three orientations in <3 min without the need for breath-holding or contrast media injection. Hundred percent of TrueFISP MRA’s were of diagnostic quality compared to only 77% of CE-MRA’s (breathing artefacts in dyspneic patients). They reported sensitivities and specificities of 93 and 100% (per examination), 96 and 100% (lobar artery PE), and 97 and 100% (segmental artery PE), respectively, for TrueFISP examinations.

Direct Thrombus Imaging Direct thrombus imaging, based on the principle that changes in MR signal intensity occur due to reproducible changes in blood clots over time, theoretically offers an ideal method for non-invasive diagnosis of PE [91]. The technique employs a heavily T1-weighted sequence which emphasises the presence of methemoglobin which has characteristically low T1 values [92–96]. As time-dependent changes in MR appearance reflect evolution of the clot, high signal is seen in new clots only. The technique also offers promise for the detection of acute DVT for the same reason [94, 95]. Further advantages include ability

19

Pulmonary MRA

261

to scan in the coronal plane, thus allowing time-efficient coverage of the lower extremity veins with two coronally overlapping acquisitions in a relatively short scan time. Moody et al., using a water-excitation only magnetisation prepared for gradient-echo sequence (inversion time chosen to suppress flowing blood) to evaluate the lower extremity veins and pulmonary vasculature for T1-bright clots reported sensitivity of 94–96% and specificity of 90–92% in a large study of patients with DVT. They also reported a study assessing the role of direct thrombus imaging in patients with PE [93, 96].

Where Does MRA Currently Fit into Diagnostic PE Algorithms and Is It Appropriate to Use It Instead of CTA? Currently, MR is evolving at a dramatic pace and the issue of sub-segmental artery visualisation (whatever its importance) will be addressed by widespread availability of isotropic or near sub-millimetre data-sets in the near future. Until this time, MRA will probably remain a second-line investigation in patients in whom CTA is contra-indicated because of the widespread availability and ease of performance of CTA. However, it is widely acknowledged that MRA is safer than CTA, although the higher number of poor quality studies is a mitigating factor. If proxy arbiters of MRA accuracy for sub-segmental embolism are accepted and if MRA and CTA were deemed equivalent or identical in terms of ability to image small arteries successfully, MR would offer benefit over CTA due to the safer profile of MR contrast agents and absence of ionising radiation. Although MR remains a less widely available modality than CT, it offers numerous additional possibilities for evaluation of the pulmonary vasculature including blood-pool imaging, perfusion imaging with either contrast-enhanced or non-contrast techniques (e.g., spin labelling), newer non-contrast approaches such as TrueFISP imaging, and direct thrombus imaging that can evaluate both the pulmonary arteries and lower extremity veins. Even newer approaches incorporating ventilation imaging, in association with MR perfusion and MRA, are coming into the clinical arena and may offer further benefit.

Pulmonary Hypertension Pulmonary hypertension poses a diagnostic challenge. Secondary causes include chronic thromboembolism, emphysema, interstitial pulmonary fibrosis, and shunts [80–82]. In less than 1% of survivors of acute PE, emboli do not resolve but remain in situ as Web-like constrictions and stenoses within arteries. In patients pre-disposed to recurrent PE, CTEPH (chronic thrombo-embolic pulmonary hypertension) occurs when approximately 60% of the pulmonary vascular bed is affected. Once mean pulmonary arterial pressures reach 30 mmHg, patients are usually severely dyspnoeic with

Fig. 19.4 Patient with pulmonary hypertension. Note enlargement of the central pulmonary arteries with normal calibre of the peripheral arteries indicating pruning. There is no evidence of intravascular filling defects and chronic thrombo-embolic pulmonary hypertension can be excluded

right-sided heart failure and anticipated 5 year survival of 30%. Pharmacological intervention is limited in these patients and surgical thrombo-endarterectomy, which offers the only hope of cure, depends on demonstration of organised thrombi in central and main pulmonary arteries. Catheter pulmonary angiography is still used for pre-operative evaluation, but gives at best limited functional information. Today, most information required for therapeutic planning is provided non-invasively by MRA (Fig. 19.4) or CTA which provides excellent information regarding the location of clots [80–82]. MRI/A can provide not only information regarding the presence, location, and size of clot but also address functional parameters of pulmonary perfusion and myocardial function [80–82]. Kreitner et al. combined morphological assessment of the pulmonary arteries with CE-MRA and functional assessment by MR with short-axis cine GRE images and ECG-gated phase contrast velocity-encoded segmented k-space acquisitions in 34 patients with CTEPH [80]. CE-MRA revealed all 533 arteries to segmental level depicted by DSA and 681 of 733 (93%) patent sub-segmental arteries shown by DSA. Functional RV imaging confirmed findings typical of CTEPH such as reduced right ventricular but normal left ventricular ejection fractions, paradoxical septal motion, and differences in flow rates through the pulmonary arteries and ascending aorta (attributed to recruitment from the bronchial circulation in CTEPH). Follow-up MRI/MRA after endarterectomy is also similar to comprehensive morphological and functional assessment.

262

J.F.M. Meaney and P. Beddy

Miscellaneous Disorders of the Pulmonary Circulation Systemic Arterial Supply to the Lung Bronchial arterial embolisation for massive and recurrent hemoptysis has since been established as a safe and effective procedure for the treatment of bronchiectasis [103, 104]. Chronic vascular obstruction by disorders such as Takayasu’s arteritis and chronic thromboembolism can also cause anastomoses between the bronchial and pulmonary arterial systems, resulting in collateralisation of the stenosed or obstructed pulmonary arterial bed by the systemic bronchial component of the dual pulmonary blood supply with subsequent hypertrophy of the bronchial arteries.

MRA of Bronchial Arteries Although bronchial artery anatomy is variable, most originate from the descending thoracic aorta close but usually around T5–T6 level at or slightly below the carina. There are several classical branching patterns as follows: 1. Two arteries on the left and one on the right which arises as an inter-costobronchial trunk (ICBT), 40%. 2. One artery on the left and one ICBT on the right, 21%. 3. Two on the left and two on the right one of which is a ICBT, 20%. 4. One on the left and two on the right one of which is a ICBT, 9.7%. Although no study has evaluated the role of MRA in patients with haemoptysis requiring bronchial artery embolisation, O’ Keefe et al. reported visualisation of the right and left bronchial arteries in 82.5 and 12.5% of subjects, respectively, in subjects undergoing pulmonary MRA for the evaluation of pulmonary vein architecture prior to radiofrequency ablation [105]. Although these results are inferior to those of MDCT [103, 104], possible explanations include the fact that the examination was carried out at lower resolution as it was tailored for the larger calibre pulmonary veins, and the 3D MRA was timed to peak enhancement within the pulmonary veins (and not the thoracic aorta) with resultant sub-optimal aortic enhancement in some patients, thus making evaluation of the bronchial artery anatomy more difficult.

Vascular Imaging in the Context of Malignancies of the Pulmonary Circulation Although most relevant information is provided by contrastenhanced CT (usually from the staging scan), MR pulmonary angiography of the pulmonary circulation enable us to

Fig. 19.5 Patient with left upper lobe lung cancer and incidental leftsided PE. (a) Axial spin-echo triggered T1-weighted image at the level of the hilum demonstrates a tumour (T) around the left hilum. (b) Coronal HASTE image confirms a large mass within the left upper lobe extending to the left hilum. (c) Axial-triggered balanced sequence (FFE) demonstrates the lung tumour, in addition to a filling defect within the left lower lobe consistent with an embolus. (d) Contrast-enhanced MRA demonstrates irregularity and narrowing of the left main pulmonary artery at the level of the mass which corresponds to a tight stenosis on the whole-volume MIP at the level of the tumour. There is extensive nonenhancement within the centre of the tumour consistent with necrosis. Note that the tumour is not visualised on the whole-volume MIP as it is not sufficiently enhancing to appear on the MIP

determine the exact location and extent of tumour mass with respect to pulmonary vessels, and tumour vascularity can be assessed (Fig. 19.5) [1]. Visualisation of the pulmonary vessels may play a crucial role for surgical planning, since the appropriate surgical approach is determined by the relation of the tumour to vital anatomy such as the main bronchi and the central pulmonary vessels. Involvement of the great vessels, major veins, and heart can be performed. Post-surgical complications, such as arterial strictures and bronchoarterial fistulas following pulmonary artery reconstruction, or clot in the pulmonary artery stump after lobectomy can be readily identified on follow-up [1]. Pulmonary artery sarcomas are rare malignancies of obscure aetiology [106–108]. Most pulmonary artery sarcomas arise from the dorsal area of the pulmonary trunk, although the tumours also may arise from the right and left pulmonary arteries, the pulmonary valve, and the right ventricular outflow tract. Because of its rarity and insidious growth characteristics, pulmonary artery sarcoma is often mistaken for pulmonary embolism, resulting in inappropriate therapy such as prolonged anticoagulation or thrombolysis [108]. Systemic symptoms and signs such as weight loss, fever, anaemia, and digital clubbing may be subtle clues to

19

Pulmonary MRA

263

diagnosis. Other characteristics, such as the absence of risk factors for deep vein thrombosis, high sedimentation rate, nodular parenchymal infiltrates, and lack of response to anticoagulation should raise the suspicion of a process other than pulmonary embolism. Cross-sectional imaging for a precise pre-operative assessment is crucial for the appropriate exploration of the pulmonary artery to ensure complete resection and reconstruction.

Imaging of Congenital Abnormalities of the Pulmonary Circulation Anomalous Origin of the Left Pulmonary Artery from the Right: “Pulmonary Artery Sling” Anomalous origin of the left pulmonary artery is usually diagnosed in infancy because of airway compression and associated tracheal or bronchial stenosis due to complete cartilage rings. This can result in obstruction, feeding problems, and respiratory tract infections. Occasionally, the abnormality may be detected as an incidental finding in an asymptomatic adult or in the adult with respiratory complaints. The aberrant left pulmonary artery originates from the right pulmonary artery and travels across the midline posterior to the distal trachea or right main bronchus, where it turns abruptly to the left, passing between the oesophagus and trachea to its destination in the left hilum. This anomalous artery has been called a sling. MRA with 3D visualisation of complicated vascular anatomy for surgical planning are vital prerequisites for appropriate surgical planning of this and similar congenital disorders [1]. Pulmonary Sequestration. In this condition, part of the lung does not communicate with the bronchial tree and the nonfunctioning lung receives its supply from the systemic circulation, usually the descending thoracic aorta. The condition is frequently asymptomatic in early life and only discovered in adulthood in patients who present with recurrent infections, haemoptysis, and bronchiectasis in whom an abnormal arterial supply to a segment of lung is visualised on imaging (Fig. 19.6).

Fig. 19.6 Pulmonary sequestration: High-resolution MRA demonstrating pulmonary artery sequestration showing an abnormal vessel (arrow) supplying the right lung extending from the abdominal aorta (Courtesy of Emily Ward, MD, Northwestern University Feinberg School of Medicine, Chicago, IL.)

smooth nature and almost invariable presence of two associated vessels, the smaller usually representing the feeding artery, the larger typically represents the (arterialise and therefore enlarged) vein. Apart from detection, MR angiography is useful for pre-therapeutic evaluation of the angioarchitecture of pulmonary AVMs, especially for assessing the number and configuration of feeding and draining vessels connected to the aneurysmal sac (Fig. 19.7) [110]. Whole body MRA may be of use in detecting extrapulmonary AVMs around the body.

Pulmonary Veins Pulmonary Arteriovenous Malformation Pulmonary arteriovenous malformation (AVM) can occur in isolation, be multiple, or be part of a systemic process where arteriovenous communications occur in the skin, mucous membranes, and other organs (hereditary hemorrhagic telangiectasia or Osler–Rendu–Weber disease) [109, 110]. Pulmonary AVMs vary in size from a few millimetres to several centimetres. They are recognised on MRA by their

It was only with the advent of pulmonary vein ablation for atrial fibrillation that there was a need to identify individual pulmonary vein tributaries, the relevant factor being the drainage pattern of the pulmonary trunks into the left atrium (defined as the distance from the ostium to the first tributary, usually longer for the upper lobe pulmonary veins than the lower lobe pulmonary veins) (Fig. 19.8) [111–113]. The fact that there is enormous variation in pulmonary vein anatomy within the lungs is largely irrelevant.

264

J.F.M. Meaney and P. Beddy

Fig. 19.7 Patient with Osler–Rendu–Weber syndrome (hereditary hemorrhagic telangiectasia). (a) Chest X-ray showing rounded retrocardiac mass on the left. (b) Axial post-contrast CTA demonstrates an arteriovenous malformation posteriorly within the left lower lobe. (c) Axial CE-MRA demonstrates an almost identical appearance. (d) Catheter angiography confirms the presence of an AVM. (e) Sagittal whole-volume MIP from selective CE-MRA performed on the left in the sagittal plane demonstrates the large posterior AVM visualised on other imaging modalities and two additional AVM’s more anteriorly. (f) Sagittal wholevolume MIP from selective CE-MRA performed on the right in the sagittal plane demonstrates several small AVMs

Pulmonary venous anatomy does not mirror pulmonary artery anatomy. The confluence of the pulmonary veins with the left atrium lies on a more posterior plane than the central pulmonary arteries. The plane of the inferior pulmonary veins is substantially more posterior to that of the superior pulmonary veins [1]. The superior pulmonary veins, which arise from the confluence of smaller tributaries, course downward, posteriorly and medially before emptying into the postero-superior corner of the left atrium. The middle lobe pulmonary vein most commonly joins the right superior pulmonary vein just before it empties into the left atrium but may drain separately into the right atrium. The lingular vein almost always empties into the left superior pulmonary vein, and only rarely directly into the left atrium. The inferior pulmonary veins describe a horizontal course for several centimetres prior to emptying into the left atrium.

Pulmonary Vein Mapping Prior to Radiofrequency Ablation for Atrial Fibrillation Paroxysms of atrial fibrillation are initiated by spontaneous discharges originating from the pulmonary veins in 90–96% of patients. In most individuals, a sleeve of left atrial myocardium extends into the pulmonary veins for a variable (2–17 mm) distance. This sleeve is longest in the superior pulmonary veins and thickest at the venoatrial junction of the left superior vein, which may explain the fact that ectopic foci of atrial fibrillation most commonly arise from the left superior pulmonary vein. Successful ablation of all electric connections to these veins can permanently cure paroxysmal atrial fibrillation. Non-invasive road-mapping with CT or MR offers benefit to the interventional cardiologist and allows careful planning

19

Pulmonary MRA

265

Anomalous Venous Drainage Anomalous Pulmonary Venous Drainage with Normal Course Within the Lung In most instances, the veins follow a normal course within the lung, it is only the “termination” of the vein that is anomalous. Examples include partially anomalous pulmonary venous drainage (almost always asymptomatic) where a right upper lobe pulmonary vein drains into the superior vena cava [114, 115].

Anomalous Pulmonary Venous Drainage with Abnormal Course Within the Lung

Fig. 19.8 Normal pulmonary veins. (a) Whole-volume MIP demonstrates overlapping arterial and venous structures and poor selective visualisation of the pulmonary veins. (b) Sub-volume MIP targeted to the left pulmonary veins demonstrates normal course of the upper and lower pulmonary veins on the left. The proximal right upper and confluence of the right lower pulmonary vein with the right atrium are also visualised. (c) A slightly more posterior MIP demonstrates excellent of the left upper, left lower, and right upper lobe pulmonary veins. (d) Coronal oblique sub-volume MIP targeted to the right lower lobe pulmonary vein demonstrates normal appearance. (e) Axial sub-volume MIP demonstrates normal configuration of the inferior pulmonary veins bilaterally

of the procedure [111–113]. Prior knowledge of the number, location, size, and configuration of all ostial locations is essential to ensure that all can be ablated. Two pulmonary veins on each side (one upper and one lower) compose the most common arrangement, but pulmonary vein anatomy is variable. Both helical CT and MRA provide adequate demonstration of pulmonary vein location, number, osteal size, branching pattern, and length of the pulmonary vein trunk prior to ablation being performed and additionally offer an ideal method for detection of procedure-related complications such as pulmonary vein stenosis. In addition, atrial or atrial appendage thrombus, an absolute contraindication to the procedure can be excluded.

An example is the “Scimitar” syndrome, where drainage of the right lower lobe is to the inferior cava, just above the diaphragm [116, 117]. This condition is usually discovered incidentally on imaging studies of the thorax, however, it may assume relevance in patients with abnormal lung function or in patients in whom resection of another lobe or lobes might result in cardio-respiratory compromise as the oxygenated blood returning through the scimitar vein never reaches the systemic circulation as it passes around in an “endless” loop from pulmonary vasculature to right atrium to pulmonary vasculature [117].

Conclusion MRA of the pulmonary vasculature is a rapidly evolving, safe, and accurate modality for assessing the pulmonary vasculature and benefits from a safety profile unmatched by other competing non-invasive modalities such as CTA. Time-resolved imaging, blood-pool contrast agents, non-contrast techniques, pulmonary ventilation and perfusion, and the potential for functional cardiac evaluation make this an exciting tool for comprehensive evaluation of the pulmonary vasculature.

References 1. Schoepf UJ, Meaney JFM. Pulmonary vasculature. In: Rubin GD, Rofsky NM, eds. CT and MR Angiography: Comprehensive Vascular Assessment. Wolters Kluwer/Lippincott Williams and Wilkins: 2008. 2. Coon WW, Willis PW. Deep venous thrombosis and pulmonary embolism: prediction, prevention, and treatment. Am J Cardiol. 1959;4:611–621. 3. Barritt DW, Jordan SC. Clinical features of pulmonary embolism. Lancet. 1961;1:729–732.

266 4. Dalen JE, Alpert JS. Natural history of pulmonary embolism. Prog Cardiovasc Dis. 1975;17:257–270. 5. Bell WR, Simon TL. Current status of pulmonary embolic disease: pathophysiology, diagnosis, prevention, and treatment. Am Heart J. 1982;103:239–262. 6. National Institutes of Health Consensus Development. Prevention of venous thrombosis and pulmonary embolism. JAMA. 1986;256:744–749. 7. Saeger W, Genzkow M. Venous thromboses and pulmonary embolisms in post-mortem series: probable causes by correlations of clinical data and basic diseases. Pathol Res Pract. 1994;190:394–399. 8. Stein PD, Henry JW, Relyea B. Untreated patients with pulmonary embolism. Chest. 1995;107:931–935. 9. Stein PD, Henry JW. Prevalence of acute pulmonary embolism among patients in a general hospital and at autopsy. Chest. 1995;108:978–981. 10. Van Beek EJR, Kuijer PMM, Bueller HR, Brandjes DPM, PMM Bossuyt, ten Cate JW. The clinical course of patients with suspected pulmonary embolism. Arch Intern Med. 1997;157:2593–2598. 11. Silverstein MD, Heit JA, Mohr DN, Petterson TM, O’Fallon M, Melton LJ. Trends in the incidence of deep vein thrombosis and pulmonary embolism. Arch Intern Med. 1998;158:585–593. 12. Ginsberg JS. Management of venous thromboembolism. N Engl J Med. 1998;335:1816–1823. 13. Stein PD, Matta F. Epidemiology and incidence: the scope of the problem and risk factors for development of venous thromboembolism. Clin Chest Med. 2010;31:611–628. 14. Stein PD, Sostman HD, Dalen JE, et al. Controversies in diagnosis of pulmonary embolism. Clin Appl Thromb Hemost. 2011;17:140–149. 15. Van der Meer FJM, Rosendaal FR, Vandenbroucke JP, Briët E. Bleeding complications in oral anticoagulant therapy. An analysis of risk factors. Arch Intern Med. 1993;153:1557–1562. 16. Zidane M, Schram MT, Planken EW, et al. Frequency of major hemorrhage in patients treated with unfractionated intravenous heparin for deep vein thrombosis or pulmonary embolism: a study in routine clinical practice. Arch Intern Med. 2000;160:2369–2373. 17. Mills SR, Jackson DC, Older RA, Heaston DK, Moore AV. The incidence, etiologies, and avoidance of complications of pulmonary angiography in a large series. Radiology. 1980;136:295–299. 18. Stein PD, Athanasoulis C, Alavi A, et al. Complications and validity of pulmonary angiography in acute pulmonary embolism. Circulation. 1992;85:462–468. 19. Hudson ER, Smith TP, McDermott VG, et al. Pulmonary angiography performed with iopamidol: complications in 1434 patients. Radiology. 1996;198:61–65. 20. Zuckerman DA, Sterling KM, Oser RF. Safety of pulmonary angiography in the 1990s. J Vasc Interv Radiol. 1996;7:199–205. 21. Van Beek EJR, Reekers JA, Batchelor D, Brandjes DPM, Peeters FLM, Buller HR. Feasibility, safety and clinical utility of angiography in patients with suspected pulmonary embolism and non-diagnostic lung scan findings. Eur Radiol. 1996;6:415–419. 22. Hull RD, Raskob GE, Ginsberg JS, et al. A non-invasive strategy for the treatment of patients with suspected pulmonary embolism. Arch Intern Med. 1994;154:289–297. 23. Becker DM, Philbrick JT, Bachhuber TL, Humphries JE. D-dimer testing and acute venous thromboembolism. Arch Intern Med. 1996;156:939–946. 24. Perrier A, Bounameaux H, Morabia A, et al. Diagnosis of pulmonary embolism by a decision analysis-based strategy including clinical probability, D-dimer levels, and ultrasonography: a management study. Arch Intern Med. 1996;156:531–536. 25. Turkstra F, Kuijer PMM, van Beek EJR, Brandjes DPM, Buller HR, ten Cate JW. Value of compression ultrasonography for the detection of deep venous thrombosis in patients suspected of having pulmonary embolism. Ann Intern Med. 1997;126:775–781.

J.F.M. Meaney and P. Beddy 26. Wells PS, Ginsberg JS, Anderson DR, et al. Use of a clinical model for safe management of patients with suspected pulmonary embolism. Ann Intern Med. 1998;129:997–1005. 27. Perrier A, Desmarais S, Miron MJ, et al. Noninvasive diagnosis of venous thromboembolism. Lancet. 1999;353:190–195. 28. Weg JG. Current diagnostic techniques for pulmonary embolism. Semin Vasc Surg. 2000;13:182–188. 29. Stein PD, Henry JW, Gottschalk A. Reassessment of pulmonary angiography for the diagnosis of pulmonary embolism: relation of interpreter agreement to the order of the involved pulmonary arterial branch. Radiology. 1999;210:689–691. 30. Quinn MF, Lundell CJ, Klotz TA, et al. Reliability of selective pulmonary arteriography in the diagnosis of pulmonary embolism. AJR Am J Roentgenol. 1987;149:469–471. 31. Wittram C, Waltman AC, Shepard JA, Halpern E, Goodman LR. Discordance between CT and angiography in the PIOPED II study. Radiology. 2007;244:883–889. 32. Qanadli SD, Hajjam ME, Mesurolle B, et al. Pulmonary embolism detection: prospective evaluation of dual-section helical CT versus selective pulmonary arteriography in 157 patients. Radiology. 2000;217:447–455. 33. Aspelin P, Aubry P, Fransson SG, Strasser R, Willenbrock R, Berg KJ; Nephrotoxicity in High-Risk Patients Study of Iso-Osmolar and Low-Osmolar Non-Ionic Contrast Media Study Investigators. Nephrotoxic effects in high-risk patients undergoing angiography. N Engl J Med. 2003;348:491–499. 34. Stein PD, Yaekoub AY, Matta F, Janjua M, Patel RM, Goodman LR, Gross ML, Denier JE. Resolution of pulmonary embolism on CT pulmonary angiography. AJR Am J Roentgenol. 2010;194:1263–1268. 35. Remy-Jardin M, Remy J, Wattinne L, Giraud F. Central pulmonary thromboembolism: diagnosis with spiral volumetric CT with the single-breathhold technique—comparison with pulmonary angiography. Radiology. 1992;185:381–387. 36. Teigen CL, Maus TP, Sheedy PF, et al. Pulmonary embolism: diagnosis with electron-beam CT. Radiology. 1993;188:839–845. 37. Goodman LR, Curtin JJ, Mewissen MW, et al. Detection of pulmonary embolism in patients with unresolved clinical and scintigraphic diagnosis: helical CT versus angiography. AJR Am J Roentgenol. 1995;164:1369–1374. 38. Prokop M. Multislice CT angiography. Eur J Radiol. 2000;36: 86–96. 39. Ghaye B, Szapiro D, Mastora I, et al. Peripheral pulmonary arteries: how far in the lung does multi–detector row spiral CT allow analysis? Radiology. 2001;219:629–636. 40. Schoepf U, Holzknecht N, Helmberger TK, et al. Subsegmental pulmonary emboli: improved detection with thin-collimation multi– detector row spiral CT. Radiology. 2002;222:483–490. 41. Schoepf UJ, Costello P. CT angiography for pulmonary embolism: state of the art. Radiology. 2004;230:329–337. 42. Novelline RA, Baltarowich OH, Athanasoulis CA, Waltman AC, Greenfield AJ, McKusick KA. The clinical course of patients with suspected pulmonary embolism and a negative pulmonary arteriogram. Radiology. 1978;126:561–567. 43. Henry JW, Relyea B, Stein PD. Continuing risk of thromboemboli among patients with normal pulmonary angiograms. Chest. 1995;107:1375–1378. 44. Forauer AR, McLean GK, Wallace LP. Clinical follow-up of patients after a negative digital subtraction pulmonary arteriogram in the evaluation of pulmonary embolism. J Vasc Interv Radiol. 1998;9:903–908. 45. Tillie-Leblond I, Mastora I, Radenne F, Sandrine Paillard, Tonnel A, Remy J, Remy-Jardin M. Risk of pulmonary embolism after a negative spiral CT angiogram in patients with pulmonary disease: 1-year clinical follow-up study. Radiology. 2002;223:461–467.

19

Pulmonary MRA

46. Oser RF, Zuckermann DA, Gutierrez FR, Brink JA. Anatomic distribution of pulmonary emboli at pulmonary angiography: implications for cross-sectional imaging. Radiology. 1996;199: 31–35. 47. Diffin DC, Leyendecker JR, Johnson SP, Zucker RJ, Grebe PJ. Effect of anatomic distribution of pulmonary emboli on interobserver agreement in the interpretation of pulmonary angiography. AJR Am J Roentgenol. 1998;171:1085–1089. 48. de Monye W, van Strijen MJL, Huisman MV, Kieft GJ, Pattynama PMT. Advances in New Technologies Evaluating the Localisation of Pulmonary Embolism (ANTELOPE) Group. Suspected pulmonary embolism: prevalence and anatomic distribution in 487 consecutive patients. Radiology. 2000;215:184–188. 49. MacFall JR, Sostman HD, Foo TKF. Thick-section, single breathhold magnetic resonance pulmonary angiography. Invest Radiol. 1992;27:318–322. 50. Wielopolski PA, Haacke EM, Adler LP. Three-dimensional MR imaging of the pulmonary vasculature: preliminary experience. Radiology. 1992;183:465–472. 51. Grist TM, Sostman HD, MacFall JR, et al. Pulmonary angiography with MR imaging: preliminary clinical experience. Radiology. 1993;189:523–530. 52. Wielopolski P, Haacke E, Adler L. Evaluation of the pulmonary vasculature with three-dimensional magnetic resonance imaging techniques. MAGMA. 1993;1:21–34. 53. Schiebler ML, Holland GA, Hatabu H, et al. Suspected pulmonary embolism: prospective evaluation with pulmonary MR angiography. Radiology. 1993;189:125–131. 54. Erdman WA, Peshock RM, Redman HC, et al. Pulmonary embolism: comparison of MR images with radionuclide and angiographic studies. Radiology. 1994;190:499–508. 55. Meaney JFM, Weg JG, Chenevert TL, et al. Diagnosis of pulmonary embolism with magnetic resonance angiography. N Engl J Med. 1997;336:1422–1427. 56. Gupta A, Frazer CK, Ferguon JM, et al. Acute pulmonary embolism: diagnosis with MR angiography. Radiology. 1999;210:353–359. 57. Oudkerk M, Van Beek EJR, Wielopolski P, et al. Comparison of contrast-enhanced MRA and DSA for the diagnosis of pulmonary embolism: results of a prospective study in 141 consecutive patients with an abnormal perfusion lung scan. Lancet. 2002;359:1643–1647. 58. Pleszewski B, Chartrand-Lefebvre C, Qanadli SD, Déry R, Perreault P, Oliva VL, Prenovault J, Belblidia A, Soulez G. Gadoliniumenhanced pulmonary magnetic resonance angiography in the diagnosis of acute pulmonary embolism: a prospective study on 48 patients. Clin Imaging. 2006;30:166–172. 59. Stein PD, Chenevert TL, Fowler SE, Goodman LR, Gottschalk A, Hales CA, Hull RD, Jablonski KA, Leeper KV Jr, Naidich DP, Sak DJ, Sostman HD, Tapson VF, Weg JG, Woodard PK; PIOPED III (Prospective Investigation of Pulmonary Embolism Diagnosis III) Investigators. Gadolinium-enhanced magnetic resonance angiography for pulmonary embolism: a multicenter prospective study (PIOPED III). Ann Intern Med. 2010;152:434–443. 60. Korosec FR, Frayne R, Grist TM, Mistretta CA. Time-resolved contrast- enhanced 3D MR angiography. Magn Reson Med. 1996;36:345–51. 61. Sodickson DK, Manning WJ. Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med. 1997;38:591–603. 62. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999;42: 952–962. 63. Ohno Y, Higashino T, Takenaka D, Sugimoto K, Yoshikawa T, Kawai H, Fujii M, Hatabu H, Sugimura K. MR angiography with sensitivity encoding (SENSE) for suspected pulmonary embolism: comparison with MDCT and ventilation-perfusion scintigraphy. AJR Am J Roentgenol. 2004;183:91–98.

267 64. Goyen M, Laub G, Ladd ME, et al. Dynamic 3D MR angiography of the pulmonary arteries in under four seconds. J Magn Reson Imaging. 2001;13:372–377. 65. Peters DC, Korosec FR, Grist TM, et al. Undersampled projection reconstruction applied to MR angiography. Magn Reson Med. 2000;43:91–101. 66. Sonnet S, Buitrago-Tellez CH, Schulte AC, Bongartz G, Bremerich J. Dose optimization for dynamic time-resolved contrast-enhanced 3D MR angiography of pulmonary circulation. Am J Roentgenol. 2003;181:1499–503. 67. Mai VM, Chen Q, Bankier AA, et al. Imaging pulmonary blood flow and perfusion using phase-sensitive selective inversion recovery. Magn Reson Med. 2000;43:793–795. 68. Hatabu H, Tadamura E, Levin DL, et al. Quantitative assessment of pulmonary perfusion with dynamic contrast-enhanced MRI. Magn Reson Med. 1999;42:1033–1038. 69. Halliburton SS, Paschal CB, Rothpletz JD, Loyd JE. Estimation and visualization of regional and global pulmonary perfusion with 3D magnetic resonance angiography. J Magn Reson Imaging. 2001;14:734–740. 70. Amundsen T, Torheim G, Kvistad KA, et al. Perfusion abnormalities in pulmonary embolism studied with perfusion MRI and ventilation-perfusion scintigraphy: an intra-modality and inter-modality agreement study. J Magn Reson Imaging. 2002;15:386–394. 71. Fink C, Bock M, Puderbach M, Schmahl A, Delorme S. Partially parallel three-dimensional magnetic resonance imaging for the assessment of lung perfusion—initial results. Invest Radiol. 2003;38:482–488. 72. Fernandez GC. Pulmonary CE-MR angiography and perfusion imaging. Eur Radiol. 2004;14(Suppl 5):M32-M33. 73. Lin YR, Wu MT, Huang TY, Tsai SY, Chung HW, Mai VM, Chen CY, Pan HB. Comparison of arterial spin labeling and first-pass dynamic contrast-enhanced MR imaging in the assessment of pulmonary perfusion in humans: the inflow spin-tracer saturation effect. Magn Reson Med. 2004;52:1291–1301. 74. The PIOPED Investigators. Value of the ventilation-perfusion scan in acute pulmonary embolism. JAMA. 1990;263:2753–2759. 75. Kauczor H-U, Chen XJ, van Beek EJR, Schreiber WG. Pulmonary ventilation imaged by magnetic resonance: at the doorstep of clinical application. Eur Respir J. 2001;17:1–16. 76. Nakagawa R, Sakuma H, Murashima S, Ishida N, Matsumura K, Takeda K. Pulmonary ventilation-perfusion MR imaging in clinical patients. J Magn Reson Imaging. 2001;14:419–424. 77. Suga K, Ogasawara N, Okada M, Matsunaga N, Arai M. Regional lung functional impairment in acute airway obstruction and pulmonary embolic dog models assessed with gadolinium-based aerosol ventilation and perfusion magnetic resonance imaging. Invest Radiol. 2002;37:281–291. 78. Moller HE, Chen XJ, Saam B, et al. MRI of the lungs using hyperpolarized noble gases. Magn Reson Med. 2002;47:1029–1051. 79. Jalali A, Ishii M, Edvinsson JM, Guan L, Itkin M, Lipson DA, Baumgardner JE, Rizi RR. Detection of simulated pulmonary embolism in a porcine model using hyperpolarized 3He MRI. Magn Reson Med. 2004;51:291–298. 80. Kreitner KF, Ley S, Kauczor HU, Mayer E, Kramm T, Pitton MB, Krummenauer F, Thelen M. Chronic thromboembolic pulmonary hypertension: pre- and postoperative assessment with breath-hold MR imaging techniques. Radiology. 2004;232:535–543. 81. Ley S, Kreitner KF, Fink C, Heussel CP, Borst MM, Kauczor HU. Assessment of pulmonary hypertension by CT and MR imaging. Eur Radiol. 2004;14:359–368. 82. Prince MR, Alderson PO, Sostman HD. Chronic pulmonary embolism: combining MR angiography with functional assessment. Radiology. 2004;232:325–326. 83. Prince MR, Zhang HL, Roditi GH, Leiner T, Kucharczyk W. Risk factors for NSF: a literature review. J Magn Reson Imaging. 2009;30:1298–1308.

268 84. Prince MR, Zhang H, Morris M, MacGregor JL, Grossman ME, Silberzweig J, DeLapaz RL, Lee HJ, Magro CM, Valeri AM. Incidence of nephrogenic systemic fibrosis at two large medical centers. Radiology. 2008;248:807–816. 85. Roditi G, Maki JH, Oliveira G, Michaely HJ. Renovascular imaging in the NSF era. J Magn Reson Imaging. 2009;30:1323–1334. 86. Ahlstrom KH, Johansson LO, Rodenburg JB, Ragnarsson AS, Akson P, Borseth A. Pulmonary MR angiography with ultra-small superparamagnetic iron oxide particles as a blood pool agent and a navigator echo for respiratory gating: a pilot study. Radiology. 1999;211:865–869. 87. Grist TM, Korosec FC, Eite S, et al. Steady-state and dynamic MR angiography with MR-325: initial experience in humans. Radiology. 1998;207:539–544. 88. Bremerich J, Roberts TP, Wendland MF, et al. Three-dimensional MR imaging of pulmonary vessels and parenchyma with NC 100150 injection (Clariscan). J Magn Reson Imaging. 2000;11:622–628. 89. Hoffmann U, Loewe C, Bernhard C, et al. MRA of the lower extremities in patients with pulmonary embolism using a blood pool contrast agent: initial experience. J Magn Reson Imaging. 2002;15:429–437. 90. Kluge A, Muller C, Hansel J, Gerriets T, Bachmann G. Real-time MR with TrueFISP for the detection of acute pulmonary embolism: initial clinical experience. Eur Radiol. 2004;14:709–718. 91. Cohen MD, McGuire W, Cory DA, Smith JA. MR appearance of blood and blood products: an in vitro study. AJR Am J Roentgenol. 1986;146:1293–1297. 92. Sostman D, Pope CF, Smith GJW, Carbo P, Core JC. Proton relaxation in experimental clots varies with the method of preparation. Invest Radiol. 1987;22:509–512. 93. Moody AR, Liddicoat A, Krarup K. Magnetic resonance pulmonary angiography and direct imaging of embolus for the detection of pulmonary emboli. Invest Radiol. 1997;32:431–440. 94. Moody AR, Pollock JG, O’Connor AR, Bagnall M. Lower-limb deep venous thrombosis direct MR imaging of the thrombus. Radiology. 1998;209:349–355. 95. Fraser DGW, Moody AR, Morgan PS, Martel AL, Davidson I. Diagnosis of lower-limb deep venous thrombosis: a prospective blinded study of magnetic resonance direct thrombus imaging. Ann Intern Med. 2002;136:89–98. 96. Moody AR, Crossley I, Moorby S, Delay G. Magnetic resonance direct thombus imaging (MRDTI) as a first line investigation for pulmonary embolism—the PDQ trial. In: Proceedings of the 11th Annual Meeting of ISMRM, Toronto, Canada, 2003. 97. Haage P, Piroth W, Krombach G, Karaagac S, Schaffter T, Gunther RW, Bucker A. Pulmonary embolism: comparison of angiography with spiral computed tomography, magnetic resonance angiography, and real-time magnetic resonance imaging. Am J Respir Crit Care Med. 2003;167:729–734. 98. Hurst DR, Kazerooni EA, Stafford-Johnson D, Williams DM, Platt JF, Cascade PN, Prince MR. Diagnosis of pulmonary embolism: comparison of CT angiography and MR angiography in canines. J Vasc Interv Radiol. 1999;10:309–318. 99. Reittner P, Coxson HO, Nakano Y, Heyneman L, Ward S, King GG, Baile EM, Mayo JR. Pulmonary embolism: comparison of gadolinium-enhanced MR angiography with contrast-enhanced spiral CT in a porcine model. Acad Radiol. 2001;8:343–350. 100. Zhang LJ, Lu L, Bi J, Jin LX, Chai X, Zhao YE, Chen B, Lu GM. Detection of pulmonary embolism comparison between dual energy CT and MR angiography in a rabbit model. Acad Radiol. 2010;17:1550–1559. 101. Seo JB, Im JG, Goo JM, Chung MJ, Moon WK, Lee KH, Kim IO. Comparison of contrast-enhanced ct angiography and gadolinium-

J.F.M. Meaney and P. Beddy enhanced MR angiography in the detection of subsegmental-sized pulmonary embolism. An experimental study in a pig model. Acta Radiol. 2003;44:403–410. 102. Ersoy H, Goldhaber SZ, Cai T, Luu T, Rosebrook J, Mulkern R, Rybicki F. Time-resolved MR angiography: a primary screening examination of patients with suspected pulmonary embolism and contraindications to administration of iodinated contrast material. AJR Am J Roentgenol. 2007;188:1246–1254. 103. Remy-Jardin M, Bouaziz N et al. Bronchial and nonbronchial systemic arteries at multi-detector row CT angiography: comparison with conventional angiography. Radiology. 2004;233:741–749. 104. Yoon YC, Lee KS, Jeong YJ, Shin SW, Chung MJ, Kwon OJ. Hemoptysis: bronchial and nonbronchial systemic arteries at 16-detector row CT. Radiology. 2005;234:292–298. 105. O’Keeffe SA, McGrath A, Farrelly CT, Meaney JFM, Rea DJ. Bronchial artery visualisation on contrast-enhanced MRA of the pulmonary vasculature. Presented at the 91st Scientific Meeting and Annual Assembly RSNA, November 2005, Chicago, IL. 106. Abul Y, Eryuksel E, Karakurt S, Yumuk F, Kodalli N, Celikel T, Ceyhan B. A malignant mesenchymal tumor (sarcoma) of the pulmonary artery presenting as a form of acute thromboembolism: educational case. Clin Respir J. 2010;4:e1-e3. 107. Tseng TW, Huang CH, Yang AH, Yu WC. Pulmonary stenosis in a patient with right pulmonary artery intima sarcoma 5 years after chemo-radiation therapy. Eur Heart J. 2010;31:2432. 108. Kim JB, Kim SH, Lim SY, Roh SY, Cho GY, Song HJ, Park S, Lee J, Kim SJ. Primary angiosarcoma of the pulmonary trunk mimicking pulmonary thromboembolism. Echocardiography. 2010;27: E23-E26. 109. Maki DD, Siegelman ES, Roberts DA, Baum RA, Gefter WB. Pulmonary arteriovenous malformations: three-dimensional gadolinium-enhanced MR angiography-initial experience. Radiology. 2001;219:243–246. 110. Ohno Y, Hatabu H, Takenaka D, Adachi S, Hirota S, Sugimura K. Contrast-enhanced MR perfusion imaging and MR angiography: utility for management of pulmonary arteriovenous malformations for embolotherapy. Eur J Radiol. 2002;41:136–146. 111. Mansour M, Holmvang G, Sosnovik D, Migrino R, Abbara S, Ruskin J, Keane D. Assessment of pulmonary vein anatomic variability by magnetic resonance imaging: implications for catheter ablation techniques for atrial fibrillation. J Cardiovasc Electrophysiol. 2004;15:387–393. 112. Haissaguerre M, Jais P, Shah DC, et al. Spontaneous initiation of atrial fibrillation by ectopic beats originating in the pulmonary veins. N Engl J Med. 1998;339:659–666. 113. Cronin P, Sneider MB, Kazerooni EA, Kelly AM, Scharf C, Oral H, Morady F. MDCT of the left atrium and pulmonary veins in planning radiofrequency ablation for atrial fibrillation: a how-to guide. AJR Am J Roentgenol. 2004;183:767–767. 114. Demos TC, Posniak HV, Pierce KL, Olson MC, Muscato M. Venous anomalies of the thorax. AJR Am J Roentgenol. 2004;182: 1139–1150. 115. Frydrychowicz A, Landgraf B, Wieben O, François CJ. Images in Cardiovascular Medicine. Scimitar syndrome: added value by isotropic flow-sensitive four-dimensional magnetic resonance imaging with PC-VIPR (phase-contrast vastly undersampled isotropic projection reconstruction). Circulation. 2010;121: e434-e436. 116. Gavazzi E, Ravanelli M, Farina D, Chiari ME, Maroldi R. Scimitar syndrome: comprehensive, noninvasive assessment with cardiovascular magnetic resonance imaging. Circulation. 2008;118:e63-e64. 117. Dahya V, Mayosi BM. Assessing scimitar syndrome - use of MRI and MRA. S Afr Med J. 2007;97:248–249.

Abdominal Aorta and Mesenteric Vessels Klaus D. Hagspiel and Patrick T. Norton

Introduction MR angiography is well established for the evaluation of the abdominal aorta and its branches and, over CT angiography, has the advantage of not involving radiation [1]. In this chapter, we review the different MR techniques utilized for the evaluation of the abdominal aorta and mesenteric vasculature as well as the range of normal and pathologic conditions encountered in these vascular beds.

MR Techniques Used for the Evaluation of the Aorta and the Mesenteric Vasculature Time-of-Flight MR Angiography Being one of the earliest MRA techniques, time-of-flight (TOF) MRA is only rarely used for the evaluation of the abdominal aorta or mesenteric vasculature today. This is due to the long scan times for 3D TOF, which preclude breathhold imaging and misregistration artifacts in breath-hold 2D TOF. Due to its higher sensitivity for through plane flow than in plane flow, assessment of the mesenteric vasculature with its tortuous course and frequently changing directions is also generally suboptimal. The triphasic nature of splanchnic arterial blood flow also requires systolic gating, which improves image quality, but also increases scan duration [1]. Furthermore, TOF techniques suffer from a tendency to overestimate the degree of stenosis.

K.D. Hagspiel, MD () • P.T. Norton, MD Division of Noninvasive Cardiovascular Imaging, Department of Radiology, University of Virginia Health System, Charlottesville, VA, USA e-mail: [email protected]

20

Phase-Contrast MRA Phase-contrast (PC) MRA allows the direct quantitative evaluation of flow direction and velocity. The proton spin phase is modulated by altering the first moment of the magnetic field in all three directions. This modulation causes spins moving in the encoded direction to acquire a phase shift which is then measured. Some of the disadvantages of the TOF techniques, such as lack of sensitivity in certain flow directions, do not apply to PC MRA. However, in order to create an MR angiogram from the phase shift within a voxel, the flow velocities have to be estimated and an appropriate velocity encoding gradient has to be chosen prior to scanning. Several groups have reported the use of 2D PC MRA with and without breath-holding as well as the use of a 2D ECG-gated cine PC technique for the functional evaluation of the mesenteric vasculature [1]. These techniques allow quantitative flow measurements (both flow velocities and flow volume). To date, PC MRA has been used to measure flow in the superior mesenteric artery (SMA) and SMV, portal vein, and azygos vein and renal arteries, etc. [2–4]. Systolically gated 3D PC MRA techniques have also been used successfully for the detection of stenoses within the proximal portions of the celiac artery and the SMA [2]. We use PC MRA extremely rarely in the abdomen, mainly only for the determination of the flow direction in the portal vein (e.g., hepatopedal vs. hepatofugal flow).

Steady-State Free Precession Steady-state free precession MRI is a new MRI technique, which uses steady states of magnetizations. Balanced SSFP MRI sequences achieve a phase of zero by applying a gradient on stationary spins between two consecutive RF pulses and returning the spin to the same phase it had before the gradients were applied. Multislice and cine versions of the SSFP sequence exist and they are useful for the assessment of aortic aneurysms and dissections (Fig. 20.1). All vendors

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_20, © Springer Science+Business Media, LLC 2012

269

270

K.D. Hagspiel and P.T. Norton

Fig. 20.1 63-Year-old male with infrarenal abdominal aortic aneurysm and bilateral iliac artery aneurysms. Anteroposterior (AP) and lateral maximum intensity projection (MIP) images of ceMRA at 3 T show the luminal width of the AAA as well as the common and internal iliac artery aneurysms. There is also a moderate left renal artery stenosis. The axial images on the right (SSFP top, fat-suppressed 3D breathold T1 VIBE bottom) show the true diameter of the aneurysm, which also contains luminal thrombus

Contrast-Enhanced 3D MRA

Fig. 20.2 Median arcuate ligament compression. Lateral MIP image shows the typical superior indentation of the celiac trunk by the median arcuate ligament. (Left ceMRA, right noncontrast NATIVE MRA)

have these sequences available (Siemens: TruFISP, GE: FIESTA, Philips: b-FFE). Iozzelli and colleagues [5] showed in a series of 35 consecutive patients undergoing ceMRA of the abdominal aorta and its branches that the addition of sagittal and axial nonbreath hold SSFP sequences resulted in an information increase over ceMRA alone.

Newer, Noncontrast MRA Techniques Recently, cardiac-triggered 3D SSFP sequences were developed for renal noncontrast MRA using Navigator gating (Fig. 20.2) [6]. This allows the scan to be acquired during regular breathing. A slab-selective inversion prepulse is applied to suppress signal from the renal parenchyma and the renal veins. We have applied this sequence occasionally for patients with suspected mesenteric ischemia and found it useful for evaluation of the proximal celiac and mesenteric arteries.

High Spatial Resolution Standard Contrast-Enhanced MRA Contrast-enhanced MRA forms the backbone of MR examinations of the mesenteric vasculature [1, 7–10]. A high-resolution 3D T1-weighted gradient-echo (GRE) pulse sequence is used in conjunction with intravenous injection of gadolinium contrast material. With this technique, images of the aorta and its branches can be acquired during a 10–20 s breath-hold. We repeat the sequence three times in order to obtain portal venous and systemic venous information. Hardware Considerations While 3D contrast-enhanced MRA can essentially be performed on any 1.5 T or 3 T MR system, gradient switching capabilities with an achievable slew rate in the region of 120 mT/m/ms or more should be considered a prerequisite for breath-hold gadolinium-enhanced MRA. Such gradient systems are standard on modern MR scanners. Images obtained during breath-holding have been shown to improve image quality significantly. As with conventional MR imaging, better image quality can be expected with the use of a phased-array coil for signal transmission and reception and parallel imaging techniques. This is particularly crucial for adequate depiction of small or distal branches. Pulse Sequences The main pulse sequence for ceMRA is a 3D Fourier transform GRE sequence. T1 weighting and background suppression is achieved by spoiling, which can be accomplished by radiofrequency or gradient techniques. Since acquisition time and spatial resolution are inversely related, the sequence needs to be adjusted relative to the breath-holding capability

20

Abdominal Aorta and Mesenteric Vessels

of a given patient. Some patients may benefit from oxygen administration and hyperventilation. The slab thickness of the 3D imaging volume, as defined by the product of section thickness and number of sections, can also be adjusted in order to tailor the acquisition time to the breath-hold interval. Typically, 60–88 sections of 1–2 mm thickness constitute the 3D slab. A field-of-view (FOV) of 350–400 mm usually suffices to cover the vascular anatomy from the celiac artery to the iliac arteries. The repetition time (TR) and echo time (TE) for the 3D GRE sequence should be kept to a minimum [9, 10]. Modern MR scanners with high performance gradients allow a TR of 3 ms or less, which translates into an acquisition time ranging between 10 and 20 s. The TE is typically in the order of 1.0–1.5 ms, short enough to counter the effects of spin-dephasing that can cause signal loss in areas of turbulence, resulting in overestimation of stenoses. The flip angle is kept around 30°. We routinely use parallel imaging in order to increase spatial resolution and/or decrease image acquisition time. Different approaches such as GRAPPA or SENSE are in use and the typical acceleration factor in out practice at 1.5 T is two [11]. Typical parameters used on our 1.5 T system (Avanto, Siemens Medical Systems, Malvern, PA) are as follows: sequential 3D FLASH sequence, excitation flip angle of 15°; matrix size 384 × 448; TR/TE 2.61/1.09 ms; FOV 390 mm; parallel imaging (GRAPPA) acceleration factor 2. The resulting uniterpolated spatial resolution is 0.8 × 1.5 × 1.5–1.9 mm with this sequence. Image orientation depends on the clinical question. For patients being evaluated for mesenteric ischemia we utilize the sagittal plane for the arterial phase scan as it allows most efficient coverage of the unpaired branches of the abdominal aorta while minimizing slice thickness. The portal and mesenteric venous phases are obtained in the coronal plane. For cases with visceral aneurysms we usually also obtain the arterial phase in coronal orientation. In general, it is desirable to image as fast as possible with 3D gadolinium MRA because this allows a higher injection rate, which translates into a tighter bolus and enhanced arterial signal. Also, faster acquisitions cause less motion-related artifact and enable a shorter breath-hold interval.

Contrast Administration and Bolus Timing Arterial signal in 3D contrast-enhanced MRA is based solely on the T1 shortening effect of the gadolinium bolus during its first pass through the vascular territory of interest. Therefore, correct timing, administration and dosing of the gadolinium bolus is critical to arterial contrast and image quality and makes the use of an automated infusion pump mandatory. An extracellular paramagnetic MR contrast agent, a gadolinium chelate, is infused via a peripheral venous access, usually an antecubital vein. In general we administer 0.1–0.2 mmol/kg gadolinium contrast material. Heverhagen et al. [12] investigated the signal time curves and image contrast of abdominal

271

aorta, vena cava, and portal vein in 60 patients prospectively who were divided into three contrast agent (Gd-DTPA) dosage groups (0.1, 0.2, and 0.3 mmol/kg). They did not find any significant differences in maximum signal enhancement between the groups; the only significant differences were found for vena cava and liver, but overall vessel conspicuity was not significantly improved with an increase of contrast agent dose. The authors concluded that the 0.1 mmol/kg dose is sufficient for high-quality abdominal MRA. Accurate bolus timing is critical to the image quality of gadolinium-enhanced MRA. Gadobenate dimeglumine (Multihance, Bracco Imaging, Milan, Italy) is an agent used by many centers for ceMRA, mainly due to its slightly higher T1 relaxivity than the other extracellular agents [13]. However, we find all current extracellular agents suitable for MRA. Gadofosveset trisodium (Ablavar, Lantheus Medical Imaging) and at the time of this writing, the role of the agent in the evaluation of the abdominal aorta and its branches is not known [14]. Ideally, peak arterial artery enhancement in the area of interest is timed to coincide with acquisition of the center of k-space, which is primarily responsible for image contrast. It is important to tailor the injection to the type of k-space mapping, e.g., sequential vs. centric. The peripheral k-space lines determine image detail, and so it is not necessary to maximize arterial contrast during this phase of data acquisition. For this reason, the effect of the gadolinium bolus need only last part of the scan duration, which allows for a reduced contrast dose and a higher injection rate. The injection rate should be adjusted to produce a contrast bolus lasting approximately 1/2 to 2/3 of the scan duration. Typically, an injection rate ranging from 1.5 to 2.5 ml/s is used for gadolinium MRA. While a tight bolus improves contrast-to-noise ratio and image quality, it does require precise timing. There are two main strategies for achieving accurate bolus timing: the test bolus technique and the automatic triggering technique. The test bolus technique is the most universally applicable because it does not require additional software. With this technique, the time interval for a small amount of gadolinium (1–2 ml) between injection into an antecubital vein and arrival in the abdominal aorta is measured by acquiring sequential images of the aorta at fixed time intervals. The automatic triggering technique for bolus timing employs a pulse sequence that automatically detects arrival of the full contrast bolus in the vessel of interest and then triggers the 3D gadolinium MRA sequence [1]. Several groups have investigated the influence of caloric stimulation on image quality. While two groups reported significantly improved image quality and contrast-to-noise ratios for both the mesenteric arterial and venous system with caloric stimulation [1, 15], another could not confirm these results [16]. The use of anticholinergic agents has not been proven to impact positively on image quality [17]. Compared to catheter angiography, state-of-the art CE 3D MRA performs favorably (as assessed by kappa values) in

272

K.D. Hagspiel and P.T. Norton

the common and proper hepatic arteries, the splenic artery, the SMA, as well as the portal, superior mesenteric, and splenic veins [18]. However, agreement is poor and catheter angiography still necessary for the evaluation of the intrahepatic arteries, the SMA branches, as well as frequently the inferior mesenteric artery (IMA) [9, 19]. While the latest parallel imaging techniques improved the ceMRA performance, it is still inferior to multidetector CTA in our experience. CeMRA was shown to be superior to DSA in the evaluation of the portal venous system [20].

Time-Resolved MRA Time-resolved MR techniques have led to a significant reduction in acquisition times, thus allowing the acquisition of vascular MRA independent of contrast travel time [21]. As such, they provide morphologic and hemodynamic flow information, similar to conventional DSA. All major equipment manufacturers have made these sequences available. We occasionally use the TWIST (time-resolved imaging with interleaved stochastic trajectories) sequence (Siemens Medical Solutions, Malvern, PA) for evaluation of the abdominal aorta and the mesenteric arteries. TWIST is a view sharing technique that undersamples the periphery of k-space to increase temporal resolution as compared to traditional tr-ceMRA. Our implementation employs the following parameters: 3 T whole body MR system (Trio, Siemens Medical Solutions, Malvern, PA), body phased-array coil, TR 2.86 ms, TE 1.18 ms, 512 × 420 matrix, central k-space sampling size 18%, peripheral sampling density 20%, parallel imaging (GRAPPA) acceleration factor 2, voxel size 1.0 × 1.0 × 4.0 mm, temporal resolution 2.2 s. Four seconds before initiation of the sequence, we inject 6–10 ml of contrast at a rate of 2 ml/s. Time-resolved MRA of the abdomen is not routinely used in our practice, and more of a problem solver. Since its main advantage is the ability to assess flow dynamics, it can be helpful to diagnose shunts due to arteriovenous fistulae or delayed portomesenteric venous filling in patients with nonocclusive mesenteric ischemia (NOMI). Kramer and colleagues [22] investigated 22 patients undergoing low-dose, time-resolved 3D MRA of the abdominal aorta and its major branches at 3 T and concluded that the technique provides rapid and important anatomic and functional information in the evaluation of the abdominal vasculature, but due to its limited spatial resolution it is inferior to ceMRA in demonstrating fine vascular details.

T1-Weighted Gradient-Echo Fat-Saturated Imaging T1-weighted breath held gradient-echo fat-saturated sequences are a part of all our mesenteric and abdominal MRA protocols. They are particularly beneficial for the

Fig. 20.3 75-Year-old male who underwent endovascular treatment of an infrarenal abdominal aortic aneurysm with an AneuRx stent graft. The AP maximum intensity projection (MIP) image of ceMRA at 1.5 T shows the lumen of the graft as well as the aneurysm sac. Three axial images taken at 20, 60, and 240 s contrast injection show progressive filling of the aneurysm sac with contrast, consistent with a type 2 endoleak. The images at 20 and 60 s are axial reformats of the ceMRA, the image at 240 s is a fat-suppressed 3D breathold T1 VIBE image

assessment of thrombi, both in the arterial and venous circulation (Fig. 20.1). They are also the most sensitive technique for the detection of endoleaks after stent graft placement (Fig. 20.3).

Black Blood Techniques We routinely use black blood MRA techniques in the abdomen for the evaluation of aortic aneurysms, dissections, and inflammatory aneurysms. The key advantage of black blood sequences is the improved visualization of the vessel wall and the perivascular structures. The typical sequences used in this setting are ECG-gated spin echo and fast spin echo sequences.

MRA at 3 T Similar to ceMRA at 1.5 T, the TR is chosen as short as possible in order to maximize T1 signal in blood while achieving minimum background signal, typically less than 3 ms. Occasionally, the TR will have to be increased in order to not exceed SAR limits. Likewise, TE should be chosen as short as possible, because susceptibility artifacts increase with field strength. At the typical values chosen for TE, 1.1– 1.3 ms, fat and water protons are out of phase at 3 T. One of the main advantages is the ability to use parallel imaging acceleration factors that are higher than those used at 1.5 T,

20

Abdominal Aorta and Mesenteric Vessels

273

and a factor of 3 or higher are routinely applied. On our 3 T system (Magnetom Trio, Siemens Medical Systems, Malvern, PA), we use a sequential 3D FLASH sequence with an excitation flip angle of 23°, a matrix size of 384 × 384, TR/TE 2.92/1.18 ms, FOV 400 mm, receiver bandwidth 590 Hz/ pixel, and a parallel imaging (GRAPPA) acceleration factor 3. Acquisition time for one dataset is 20 s and the achieved uninterpolated spatial resolution is 1.0 × 1.0 × 1.2 mm. One of the advantages of a 3 T system is the ability to reduce the contrast dose due to the inherently higher signal on these systems. However, to date there is no convincing evidence for the diagnostic superiority of ceMRA at 3 T over 1.5 T for the abdominal vasculature.

Anatomy of the Abdominal Aorta and the Mesenteric Vasculature Abdominal Aorta The abdominal aorta, the continuation of the thoracic aorta, is the largest artery in the abdomen. It originates at the diaphragm and extends to its bifurcation where it terminates in the common iliac arteries. It gives off a number of paired and unpaired visceral and parietal branches and its diameter decreases from proximal to distal from about 25 to 20 mm in the average adult. The visceral branches include from proximal to distal the celiac, superior mesenteric, middle suprarenal, renal, gonadal, and inferior mesenteric arteries. The parietal branches are the inferior phrenic, lumbar, and median sacral arteries.

Mesenteric Arterial System The blood supply to the intestinal tract is derived from the three major anterior branches of the abdominal aorta: the celiac artery (CA), the SMA, and the IMA [23]. ceMRA allows the detailed assessment of the normal and abnormal vascular anatomy in the majority of cases.

Celiac Artery The celiac artery arises from the ventral surface of the aorta at the T12-L1 interspace. It supplies the upper abdominal viscera. In 65% of patients, the celiac artery has the “classic” branching pattern into three branches, the splenic, common hepatic, and left gastric arteries. In 35% of patients, the branching pattern of the celiac artery can be quite variable. The splenic, common hepatic, or left gastric artery may arise directly from the aorta or from the SMA (Fig. 20.4). The proper hepatic artery typically divides into the left and right hepatic artery. This branching pattern is present in 50% of all individuals. The remaining 50% have replaced or accessory

Fig. 20.4 70-Year-old male with postprandial abdominal pain. AP and lateral maximum intensity projection (MIP) images of ceMRA at 3 T show the atherosclerotic native aorta as well as a severe celiac trunk stenosis, an IMA occlusion and a severe left renal artery stenosis. There is also a replaced right hepatic artery of the superior mesenteric artery (arrowhead)

hepatic arterial branches (Fig. 20.4). The gastroduodenal artery arises from the common hepatic artery in approximately 75% of patients and usually has two main branches: the superior pancreaticoduodenal artery and the right gastroepiploic artery. The superior pancreaticoduodenal artery forms an anastomotic arcade with the inferior pancreaticoduodenal artery [23].

Superior Mesenteric Artery The SMA typically arises from the ventral aspect of the aorta approximately 1 cm below the origin of the celiac artery [23]. Rarely, a single celiomesenteric trunk arises directly from the aorta. The inferior pancreaticoduodenal artery is typically the first branch of the SMA. The jejunal and ileal artery branches usually originate from the left side of the SMA. A distinguishing feature of the jejunal and ileal branches is the presence of arcades which anastomose with adjacent branches. The most distal arcades run along the mesenteric border of the bowel and give off the staight vasa rectae which reach the antimesenteric border [23]. The arcades are not routinely visualized with MRA [24]. The SMA gives off three right sided branches. These are the middle colic artery, the right colic artery, and the ileocolic artery. Inferior Mesenteric Artery The IMA arises from the ventral aspect of the aorta approximately at the level of the third lumbar vertebral body, and

274

K.D. Hagspiel and P.T. Norton

Magnetic Resonance Angiography of the Abdominal Aorta Abdominal Aortic Aneurysm

Fig. 20.5 61-Year-old male with saccular splenic artery aneurysm seen here on volume rendered image of ceMRA at 3 T

measures between 1.2 and 5.5 mm in diameter at its origin. The first branch of the IMA is an ascending branch which can consist of either the left colic artery alone or a common trunk of the left colic and sigmoid arteries. The IMA then gives off the sigmoid branches (not originating from the left colic artery) [23]. The IMA terminates in the superior hemorrhoidal artery. Knowledge of collateral pathways between the mesenteric arteries is important when evaluating patients for mesenteric ischemia. More than 50 collateral pathways in the large and small bowels have been described [23]. The Marginal Artery of Drummond is situated along the mesenteric border of the colon and formed by the anastomotic continuation between the right, middle, and left colic arteries. The Arc of Riolan is situated more centrally within the mesentery and is an inconstant anastomosis between the middle and left colic arteries. The meandering artery is also situated within the mesentery, and is a very large tortuous vessel communicating between the SMA and IMA. It potentially represents an enlarged Arc of Riolan (Moskowitz) (Fig. 20.5). The Arc of Buehler is a proximal connection between the CA and SMA. The dominant collateral pathway between the celiac artery and the SMA is via the gastroduodenal artery and the pancreaticoduodenal arcades (Fig. 20.4). However, due to anatomical variants, those pathways may be altered significantly [23]. The superior hemorrhoidal artery can also become an important collateral pathway as it communicates in the rectal wall with the inferior hemorrhoidal artery, a branch of the internal iliac or hypogastric artery.

Currently, the three noninvasive modalities which are used in the assessment of the abdominal aorta are ceMRA, CTA, and ultrasound. In the acute setting, CTA is the preferred modality of choice for evaluation of aortic dissection due to short exam time and wide availability. The most frequent indication for MRA of the abdominal aorta in our practice is for the diagnosis and follow-up of abdominal aortic aneurysms (AAAs) (Fig. 20.1). AAAs are divided into infrarenal (at least a 10 mm neck of normal aorta below the lowest renal artery, juxtarenal (less than 10 mm), or suprarenal types). MRA can reliably determine the size and type of aneurysms, determine the presence of atheroma or thrombus, and define the involvement of side branches. MRA has been shown to be useful for the planning of endograft repair of aortic aneurysms and dissections [25]. In a direct comparison between CTA and MRA for the assessment of AAA prior to endovascular treatment, data sets from both modalities provided precise and reliable volumetric measurements [25]. MRA can also be used for the postprocedural assessment of endograft complications, including endoleaks and stent failure (Fig. 20.3), after the repair of aortic aneurysm and aortic dissection. An endoleak occurs when the treated aneurysm sac becomes re-pressurized due to leaking of blood around the endograft. There are several types of endoleaks: Type 1 due to leakage around the ends of the graft, Type 2 due to back flow through side branches of the aorta which were excluded, Type 3 from technical failure of the graft, or Type 4 due to high porosity of the graft. Type 5 endoleaks are defined as enlargement of the aneurysm sac in the absence of an endoleak demonstrable by imaging. This re-pressurization can lead to progressive growth of the aneurysm sac and potentially rupture. Definitive diagnoses of complications including endoleak are obtainable with either modality. However, there is contradicting evidence in the literature as to which is most sensitive for the detection of endoleaks [26]. MRA surveillance of endograft complications is not feasible with stainless steel endografts because of the large artifacts produced; additionally, there is a theoretical risk of stent migration in these grafts (Fig. 20.3). While CTA is the preferred modality for the postprocedural assessment of endografts, MRA plays an important role for the assessment of endografts in patients with mild to moderate (GFR > 30 mg/min) renal insufficiency.

Aortic Occlusive Disease Aortic occlusive disease (Figs. 20.6 and 20.7), also known as Leriche Syndrome, is defined as atherosclerotic occlusive

20

Abdominal Aorta and Mesenteric Vessels

275

frequently, are the consequence of penetrating ulcers or thrombosed dissections. If acute, the hallmark MR finding is high-signal intensity material in the aortic wall on T1-weighted sequences.

Aortitis

Fig. 20.6 55-Year-old female with aortic occlusive disease and prior aortobifemoral bypass graft. AP and lateral maximum intensity projection (MIP) images of ceMRA at 3 T show the native aorta as well as the graft. There is a stenosis at aortic anastomosis as well as a severe celiac trunk stenosis

disease involving the abdominal aorta. The iliac arteries are usually also involved. Classical symptoms are buttock, thigh, and leg claudication and erectile dysfunction in males. Therapy depends on the extent of disease. MRA is well suited to demonstrate the site and extent of the occlusions and to identify target vessels for revascularization [13, 25]. Venkataraman and colleagues [27] examined 79 patients with state-of-the-art ceMRA and reported sensitivities of 100% in the aortic segments, 100% in the common iliac, and 93.8% in the external iliac arteries, the specificities were 100%, 89.7%, and 95.2%, respectively.

Aortic Dissection In the abdomen, aortic dissections are usually extensions of dissections beginning more proximal. Dissections isolated to the abdominal aorta are rare. The main task for the assessment of dissections of the abdominal aorta is to determine involvement of the renal and visceral branches as well as the iliac/femoral arteries. High-resolution mRA is well suited for this task, and is occasionally complemented by SSFP techniques. Intramural hematoma can be either caused by rupture of the vasa vasorum without intimal tear or, more

Aortitides can be divided into those with infective and those with noninfective etiology. Infective aortitides (also called mycotic aneurysms) have a different morphology than the classical atherosclerotic aneurysms. They tend to be lobulated, saccular, and frequently thick-walled. Often, there is associated infiltration of the surrounding fat planes, all features absent from atherosclerotic aneurysms. Infective aortitides are more frequent in immunocompomised patients and occur more frequently in intravenous drug abusers or patients with endocarditis. The main noninfective vasculitides are Takayasu arteritis and Giant cell arteritis. Takayasu’s arteritis (TA) is an idiopathic inflammatory vascular disorder that typically involves the thoracoabdominal aorta and its branches and the pulmonary arteries. The disease is more prevalent in Asia, and typically affects women in their 20s and 30s. The MRI MRA findings of TA in the early phase are vascular wall thickening and enhancement, and stenoses, occlusions, and aneurysm formation in the later stages [28]. Giant cell arteritis typically affects patients over 50 years of age. There is a predilection for affecting the upper extremities, typically in the form of long stenoses. In the aorta, aneurysms are more common than stenoses, and there is concentric wall thickening with edema resulting in high-signal intensity on T2-weighted and STIR sequences as well as significant contrast enhancement.

Congenital Anomalies Abdominal coarctation is a rare entity with some overlap to midaortic dysplastic syndrome. Acquired abdominal coarctation can be related to umbilical artery catheterization in newborns (Seifert Conn Med 2009). Neurofibromatosis is another entity known to lead to stenoses of the abdominal aorta and its branches.

Magnetic Resonance Angiography of the Mesenteric Arterial System Mesenteric Ischemia Acute Mesenteric Ischemia Acute interruption of the blood supply to the small bowel and/or colon is a catastrophic event, and carries a morbidity

276

K.D. Hagspiel and P.T. Norton

Fig. 20.7 76-Year-old male with classic symptoms of mesenteric ischemia. AP and lateral maximum intensity projection (MIP) images of ceMRA at 3 T show the severely diseased native aorta with occluded right iliac arteries and status poststenting of the left iliac vessels and a left to right fem-fem bypass. There are severe celiac trunk, SMA and bilateral renal artery stenoses, and an IMA occlusion. AP and lateral DSA confirms these findings

and mortality rate exceeding 60%. The four major causes of acute mesenteric ischemia (AMI) are SMA embolus, SMA thrombosis, mesenteric venous thrombosis (MVT), and nonocclusive mesenteric vasoconstriction [23]. Aortic dissections have also been reported to cause AMI on rare occasions [23]. Due to the serious clinical status and urgent need for a diagnosis, MRA is only rarely performed in this setting.

Acute SMA Embolism Acute emboli to the SMA account for approximately 40–50% of all episodes of AMI [23]. These patients typically have a clinical history of cardiovascular disease. The majority of

emboli in the SMA lodge just beyond the origin of the middle colic artery. The angiographic hallmark of an embolic occlusion is the abrupt termination of the vessel (cut-off-sign). DSA is the diagnostic modality of choice and has been shown to be extremely accurate, but MRA is also capable of demonstrating these acute occlusions. In one study, ceMRA was compared with DSA in a porcine acute embolic segmental mesenteric ischemia model. Sensitivity and specificity for the two modalities were 91%/100% and 80%/90%, respectively [29]. In patients with prior embolic events, recanalized vessels may be seen. The MRA appearance of a septic embolus and mycotic pseudoaneurysm of the SMA in a patient with enterococcal endocarditis has also been published [30].

20

Abdominal Aorta and Mesenteric Vessels

277

Acute Mesenteric Artery Thrombosis Acute mesenteric arterial thrombosis is typically associated with a preexisting atherosclerotic lesion [23]. In up to 50% of cases, a history of intestinal angina can be obtained [23]. In contrast to the abrupt catastrophic onset of symptoms associated with an embolus to the SMA, the abdominal pain and symptoms associated with acute mesenteric arterial thrombosis may be more insidious due to the development of collateral circulation. Occlusion of the SMA is typically within the first 2 cm of its origin (in contrast to acute embolic occlusions), and there is typically no defined meniscus or intraluminal filling defect. MRA can show these findings, in addition to the visualization of collateral vessels [23]. Mesenteric Venous Thrombosis MVT accounts for approximately 5–15% of all cases of AMI [23]. The most common associated risk factors are portal hypertension, hypercoagulable state, trauma, intraabdominal inflammatory diseases, the use of oral contraceptives, and recent surgery affecting the portomesenteric venous system, especially splenectomy [23]. AMI develops when MVT is associated with a lack of adequate venous collaterals resulting in the development of intestinal mucosal edema and subsequent arterial hypoperfusion. The clinical presentation of patients with acute MVT is characterized by pain out of proportion to the physical findings. Dual-phase contrast-enhanced 3D MRA has been demonstrated to be highly accurate for the evaluation of SMV and PV thrombosis [17]. We use T1-weighted breathhold fat-suppressed 3D sequences for the detection of MVT (Fig. 20.8). Nonocclusive Mesenteric Ischemia NOMI is thought to be responsible for approximately 25% of cases of AMI with a mortality rate as high as 70% [23]. NOMI usually develops during an episode of cardiogenic shock or a state of hypoperfusion in which excessive sympathetic activity results in secondary vasoconstriction of the mesenteric arteries. The diagnosis of this entity with MRA has not been described in humans, but Li and coworkers successfully used SMV HbO2 and volume flow rate measurements to diagnose and monitor mesenteric ischemia due to hemorrhagic shock in an in vivo canine model [31]. Aortic Dissection Approximately 5% of patients with aortic dissection develop mesenteric ischemia as a complication of the dissection process [23]. Both 2D TOF and gadolinium-enhanced techniques have been used to diagnose this condition although not usually in the acute phase. MRA has been shown to allow to classify the dissection, define entry and reentry points, differentiate thrombus from slow flow, and evaluate branch vessel involvement [31] (Figs. 20.2 and 20.9). Isolated dissections

Fig. 20.8 55-Year-old male with hypercoagulable state and mesenteric vein thrombosis. Coronal thin slab MIP image of venous phase of ceMRA at 1.5 T shows the thrombus in the superior mesenteric vein. Axial T1-weighted fat-suppressed 3D VIBE image also shows the nonocclusive thrombus

of the SMA either in association with cystic degeneration or as a complication of catheter angiography are extremely rare [7]. The CE 3D MRA appearance has been described in one case [32].

Chronic Mesenteric Ischemia 3D ceMRA is a well-established screening technique for patients suspected of having chronic mesenteric ischemia (CMI). It is particularly suitable for the detection of the more proximal atherosclerotic occlusive in the CA and SMA as well as the assessment of patients with chronic aortic dissection.

Atherosclerotic CMI CMI is almost always caused by severe atherosclerotic disease and characterized by a classical clinical triad of postprandial abdominal pain, weight loss, as well as the desire of

278

K.D. Hagspiel and P.T. Norton

Fig. 20.9 48-Year-old male with type B aortic dissection. The AP maximum intensity projection (MIP) image of ceMRA at 1.5 T shows the dissection to extend into the bilateral iliac arteries. The axial reformats of the ceMRA show the dissecting membrane extend into the SMA. The patient subsequently underwent surgical repair (data not shown)

arteries in up to 67% of subjects aged 80 years or older [23]. While atherosclerosis of the mesenteric branches is frequent, CMI is relatively uncommon (Fig. 20.8), which is a function of the rich mesenteric collateral circulation. This collateral network in the mesenteric system makes it difficult to estimate the degree of mesenteric vascular stenosis necessary to cause intestinal angina [23]. It is generally felt that at least two of the three main vessels have to be affected either by occlusive or stenotic disease in order to produce clinical symptoms, although exceptions to this rule exist (Fig. 20.5). CMI in the setting of proximal or segmental mesenteric artery stenosis or occlusion in only one affected vessel is rare, but can occur. Several groups have investigated the flow dynamics in patients with CMI. Cine cardiac-gated PC MRA has been used to show that flow rates in the SMA and SMV correlate well and that patients with CMI show a significantly reduced rate of postprandial flow augmentation in the SMV compared to normals. It has furthermore been demonstrated that measurements of the percent flow change in the SMA 30 min after a meal provide the best discriminator between patients with and without CMI [1]. Morphological imaging of patients suspected of having CMI using ceMRA is the diagnostic mainstay in almost all MR centers worldwide. The evaluation of patients with suspected CMI is the most frequent indication for mesenteric MRA in our institution. Published results for the evaluation of patients with CMI exist for 3D PC MRA [1, 2] and ceMRA [1, 2, 9, 10, 16]. These papers report on the morphological evaluation of the proximal mesenteric vessels. Results for systolically gated 3D PC MRA are disappointing, because only 66% of all stenoses were detected using catheter angiography as the gold standard [2]. These authors also reported false-positive results. CE 3D MRA performs consistently better with sensitivities and specificities of 100 and 95% in one small series

Fig. 20.10 71-Year-old female with symptoms of mesenteric ischemia and biopsy proven ischemic changes in the descending colon. Lateral MIP image of ceMRA at 3 T shows the severely stenosed IMA. The CA

and SMA were patent (data not shown). Lateral DSA (middle image) confirmed the stenosis. CTA was used to follow the patient and assess stent patency (right image)

the patient to avoid food (Figs. 20.4, 20.7, and 20.10). With advanced age, the abdominal aorta and mesenteric arteries are frequently involved with atherosclerosis. Autopsy studies show significant stenoses of the mesenteric and celiac

20

Abdominal Aorta and Mesenteric Vessels

of 14 patients with angiographic or surgical correlation [9]. CE 3D MRA is also well suited for the evaluation of patients with aortic occlusion (Leriche Syndrome) [33] and mesenteric involvement [23]. Due to its small size, the IMA is more challenging to assess with MRA due to its small size. As was shown in a study, interobserver variability was good to excellent for the celiac and superior mesenteric arteries, but poor for the IMA (kappa 0.90, 0.92, and 0.48, respectively) [18]. These results all refer to stenoses or occlusions of the proximal mesenteric arteries. Due to small size, the distal and side branches pose more of a diagnostic challenge to MRA. For example, patients suspected of having a small to medium artery vasculitis should undergo catheter angiography.

279

Median Arcuate Ligament Syndrome Median arcuate ligament syndrome is caused by extrinsic compression of the CA and/or the celiac neural plexus by the central tendon of the crura of the diaphragm [1, 7]. The classic angiographic findings are best seen on a lateral view of the aorta and consist of a smooth indentation of the superior aspect of the proximal celiac artery (Fig. 20.2). This indentation is classically more marked on expiration than on inspiration [23]. This entity has been demonstrated both with contrast and noncontrast techniques [17] (Fig. 20.4). Median arcuate ligament compression can occasionally also be seen in the proximal SMA and even renal arteries.

Unlike in aneurysms of large vessels, atherosclerosis is not considered to be the primary etiological factor for splenic aneurysms [34] (Fig. 20.5). Degenerative aneurysm formation from underlying medial fibrodysplasia is one possible mechanism. For unclear reasons but perhaps due to increased splenic artery perfusion, patients with portal hypertension and splenomegaly have also been found to have a higher incidence of splenic artery aneurysms. Because of this, aneurysms are frequently discovered in patients being evaluated for liver transplantation. Inflammatory pseudoaneurysms are almost always associated with pancreatitis and the presence of pseudocysts. Very rarely, pseudoaneurysms are caused by other regional inflammatory diseases such as peptic ulcer disease. Polyarteritis nodosa is another known inflammatory etiology, although the splenic artery is less commonly involved than in other locations. Posttraumatic splenic artery pseudoaneurysm formation from penetrating injuries is uncommon. Splenic artery injury from rapid deceleration can result in damage to the intima and elastic lamina [24] although this is difficult to assess given that identification of aneurysms on trauma CTs is often inconclusively attributable to the traumatic even itself. Iatrogenic trauma secondary to postoperative anastomotic leakage after pancreatic surgery (especially pancreatoduodenectomy) is another well known cause of pseudoaneurysm formation [34]. Splenic artery aneurysm formation during pregnancy or in multiparae is perhaps the most clinically important given the high associated mortality for both the mother and fetus. Mortality rates are as high as 75% in pregnant mothers. ceMRA has been demonstrated to be well suited for the diagnosis of aneurysms in the mesenteric circulation [1, 7].

Aortic Dissection MRA is an excellent modality to follow chronic dissections without the need for ionizing radiation [1, 7] (Fig. 20.9).

MRA After Surgical or Endovascular Therapy of Mesenteric Ischemia

Nonatherosclerotic Vascular Pathogies of the Mesenteric Vasculature Fibromuscular Dysplasia FMD is a rare, but well-recognized cause of CMI. Our group published a case of FMD of the SMA which showed the classical string of beads appearance on a ceMRA [7].

Vasculitis CMI has been described as one of the protean manifestations of vasculitides, especially Takayasu’s arteritis [1, 7]. MRA bears promise in this application because of its ability to assess both luminal and vascular wall changes. The technique is valuable both for the assessment of the aorta and its side branches. Stenoses, occlusions, vascular wall thickening, wall enhancement, and increased signal on T2 and STIR images have been reported in vasculitides [28].

CeMRA is well established for the noninvasive follow-up of patients after surgical treatment or percutaneous transluminal angioplasty of mesenteric arterial stenoses and occlusions. However, it is not suited for follow-up of patients who received stents due to the magnetic susceptibility artifacts caused by these devices. In these cases, CTA is a suitable cross-sectional modality if vascular ultrasound is nonconclusive [1, 23] (Fig. 20.10).

Mesenteric Aneurysms The exact mechanism causing mesenteric aneurysms is not clearly understood. They can be divided into four categories: degenerative, inflammatory, posttraumatic, and pregnancy related.

MRA in Transplant Surgery MRI is a well-established modality for the preoperative evaluation of liver and kidney transplant donors [1] as well as the postoperative evaluation of liver, kidney, and pancreas

280

transplant recipients [35, 36]. Our center has extensive experience in the evaluation of pancreatic allograft recipients. Pancreatic transplantation is increasingly utilized for the treatment of type 1 diabetes mellitus and is commonly performed in conjunction with kidney transplantation. Two variations exist: the original procedure is the systemic-bladder drainage (SBD) type which consists of intraperitoneal placement of the whole pancreas into the pelvis and anastomosis of the transplant splenic and superior mesenteric arteries to the recipient’s iliac arteries via a y-graft formed from the donor’s common, internal, and external iliac arteries. Pancreatic venous outflow with this type of graft is into the systemic iliac veins. Drainage of the exocrine secretions is into the urinary bladder using an interposition duodenal segment [36]. The second technique is the portal enterically drained pancreas transplant (PED) technique. As in the SBD technique, the pancreatic allograft is placed intraperitoneally, but higher in the recipient’s abdomen. The allograft splenic and superior mesenteric arteries are anastomosed to the iliac arteries, but due to the higher position in the abdomen via a much longer y-graft. Pancreatic venous outflow is achieved by anastomosing the transplant portal vein with the recipient superior mesenteric vein, which result of more physiologic release of insulin into the portal venous circulation. The exocrine pancreatic secretions are drained into a small bowel loop [36]. The most common cause for postoperative pancreatic transplantation dysfunction after rejection is vascular complications. Arterial and venous allograft thrombosis, stenoses, kinks as well as aneurysm formation, and hemorrhage can be encountered and reliably diagnosed with ceMRA [35].

Conclusion MR angiography is a robust clinical diagnostic modality used in many centers worldwide for the evaluation of the abdominal aorta and its branches. Due to the availability of both contrast-enhanced and noncontrast MRA techniques, in conjunction with MR sequences allowing morphological and functional assessment, tissue characterization, flow measurement, MR urography, and MR pancreaticocholangiography, it is the most versatile imaging method available for the assessment of the full range of abdominal aortic and mesenteric pathologies.

References 1. Hagspiel KD, Leung DA, Angle JF, Spinosa DJ, Pao DG, de Lange EE, Butty S, Matsumoto AH. MR angiography of the mesenteric vasculature. Radiol Clin North Am. 2002;40:867–886. 2. Wasser MN, Geelkerken RH, Kouwenhoven M, et al. Systolically gated 3D phase contrast MRA of mesenteric arteries in suspected mesenteric ischemia. J Comput Assist Tomogr. 1996;20:262–268.

K.D. Hagspiel and P.T. Norton 3. Debatin JF, Zahner B, Meyenberger C, et al. Azygos blood flow: phase contrast quantitation in volunteers and patients with portal hypertension pre- and postintrahepatic shunt placement. Hepatology. 1996;24:1109–1115. 4. Burkart DJ, Johnson CD, Ehman RL, Weaver AL, Ilstrup DM. Evaluation of portal venous hypertension with cine phase-contrast MR flow measurements: high association of hyperdynamic portal flow with variceal hemorrhage. Radiology. 1993;188:643–648. 5. Iozzelli A, D’Orta G, Aliprandi A, Secchi F, Di Leo G, Sardanelli F. The value of true-FISP sequence added to conventional gadolinium-enhanced MRA of abdominal aorta and its major branches. Eur J Radiol. 2009;72:489–493. 6. Wyttenbach R, Braghetti A, Wyss M, et al. Renal artery assessment with nonenhanced steady-state free precession versus contrastenhanced MR angiography. Radiology. 2007;245:186–195. 7. Shih MC, Hagspiel KD. CTA and MRA in mesenteric ischemia: part 1, Role in diagnosis and differential diagnosis. AJR Am J Roentgenol. 2007;188:452–461. 8. Shih MC, Angle JF, Leung DA, et al. CTA and MRA in mesenteric ischemia: part 2, Normal findings and complications after surgical and endovascular treatment. AJR Am J Roentgenol. 2007;188:462–471. 9. Meaney JF, Prince MR, Nostrant TT, Stanley JC. Gadoliniumenhanced MR angiography of visceral arteries in patients with suspected chronic mesenteric ischemia. J Magn Reson Imaging. 1997;7:171–176. 10. Baden JG, Racy DJ, Grist TM. Contrast-enhanced three-dimensional magnetic resonance angiography of the mesenteric vasculature. J Magn Reson Imaging. 1999;10:369–375. 11. Heverhagen JT, Reitz I, Pavlicova M, Levine AL, Klose KJ, Wagner HJ. The impact of the dosage of intravenous gadolinium-chelates on the vascular signal intensity in MR angiography. Eur Radiol. 2007;17:626–637. 12. Wikström J, Wasser MN, Pattynama PM, et al. Gadobenate dimeglumine-enhanced magnetic resonance angiography of the pelvic arteries. Invest Radiol. 2003;38:504–515. 13. Nissen JC, Attenberger UI, Fink C, Dietrich O, Rohrer M, Schoenberg SO, Michaely HJ. Thoracic and abdominal MRA with gadofosveset: influence of injection rate on vessel signal and image quality. Eur Radiol. 2009;19:1932–1938. 14. Hany TF, Schmidt M, Schoenenberger AW, Debatin JF. Contrastenhanced three-dimensional magnetic resonance angiography of the splanchnic vasculature before and after caloric stimulation. Original investigation. Invest Radiol. 1998;33:682–686. 15. Shirkhoda A, Konez O, Shetty AN, Bis KG, Ellwood RA, Kirsch MJ. Mesenteric circulation: three-dimensional MR angiography with a gadolinium-enhanced multiecho gradient-echo technique. Radiology. 1997;202:257–261. 16. Kopka L, Rodenwaldt J, Vosshenrich R, et al. Hepatic blood supply: comparison of optimized dual phase-contrast-enhanced threedimensional MR angiography and digital subtraction angiography. Radiology. 1999;211:51–58. 17. Carlos RC, Stanley JC, Stafford-Johnson D, Prince MR. Interobserver variability in the evaluation of chronic mesenteric ischemia with gadolinium-enhanced MR angiography. Acad Radiol. 2001;8:879–887. 18. Ernst O, Asnar V, Sergent G, Lederman E, Nicol L, Paris JC, Hermine CL. Comparing contrast-enhanced breathhold MR angiography and conventional angiography in the evaluation of mesenteric circulation. AJR Am J Roentgenol. 2000;174:433–439. 19. Vosshenrich R, Engeroff B, Obenauer S, Grabbe E. [Contrastenhanced MR angiography of the arterial and portovenous system of the liver with varying concentrations of contrast medium] Rofo. 2003;175:1239–1243. 20. Barger AV, Block WF, Toropov Y, Grist TM, Mistretta CA. Timeresolved contrast-enhanced imaging with isotropic resolution and

20

21.

22.

23. 24.

25.

26.

27. 28.

Abdominal Aorta and Mesenteric Vessels broad coverage using an undersampled 3D projection trajectory. Magn Reson Med. 2002;48:297–305. Kramer U, Fenchel M, Laub G, et al. Low-dose, time-resolved, contrast-enhanced 3D MR angiography in the assessment of the abdominal aorta and its major branches at 3 Tesla. Acad Radiol. 2010;17:564–576. Hagspiel KD, Angle JF, Spinosa DJ, Matsumoto AH. Mesenteric ischemia: angiography and endovascular interventions. In: Longo W, GJ Peterson, DL Jacobs, eds. Intestinal Ischemia Disorders: Pathophysiology and Management. St Louis, MO, Quality Medical Publishing, Inc.; 1999:105–154. Meaney JFM. Non-invasive evaluation of the visceral arteries with magnetic resonance angiography. Eur Radiol. 1999;9:1267–1276. Lutz AM, Willmann JK, Pfammatter T, Lachat M, Wildermuth S, Marincek B, Weishaupt D. Evaluation of aortoiliac aneurysm before endovascular repair: comparison of contrast-enhanced magnetic resonance angiography with multidetector row computed tomographic angiography with an automated analysis software tool. J Vasc Surg. 2003;37:619–627. Pitton MB, Schweitzer H, Herber S, et al.MRI versus helical CT for endoleak detection after endovascular aneurysm repair. AJR Am J Roentgenol. 2005;185:1275–1281. Venkataraman S, Semelka RC, Weeks S, Braga L, Vaidean G. Assessment of aorto-iliac disease with magnetic resonance angiography using arterial phase 3-D gradient-echo and interstitial phase 2-D fat-suppressed spoiled gradient-echo sequences. J Magn Reson Imaging. 2003;17:43–53. Gotway MB, Araoz PA, Macedo TA, et al. Imaging findings in Takayasu’s arteritis. AJR Am J Roentgenol. 2005;184:1945–1950. Chan FP, Li KC, Heiss SG, Razavi MK. A comprehensive approach using MR imaging to diagnose acute segmental mesen-

281

29.

30.

31.

32.

33.

34.

35.

36.

teric ischemia in a porcine model. AJR Am J Roentgenol. 1999;3:523–529. Cantisani V, Carbone I, Marcelli G, Venditti F, Bezzi M. A septic embolism and mycotic pseudoaneurysm of the superior mesenteric artery in a patient with enterococcal endocarditis. Their diagnosis by computed tomography and magnetic resonance angiography. Radiol Med (Torino). 2000;99:285–287. Li KC, Pelc LR, Puvvala S, Wright GA. Mesenteric ischemia due to hemorrhagic shock: MR imaging diagnosis and monitoring in a canine model. Radiology. 1998;206:219–225. Leung DA, Schneider E, Kubik-Huch R, Marincek B, Pfammatter T. Acute mesenteric ischemia caused by spontaneous isolated dissection of the superior mesenteric artery: treatment by percutaneous stent placement. Eur Radiol. 2000;10:1916–1919. Ruehm SG, Weishaupt D, Debatin JF. Contrast-enhanced MR angiography in patients with aortic occlusion (Leriche Syndrome). J Magn Reson Imaging. 2000:11:401–410. Shin TA, Hagspiel KD. Visceral artery aneurysms: endovascular management. In: Kandarpa K, ed. Peripheral Vascular Interventions. Lippincott Williams & Wilkins, Philadelphia, PA, 2008:263–277. Stafford-Johnson DB, Hamilton BH, Dong Q, et al. Vascular complications of liver transplantation: evaluation with gadoliniumenhanced MR angiography. Radiology. 1998;207;153. Hagspiel KD, Nandalur K, Pruett TL, et al. Evaluation of vascular complications of pancreas transplantation with high-spatial-resolution contrast-enhanced MR angiography. Radiology. 2007;242: 590–599. Hagspiel KD, Nandalur K, Burkholder B, et al. Contrast-enhanced MR angiography after pancreas transplantation: normal appearance and vascular complications. AJR Am J Roentgenol. 2005;184: 465–473.

wwwwwwwwwwww

Renal Vascular Diseases Tim Leiner and Henrik Michaely

Introduction Renal artery stenosis (RAS) is a relatively common and wellknown condition and potentially curable cause of secondary hypertension [1]. The main clinical syndromes associated with RAS are renovascular hypertension (RVH), ischemic nephropathy, proteinuria, and flash pulmonary edema [2]. Diagnosis and management of RAS remains an important clinical problem especially considering that the prevalence of RAS is increasing, mainly due to greater awareness of the long-term deleterious consequences of untreated RVH and the increase in patients with diabetes mellitus [3]. In recognition of this problem, it is important to noninvasively identify patients with RAS in whom an intervention might delay the decline or even improve renal function. However, whereas it was formerly thought that the mere presence and subsequent endovascular treatment of RAS of more than 50% luminal narrowing would improve renal function, several randomized trials have shown this is not necessarily the case and only certain patients will benefit from intervention [4–9]. Recent advances in MR gradient hardware in combination with the migration to 3.0 T now enable the acquisition of isotropic submillimeter spatial resolution datasets, facilitating the detection of renal artery narrowing with high accuracy [10, 11]. Although these advances are rather evolutionary, the real value of MR lies in the fact that it is now also possible to demonstrate the functional consequences of RAS such as a decline in renal perfusion and glomerular filtration.

T. Leiner, MD, PhD () Department of Radiology, Utrecht University Medical Center, Heidelberglaan 100, Utrecht NL-3584CX, The Netherlands e-mail: [email protected] H. Michaely, MD, PhD Institute of Clinical Radiology and Nuclear Medicine, University Medical Center Mannheim, Theodor-Kutzer-Ufer 1-3, 68167 Mannheim, Germany

21

The unique capability of MRI to study not only the anatomy of the renal arteries and kidneys but also a variety of physiological phenomena have culminated into a set of tools that are of high value for this particular clinical problem. In the present review, we discuss recent advances in contrast-enhanced MRA of the renal arteries and kidneys, and how they can be applied to improve the diagnostic workup of suspected RAS.

Renal Artery MR Imaging Renal artery MR imaging is primarily focused on the detection of RAS. Atherosclerosis accounts for 70–90% of cases of RAS and usually involves the ostium and proximal third of the main renal artery [1, 12]. Fibromuscular dysplasia (FMD) is a collection of vascular diseases that affects the intima, media, and adventitia and is responsible for 10–30% of cases of RAS [1, 12, 13]. Although intraarterial digital subtraction angiography (IA-DSA) is traditionally regarded as the definitive test to diagnose the presence of RAS, both the invasive nature of IA-DSA and the difficulty in assessing the pathophysiological significance of stenotic lesions have encouraged the search for widely available non- or minimally invasive diagnostic tests. In addition, IA-DSA is by no means a perfect test for the detection of RAS as it is subject to substantial interobserver variation [14]. With the introduction of high spatial-resolution contrast-enhanced magnetic resonance angiography (CE-MRA), a reliable alternative for IA-DSA has emerged.

Anatomical Considerations The renal arteries arise from the abdominal aorta and assume a dorsoinferolateral course until they enter the kidney at the renal hilum. In about one third of the general population, there are variations in number, location, and branching

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_21, © Springer Science+Business Media, LLC 2012

283

284

T. Leiner and H. Michaely

Fig. 21.1 63-Year-old male patient suffering from refractory hypertension. Whole volume maximum intensity projection (a) depicts renal arterial anatomy with exquisite detail, clearly demonstrating the early bifurcation of the right renal artery. Thin-slab sliding MIPs of the left (b) and right (c) renal arteries allow for more detailed inspection. Phase-contrast imaging (axial MIP; d) confirms brisk flow through to both kidneys. Both renal arteries were considered normal

patterns of the renal arteries, with over 30% of subjects having one or more accessory renal arteries [15, 16]. This is clinically important because RAS in an accessory renal artery can, albeit rarely, also be responsible for RVH [17]. In healthy adults, the mean maximum diameter of the renal artery lumen is about 5–6 mm. The diameter of accessory renal arteries is highly variable but is generally equal to or smaller than the main renal artery [15]. There is no consensus on what constitutes a “significant” stenosis but most authors use a reduction in luminal diameter of ³50% as the cut-off point [18]. Precise knowledge of renal arterial anatomy is important because it determines the spatial resolution of the MR imaging sequence in order to reliably differentiate a stenosed from a healthy vessel. At least three pixels are needed across the lumen of an artery to quantify the degree of stenosis with an error of less than 10% [19]. Taking into account the average diameter of 5–6 mm, the spatial resolution of any given imaging technique must ideally be in the order of 1.0 × 1.0 × 1.0 mm3. In addition to the arterial supply, it is important to evaluate renal size, cortical thickness, and corticomedullary differentiation and to compare these parameters with the contralateral kidney [1].

MR Angiography for Detection of RAS Because of the relative ease of use and high reliability contrastenhanced (CE) MRA remains the preferred MR technique

for the detection of RAS [20]. In addition, phase-contrast MR flow measurements should be obtained to supplement the anatomical information (see Section on “Evidence-Based Indications and Applications for Functional Renal Imaging”). Also some of the newer, noncontrast-enhanced MRA techniques that have recently been shown to be promising for the detection of RAS are discussed. In CE-MRA, the renal arteries are imaged in the coronal plane during initial arterial passage of 0.1–0.3 mmol/kg 0.5 M extracellular gadolinium chelate contrast medium. Because of the increase in T1 of tissue at 3.0 T, the contrast dose can be reduced relative to 1.5 T [21]. For best results, patients are required to hold their breath during the acquisition, which typically lasts about 10–20 s, depending on the resolution and other technical factors related to system performance. Contrast medium is injected at speeds up to 3.0 mL/s, followed by 25 mL saline flush, and spatial resolution in current reports is typically in the order of 1.0 × 1.0 × 1.5–2.0 mm3 (craniocaudal/frequency direction × left-right/ phase-encoding direction × anteroposterior/slice direction) or better. Using this approach, the abdominal aorta and renal arteries, including accessory arteries, can be visualized with high accuracy (Figs. 21.1 and 21.2). Arteries can usually be evaluated down to the proximal part of the segmental arteries. Distal segmental and interlobar branches cannot be evaluated reliably at this time [20, 22]. To obtain a study with maximum arterial and minimum venous enhancement, it is very important to ensure careful synchronization of peak arterial contrast concentration with sampling of central

21

Renal Vascular Diseases

Fig. 21.2 50-Year-old female suffering from unexplained hypertension with underlying scleroderma and known peripheral arterial disease. In (a), the whole volume maximum intensity projection suggests a stenosis in the left renal artery of moderate to severe intensity. The subvolume MIP, however, suggests a mild stenosis (arrowhead in b). Supplemental phase-contrast imaging confirms relatively preserved flow in the left renal artery confirming mild instead of severe stenosis (arrowhead in c)

k-space profiles [23]. Formerly, this was typically done by performing a timing sequence with 1–2 mL of contrast medium prior to the contrast-enhanced acquisition. This approach has been superseded by the use of real-time bolus monitoring software. With the latter technique, the entire contrast bolus is injected and simultaneously monitored by either the operator or the MR scanner, and when sufficient enhancement is present in the descending aorta the MR fluoroscopy sequence is aborted, the patient is given a breathhold command and the 3D CE-MRA acquisition is started. A third option, available on the most advanced systems, is the use of a time-resolved technique to obtain a series of high spatial-resolution 3D volumes by using view-sharing techniques [11]. Because of the risk for motion artifacts, it is important to limit breath-hold length. Motion artifacts may occur when patients are unable to sustain the breath-hold because the acquisition is too long, and because even while performing a breath-hold the kidneys are subject to linear caudocranial motion [22]. Application of parallel imaging technology has lead to higher spatial-resolution CE-MRA and shorter acquisition durations [24]. Both nonenhanced (two-dimensional time-of-flight [TOF] and phase-contrast [PC]) as well as contrast-enhanced (CE)

285

MRA techniques have been investigated for detection of RAS. The results of these studies were summarized in a meta-analysis by Vasbinder et al. Of the 306 studies published up to mid 2000, in which the utility of renal MRA was investigated, 18 studies were performed because of clinical suspicion of RVH, used explicitly defined criteria for the presence of RAS, and used IA-DSA as the standard of reference test. Reported sensitivities and specificities for the detection of atherosclerotic RAS in these CE-MRA studies are uniformly high [18]. The reported sensitivity of CE-MRA for the visualization of accessory renal arteries is over 90% [25, 26]. At present, a single study has been published that specifically investigated the utility of CE-MRA for the detection of FMD. Willoteaux et al. [27] retrospectively analyzed the accuracy of CE-MRA in comparison to IA-DSA in 25 subjects with angiographically proven FMD. The authors evaluated the sensitivity and specificity and accuracy of CE-MRA for the detection of FMD-associated stenosis, “string-of-pearls” sign and aneurysm formation. Although the sensitivity for FMD-associated stenosis in their study was only 68%, the sensitivity for the detection of the “string-of-pearls” sign and aneurysm formation were 95% and 100%, respectively. The overall sensitivity and specificity for the diagnosis FMD were 97% and 93%. However, although overt cases of FMD can be diagnosed with CE-MRA (Fig. 21.3a, b), the general opinion is that CE-MRA is currently not able to detect FMD with high accuracy in the presence of only subtle anatomical changes. The favorable results of the aforementioned CE-MRA studies are, however, in contrast with the results of a large multicenter study from The Netherlands in which the validity of CE-MRA and CTA were prospectively investigated in 356 patients suspected of having RVH, using IA-DSA as the standard of reference. Two panels of three observers judged CE-MRA and CTA exams, and were blinded for each other results and the results of all other imaging modalities. Overall, sensitivity ranged from 61 to 69% for CTA, and 57 to 67% for CE-MRA. Specificity ranged from 89 to 97% for CTA and 77 to 90% for CE-MRA [16]. Additional analyses revealed that selecting a subgroup of patients with high prevalence of RAS could increase the diagnostic performance of both tests, but not to levels as commonly reported in the literature. Possible explanations for these discrepant findings are suboptimal technique, low overall disease prevalence, a high proportion of patients with FMD, and imperfect standard of reference [16]. Strikingly similar results were obtained in a more recent multicenter trial by Soulez et al. [28] who investigated the diagnostic performance of gadobenate dimeglumine for the detection of RAS using IA-DSA as the reference standard in 268 patients with hypertension and/or suspected RAS. Sensitivity, specificity, and accuracy for the detection of 51% or more RAS on patient level in this study ranged from 65.2 to 79.9%, 81.3 to 91.4%, and 73.6 to

286

T. Leiner and H. Michaely

Fig. 21.3 61-Year-old male patient with therapy-resistant hypertension. In (a), a whole volume maximum intensity projection of the abdominal aorta and renal arteries show the typical “string-of-pearls” sign suggestive of fibromuscular dysplasia (FMD). In (b), a zoomed, thin-slab maximum intensity projection is shown of the area enclosed by the white box in (a). There is a high-grade stenosis at the origin of the right renal artery (arrow). In addition, both renal arteries show the “string of pearls” sign characteristic for FMD (arrowheads). Multiplanar reformations (viewed in transverse orientation) are helpful to confirm location and severity of high-grade stenosis in the right main renal

artery (c, arrow), as well the string-of-beads sign in the distal right (c, arrowhead) and left renal arteries (d, arrowhead). In (e), the corresponding intraarterial digital subtraction angiogram is shown, which confirms the MRI findings. Diagnostic accuracy can be increased further by measuring reduction in cross-sectional area in a stenosis (as opposed to reduction in diameter) as shown in (f) (right renal artery). This functionality is available on most commercially available postprocessing workstations (example was made with Philips EasyVision Software, R 4.0, Philips Medical Systems, Best, The Netherlands). Ao aorta, L left, R right

83.8%, respectively. A more recently published multicenter study in 395 patients with suspected or known RAS by Garovic et al. [29] reported somewhat better results with sensitivities and specificities of three blinded observers ranging between 81 and 86% compared to IA-DSA.

Newer Noncontrast-Enhanced MRA Techniques The newer noncontrast-enhanced techniques comprise arterial spin labeling (ASL), and several variants of balanced steady-state free precession (bSSFP). An extensive and

21

Renal Vascular Diseases

excellent review of these techniques has been published by Wilson et al. [30]. Below the principles of these techniques are briefly discussed as well as data on their diagnostic accuracy. In ASL, two acquisitions are performed. First, spins in blood upstream from the kidney are labeled by an inversion prepulse. After a prespecified delay, the volume of interest is imaged, typically with a bSSFP readout. Subsequently, a nonlabeled acquisition is performed and the labeled acquisition is subtracted from the nonlabeled acquisition, creating an image of the renal arteries, or, if the delay is sufficiently long, a map of renal blood flow. There is only limited data on the diagnostic accuracy of ASL. Spuentrup et al. [31] were the first to describe this method and to successfully image seven patients with suspected renal arterys stenosis. Fenchel et al. [32] imaged 18 patients with and without RAS and found a good correlation with nuclear imaging. Balanced SSFP is another promising noncontrast-enhanced MRA technique and can be combined with fat-saturation preparation, ECG-triggering and respiratory motion correction using either breath holding or navigator gating [30]. Several small proof-of-concept studies have reported encouraging results using bSSFP sequences for detection of RAS, with sensitivity, specificity, and accuracy compared to IA-DSA and CE-MRA generally in the 90% range [33–39]. Two somewhat larger and very recent studies in 67 [40] and 60 patients [41] also report encouraging results with sensitivities and specificities between 82–94% (compared to CE-MRA) [40] and 99–100% (compared to 64-slice CTA) [41], respectively. However, there still are no large studies comparing either ASL or bSSFP renal artery imaging with established standard of reference, IA-DSA. It is likely that without such studies these techniques will remain experimental.

287

The Use of Gadolinium-Based Contrast Media in Patients with Impaired Renal Function Because of the extensive experience with CE-MRA and the superior results obtained with this technique compared to nonenhanced MRA, it remains the preferred technique for the detection of RAS, as well as the study of renal perfusion (see below). Administration of gadolinium-based contrast agents (GdBCA) in patients with acute or chronic severely impaired renal function (CKD 4 and 5) [44] may, however, lead to nephrogenic systemic fibrosis (NSF). NSF is a rare idiopathic systemic fibrosing disorder characterized by pain, dermopathy, and joint contractures [45]. To date, virtually all unconfounded cases of NSF have been associated with the administration of the linear group of GdBCA [gadodiamide (Omniscan®, GE Healthcare, Chalfont-St. Giles, UK), gadopentetate dimeglumine (Magnevist®, Bayer Schering Pharma, Berlin, Germany), and gadoversetamide (OptiMark, Mallinckrodt Inc, St. Louis, MO)]. Therefore, these agents are contraindicated in patients with severe renal impairment. The macrocyclic agents gadobutrol (Gadovist® BayerSchering Pharma, Berlin, Germany), Gd-HPDO3A (ProHance®; Bracco, Milan, Italy) and Gd-DOTA (Dotarem®; Guerbet, Aulnay-sur-Bois, France) [46] on the other hand have not been convincingly associated with NSF, and some experts deem them to be safe in patients with severely impaired renal function [47]. Although the exact pathogenesis of NSF is a rapidly moving target, we would advise caution at present in this group of patients. In patients with normal or mildly decreased renal function (CKD stages 1–3), GdBCA can however be used without problems [44]. For current guidelines, the reader is referred to the Web sites of the European Medicines Agency and the United States Food and Drug Administration [48, 49].

Postprocessing and Display of MRA Data Because MRA datasets are typically truly three-dimensional (3D), they can be viewed from an infinite number of angles after acquisition without the need for additional injections of contrast medium. Datasets are usually displayed using raycasting algorithms such as (targeted) maximum intensity projection (MIP), shaded surface display (SSD), or volume rendering (VR). Although these techniques are useful to get an overview of renal arterial anatomy, the final diagnosis should always be made by combining review of original partitions, multiplanar reformations along the vessel axis, and reformations orthogonal to the stenosis (Fig. 21.3c–f). The latter postprocessing technique significantly increases diagnostic accuracy [42]. In addition, recent developments in quantitative vessel analysis are promising with regard to reduction of intra- and interobserver variation for measuring renal artery geometry [43].

Assessment of the Functional Significance of RAS MRI has the advantage of not only providing MR-angiographic data with high spatial resolution and high image contrast, but also offers the possibility of acquiring functional data. Functional data allow characterization of blood flow in the renal artery and vein [50], to measure renal parenchymal blood flow with [51–54] or without contrast agent [55, 56], and to determine glomerular filtration rate (GFR) as well as the split renal function [57, 58]. Functional renal imaging techniques have been studied extensively in preclinical and volunteer studies. Meanwhile, these techniques are increasingly applied clinically as the combination of MRA and functional renal imaging techniques allows for comprehensively assessing renal morphology and renal function. For example, in many cases, the hemodynamic relevance of a

288

RAS cannot be assessed correctly, particularly in intermediategrade RAS. In these cases, measurement of renal artery flow and renal parenchymal perfusion will improve assessment of the hemodynamic relevance of a RAS. Other applications of functional renal imaging sequences include the detection of renoparenchymal damage in the absence of renovascular disease.

Technical Concepts Phase-Contrast MR Flow Measurements Phase-contrast (PC) MR flow measurements require the acquisition of two ECG-gated corresponding phase images one of which does not contain flow information (flowcompensated image) while the other, flow-sensitive image is subject to a phase shift proportional to the flow velocity. Preferably the imaging plane is chosen perpendicular to the axis of the vessel. Velocity information can be extracted by subtracting the two phase images. The velocity encoding (VENC) sensitivity has to be chosen beforehand and is typically set to 75 cm/s for the renal arteries. Choosing a lower VENC may lead to aliasing of the flow measurements while a higher VENC may lead to underestimation of the measured flow values. PC MR flow measurements can be acquired using standard gradient-echo-sequences (GRE) with resulting measurement times of 3–5 min depending on the cardiac cycle [50]. Faster approaches for flow measurements include the use of interleaved echoplanar imaging sequences (EPI) and spatio-temporal undersampling such as k-t-BLAST [59, 60]. ASL Perfusion Measurements ASL techniques allow measurement of renal blood flow without the administration of contrast agent. As explained above, the basic principle of ASL techniques is to magnetically label blood outside of the imaging plane. The background tissue in the imaging plane is then selectively suppressed. After the inflow time TI in which the magnetically labeled blood enters the imaging plane, the image acquisition takes place. Typically a flow-sensitive alternating inversion recovery (FAIR) technique is used in combination with a TrueFISP or HASTE-readout [56, 61]. ASL techniques can display renal perfusion but cannot be used to determine the renal function as the signal from inflowing blood continuously decays. At higher field strengths with prolonged T1-times, ASL techniques benefit twofold: the inherently SNR-low ASL techniques benefit from the higher baseline SNR at, e.g., 3.0 T but also from the prolonged T1-times of blood which leads to a slower signal decay of the labeled blood [55]. Depending on the technique used, the acquisition of ASL-data takes between 30 s and 5 min.

T. Leiner and H. Michaely

Dynamic Contrast-Enhanced MRI Dynamic contrast-enhanced (DCE) MRI of the kidneys has been first reported more than 20 years ago [62]. In DCE, fast repetitious imaging of the kidney is performed during the first pass of the contrast agent. The temporal resolution has to be chosen high enough to monitor the arrival of the contrast agent in the kidneys. A recent study investigating the minimal temporal resolution found that values of 4 s or lower provide best results [63]. To achieve this high temporal resolution, fast heavily T1-weighted two-dimensional sequences such as TurboFLASH (Fast Low Angle Shot), SPGR (Spoiled Gradient Recalled Echo), or FFE (Fast Field Echo) have to be employed [54, 64]. These sequences provide a sufficiently high temporal resolution but only a limited spatial coverage due to their two-dimensional character. In addition, as these sequences are designed for maximal speed their SNR is inherently low. Therefore, two trends have clearly evolved in the past years: three-dimensional DCE imaging and 3.0 T imaging. Three-dimensional sequences that are typically used for perfusion imaging are VIBE (volume-interpolated breath-hold examination) and TWIST (time-resolved MRA with stochastic trajectories) [57, 65, 66]. Three-dimensional sequences offer higher SNR than comparable two-dimensional sequences and allow covering the entire kidneys without intersection gaps. Typically, the slice thickness of three-dimensional sequences is thinner (3–5 mm) compared to two-dimensional approaches (5–8 mm) [21, 67]. Similarly the in-plane resolution of three-dimensional sequences is superior to that of two-dimensional sequences. These advantages of three-dimensional DCE-sequences have to be balanced against their longer acquisition times with resulting lower achievable temporal resolution. Parallel imaging and echo-sharing have however significantly decreased the acquisition times of three-dimensional data sets without major degradation of the image quality. While echo-sharing works equally well at both field strengths (1.5 T and 3.0 T), the application of parallel imaging is more beneficial at 3.0 T where the inherent SNR loss with parallel imaging is counterbalanced by the SNR gain at 3.0 T (Fig. 21.4) [21]. The perfusion sequence and the administration of the contrast agent should be started at the same time in order to acquire several images before arrival of the contrast agent. These images serve as baseline images for the postprocessing of the perfusion data. The amount of contrast agent and the contrast agent differ between the published studies. Apart from the protein-binding contrast agents, all other contrast agents are freely filtered at the glomerulus without neither being excreted nor being reabsorbed in the tubuli (Fig. 21.5) [68]. Therefore, widely used standard contrast agents such as gadodiamide (Omniscan®, GE Healthcare) or Gd-DTPA (Magnevist, Bayer Healthcare) can be utilized to determine the GFR. A fast injection rate of at least 2 mL/s followed by a saline chaser of 30 mL at the same injection rate should be

21

Renal Vascular Diseases

289

Fig. 21.4 Comparison of 1.5 T (upper row) and 3.0 T (lower row) MR perfusion measurements of the kidneys. Despite the application of parallel imaging at 3.0 T which enabled the acquisition of 5 simultaneous slices per se cond instead of 4 slices per second as at 1.5 T, the signal intensity of the kidney is favorable at 3.0 T which can be particularly appreciated in the late phase 120 s postcontrast agent injection

chosen to allow for a compact contrast agent bolus even after the pulmonary passage. A compact contrast agent bolus will yield maximal T1-shortening during the first pass and hence yield optimal enhancement of the aorta and of the kidneys. The amount of contrast agent to be injected for DCE-MRI should be chosen as low as possible to avoid T2*-related signal nonlinearity but high enough to allow for good delineation of the kidneys even in the equilibrium phase. Particularly in view of the potential side effects of gadoliniumchelates that have been discussed above, the administered amount of Gadolinium ought to be minimized [69]. In the literature, different amounts of contrast agents were used ranging from 2 to 7 mL of standard 0.5 M extracellular contrast agent [57, 64, 70]. A dose of 4 mL of 0.5 M contrast agent seems to be a good compromise between sufficient enhancement of the kidneys and preserved signal linearity. Slightly protein-binding contrast agents were found to lead to erroneously low filtration parameters in one study and hence seem not to be optimal tracers for DCE-MRI [53]. At 3.0 T, the longer T1-times of the background tissue in combination with only slightly reduced contrast agent relaxivity enable a significantly better image contrast [21]. The effect of this combination is most apparent in the later phases of the DCE-MRI measurements after the first pass when the contrast agent accumulates in the medulla and is being excreted renally [21]. Given the current technical capabilities and the evidencebased knowledge about renal DCE-MRI, the following recommendations can be given: • The field strength should be at least 1.5 T and a phasedarray surface coil or matrix coil for optimal SNR is recommended. • 2D-GRE (SR-TurboFLASH) or 3D-GRE (TWIST > VIBE) are the preferred sequence types and parallel imaging should only be used with 3D-sequences or at 3.0 T.

• The slice orientation should be in the oblique coronal plane, in case of 2D-sequences one additional axial slice should be used to measure the arterial input function. • The temporal resolution should be at least 4 s and total measurement time at least 180 s. • Spatial resolution of 2.5 × 2.5 mm² in-plane seems sufficient with slice thickness between 4 and 8 mm. • 4 mL of 0.5 M extracellular standard contrast agent should be injected at 3–4 mL/s. Before functional renal parameters can be derived from DCE-MRI, the data have to be postprocessed. As none of the vendors is currently offering a standardized software tool for this purpose, several self-written software packages from different sites are in use worldwide. Postprocessing of the data is challenging in a few regards. The overall amount of data is rather big (>200 MB for one DCE-MRI) which is a minor hurdle. Major problems are the motion of the kidneys during the respiratory cycle as well as the deformation of the kidneys throughout the measurement. As respiration-dependent motion is regular in its pattern, it can be corrected during the postprocessing [71, 72]. In addition, automated and semiautomated segmentation algorithms are becoming available which will further facilitate the postprocessing of DCEMRI datasets [71, 73]. The correction of the renal deformation is currently the focus of several research groups but the issue has not been solved sufficiently so far. To derive functional renal parameters from the MR data, signal-intensity–time curves have to be derived from the kidneys as well as from the aorta. Based on multicompartmental models, functional parameters describing the first pass (mean transit time, plasma volume, and plasma flow) as well as parameters characterizing the renal excretory function can be calculated (Figs. 21.6 and 21.7) [53, 74, 75]. From the latter, the split renal function can be calculated with good correlation to scintigraphy [57]. Different approaches for the determination

290

Fig. 21.5 Schematic semiquantitative perfusion curves. In (a), measured signal intensity versus time curves of a healthy volunteer derived from a region of interest in the renal cortex (black line) and the renal medulla (red line) are shown. The cortical curve shows a steep upslope with a distinct peak, a recirculation peak, and a subsequent signal decay which represents the excretion of the contrast agent from the cortex. In contrast, the medullary curve is characterized by a less distinct first pass part with only moderate signal increase initially after the contrast agent injection. However, an increase in the medullary signal intensity can be seen over time representing the accumulation of the contrast agent in the medulla and the collecting system in particular. To derive semiquantitative perfusion parameters from measured signal intensity values, a function fitted to the measured curves has to be found which is presented in (b). From the distinct curve parameters such as the peak signal intensity, the upslope, the time to peak, and the full width at half maximum (a parameter related to mean transit time) can be calculated. These semiquantitative perfusion parameters reflect the renal perfusion but they are dependent on inflow effects, bolus length, cardiac function, etc. Therefore, quantitative approaches which take the variation of the input function into account and are thus not dependent on any of the above-mentioned factors are preferred (Reprinted with permission from Michaely et al. [11].)

of the renal first pass have been published based on the uptake of gadolinium chelate in the kidneys [76, 77].

Evidence-Based Indications and Applications for Functional Renal Imaging MR-PC-flow measurements of the renal arteries have been extensively studied. Their main application is to detect and

T. Leiner and H. Michaely

Fig. 21.6 Schematic representation of the two-compartment model: blood flowing into the kidney is used as input function. The parameters derived from the two-compartment model are the plasma flow (FP) and plasma volume (VP) which characterize the first pass distribution of the blood. From the cortex-enhanced blood as the endogenous tracer is then distributed into the tubular system which is characterized by the filtration parameters tubular flow (FT) and the tubular volume (TV)

grade renal artery stenoses [78]. The flow profile derived from the renal artery shows characteristic changes with increasing degree of renal artery narrowing. If used in combination with high spatial-resolution MRA of the renal arteries, MR-PC-flow measurements minimize interobserver variability and lead to an improved accuracy, as has been demonstrated in a tricenter study of 43 renal arteries [79]. In combination with ASL, perfusion measurements can be employed to differentiate healthy kidneys from abnormal kidneys. In a study with 24 volunteers and 46 patients with suspected RAS, a specificity of 99% and sensitivity of 69% with a positive/negative predictive value of 97%/84% was achieved for the separation of healthy kidneys from kidneys with vascular, parenchymal, or combined disease [61]. Finally, a recent study by Ritt et al. [80] compared ASLbased kidney perfusion with renal plasma flow as determined with para-aminohippuric acid (PHA) plasma clearance before and after 2 weeks of 80 mg daily antihypertensive therapy with the angiotensin-II receptor blocker telmisartan. A good correlation was found between both methods at both timepoints, even though absolute perfusion values differed significantly with both techniques (mean ASL-derived flow of 253 ± 20 mL/min/100 g versus mean PAH-derived flow of 313 ± 47 mL/min/100 g).

21

Renal Vascular Diseases

291

Fig. 21.7 MIP view of a contrast-enhanced MRA of a normal renal artery (left image). In the color-coded parameter maps derived from the MR-perfusion measurements, normal plasma flow and tubular flow with homogenous distribution throughout the kidney can be appreciated

Most publications on DCE-MRI have focused on technical concepts and postprocessing. At this time, there are only a limited number of clinical papers on the topic available. An initial study focused on the parenchymal blood flow changes in patients with RAS [11]. In this study, with 73 patients with semiquantitative postprocessing, significant differences in blood flow parameters between patients without RAS and those with significant (>75%) RAS were found. Patients with intermediate RAS showed nonsignificantly decreased perfusion parameters. In this study and in another study by Michoux and colleagues, significant correlations between raising serum creatinine levels and decreasing renal perfusion parameters were found as well indicating that renal first pass perfusion parameters may reflect – at least to a certain degree – renal function [54, 77]. A smaller study by Vallee and colleagues [54] included four renal transplants with RAS and seven renal transplants with renal failure and reported on significantly reduced blood flow in transplants with either RAS or with renal failure where the latter showed a smaller residual perfusion. In a further study, the same group investigated 30 patients with normal renal function or chronic renal failure the cortical and medullary perfusion [77]. Significant reductions of the cortical perfusion, the medullary perfusion, and of the accumulation of contrast media in the medulla were found in the presence of renal failure. Similarly, detection of segmental perfusion deficits in renal transplants with focal rejection was also reported with DCE-MRI [81]. A further application of DCE-MRI is the evaluation of kidneys after stent placement [11, 64] when the renal artery cannot be assessed with CE-MRA due to stent-induced susceptibility artifacts (Fig. 21.8). DCE-MRI allows demonstrating normalized perfusion parameters after successful stent placement and hence proves the patency of the stent.

Future Applications Several clinical trials aimed at elucidating the optimal treatment strategy in patients with RAS have recently been published or are in the final phase of follow-up. The conventional wisdom, i.e., to dilate and/or stent in the presence of 50% or greater RAS, was greatly challenged in 2000 with the appearance of the results of the study by van Jaarsveld et al. [8], who found no difference at 12-months follow-up in 106 patients with RAS and hypertension randomized between medical antihypertensive therapy and angioplasty. The Dutch benefit of stent placement and blood pressure and lipid-lowering for the prevention of progression of renal dysfunction caused by atherosclerotic ostial stenosis of the renal artery (STAR) trial also reported a lack of effect of stenting. In 140 patients with impaired renal function and RAS >50%, stent placement had no clear effect on progression of impaired renal function but was associated with significant procedurerelated complications [82]. Another large study that provides evidence against renal artery revascularization is the randomized but unblinded ASTRAL trial that was recently published in the New England Journal of Medicine [83]. The ASTRAL trial found no evidence of clinical benefit from renal artery revascularization over optimized medical treatment in 806 patients with suspected atherosclerotic RAS followed for a median of 34 months. On the other hand, substantial risks were reported in 23/403 (5.7%) patients randomized to the interventional arm of the study [83]. A final study that is currently still underway is the 1,000-patient cardiovascular outcomes with renal atherosclerotic lesions (CORAL) study [84], which aims to address a similar question to ASTRAL. Results of this trial are expected in 2011.

292

Fig. 21.8 MRA and perfusion measurements pre- and postintervention. In (a), a coronal thin-slab MIP of the MRA (1.5 T, 1 × 0.9 × 1 mm³, PAT factor 2) of a 55-year-old male patient with hypertension who presented with a proximal high-grade stenosis of the right renal artery is shown. After stenting (left image), the stent artifact disrupts the local magnetic field so that no vascular enhancement can be appreciated at this site. In order to assess the renal parenchymal blood flow and success of the intervention renal perfusion measurement were performed at the baseline visit (b, right column) as well as after the intervention (b, left column). Before the intervention delayed perfusion of the affected,

Common to all of these trials is selection of patients purely based on the presence of RAS, not taking into account the functional consequences of the stenosis nor the degree of renal impairment. Future trials using combined MRA/renal MRI protocols will have to be performed to investigate whether such a protocol leads to optimized selection of patients in whom revascularization therapy is indeed beneficial. Apart from the above, the detection and differentiation of renoparenchymal disease independent from the presence of RAS may be another suitable indication for functional MR imaging techniques (Fig. 21.9). Larger single-center studies on this topic are currently being undertaken. Finally, it is worthwhile to mention a new therapeutic concept in the treatment of therapy-resistant RVH. Recently, a very encouraging multicenter safety and proof-of-principle study employing renal sympathetic denervation in 50 patients was published in The Lancet [85]. This novel therapeutic concept seems to be much more efficacious compared to simple renal artery dilatation and/or stent placement. Krum et al. [85] found a mean reduction of 27 and 17 mmHg in systolic and diastolic office blood pressure at 12 months follow-up. If this effect were sustained in larger and randomized trials, it would certainly constitute a revolution in treatment of RVH.

T. Leiner and H. Michaely

right side could be appreciated with hypo-enhancement compared to the left side. After the intervention there was again a bilateral regular enhancement of both kidneys. The perfusion changes of the affected right kidney can be visualized semiquantitatively by using signal intensity versus time curves as done in this example (c). Comparing the preinterventional signal (red line) with the postinterventional signal (black line), a significant change can be appreciated. The impaired perfusion of the right kidney before intervention is reflected by the slower upslope and the delayed and lowered peak signal intensity in the semiquantitative assessment (Reprinted with permission from Michaely et al. [88].)

Conclusions Although IA-DSA is still regarded as the most accurate test for anatomical detection of RAS, MRA is an attractive noninvasive alternative in the diagnostic workup of patients suspected of having RVH. In addition to anatomical diagnosis of RAS, CE-MRA enables precise quantification of the degree of renal impairment using MR perfusion sequences. Additional studies are needed to establish reference values, to determine the optimal postprocessing protocols, and to investigate whether such a comprehensive MR imaging protocol improves selection of patients in whom revascularization is beneficial. Because the prevalence of RAS among patients with hypertension is low, the cost-effectiveness of any diagnostic strategy is sensitive to the pretest probability of RVH. Therefore, careful clinical evaluation in order to achieve a pretest probability of at least 20% is an essential component in the workup of patients suspected of having RVH [86, 87]. Because missing RVH may have serious consequences, the most important requirement for an alternative test is that it has high sensitivity. The combination of renal artery imaging and assessment of renal perfusion will undoubtedly lead to better selection of patients who will benefit from interventional therapy.

21

Renal Vascular Diseases

Fig. 21.9 Different time frames from the TurboFLASH perfusion study in a healthy volunteer who nicely demonstrates a good corticomedullary differentiation and a normal excretory function of the kidney (a). In contrast, different time frames of the same sequence in a patient with hypertension show very poor cortical enhancement after contrast agent administration, also no excretion can be demonstrated (b). In the absence of renal artery stenosis as can be seen on this thin MIP of the

References 1. Safian RD, Textor SC. Renal-artery stenosis. N Engl J Med. 2001;344:431–442. 2. Textor SC, Wilcox CS. Renal artery stenosis: a common, treatable cause of renal failure? Annu Rev Med. 2001;52:421–442. 3. Crook ED. The role of hypertension, obesity, and diabetes in causing renal vascular disease. Am J Med Sci. 1999;317:183–188. 4. Cheung CM, Hegarty J, Kalra PA. Dilemmas in the management of renal artery stenosis. Br Med Bull. 2005;73–74:35–55. 5. Ives NJ, Wheatley K, Stowe RL, et al. Continuing uncertainty about the value of percutaneous revascularization in atherosclerotic renovascular disease: a meta-analysis of randomized trials. Nephrol Dial Transplant. 2003;18:298–304. 6. Plouin PF, Chatellier G, Darne B, Raynaud A. Blood pressure outcome of angioplasty in atherosclerotic renal artery stenosis: a randomized trial. Essai Multicentrique Medicaments vs Angioplastie (EMMA) Study Group. Hypertension. 1998;31:823–829. 7. van de Ven PJ, Kaatee R, Beutler JJ, et al. Arterial stenting and balloon angioplasty in ostial atherosclerotic renovascular disease: a randomised trial. Lancet. 1999;353:282–286.

293

arterial phase MRA, the patient underwent biopsy which revealed extensive nephrosclerosis (c). Comparison of the cortical SIVT curve of the healthy volunteer (green, upper line) and of the patient (red lower line) shows a normal SIVT curve for the volunteer (d). The patient’s SIVT is grossly pathologic with a delayed upslope, a lacking first pass peak, and a diminished overall signal (Reprinted with permission from Michaely et al. [81].)

8. van Jaarsveld BC, Krijnen P, Pieterman H, et al. The effect of balloon angioplasty on hypertension in atherosclerotic renal-artery stenosis. Dutch Renal Artery Stenosis Intervention Cooperative Study Group. N Engl J Med. 2000;342:1007–1014. 9. Webster J, Marshall F, Abdalla M, et al. Randomised comparison of percutaneous angioplasty vs continued medical therapy for hypertensive patients with atheromatous renal artery stenosis. Scottish and Newcastle Renal Artery Stenosis Collaborative Group. J Hum Hypertens. 1998;12:329–335. 10. Leiner T. Magnetic resonance angiography of abdominal and lower extremity vasculature. Top Magn Reson Imaging. 2005;16:21–66. 11. Michaely HJ, Sourbron S, Dietrich O, Attenberger U, Reiser MF, Schoenberg SO. Functional renal MR imaging: an overview. Abdom Imaging. 2006;32:758–771. 12. Detection, evaluation, and treatment of renovascular hypertension. Final report. Working Group on Renovascular Hypertension. Arch Intern Med. 1987;147:820–829. 13. Slovut DP, Olin JW. Fibromuscular dysplasia. N Engl J Med. 2004; 350:1862–1871. 14. Schreij G, de Haan MW, Oei TK, Koster D, de Leeuw PW. Interpretation of renal angiography by radiologists. J Hypertens. 1999;17:1737–1741.

294 15. Kaufman JA. Renal Arteries. In: Kaufman JA, Lee MJ, eds. Vascular and Interventional Radiolgy. The Requisites. Philadelphia: Mosby; 2004:323–349. 16. Vasbinder GB, Nelemans PJ, Kessels AG, et al. Accuracy of computed tomographic angiography and magnetic resonance angiography for diagnosing renal artery stenosis. Ann Intern Med. 2004; 141:674–682; discussion 682. 17. Bude RO, Forauer AR, Caoili EM, Nghiem HV. Is it necessary to study accessory arteries when screening the renal arteries for renovascular hypertension? Radiology. 2003;226:411–416. 18. Vasbinder GB, Nelemans PJ, Kessels AG, Kroon AA, de Leeuw PW, van Engelshoven JM. Diagnostic tests for renal artery stenosis in patients suspected of having renovascular hypertension: a metaanalysis. Ann Intern Med. 2001;135:401–411. 19. Hoogeveen RM, Bakker CJ, Viergever MA. Limits to the accuracy of vessel diameter measurement in MR angiography. J Magn Reson Imaging. 1998;8:1228–1235. 20. Schoenberg SO, Prince MR, Knopp MV, Allenberg JR. Renal MR angiography. Magn Reson Imaging Clin N Am. 1998;6:351–370. 21. Michaely HJ, Kramer H, Oesingmann N, et al. Intraindividual Comparison of MR-Renal Perfusion Imaging at 1.5 T and 3.0 T. Invest Radiol. 2007;42:406–411. 22. Vasbinder GB, Maki JH, Nijenhuis RJ, et al. Motion of the distal renal artery during three-dimensional contrast-enhanced breathhold MRA. J Magn Reson Imaging. 2002;16:685–696. 23. Maki JH, Chenevert TL, Prince MR. Three-dimensional contrastenhanced MR angiography. Top Magn Reson Imaging. 1996;8: 322–344. 24. Weiger M, Pruessmann KP, Kassner A, et al. Contrast-enhanced 3D MRA using SENSE. J Magn Reson Imaging. 2000;12:671–677. 25. Bakker J, Beek FJ, Beutler JJ, et al. Renal artery stenosis and accessory renal arteries: accuracy of detection and visualization with gadolinium-enhanced breath-hold MR angiography. Radiology. 1998;207:497–504. 26. Shetty AN, Bis KG, Kirsch M, Weintraub J, Laub G. Contrastenhanced breath-hold three-dimensional magnetic resonance angiography in the evaluation of renal arteries: optimization of technique and pitfalls. J Magn Reson Imaging. 2000;12:912–923. 27. Willoteaux S, Faivre-Pierret M, Moranne O, et al. Fibromuscular dysplasia of the main renal arteries: comparison of contrastenhanced MR angiography with digital subtraction angiography. Radiology. 2006;241:922–929. 28. Soulez G, Pasowicz M, Benea G, et al. Renal artery stenosis evaluation: diagnostic performance of gadobenate dimeglumine-enhanced MR angiography – comparison with DSA. Radiology. 2008;247:273–285. 29. Garovic VD, Achauer MA, Kittner T, Horak D, Sheng R, Stanson AW. Comparison of gadodiamide-enhanced MR angiography to intraarterial digital subtraction angiography for evaluation of renal artery stenosis: results of a phase III multicenter trial. J Magn Reson Imaging. 2010;31:390–397. 30. Wilson GJ, Maki JH. Non-contrast-enhanced MR imaging of renal artery stenosis at 1.5 tesla. Magn Reson Imaging Clin N Am. 2009;17:13–27. 31. Spuentrup E, Manning WJ, Bornert P, Kissinger KV, Botnar RM, Stuber M. Renal arteries: navigator-gated balanced fast field-echo projection MR angiography with aortic spin labeling: initial experience. Radiology. 2002;225:589–596. 32. Fenchel M, Martirosian P, Langanke J, et al. Perfusion MR imaging with FAIR true FISP spin labeling in patients with and without renal artery stenosis: initial experience. Radiology. 2006;238:1013–1021. 33. Katoh M, Buecker A, Stuber M, Gunther RW, Spuentrup E. Freebreathing renal MR angiography with steady-state free-precession (SSFP) and slab-selective spin inversion: initial results. Kidney Int. 2004;66:1272–1278.

T. Leiner and H. Michaely 34. Katoh M, Spuentrup E, Stuber M, Hoogeveen R, Gunther RW, Buecker A. Free-breathing renal magnetic resonance angiography with steady-state free-precession and slab-selective spin inversion combined with radial k-space sampling and water-selective excitation. Magn Reson Med. 2005;53:1228–1233. 35. Wyttenbach R, Braghetti A, Wyss M, et al. Renal artery assessment with nonenhanced steady-state free precession versus contrastenhanced MR angiography. Radiology. 2007;245:186–195. 36. Coenegrachts KL, Hoogeveen RM, Vaninbroukx JA, et al. Highspatial-resolution 3D balanced turbo field-echo technique for MR angiography of the renal arteries: initial experience. Radiology. 2004;231:237–242. 37. Herborn CU, Watkins DM, Runge VM, Gendron JM, Montgomery ML, Naul LG. Renal arteries: comparison of steady-state free precession MR angiography and contrast-enhanced MR angiography. Radiology. 2006;239:263–268. 38. Maki JH, Wilson GJ, Eubank WB, Glickerman DJ, Millan JA, Hoogeveen RM. Navigator-gated MR angiography of the renal arteries: a potential screening tool for renal artery stenosis. AJR Am J Roentgenol. 2007;188:W540-546. 39. Maki JH, Wilson GJ, Eubank WB, Glickerman DJ, Pipavath S, Hoogeveen RM. Steady-state free precession MRA of the renal arteries: breath-hold and navigator-gated techniques vs. CE-MRA. J Magn Reson Imaging. 2007;26:966–973. 40. Glockner JF, Takahashi N, Kawashima A, et al. Non-contrast renal artery MRA using an inflow inversion recovery steady state free precession technique (Inhance): comparison with 3D contrastenhanced MRA. J Magn Reson Imaging. 2010;31:1411–1418. 41. Jun-ling X, Da-Peng S, Yong-Li L, Ji-liang Z, Shao-cheng Z, Hao S. Non-enhanced MR angiography of renal artery using inflowsensitive inversion recovery pulse sequence: A prospective comparison with enhanced CT angiography. Eur J Radiol. 80:e57–63. 42. Schoenberg SO, Rieger J, Weber CH, et al. High-spatial-resolution MR angiography of renal arteries with integrated parallel acquisitions: comparison with digital subtraction angiography and US. Radiology. 2005;235:687–698. 43. van Assen HC, Vasbinder GB, Stoel BC, Putter H, van Engelshoven JM, Reiber JH. Quantitative assessment of the morphology of renal arteries from X-ray images: quantitative vascular analysis. Invest Radiol. 2004;39:365–373. 44. Marckmann P. Nephrogenic systemic fibrosis: epidemiology update. Curr Opin Nephrol Hypertens. 2008;17:315–319. 45. Cowper SE, Rabach M, Girardi M. Clinical and histological findings in nephrogenic systemic fibrosis. Eur J Radiol. 2008;66: 191–199. 46. Idee JM, Port M, Medina C, et al. Possible involvement of gadolinium chelates in the pathophysiology of nephrogenic systemic fibrosis: A critical review. Toxicology. 2008;248:77–88. 47. Thomsen HS, Marckmann P. MRI contrast media are used to improve visualization of abnormal structures or lesions in various parts of the body. Introduction. Eur J Radiol. 2008;66:153–159. 48. European Medicines Agency. European Medicines Agency makes recommendations to minimise risk of nephrogenic systemic fibrosis with gadolinium-containing contrast agents. http://www.ema. europa.eu/docs/en_GB/document_library/Press_release/2009/11/ WC500015569.pdf. Accessed September 20, 2010. 49. United States Food and Drug Administration. FDA Drug Safety Communication: New warnings for using gadolinium-based contrast agents in patients with kidney dysfunction. http://www.fda.gov/ drugs/drugsafety/ucm223966.htm. Accessed September 20, 2010. 50. Schoenberg SO, Just A, Bock M, Knopp MV, Persson PB, Kirchheim HR. Noninvasive analysis of renal artery blood flow dynamics with MR cine phase-contrast flow measurements. Am J Physiol. 1997;272:H2477–2484.

21

Renal Vascular Diseases

51. Montet X, Ivancevic MK, Belenger J, et al. Noninvasive measurement of absolute renal perfusion by contrast medium-enhanced magnetic resonance imaging. Invest Radiol. 2003;38:584–592. 52. Schoenberg SO, Aumann S, Just A, et al. Quantification of renal perfusion abnormalities using an intravascular contrast agent (part 2): results in animals and humans with renal artery stenosis. Magn Reson Med. 2003;49:288–298. 53. Sourbron SP, Michaely HJ, Reiser MF, Schoenberg SO. MRImeasurement of perfusion and glomerular filtration in the human kidney with a separable compartment model. Invest Radiol. 2008; 43:40–48. 54. Vallee JP, Lazeyras F, Khan HG, Terrier F. Absolute renal blood flow quantification by dynamic MRI and Gd-DTPA. Eur Radiol. 2000;10:1245–1252. 55. Boss A, Martirosian P, Graf H, Claussen CD, Schlemmer HP, Schick F. High resolution MR perfusion imaging of the kidneys at 3 Tesla without administration of contrast media. Rofo. 2005;177:1625–1630. 56. Martirosian P, Klose U, Mader I, Schick F. FAIR true-FISP perfusion imaging of the kidneys. Magn Reson Med. 2004;51:353–361. 57. Lee VS, Rusinek H, Noz ME, Lee P, Raghavan M, Kramer EL. Dynamic three-dimensional MR renography for the measurement of single kidney function: initial experience. Radiology. 2003;227: 289–294. 58. Teh HS, Ang ES, Wong WC, et al. MR renography using a dynamic gradient-echo sequence and low-dose gadopentetate dimeglumine as an alternative to radionuclide renography. AJR Am J Roentgenol. 2003;181:441–450. 59. Baltes C, Kozerke S, Hansen MS, Pruessmann KP, Tsao J, Boesiger P. Accelerating cine phase-contrast flow measurements using k-t BLAST and k-t SENSE. Magn Reson Med. 2005;54:1430–1438. 60. Bock M, Schoenberg SO, Schad LR, Knopp MV, Essig M, van Kaick G. Interleaved gradient echo planar (IGEPI) and phase contrast CINE-PC flow measurements in the renal artery. J Magn Reson Imaging. 1998;8:889–895. 61. Michaely HJ, Schoenberg SO, Ittrich C, Dikow R, Bock M, Guenther M. Renal Disease: Value of Functional Magnetic Resonance Imaging With Flow and Perfusion Measurements. Invest Radiol. 2004;39:698–705. 62. Pettigrew RI, Avruch L, Dannels W, Coumans J, Bernardino ME. Fast-field-echo MR imaging with Gd-DTPA: physiologic evaluation of the kidney and liver. Radiology. 1986;160:561–563. 63. Michaely HJ, Sourbron SP, Buettner C, Lodemann KP, Reiser MF, Schoenberg SO. Temporal Constraints in Renal Perfusion Imaging With a 2-Compartment Model. Invest Radiol. 2008;43:120–128. 64. Michaely HJ, Schoenberg SO, Oesingmann N, et al. Renal artery stenosis: functional assessment with dynamic MR perfusion measurements – feasibility study. Radiology. 2006;238:586–596. 65. Gandy SJ, Sudarshan TA, Sheppard DG, Allan LC, McLeay TB, Houston JG. Dynamic MRI contrast enhancement of renal cortex: a functional assessment of renovascular disease in patients with renal artery stenosis. J Magn Reson Imaging. 2003;18:461–466. 66. Song T, Lee VS, Rusinek H, Wong S, Laine AF. Integrated four dimensional registration and segmentation of dynamic renal MR images. Med Image Comput Comput Assist Interv. 2006;9:758–765. 67. Huang AJ, Lee VS, Rusinek H. Functional renal MR imaging. Magn Reson Imaging Clin N Am. 2004;12:469–486, vi. 68. Carvlin MJ, Arger PH, Kundel HL, et al. Use of Gd-DTPA and fast gradient-echo and spin-echo MR imaging to demonstrate renal function in the rabbit. Radiology. 1989;170:705–711. 69. Grobner T. Gadolinium – a specific trigger for the development of nephrogenic fibrosing dermopathy and nephrogenic systemic fibrosis? Nephrol Dial Transplant. 2006;21:1104–1108.

295 70. Lee VS, Rusinek H, Johnson G, Rofsky NM, Krinsky GA, Weinreb JC. MR renography with low-dose gadopentetate dimeglumine: feasibility. Radiology. 2001;221:371–379. 71. Attenberger UI, Sourbron SP, Notohamiprodjo M, et al. MR-based semi-automated quantification of renal functional parameters with a two-compartment model – an interobserver analysis. Eur J Radiol. 2008;65:59–65. 72. Song T, Lee VS, Rusinek H, Kaur M, Laine AF. Automatic 4-D registration in dynamic MR renography based on over-complete dyadic wavelet and Fourier transforms. Med Image Comput Comput Assist Interv. 2005;8:205–213. 73. Rusinek H, Boykov Y, Kaur M, et al. Performance of an automated segmentation algorithm for 3D MR renography. Magn Reson Med. 2007;57:1159–1167. 74. Annet L, Hermoye L, Peeters F, Jamar F, Dehoux JP, Van Beers BE. Glomerular filtration rate: assessment with dynamic contrastenhanced MRI and a cortical-compartment model in the rabbit kidney. J Magn Reson Imaging. 2004;20:843–849. 75. Lee VS, Rusinek H, Bokacheva L, et al. Renal function measurements from MR renography and a simplified multicompartmental model. Am J Physiol Renal Physiol. 2007;292:F1548-1559. 76. Martin DR, Sharma P, Salman K, et al. Individual kidney blood flow measured with contrast-enhanced first-pass perfusion MR imaging. Radiology. 2008;246:241–248. 77. Michoux N, Montet X, Pechere A, et al. Parametric and quantitative analysis of MR renographic curves for assessing the functional behaviour of the kidney. Eur J Radiol. 2005;54:124–135. 78. Schoenberg SO, Knopp MV, Bock M, et al. Renal artery stenosis: grading of hemodynamic changes with cine phase-contrast MR blood flow measurements. Radiology. 1997;203:45–53. 79. Schoenberg SO, Knopp MV, Londy F, et al. Morphologic and functional magnetic resonance imaging of renal artery stenosis: a multireader tricenter study. J Am Soc Nephrol. 2002;13:158–169. 80. Ritt M, Janka R, Schneider MP, et al. Measurement of kidney perfusion by magnetic resonance imaging: comparison of MRI with arterial spin labeling to para-aminohippuric acid plasma clearance in male subjects with metabolic syndrome. Nephrol Dial Transplant. 2010;25:1126–1133. 81. Michaely HJ, Herrmann KA, Nael K, Oesingmann N, Reiser MF, Schoenberg SO. Functional renal imaging: nonvascular renal disease. Abdom Imaging. 2007;32:1–16. 82. Bax L, Woittiez AJ, Kouwenberg HJ, et al. Stent placement in patients with atherosclerotic renal artery stenosis and impaired renal function: a randomized trial. Ann Intern Med. 2009;150: 840–848. 83. Wheatley K, Ives N, Gray R, et al. Revascularization versus medical therapy for renal-artery stenosis. N Engl J Med. 2009;361: 1953–1962. 84. Murphy TP, Cooper CJ, Dworkin LD, et al. The Cardiovascular Outcomes with Renal Atherosclerotic Lesions (CORAL) study: rationale and methods. J Vasc Interv Radiol. 2005;16:1295–1300. 85. Krum H, Schlaich M, Whitbourn R, et al. Catheter-based renal sympathetic denervation for resistant hypertension: a multicentre safety and proof-of-principle cohort study. Lancet. 2009;373:1275–1281. 86. Mounier-Vehier C, Haulon S, Devos P, et al. Renal atrophy outcome after revascularization in fibromuscular dysplasia disease. J Endovasc Ther. 2002;9:605–613. 87. Mounier-Vehier C, Haulon S, Lions C, et al. Renal atrophy in atherosclerotic renovascular disease: gradual changes 6 months after successful angioplasty. J Endovasc Ther. 2002;9:863–872. 88. Michaely HJ, Dietrich O, Reiser MF, Schoenberg SO. Neue konzepte für die funktionelle MRT bei nierenerkrankungen. Der Nephrologe. 2006;1:40–49.

wwwwwwwwwwww

MRA: Upper Extremity and Hand Vessels Ruth P. Lim and Vivian S. Lee

Introduction

Vascular Anatomy

Catheter angiography is the acknowledged clinical gold standard for upper extremity arterial assessment. However, it is invasive and can be particularly challenging in the upper extremity, as iodinated contrast injection can initiate vasospasm and pain, particularly in young adults. MR evaluation of the upper extremity and hand vessels can provide a noninvasive comprehensive assessment, particularly since the advent of gadolinium-enhanced MRA (Gd MRA) in the early 1990s [1]. Indications for upper extremity MR angiography or venography are varied, including atherosclerosis, trauma, thromboembolic phenomena, and vasculitides. Benefits of MRA include lack of ionizing radiation, no requirement for iodinated contrast, and ability to acquire functional information including flow direction and velocity. Advances in magnet and coil technology have enabled continued improvements in spatial and/or temporal resolution. For successful imaging, a clear understanding of vascular anatomy, patient preparation, imaging protocols, and potential pathology is required. Some potential challenges that are faced when evaluating the arm are small caliber distal vessels, anatomic variants, relatively slow blood flow, and short arteriovenous transit times. This chapter begins with a review of vascular anatomy from the axilla to the fingertips. Scanning technique is discussed, including patient positioning and appropriate coil selection. Next, basic sequences used in MRA and MRV are reviewed in the context of upper extremity imaging, describing both contrast-enhanced and noncontrast-enhanced angiographic techniques, and strategies to optimize image quality. Pitfalls in image acquisition and interpretation are reviewed. Clinical indications for upper extremity and MRA and MRV are presented, and finally, areas under exploration are discussed.

Arterial Anatomy [2]

R.P. Lim, MBBS, MMed, FRANZCR () • V.S. Lee, MD, PhD, MBA Department of Radiology, New York University Langone Medical Center, New York, NY, USA e-mail: [email protected]

22

The axillary artery supplies the upper extremity, arising as the continuation of the subclavian artery at the lateral border of the first rib. The axillary artery gives off body wall and anterior and posterior circumflex humeral arteries, and then continues as the brachial artery at the inferior border of teres major. Branches in the arm include the profunda brachii artery, equivalent to the profunda femoris artery in the leg, continuous with the radial collateral artery, and ulnar collateral arteries that anastomose around the elbow with recurrent collateral vessels from the forearm. The brachial artery terminates by bifurcating into radial and ulnar arteries at the level of the neck of the radius. In the forearm, the ulnar artery is generally a larger caliber vessel than the radial artery, and gives rise to the common interosseous artery, which in turn divides into anterior and posterior interosseous arteries that supply the flexor and extensor compartments of the forearm. The ulnar and radial arteries give rise to palmar and dorsal carpal branches that anastomose around the wrist. Hand arterial anatomy can be arbitrarily divided by radial and ulnar artery supply. The radial artery typically runs volar to dorsal within the anatomical snuffbox at the level of the trapezium. It gives rise to the arteria radialis indicis, supplying the radial side of the index finger, and the princeps pollicis artery, which divides into two palmar digital branches to supply the thumb. The radial artery then terminates in the deep palmar arch where it anastomoses with the deep branch of the ulnar artery. The deep palmar arch gives rise to three palmar metacarpal arteries that anastomose with common palmar digital arteries from the superficial palmar arch. The ulnar artery continues in the hand as the superficial palmar arch, located approximately 1 cm distal to the deep palmar arch, at the same level as the distal border of the outstretched thumb. This is often incomplete, but if complete, anastomoses

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_22, © Springer Science+Business Media, LLC 2012

297

298

R.P. Lim and V.S. Lee

Fig. 22.1 Line drawing of the arterial anatomy of the upper limb, demonstrating major vessels of the arm and forearm (left), and hand arterial anatomy in detail (right). Arteries in the hand arising from the radial

artery and deep arch are depicted in burgundy, and those arising from the ulnar artery and superficial arch are shown in red

with the superficial palmar branch of the radial artery. Three common palmar digital arteries arise from the superficial arch, and one proper palmar digital artery that supplies the ulnar side of the little finger. The common palmar digital arteries subsequently divide into proper palmar digital arteries to the ulnar side of the index finger, the middle and ring fingers, and the radial side of the little finger. Conventional arterial anatomy of the upper extremity is depicted in Fig. 22.1. Arterial anatomic variants are not uncommon [3, 4]. These include a brachioradial artery, referring to a high origin of the radial artery from the axillary or proximal to mid brachial artery, found in up to 13.8% in one large cadaver series, and a much less common brachioulnar artery. Variants restricted to the forearm vessels are relatively rare, and include absence or duplications of the radial or ulnar arteries. Variations in the hand are very common, particularly of the superficial palmar arch, where nine variants have been described. A complete arch, where the superficial arch provides supply to all five digits, is seen in over two-thirds of

subjects. Less commonly, arterial supply to one or more digits is exclusively from the radial artery, or from a persistent median artery, located between the ulnar and radial arteries. The interested reader is referred to the work of Coleman and Anson [4].

Venous Anatomy [2] Venous drainage from the upper limb can be divided into superficial and deep venous systems, again paralleling lower limb vascular anatomy. The deep system predominantly drains the arm and forearm, and the deep veins usually run as dual venae comitantes with the major arteries. The superficial venous system provides most of the drainage of the hand in addition to the subcutaneous tissue of the upper limb. It begins distally as the palmar digital veins, draining into a dorsal venous plexus. In the forearm, the venous plexus drains into cephalic vein laterally and the basilic vein medially,

22

MRA: Upper Extremity and Hand Vessels

Fig. 22.2 Line drawing of the superficial venous anatomy of the upper limb. Deep veins (not shown) follow the arteries, generally as paired venae comitantes. The cephalic vein is the main superficial vein laterally, and the basilic vein provides superficial drainage medially. Anatomic variations in superficial venous drainage are common, particularly in the antecubital region

and variably, median forearm and median cubital veins runs between the two along the volar forearm. While the cephalic vein remains superficial until its termination in the axillary vein beyond the clavipectoral fascia, the basilic vein passes through the deep fascia in the mid arm, becoming the axillary vein at the lower margin of teres major. Superficial venous drainage is summarized in Fig. 22.2.

Scanning Patient Positioning Patient positioning and comfort are integral components of achieving high-quality MR images. For above-elbow imaging, patients can be imaged supine. Below the elbow, in order to minimize wrap artifact from adjacent body structures, it is desirable to isolate the arm or arms if a bilateral examination is required. If feasible for the patient, prone positioning with one or both arms outstretched above the head (Fig. 22.3) will minimize wrap [5]. In order to maintain patient comfort, the arms can be raised to shoulder level by propping them up with supporting padding. In elderly patients or patients with limited arm extension, a less rigorous alternative is to remain

299

in a supine position, with arms by their sides or resting in neutral position on the thigh. In either position, the arm of interest should be centered as close to the bore of the magnet as possible, where the magnetic field is most homogeneous. This might involve positioning the patient’s torso laterally within the bore when feasible. A final alternative, for smaller patients unable to lie prone is to lie in a lateral decubitus position with the arm above the head. One other consideration with positioning is to minimize rotation of the upper extremity, to minimize slice or partition number, and also to simplify image interpretation. For example, the forearm should ideally be positioned in full pronation or supination, and for the hand, digits should be extended and immobilized as much as possible. This may be achieved by taping the hand to a rigid support, such as an arm board as is used with peripheral intravenous cannulae. The patient should be consulted and made comfortable prior to commencement of scanning, which may include additional supportive padding, blankets, and oxygen for dyspnea. Keeping the patient warm is important to prevent vasoconstriction, particularly if the digital arteries are of interest. Patients should be warned against motion, particularly with respect to the target area.

Coil Selection Coil selection will depend on which area is of clinical interest and the patient position selected. Surface receive coils are desirable in order to maximize signal to noise ratio, as there is no requirement for signal reception from deep tissues with upper extremity imaging. For the hand, there are commercially available multichannel hand coils, however knee, head, or a surface-phased array coil placed over or wrapped around the hand(s) can be substituted in practice and yield diagnostic quality results. If large field of view coverage is desired, for example, from the axillary level to the fingertips, overlapping surface-phased array coils as are used for body and cardiac applications can be positioned over the area of interest and combined with spinal coils and the inherent body coil of the magnet as required for signal reception.

Key Sequences Localizer Bright blood localizer images are desirable, for example, balanced steady-state free precession (SSFP) images, for adequate image planning. More than one set of multiplanar localizing images may be required for upper extremity imaging, because of variability of patient positioning within the magnet and desired coverage.

300

R.P. Lim and V.S. Lee

Fig. 22.3 Examples of patient positioning and coils. (a) Patient lying prone with arm extended above head and hand imaged within a knee coil. Dedicated hand, elbow, or head coils may be substituted depending on availability and patient hand size. Smaller patients can also lie in a lateral decubitus position with contralateral arm down, if prone positioning is not possible. (b) Patient lying prone with both arms extended above the head, using a 6-element surface-phased array coil, helpful for

bilateral forearm and hand imaging, for example, in suspected emboli or vasculitis. (c) Patient lying supine with arm of interest placed as close to the center of the bore as possible using two 6-element surface phased array coils. This is useful for assessment of the arm, but can also be employed for forearm and hand assessment in frail patients where prone imaging is not possible

2D Fast Spin Echo Imaging

surrounding structures without the distraction of bright blood signal. T2-weighted FSE imaging is therefore a useful sequence when it is important to evaluate the vessel wall, as in vasculitis or atherosclerosis. Use of nonselective and selective 180° inversion prepulses is often used to ensure that blood is nulled in the slice of interest. This technique is often used in cardiac imaging where imaging during diastole is desirable to minimize cardiac motion but slow blood flow

Since spin echo imaging involves an initial 90° excitation pulse, followed by one (spin echo) or more [fast spin echo (FSE)] 180° refocusing pulses, blood that has moved through a slice will appear black, as it does not experience both of these radiofrequency pulses. This effect can be exploited in vascular imaging to enable assessment of the vessel wall and

22

MRA: Upper Extremity and Hand Vessels

makes blood nulling more difficult. Nulling of blood signal, or black-blood imaging, is dependent on blood flow exceeding a threshold velocity proportional to slice thickness and inversely proportional to half the echo time [6]. Blood moving slower than this velocity will not be nulled, as it will be exposed to both RF pulses, and therefore will appear bright. This phenomenon can be used to differentiate highflow from slow-flow vascular malformations or lymphatic malformations, as high-flow vascular malformations appear dark on T2-weighted FSE imaging, unlike the other two pathologies. However, a potential pitfall is that blood moving within the imaging plane will similarly appear bright, and this can lead to incorrect diagnosis of slow flow. As such, bright blood imaging is preferable for luminal assessment. As vessels are relatively small caliber in the upper extremity, multishot FSE is preferable to single-shot FSE imaging, providing higher signal to noise ratio which can be used to maximize spatial resolution. Fat suppression is desirable to increase conspicuity of pathologic findings.

Noncontrast-Enhanced MRA Sequences: Phase-Contrast and Time-of-Flight MRA Although Gd MRA has largely superseded traditional noncontrast-enhanced bright blood MRA techniques for the upper extremity, they can be considered when gadolinium is contra-indicated and can also provide limited functional information. Phase-contrast MRA is rarely used in clinical practice because of the inherent requirement for multiple gradients and because extremity MRA usually demands a large field of view, both of which contribute to impractical acquisition times. Time-of-flight MRA (TOF MRA) is briefly discussed in the context of the upper extremity. TOF MRA for upper extremity imaging should be considered when only relatively small coverage is desired, as the need to acquire images perpendicular to flow directly impacts imaging times. When used to image the hand, 2D TOF MRA is preferred, due to slow arterial flow, but oblique rather than perpendicular positioning will enable visualization of all vessels, as the palmar arches are perpendicular in orientation to the radial, ulnar, and digital vessels [7]. Strategies to maximize flow-related enhancement include: (a) lengthening repetition time, at the cost of increased imaging time and poorer background suppression; (b) using a lower flip angle, which worsens background suppression; (c) minimizing voxel size and slice thickness to combat intravoxel dephasing and through-plane saturation effects, at the cost of decreased SNR and longer acquisition; (d) minimizing TE; and (e) use of flow compensation gradients, which also minimize intravoxel dephasing but increase achievable echo times. These strategies can equally be applied to MR venography, where

301

flow is also slow. It can clearly be seen that the potential benefits and disadvantages must be carefully balanced and tailored to the target vascular bed. Newly described noncontrast techniques under development are discussed at the end of this chapter.

Gadolinium-Enhanced MRA Unlike the flow-related techniques described above, intravascular contrast is generated with a gadolinium chelate, an exogenous contrast agent that shortens the T1 relaxivity of blood when injected, thereby producing “bright blood” images on the first pass of the contrast agent. Crucial components to consider when performing Gd MRA include contrast dosage, timing and rate, and tailoring sequence parameters for desired spatial resolution and acquisition times.

Contrast Dosage Contrast dosage is weight-based, and single to double dose (0.1–0.2 mmol/kg) for standard extracellular gadolinium chelates is suitable for upper extremity MRA. If the entire upper extremity is desired, two-station MRA can be performed, in which case a two injection approach is generally preferable to a bolus chase approach, as this allows for differences in positioning of the arm versus the arm and forearm. Using the two injection approach, it is advisable to complete one station in its entirety before acquiring the “precontrast” mask images for the second station to allow for more accurate image subtraction, and elimination of venous contamination and bright background signal from the first injection. Using a smaller first and larger second contrast dose, for example, 0.08 mmol/kg followed by 0.12 mmol/kg, will minimize the effect of the first injection when reviewing source images from the second station. Gd MRA Timing Accurate timing of the postcontrast acquisition is essential for pure arterial phase imaging. The digital arteries are relatively slow to fill, and sufficient delay between injection and acquisition is required to ensure their opacification. Conversely, too delayed an acquisition will result in venous contamination, as arteriovenous transit times decrease moving proximal to distal within the upper extremity. For forearm and hand MRA, a blood-pressure cuff can be applied proximally and inflated to subsystolic pressure to retard venous filling. Appropriate timing can be approached a number of ways. An empiric time delay of 15 s has been reported to yield diagnostic results for the hand [8]. However, with differences in individuals’ circulation times, contrast dosages and injection rates, a bolus tracking or test bolus

302

R.P. Lim and V.S. Lee

approach is preferable. This can be centered on the brachial artery at the antecubital level. Using the test bolus approach, an accurate scan delay can be calculated by determining the

Scan delay = Time to peak enhancement +

For sustained arterial enhancement during image acquisition, it is desirable for duration of enhancement to approximate imaging time. This can be achieved with slower injection rates, of the order of 2 ml/s, followed by a saline chaser at the same injection rate. A final alternative is to use a time-resolved approach, with multiple measures. This obviates the need for estimation of appropriate timing, provided image acquisition commences before contrast has reached the imaging field of view. With sufficient temporal resolution, time-resolved imaging has the additional benefit of hemodynamic information, including speed of vascular enhancement.

Optimizing Gd MRA Parameters Sequence parameters need to be tailored to the upper extremity. There are competing demands for high spatial resolution, as digital arteries are of the order of 1 mm in diameter, and for relatively short acquisitions to minimize venous contamination. Ideally, voxel sizes in the 1.5 mm range for the arm and forearm, and 1 mm or less for the hand, are desirable. However, for the hand in particular, spatial resolution may be limited by signal to noise ratio, and time saving techniques described below will also affect SNR. Use of high-relaxivity contrast agents, such as gadobenate dimeglumine or gadofosvoset, and imaging at 3 T can mitigate this problem, to be discussed at the end of this chapter. One strategy that can shorten acquisitions is parallel imaging when multiple receiver coils are available in the phase encoding direction. Receiver coil data is used to provide spatial localization information, decreasing the number of required phase encoding lines, and facilitating large field of view rapid coverage. Image reconstruction techniques for parallel imaging include sensitivity encoding (SENSE) [10], simultaneous acquisition of spatial harmonics (SMASH) [11], and generalized autocalibrating partially parallel acquisitions (GRAPPA) [12]. With 3D Gd MRA, the slice or partition direction is essentially a second-phase encoding direction. The frequency encoding direction should be placed along the longest plane of the desired field of view, as this will not affect imaging time. For upper extremity MRA, an acquisition that is coronal to the outstretched hand and parallel to the plane of the radius and ulna will ensure the two shorter dimensions (transaxial to the upper extremity), fall along the slice and

time at which the midpoint of arterial enhancement will coincide with time at which the center of k-space (determining image contrast) is acquired [9]: 1 Injection time - Time tok - space center. 2

phase encode directions. Ideally, the slice direction should be along the shortest dimension, to maximize the use of rectangular field of view in the phase encode direction, at no cost to spatial resolution. Partial Fourier and/or zero interpolation can be employed in phase encode or slice directions, although this will decrease true spatial resolution of the acquisition. Centric encoding, where central lines of k-space are acquired early, before peripheral lines, can be employed to good effect by collecting central lines of k-space before contrast reaches the venous system. This allows more time for peripheral k-space to be filled after image contrast data has been collected. In this way, relatively long acquisition times of 30–40 s can provide high-resolution arterial phase hand MR angiograms [8, 13].

Time-Resolved Gd MRA Echo-sharing or keyhole imaging strategies can be used to further decrease imaging times and provide some hemodynamic information. The basic premise is that with multiple measures, the center of k-space is fully sampled with every acquisition; however, the periphery of k-space is undersampled, with uncollected data interpolated between consecutive measures. Various trajectories have been described in order to undersample k-space, including time-resolved imaging of contrast kinetics (TRICKS) [14] and time-resolved imaging with stochastic trajectories (TWIST), but all allow sub-10 s temporal resolution, which can be particularly important in the assessment of vascular malformations. As image acquisitions are faster, no more than single dose of contrast is required, and early arterial through to venous phase images can be acquired without a test bolus. A potential disadvantage of this approach is that the acquisition of multiple measures can generate a large amount of data, and reconstruction time can slow down the examination overall [5]. Also, Gibbs ringing artifact may be problematic with such techniques, where there are sharp transitions in MR signal between center and periphery of k-space acquisition, exacerbated by high contrast injection rates and aggressive k-space undersampling. As there is inevitably some loss of signal to noise ratio and subsequently spatial resolution with time-resolved MRA, it is most effective when hemodynamic information is desired, for example, in the evaluation of vascular malformations or fistulas.

22

MRA: Upper Extremity and Hand Vessels

303

Table 22.1 Overview of key sequences used in clinical vascular imaging Sequence 2D FSE

Description Blood moving through the imaging slice not experiencing both 90° (excitation) and 180° (refocusing) pulses or double inversion recovery prepulses null blood signal for black-blood images

TOF MRA

Multiple excitation pulses lead to saturation of stationary tissue, while fresh inflowing blood appears relatively bright

PC MRA

Dephasing and rephasing gradients cause phase shifts in moving protons in blood, resulting in a bright blood angiographic image Injection of gadolinium chelate causes rapid T1 relaxation of arterial blood on first-pass T1-weighted gradient echo imaging

Gd MRA

Time-resolved Gd MRA

Multiple Gd MRA measures are acquired utilizing image acceleration techniques including parallel imaging and keyhole imaging

Strengths Relatively rapid if a single shot (SSFSE) approach is employed Enables evaluation of vessel wall and surrounding structures Helpful in evaluating vascular malformations; flow voids will be seen with high-flow malformations Using saturation prepulses, direction of blood flow can be determined Bright blood luminal assessment without exogenous contrast Bright blood luminal assessment without exogenous contrast Excellent background suppression Independent of flow Rapid 3D volumetric coverage with good spatial resolution Multiple phase dynamic images can be obtained, including venous phase MRV Timing run not required Faster acquisitions allow for lower doses of contrast Provides hemodynamic information

Gadolinium-Enhanced MRV Although noncontrast methods of MR venography are available, particularly TOF MRA and, more recently, balanced SSFP imaging, gadolinium-enhanced 3D MRV provides a large field of view, rapid assessment. Gadolinium-enhanced MRV (Gd MRV) can be performed using a direct approach, where dilute gadolinium contrast is injected into the venous system of the extremity of interest distal to the area of concern. More commonly, indirect Gd MRV is performed, where image acquisition occurs after intravenously injected gadolinium has recirculated to opacify the entire venous system, usually approximately 3 min following injection. With the indirect approach, there are no restrictions to the intravenous access site, and any part of the venous system can be rapidly imaged. Image subtraction can highlight the venous system, particularly if arterial phase imaging is performed first. Similar to Gd MRA, a T1-weighted spoiled gradient echo sequence is used. Recirculation within the venous system necessarily dilutes the contrast, with less T1 relaxation of venous blood compared with a first-pass arterial examination. For this reason, double dose (0.2 mmol/kg) of contrast is also desirable for Gd MRV, provided renal function is normal.

Weaknesses Slow or in-plane blood flow may not be nulled, which may be incorrectly interpreted as intravascular thrombus 2D approach limits assessment to the acquisition plane

Must be acquired perpendicular to blood flow, impacting acquisition time Slow or in-plane flow causes saturation of blood which may lead to stenosis overestimation Relatively long TE increases intravoxel dephasing and susceptibility artifact Poor coverage: multidirectional gradients increase imaging time

Low risk of adverse reaction to gadolinium Risk of Nephrogenic Systemic Fibrosis in renal failure patients Susceptibility artifact can be problematic when metallic implants including vascular stents are present, particularly at higher field strength Decreasing acquisition may compromise spatial resolution Multiple measures and acceleration techniques require longer reconstruction times Large amount of data generated

Use of frequency selective fat saturation, and a lower flip angle can improve soft tissue contrast, valuable for Gd MRV. The strengths and potential disadvantages of T2-weighted dark blood FSE images, 2D TOF MRA, and Gd MRA are presented in Table 22.1, followed by suggested imaging parameters for key sequences in Table 22.2.

Pitfalls There are a number of potential pitfalls that may be encountered in upper extremity MRA. These include inaccurate timing, motion artifact, flow-related artifacts, pseudostenosis, vascular mimics, and nonvisualization of extraluminal pathology [15].

Inaccurate Timing As discussed above, accurate timing for upper extremity MRA can be challenging. Distal arteries will not be opacified with too early an acquisition, and Maki artifact may also arise, where larger structures such as the center of large

TR (ms) 4,000

20

3.3

3.3

Sequence FST2w FSE

2D TOF MRA (for neck)

Gd MRA

FST1wGRE

1.2

1.3

5

TE (ms) 70

12

25

40

Flip angle (º) 180

20 s

20–30 s Breath hold (<25 s) if central vessel assessment required 30 s

1.2 × 2.3 mm2 10 mm Slice thickness

Hand: £1 × 1 × 1 mm3 Arm: £1.5 × 1.5 × 1.5 mm3

Hand: £1 × 1 × 1 mm3 Arm: £1.5 × 1.5 × 1.5 mm3

Acquisition time 3 min

Desired voxel size 1 × 1 mm2 Hand 1.5 × 1.5 mm2 Arm 3 mm Slice thickness

Frequency selective fat suppression Perform at 3–5 min for good venous imaging

Other 2D fat-suppressed imaging to evaluate the speed of blood flow, extraluminal structures, and extent of deep involvement for vascular malformations Rapid low-resolution images to determine the direction of flow Run once with saturation prepulse above slices and once below Timing run to ensure optimal arterial imaging

490

430

180

2

2

2

Bandwidth (Hz/ Parallel imaging voxel) factor 130 2

Table 22.2 Sample imaging parameters for T2-weighted fat-suppressed 2D FSE sequence (FST2w FSE), 2D time-of-flight (TOF) MRA, 3D T1-weighted gradient echo contrast-enhanced MRA (Gd MRA) and a fat-suppressed T1-weighted 3D gradient echo sequence useful for venous imaging in addition to the standard MRA sequence (FST1wGRE)

304 R.P. Lim and V.S. Lee

22

MRA: Upper Extremity and Hand Vessels

caliber vessels are unenhanced if the center of k-space is collected prior to arrival of the contrast bolus. Venous contamination is problematic with too late an acquisition, and can make interpretation far more challenging, since deep venous system venae comitantes run with the arteries. As discussed above, a timing run or time-resolved imaging will ensure high-quality arterial phase images.

Motion Artifact Motion artifact is particularly problematic for hand MRA, as even slight digital motion will greatly affect visualization of small caliber digital arteries on subtraction images. As described previously, spending time on patient positioning and digital immobilization is vital to minimize motion during or between acquisitions.

Flow-Related Artifacts Flow-related artifacts can cause overestimation of stenosis on TOF MRA or Gd MRA, as turbulent flow leads to spin dephasing and subsequent signal loss. High spatial resolution will minimize voxel size and subsequently intravoxel dephasing. Minimizing TE will also minimize intravoxel dephasing. Conversely, for black-blood FSE sequences, slow or turbulent flow may lead to misdiagnosis of intravascular thrombus or even dissection due to unwanted rephasing of blood spins leading to bright intravascular signal. Correlation of black-blood imaging with bright blood imaging will enable differentiation of slow flow from true intraluminal pathology.

Pseudostenosis There are a number of causes of pseudostenosis. These include susceptibility from metallic clips, vascular stents, concentrated gadolinium, or mural calcification, which may cause T2 or T2* shortening and signal loss. Review of source images will alert the reader to the presence of potential causes of susceptibility. To avoid concentrated gadolinium causing susceptibility, contrast injection should be via the contralateral arm, or if access can only be obtained in the ipsilateral arm, consider contrast dilution. Pseudostenosis can also be caused by exclusion of a vessel from the field of view, which can be prevented by careful positioning of the imaging volume. Volume averaging is another potential culprit, which can be minimized by high spatial resolution imaging.

305

Vascular Mimics On both TOF MRA and Gd MRA, other tissues with short T1 will also appear bright. This includes fat, marrow, or subacute blood products such as methemoglobin. These can potentially mimic vascular lesions. Subtraction images and review of source precontrast images are helpful for differentiating vascular enhancement from an intrinsically T1 hyperintense structure or lesion. This is a recognized potential pitfall of TOF MRA, where reference to a precontrast T1-weighted sequence can prevent misdiagnosis.

Nonvisualization of Extraluminal Pathology Extraluminal pathology can potentially be missed if interpretation relies exclusively on subtracted images. This may include dissection, a thrombosed aneurysm, intramural hematoma, atherosclerotic plaque, or a mass causing extrinsic compression of a vessel. T2-weighted FSE imaging and source T1-weighted pre- and postcontrast imaging will enable diagnosis of extraluminal abnormalities. Potential pitfalls in vascular imaging are summarized in Table 22.3, with examples of common pitfalls provided in Fig. 22.4.

Clinical Indications for Upper Extremity MRA and MRV Arterial ischemia is the main indication for evaluating the upper extremity. This may be due to a number of etiologies, including atherosclerotic thromboembolic disease, trauma, vasospasm, vasculitis, or vascular malformations. Venous or arterial compression may cause vascular disturbance in the thoracic outlet syndrome. Upper extremity MRV is occasionally required to diagnose venous thrombosis. Vascular conditions of the upper extremity are discussed, with reference to potential contribution of MRA and MRV in clinical management.

Atherosclerotic Thromboembolic Disease Steno-Occlusive Disease (Fig. 22.5) Atherosclerotic plaque most commonly involves the proximal feeding arteries to the upper extremity, particularly the subclavian and axillary arteries. Clinical presentation of upper extremity chronic ischemia includes arm claudication, or less commonly, rest pain or evidence of tissue loss. Subclavian steal syndrome is a well-documented clinical syndrome where high-grade stenosis or occlusion of the subclavian artery proximal to the vertebral artery origin or

306

R.P. Lim and V.S. Lee

Table 22.3 Pitfalls in vascular MRA Pitfall Inaccurate timing

Motion artifact

Flow-related artifact

Pseudostenosis

Vascular mimics

Nonvisualization of extraluminal pathology

Description First-pass contrast-enhanced MRA depends on having peak arterial contrast coinciding with acquisition of center of k-space Too early: poor arterial opacification Too late: venous contamination Patient motion leading to image blurring

Turbulent flow can increase intravoxel dephasing leading to overestimation of stenosis on Gd MRA and TOF MRA Slow or in-plane flow on TOF MRA and dark blood FSE imaging leading to overestimation of stenosis or intraluminal plaque, thrombus, or dissection Apparent narrowing or occlusion of a vessel that is patent in reality Causes: 1. Susceptibility artifact from concentrated Gd, vascular clips or stents, calcification, etc. 2. Exclusion of vessel from imaging field of view Structures that have short T1 relaxation times can mimic a vascular abnormality (e.g., aneurysm) on source Gd MRA or TOF MRA Extraluminal pathology, such as dissection, plaque, intramural hematoma, or thrombosed aneurysm may be missed if subtraction images are only reviewed

brachiocephalic trunk leads to diversion of blood from the Circle of Willis to the involved arm, via retrograde through the vertebral artery. The left arm is much more commonly involved than the right. Neurologic symptoms in addition to arm symptoms may arise from “steal” of blood from the vertebrobasilar system, resulting in syncope, dizziness, or less commonly, posterior circulation stroke. Gd MRA can depict the primary stenosis or occlusion. The steal phenomenon can be confirmed by a number of means at MRI. TOF MRA in the neck with a saturation prepulse cranial to the imaging volume or phase-contrast MRI can demonstrate the absence of antegrade flow in the ipsilateral vertebral artery. Timeresolved imaging can also be used to confirm delayed filling of the subclavian artery distal to the site of disease. Less commonly, the brachial or infrabrachial arteries are affected. Patients with upper extremity atherosclerotic disease tend to be a decade younger at presentation (50–60 years of age), with a slight female predominance, whereas lower extremity peripheral arterial disease more commonly affects males. Diabetes is notably less common in upper extremity disease (40% compared with 83% in an upper

Solutions Timing run

Bolus tracking Time-resolved imaging Make patient comfortable Immobilize region of interest Minimize acquisition time Minimize voxel size and TE to minimize intravoxel dephasing For dark blood imaging, correlate with a bright blood sequence to evaluate the lumen Review source images for contrast-enhanced MRA for sources of susceptibility artifact Inject contrast via contralateral arm to avoid concentrated venous contrast causing susceptibility Careful image planning to ensure vessels are entirely within the field of view Review precontrast T1 weighted or subtraction Gd MRA images Review source pre- and postcontrast Gd MRA images Obtain T2-weighted FSE images for alternative tissue contrast and blood nulling Use a low flip angle T1-weighted fat-suppressed sequence for improved soft tissue contrast that will enable extraluminal assessment on bright blood imaging

extremity versus lower extremity bypass population) [16]. As upper extremity exertional pain can significantly impact quality of life, such patients benefit from intervention. High patency rates and low mortality of upper extremity surgical bypass have been reported, except in end-stage renal failure patients, where mortality is high. MRA provides a useful road map for preoperative planning, and can aid decisionmaking with regard to suitability for percutaneous intervention (angioplasty or stent placement).

Embolic Disease Emboli to the upper extremity are not uncommon, accounting for up to 20% of peripheral emboli [17]. The most common embolic source is cardiac thrombus, particularly in the setting of atrial fibrillation. Emboli from the subclavian artery in thoracic outlet disease, or aneurysms related to the superficial palmar arch, are other potential sources. Upper extremity embolic disease may manifest with pain, pallor, cyanosis, parasthesias, and in the acute setting, catheter angiography may be preferable if intraarterial thrombolysis is being considered. However, with a more indolent presentation, MRA

22

MRA: Upper Extremity and Hand Vessels

307

Fig. 22.4 Examples of potential pitfalls in upper extremity vascular MRA. (a) MIP subtraction MRA showing extensive venous contamination due to inaccurate timing of the acquisition. (b) Motion causes misregistration artifact on subtraction MIP image in this patient with limited scleroderma, with visible skin and bone edges (arrows).

(c) Susceptibility artifact caused by concentrated intravenous gadolinium on subtraction MIP Gd MRA image performed with arms up, leading to apparent disease of the subclavian artery (arrowhead) and vein (open arrow), both of which are shown to be patent on subsequent equilibrium phase thin MIP reconstructions imaged in the same position (d and e)

provides a comprehensive noninvasive assessment. Using a two-station approach, the forearm and hand can be evaluated with the first injection, and the patient can be repositioned for evaluation of the heart and central vessels for an embolic source, firstly with first-pass Gd MRA, and then with a lower flip angle fat-suppressed 3D sequence to evaluate the cardiac chambers.

Trauma Either blunt or penetrating trauma may result in vascular injury to the upper extremity. Potential complications of trauma include: ischemia secondary to thrombosis, stenosis or vasospasm; transection; hemorrhage; and false or true aneurysm formation. If there is acute vascular compromise,

308

R.P. Lim and V.S. Lee

Fig. 22.5 Steno-occlusive disease involving the upper limb. (a) 70-Yearold female with chronically occluded left subclavian, axillary, and proximal brachial arteries, with the development of corkscrew collateral thyrocervical trunk/suprascapular vessels (arrow) across the occlusion. (b) 79-Year-old female with proximal left subclavian high-grade stenosis/occlusion (open arrow) causing subclavian steal. Time-offlight images with saturation prepulse (c) below and (d) above the imaging plane demonstrate retrograde flow of the left vertebral artery flowing cranial to caudal (arrowhead), in the same direction as the internal jugular veins (open arrowheads)

MRA should not be performed as a first-line examination, however it is often helpful in the subacute setting, providing a 3D overview of the arterial system.

Hypothenar Hammer Syndrome One well-characterized disorder induced by repetitive blunt trauma to the carpus is hypothenar hammer syndrome. This is classically described in manual laborers from jackhammer use, but may occur in athletes or other individuals where the hypothenar eminence is repetitively subjected to blunt force. The terminal ulnar artery in the canal of Guyon, and its continuation, the superficial palmar arch, are vulnerable to injury where they lie in close proximity to the hook of the hamate. The ulnar artery becomes compressed between the hook of the hamate and an external force, leading to intimal injury and disruption of the internal elastic lamina. Thrombosis, distal microemboli, or aneurysm formation may develop (Fig. 22.6). Clinical presentations include Raynaud syndrome, rest pain, or digital ulcers as a result of digital microemboli,

Fig. 22.6 Hypothenar hammer syndrome. 36-Year-old tennis player with pulsatile mass and recent Raynaud’s phenomenon of the right hand. The pulsatile mass was marked with cutaneous Vitamin E capsules. (a) Subtraction MIP MRA demonstrates an aneurysm associated with the superficial arch where it arises from the ulnar artery (arrow). The ring finger radial proper digital artery is not visualized beyond the middle phalanx, suggestive of occlusion (thin arrow). (b) Sagittal T1-weighted image demonstrates close proximity of the aneurysm (arrow) to the hook of the hamate (arrowhead). (c) Axial fat-suppressed T2-weighted FSE and (d) T1-weighted postcontrast fat-suppressed gradient echo images demonstrate low signal within the pseudoaneurysm. Partial thrombosis of the aneurysm is responsible for incomplete enhancement of the aneurysm sac (open arrow) and may be a source of embolus for the ring finger digital artery occlusion

22

MRA: Upper Extremity and Hand Vessels

309

Fig. 22.7 16-Year-old male status postremote knife trauma with right medial arm mass (arrow). (a) Axial fat-suppressed T2-weighted imaging demonstrates a T2 hyperintense mass in close proximity to the brachial neurovascular bundle. (b–d) Progressive time-resolved

contrast-enhanced sagittal MIP MRA images demonstrating vascular nature of the mass compatible with a venous varix, which fills via a fistulous connection (open arrow), with early appearance of the brachial vein at this level (arrowheads)

or a pulsatile palmar mass secondary to aneurysm formation. Typically, the fourth and fifth digits are involved, but any digit may be involved in patients with a complete superficial arch. MRA can identify the presence of stenosis, occlusion, or aneurysm formation at the site of trauma, and presence of distal disease. T2-weighted FSE imaging is helpful to identify an aneurysm, where there is turbulent or slow flow if the aneurysm is patent, or thrombus if occluded. Gd MRA should be performed with a small field of view restricted to the hand, and careful positioning and immobilization for accurate evaluation of the digital arteries. In this setting, use of a highrelaxivity contrast agent or imaging at 3 T should be considered to enable greater spatial resolution imaging while maintaining signal to noise ratio. Management of hypothenar hammer syndrome ranges from conservative to surgical [18]. Conservative measures include cessation of the offending activity, smoking cessation, and use of calcium channel blockers for vasodilation. Surgical measures may include ligation of the ulnar artery to prevent further emboli, aneurysm resection, or segmental resection and interposition grafting, if there is inadequate collateral flow between ulnar and radial circulations. Cervical sympathectomy may be performed when revascularization is not possible.

Arteriovenous Fistula Traumatic arteriovenous fistulas (AVF) may occur in the upper extremity, particularly in the setting of penetrating trauma from knife or shotgun injury. Iatrogenic AVF of the upper extremity has also been reported. In one large series of traumatic AVF, 22% affected the upper limb [19]. Clinical features include a machinery murmur at the fistula site and venous engorgement. Rarely, distal ischemia from “steal” of blood into the venous system may occur. MRA is helpful for diagnosis of occult cases of AVF not clinically apparent or suspected at the time of injury. In this setting, time-resolved Gd MRA is helpful to demonstrate early venous enhancement before opacification of more distal arteries or veins. Engorged venous channels can be fully assessed if present. The fistula site can be evaluated from various planes for effective presurgical planning (Fig. 22.7).

Raynaud’s Phenomenon Raynaud’s phenomenon, an abnormal vasospastic response of the digital arteries to cold or emotional stress, is common, affecting an estimated 2.5–21% of the general population,

310

R.P. Lim and V.S. Lee

Fig. 22.8 53-Year-old male with a history of Wegener’s granulomatosis, presenting with ischemia of the right thumb, index, and middle fingers. (a–d) Time-resolved MRA of forearm and hand demonstrate delayed filling of the radial artery (arrow) which is occluded at the

wrist. The ulnar artery and superficial palmar arch provide supply to the radial side of the hand and the deep palmar arch (arrowhead), however, the digital arteries of the index and radial side of the middle fingers are not seen, and a princeps pollicis artery to the thumb is not identified

depending on geographic location. It classically presents with pallor followed by cyanosis and erythema of the affected fingers. In most cases, it is an idiopathic functional condition with a benign clinical course, known as primary Raynaud’s phenomenon. However, it may also be an occupationally induced phenomenon, drug-related, or the heralding sign of an underlying connective tissue disease, and it is especially prevalent in scleroderma. If an underlying etiology is found, then it is characterized as secondary Raynaud’s phenomenon. With regard to imaging of Raynaud’s phenomenon, the act of catheter angiography itself can induce vasospasm, making exclusion of a structural abnormality difficult, even with the use of vasodilators. MRA can therefore be very helpful in noninvasively evaluating for such structural abnormalities as emboli or stenoses. There are no definitive, specific findings of primary Raynaud’s phenomenon on angiography, although tapering of the proper digital arteries and fingertip capillary congestion have been described [8]. Physiologic stress testing to differentiate between a functional and structural abnormality has been described to differentiate primary from secondary Raynaud’s phenomenon, and noncontrast-enhanced hand MRA techniques offer the promise of a combined anatomic and functional assessment [20].

Vasculitides Vasculitis, where there is inflammation involving the blood vessels, can be classified by size of the vessels involved. Examples of large vessel vasculitis include Takayasu arteritis and giant cell arteritis, and examples of medium to small vessel vasculitis include polyarteritis nodosa (PAN) and Wegener’s granulomatosis (Fig. 22.8). Vasculitis is also a feature of the connective tissue disorders, including systemic lupus erythematosus, scleroderma, and rheumatoid arthritis. The upper limb is less commonly involved in vasculitis. However, it may be indirectly involved in Takayasu’s arteritis, which has a predilection for the great vessels, for which imaging of the large central arteries is important. Vasculitis may directly involve the arteries of the upper limb in PAN or scleroderma, and patients may present with distal ischemia and digital ulcers, particularly in the latter condition. Endothelial damage from vasculitis leads to intimal hyperplasia, and ultimately obliterative endarteritis. This affects the digital arteries in both scleroderma and PAN. Imaging features include tapering, irregular segmental narrowing, or frank occlusion of the digital arteries. High-resolution MRA targeting the hand is required for accurate evaluation.

22

MRA: Upper Extremity and Hand Vessels

One particular challenge in the connective tissue disease population, particularly in scleroderma and rheumatoid arthritis, is joint and skin contractures that may preclude standard positioning, however, supine imaging can provide diagnostic images as previously discussed. Buerger disease is a debilitating condition that is associated with tobacco use, which is thought to trigger an autoimmune destructive vasculitic response. It is also known as thromboangiitis obliterans and has a predilection for the distal extremities, both upper and lower. Both arteries and veins can be affected. Digital ulcers and gangrene may develop. Interestingly, other there is no visceral organ involvement in this condition. Imaging features include distal stenoses or occlusions, with relative preservation of the inflow vessels [5]. Corkscrew collateral vessels are also an imaging feature, although this is a nonspecific finding that is seen in other small vessel diseases. Cessation of smoking is imperative to prevent further progression of the disease.

Vascular Malformations MRA is particularly helpful in the evaluation of vascular malformations, as these often present in the pediatric to young adult population, where the lack of ionizing radiation with magnetic resonance imaging is desirable. Vascular malformations are considered congenital lesions that persist beyond infancy, unlike infantile hemangiomas that classically regress [21]. Vascular malformations have a prevalence of approximately 1.5% of the population and can be divided into venous, capillary, arteriovenous, or mixed malformations, with helpful imaging features in most cases. Lymphatic malformations will also be discussed, as imaging is often required to differentiate between lymphatic and vascular malformations. Venous and lymphatic malformations are important to identify as diagnosis may obviate the need for catheter angiography. For simplicity, vascular malformations are often simply divided into low- versus high-flow malformations, as this distinction is important for subsequent management. Vascular malformations are seen in a number of syndromes, including Klippel–Trenaunay syndrome, characterized by capillary malformations and superficial varicose veins, and Parkes–Weber syndrome, where AVF and lymphatic malformations are present [22]. In addition to Gd MRA, T1- and T2-weighted FSE imaging is helpful in making the appropriate diagnosis, and also in evaluating the extent of involvement of superficial versus deep (muscular or osseous) structures. Venous and lymphatic malformations will tend to appear intermediate on T1-weighted imaging, and strikingly hyperintense on T2-weighted imaging, with fluid–fluid levels often seen secondary to proteinaceous or hemorrhagic content. Venous or slow-flow malformations can be distinguished from

311

lymphatic malformations by progressive enhancement of serpiginous venous channels; such enhancement will not be apparent in lymphatic malformations. Also, phleboliths may be associated with low-flow venous or capillary malformations, appearing variable in signal intensity on T1-weighted images, and low in signal intensity on long TE imaging. Capillary malformations are superficial small vessel ectatic lesions that are not always apparent at imaging and better diagnosed clinically. However, in mixed malformations, capillary malformations can be associated with deeper vascular malformations, and therefore imaging may be indicated to exclude a co-existing deep malformation. Lastly, arteriovenous malformations have an arterial component, with abnormal communication between arterial and venous structures, bypassing the capillary bed. Arteriovenous malformations occur less commonly than venous malformations, which account for up to two-thirds of extremity vascular malformations [21]. By definition, arteriovenous malformations have a nidus, a tangled core of abnormal vessels with feeding arteries and draining veins, distinguishing them from AVF (Fig. 22.9). Arteriovenous malformations can cause high-output cardiac failure, limb overgrowth, thromboembolism, hemorrhage secondary to rupture, and, because there is bypass of distal tissue, patients can also present with ischemic symptoms. By nature of their high flow, arteriovenous malformations will cause flow voids on T2-weighted FSE imaging, distinguishing them from low flow or lymphatic malformations. Gd MRA will demonstrate morphology as well as rapid enhancement of both feeding arteries, nidus, and draining veins. Time-resolved imaging has been proven to be valuable in demonstrating early enhancement of arteriovenous malformations during the arterial phase, and distinguishing them from venous malformations [23]. Features of venous malformations include dilated venous spaces, lack of flow voids, and late lesion enhancement occurring greater than 6 s after the arterial phase. Gd MRA can also evaluate such potential complications as thrombus, aneurysm, or varix formation, as well as define relationships of the malformation with adjacent bone and soft tissue.

Thoracic Outlet Syndrome Thoracic outlet syndrome refers to compression of the neurovascular bundle as it passes through the thoracic outlet. Neurogenic thoracic outlet syndrome accounts for the majority of presentations. However in less than 5% of cases, venous or more rarely arterial compression can cause clinical symptoms related to venous congestion or arterial ischemia, respectively. Venous compression can cause thrombosis, venous engorgement, and discomfort, and arterial compression may lead to thromboembolism or stenosis, with clinical

312

R.P. Lim and V.S. Lee

Fig. 22.9 52-Year-old male with left arm arteriovenous malformation. (a) Arterial phase imaging demonstrates markedly dilated and tortuous left subclavian and axillary arteries (arrow), as well as development of a thyrocervical trunk collateral vessel (arrowhead). (b) Early venous phase imaging demonstrating tangle of vessels in the axillary region compatible with nidus of the AVM (circle), and multiple deep and superficial draining veins (open arrowheads). (c) 3D Volume-rendered image which can be reconstructed in multiple planes, helpful for prein-

tervention planning. (d) Fat-suppressed coronal T2-weighted FSE image demonstrating the effect of slow or in-plane flow on blood nulling, with bright signal in the left subclavian artery beyond an acute kink in the vessel (thin arrow). (e) Fat-suppressed low flip angle 3D T1-weighted gradient echo image, with good visualization of nonocclusive thrombus within the aneurysmal axillary artery (open arrow) and vessel wall, due to improved soft tissue contrast compared with higher flip angle standard MRA/MRV images

features including pain, pallor, and pulselessness. There are three potential sites of extrinsic compression at the thoracic outlet moving from medial to lateral: the interscalene triangle, bounded by scalenus anterior, scalenus medius, and the first rib, and containing the subclavian artery and the trunks of the brachial plexus; the costoclavicular space between the medial half of the clavicle and the first rib, containing the subclavian artery, subclavian vein, and the cords of the brachial plexus; and, less commonly, the subpectoral space, deep to the pectoralis minor muscle below the coracoid process, through which the axillary artery, axillary vein, and

cords of the brachial plexus run (Fig. 22.10). Abduction and external rotation of the arm at the shoulder can exacerbate symptoms by narrowing these spaces. Arterial compression most commonly occurs in the costoclavicular space, and potential causes of narrowing of this space include a cervical rib, malunited clavicular fracture, fibromuscular bands, or repetitive movements. If a structural cause is found, surgical decompression is advocated for definitive management. MR imaging can aid both anatomic and functional assessment. If thoracic outlet syndrome is suspected, patients should be positioned supine with arms above the head in a

22

MRA: Upper Extremity and Hand Vessels

Fig. 22.10 Potential sites of narrowing in the thoracic outlet syndrome on sagittal T1-weighted FSE imaging from medial to lateral. (a) Interscalene triangle, bounded by scalenus anterior (a) anteriorly, and scalenus medius (m) and scalenus posterior posteriorly. It contains the subclavian artery (arrow) and trunks of the brachial plexus (arrowhead). (b) Costoclavicular space, bounded by the medial clavicle

313

(c) and subclavius muscle (s) and first rib (open arrowhead), containing the subclavian artery (arrow), subclavian vein (thin arrow), and cords of the brachial plexus (arrowhead). (c) Subpectoral space, deep to pectoralis minor (p) and overlying serratus anterior (sa), containing the axillary artery (arrow), axillary vein (thin arrow), and cords of the brachial plexus (arrowhead)

Venous Thrombosis

Fig. 22.11 Two different patients with vascular thoracic outlet syndrome. (a) Arms up and (b) arms down volume-rendered images of a 31-year-old female with right arm parasthesia. Right subclavian artery stenosis (arrow) is apparent when patient performs the Wright maneuver, with alleviation of compression when arms are in neutral position. (c) Arms up and (d) arms down thin MIP venous phase images showing reversible compression of the left subclavian vein (open arrow) in this 45-year-old male with left arm pain

Wright maneuver. Single dose Gd MRA in this position can evaluate for the presence of stenosis or occlusion and its complications. If positive, repeat Gd MRA with the patient’s arms in neutral position can evaluate the degree to which positioning exacerbates narrowing (Fig. 22.11). Sagittally oriented T1 weighted imaging is helpful for evaluating the extent of compression of the subclavian artery in the costoclavicular or interscalene spaces.

Although less common than lower extremity deep venous thrombosis (DVT), upper extremity venous thrombosis is another potential clinical application. Subclavian, axillary, and brachial vein thrombosis is included in the definition of upper extremity DVT. It accounts for an estimated less than 10% of all DVTs, perhaps due to higher upper limb blood flow rates, less stasis, and relatively lesser effects of gravity in comparison to the lower limb [24]. However, it has important implications including risk of pulmonary embolism, seen in 30% of patients, loss of vascular access, and postthrombotic syndrome, manifest by arm swelling or pain that is particularly debilitating in the dominant arm [25]. Upper limb DVT may be classified as primary DVT, including idiopathic or exertionally induced DVT, as in the Paget–Schroetter syndrome. In 70% of cases of upper limb DVT, it is secondary to factors including interventions, including central venous catheter insertion or surgery, or hypercoagulable states including immobilization, malignancy, or genetic predisposition, as in factor V Leiden or prothrombin 2010A mutations. Management of upper extremity DVT is not standardized, but usually consists of anticoagulation. If a central venous catheter is present, removal or replacement at a different site is usually warranted. More aggressive therapy may include catheter-directed thrombolysis, thrombectomy, or even angioplasty or stent/SVC filter placement. There are no randomized control trials demonstrating added benefit of thrombolysis over anticoagulation in the upper extremity. As described previously, venous imaging is easily obtained 3–5 min following arterial imaging without requirement for additional contrast. Low flip angle T1-weighted spoiled gradient echo imaging provides good contrast between vessels and adjacent soft tissue that is helpful for venous assessment. As the deep forearm veins are small in caliber, examination should be focused on the arm and central veins, and supine

314

R.P. Lim and V.S. Lee

Table 22.4 Clinical indications for upper extremity MRA and MRV Clinical indication Subclavian steal syndrome

Upper extremity emboli Hypothenar hammer syndrome

Arteriovenous fistula Raynaud’s Phenomenon Vasculitis

Vascular malformations

Thoracic outlet syndrome

Venous thrombosis

Imaging suggestions Evaluate arch and great vessels Evaluate flow direction in vertebral arteries with TOF MRA with a saturation prepulse, phase-contrast MRI, or time-resolved MRA Use a two-station protocol to evaluate proximal vessels from the aortic arch to the fingertips. Begin with distal station where venous contamination is least desirable Evaluate the hand using a small field of view High-resolution imaging for full evaluation of the arteries of the hand, including the proper digital arteries to evaluate for distal emboli Include high-resolution fat-suppressed T2-weighted FSE images to evaluate for aneurysm Use time-resolved MRA with good temporal resolution to identify early venous filling Evaluate the hand using a small field of view for high resolution Evaluate proximal vessels for large vessel arteritis (e.g., Takayasu arteritis or giant cell arteritis) Focus on the hand for small vessel vasculitis (e.g., polyarteritis nodosa, scleroderma, or Buerger disease) Use T1-weighted and fat-suppressed T2 FSE images for evaluation of extent and differentiation of high from low flow vascular malformations Time-resolved imaging enables characterization of arterial inflow, nidus, and venous drainage Perform Gd MRA with arms above the head; if subclavian artery and vein are normal in this position, vascular thoracic outlet syndrome can be excluded. If abnormal, repeat with arms in neutral position to evaluate for fixed stenosis Sagittal T1-weighted FSE imaging through the thoracic outlet to evaluate for interscalene or costoclavicular space compression Obtain venous imaging at 3–5 min postinjection Low flip angle T1-weighted spoiled gradient echo imaging provides good soft tissue contrast for venous assessment

imaging with surface-phased array coils provides a good assessment of the brachial veins as far proximally as the superior vena cava and right atrium, which is particularly helpful in patients with central venous catheters who are at increased risk of upper extremity DVT.

Paget–Schroetter Syndrome Paget–Schroetter syndrome can be considered a specific outcome of venous thoracic outlet syndrome. Also known as effort-induced venous thrombosis, it is often found in young, athletic adults who perform repetitive shoulder movements involving hyperabduction and external rotation at the shoulder, such as weightlifting. It is defined as venous thrombosis of the subclavian or axillary veins as a result of venous compression in the costoclavicular space. Up to 75% of cases can be linked to a recent history of strenuous exertion. As the underlying cause is usually venous compression from thoracic outlet syndrome, imaging in both arm extension and external rotation, followed by in the neutral position is useful to demonstrate the culprit site. Gd MRA and later phase MRV should be performed in both positions to make the diagnosis and exclude any arterial disease. T1-weighted FSE images again may be helpful in defining a structural cause or pinpointing the anatomic space involved. Other potential causes of arm engorgement such as an axillary mass or lymphedema can be evaluated with postcontrast images at a

delay of approximately 3 min using a low flip angle 3D fat-suppressed T1-weighted sequence. Excellent venous opacification will also be achieved simultaneously. Management of Paget–Schroetter syndrome is a twostage process. First, the acute venous thrombosis must be managed. Second, the underlying cause must be addressed to prevent future events, with surgical decompression, for example, with rib or clavicle resection, required. Previous descriptions of percutaneous intervention demonstrate high failure rates, with complications of stent fracture and recurrent stenosis and thrombosis. Clinical indications for upper extremity vascular imaging are summarized in Table 22.4, including suggestions for image acquisition.

Future Directions There have been astounding advances in MRI, including improvements in magnet design, RF transmit and receive technology, image acceleration and reconstruction techniques, which have been applied to MR angiography. These have led to advances in image quality, with submillimeter spatial resolution or subsecond temporal resolution now within the bounds of reality. Simultaneous to hardware and software developments, there has been exploration new

22

MRA: Upper Extremity and Hand Vessels

contrast alternatives, to some extent spurred on by the recognition of gadolinium as a causative agent in the development of Nephrogenic Systemic Fibrosis in patients with moderate to severe renal failure [26]. Some areas of interest relative to upper extremity MR angiography include: use of higher T1 relaxivity gadolinium agents and gadolinium alternatives, image acceleration (compressed sensing, HYPR VIPR), high-field imaging, and novel noncontrast-enhanced MRA techniques. Each of these areas will be briefly discussed in turn.

Contrast Agents A number of gadolinium contrast agents are available. Gadobenate dimeglumine (Gd-BOPTA) is an extracellular agent with a linear ionic structure that has higher T1 relaxivity than other standard gadolinium contrast agents. It also weakly binds to protein and therefore this can translate into brighter intravascular signal for the same volume as standard contrast agents on T1-weighted imaging, or smaller doses for the same amount of vascular to soft tissue contrast. One drawback of extracellular agents is that they do not remain within the vascular pool, and therefore cannot be used for steady-state imaging. Blood pool contrast agents address this issue by remaining in the circulation for longer periods of time. Blood pool contrast agents may be gadolinium-based, but other exogenous agents, particularly iron oxide-based agents have been described. These allow for steady state in addition to first-pass dynamic imaging, which theoretically should improve venous evaluation. It also has exciting implications for functional assessment of viscera and neoplasms, including perfusion. An example of a gadolinium-based blood pool agent is gadofosvoset trisodium, which has recently been FDA-approved and has been used in clinical practice outside of the USA. It has a T1 relaxivity up to ten times that of gadopentetate dimeglumine, a standard extracellular contrast agent at 0.5 T, and therefore much lower doses (0.03 mmol/kg) can be used. It can potentially improve visualization of small vessels such as the digital arteries. Ferumoxytol is an example of an ultrasmall superparamagnetic iron oxide (USPIO)-based blood pool agent. As it causes marked T2 shortening in addition to T1 shortening, dual contrast imaging can be obtained or complete vascular evaluation, whereby blood will appear bright on T1-weighted imaging and dark on T2-weighted imaging. Li et al. demonstrated potential application of the dual contrast mechanism for DVT assessment [27]. Another potential application of USPIOs for vascular contrast is in the renal failure population. For example, venous mapping or hemodialysis fistula assessment could be safely performed without concern for Nephrogenic Systemic Fibrosis.

315

Image Acceleration In addition to parallel imaging, other advanced image acceleration techniques have been developed that could greatly impact clinical MR angiography. One such technique is vastly undersampled imaging with projections (VIPR), which uses a radial acquisition to undersample k-space by factors as great as 100 times or more. When combined with highly constrained back projection (HYPR), the loss in SNR from undersampling can be enhanced by reducing the g-factor associated with aggressive use of VIPR, allowing for marked improvements in spatial and temporal resolution [28]. Another example of advanced image acceleration is compressed sensing, whereby image sets which have relatively sparse data, of which MRA is a good example, can be markedly undersampled in random fashion. Images can then be recovered using a nonlinear scheme with good fidelity [29]. Clinical evaluation and validation are still required for these techniques, however, they offer exciting potential for imaging the digital vessels, for example.

High-Field Imaging Imaging at high magnetic field strength of 3 T or greater can be challenging, due to technical factors including increased magnetic susceptibility, B1 field inhomogeneity, increased radiofrequency tissue deposition, and increased chemical shift artifact. Patients with metallic implants including vascular stents or orthopedic hardware should be identified, and examination at lower field strength may optimize image quality in this population. However, high-field imaging has theoretical benefits for MRA of increased signal to noise ratio and longer soft tissue T1 relaxation times result in increased contrast between enhancing and nonenhancing structures for Gd MRA, and between saturated and unsaturated spins for TOF imaging. These benefits can be translated into higher spatial resolution imaging and improved vessel contrast. Wang et al. reported initial experience with high-resolution TOF MRA of the digits in a series comparing healthy controls with patients with systemic sclerosis [30]. Using a custom-made finger receiver coil, 16 subjects were imaged with in-plane spatial resolution of 0.16 × 0.22 mm2, enabling quantitative evaluation of digital artery caliber. Improved vessel contrast and spatial resolution can be envisaged at increasing field strength, as has been demonstrated for intracranial TOF MRA at 7 T [31].

Developing Noncontrast-Enhanced Techniques Concerns regarding Nephrogenic Systemic Fibrosis have led to increased interest in contrast-free imaging alternatives

316

R.P. Lim and V.S. Lee

diastolic images to yield a bright blood angiogram. This was used to image a small group of Raynaud’s Phenomenon patients with impressive results in comparison to timeresolved Gd MRA [34]. Further validation of the technology and techniques described above will determine their place in clinical upper extremity imaging. Acknowledgments We wish to thank Martha Helmers for her expert illustration and assistance with preparation of the figures.

References

Fig. 22.12 ECG-triggered noncontrast-enhanced MRA imaging of the hand. (a) Variable flip angle 3D FSE MRA image of the left hand of a healthy volunteer at 3 T. Note good visualization of the proper digital arteries to the distal phalanges (arrow). (b) Flow-sensitized dephasing prepared SSFP MRA at 1.5 T of a 52-year-old female with digital ulcers of the right fourth and fifth fingers. There is good visualization of the palmar vessels (arrowheads), superior to contrast-enhanced MRA (c), where inaccurate timing results in venous phase imaging. Images (b) and (c) courtesy of Drs Zhaoyang Fan, Debiao Li, James Carr, and John Sheehan, Northwestern Memorial Hospital, Chicago

(Fig. 22.12). Several approaches have been applied to the upper extremities, focused on hand vascular imaging in particular. An FSE-based technique, known as fresh blood imaging (FBI), has been described by Miyazaki et al., exploiting differences in fast systolic and slow diastolic arterial flow [32]. Signal dephasing occurs in systole while bright arterial signal is obtained in diastole, with subtraction of the images yielding a bright blood MRA. A modified variable flip angle FSE MRA approach has been applied to hand imaging, allowing for greater flow sensitivity for slow arterial digital flow [20]. This was combined with a functional temperature challenge that depicted vessel reactivity in healthy subjects and patients with systemic sclerosis at 3 T. Another promising subtraction technique, based on the FBI approach combined with arterial spin labeling, uses an additional time-spatial labeling inversion pulse to selectively label arteries upstream of the region of interest, with improved digital artery visualization compared with 2D TOF MRA [33]. Finally, a balanced SSFP technique, flow-sensitized dephasing prepared balanced SSFP MRA (FSD SSFP MRA), has been described that uses gradient pulses to dephase fastflowing blood which are only played out in systole, but not in diastole. Again, systolic phase images are subtracted from

1. Prince MR, Yucel EK, Kaufman JA, Harrison DC, Geller SC. Dynamic gadolinium-enhanced three-dimensional abdominal MR arteriography. J Magn Reson Imaging. 1993;3:877–881. 2. Sinnatamby CS. Upper limb. In: Last’s Anatomy: Regional and Applied. London: Churchill Livingstone, 1999; 35–106. 3. Rodriguez-Niedenfuhr M, Vazquez T, Nearn L, Ferreira B, Parkin I, Sanudo JR. Variations of the arterial pattern in the upper limb revisited: a morphological and statistical study, with a review of the literature. J Anat. 2001;199(Pt 5):547–566. 4. Coleman SS, Anson BJ. Arterial patterns in the hand based upon a study of 650 specimens. Surg Gynecol Obstet. 1961;113:409–424. 5. Stepansky F, Hecht EM, Rivera R, et al. Dynamic MR angiography of upper extremity vascular disease: pictorial review. Radiographics. 2008;28(1):e28. 6. Lee VS. Cardiovascular MRI: Physical Principles to Practical Protocols. Philadelphia: Lippincott Williams and Wilkins; 2006. 7. Rofsky NM. MR angiography of the hand and wrist. Magn Reson Imaging Clin N Am. 1995;3:345–359. 8. Connell DA, Koulouris G, Thorn DA, Potter HG. Contrast-enhanced MR angiography of the hand. Radiographics. 2002;22:583–599. 9. Earls JP, Rofsky NM, DeCorato DR, Krinsky GA, Weinreb JC. Breath-hold single-dose gadolinium-enhanced three-dimensional MR aortography: usefulness of a timing examination and MR power injector. Radiology. 1996;201:705–710. 10. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999;42:952–962. 11. Sodickson DK, Manning WJ. Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med. 1997;38:591–603. 12. Griswold MA, Jakob PM, Heidemann RM, et al. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med. 2002;47:1202–1210. 13. Brauck K, Maderwald S, Vogt FM, Zenge M, Barkhausen J, Herborn CU. Time-resolved contrast-enhanced magnetic resonance angiography of the hand with parallel imaging and view sharing: initial experience. Eur Radiol. 2007;17:183–192. 14. Korosec FR, Frayne R, Grist TM, Mistretta CA. Time-resolved contrast-enhanced 3D MR angiography. Magn Reson Med. 1996;36: 345–351. 15. Lee VS, Martin DJ, Krinsky GA, Rofsky NM. Gadoliniumenhanced MR angiography: artifacts and pitfalls. AJR Am J Roentgenol. 2000;175:197–205. 16. Hughes K, Hamdan A, Schermerhorn M, Giordano A, Scovell S, Pomposelli F, Jr. Bypass for chronic ischemia of the upper extremity: results in 20 patients. J Vasc Surg. 2007;46:303–307. 17. Banis JC Jr, Rich N, Whelan TJ Jr. Ischemia of the upper extremity due to noncardiac emboli. Am J Surg. 1977;134:131–139.

22

MRA: Upper Extremity and Hand Vessels

18. McCready RA, Bryant MA, Divelbiss JL. Combined thenar and hypothenar hammer syndromes: case report and review of the literature. J Vasc Surg. 2008;48:741–744. 19. Robbs JV, Carrim AA, Kadwa AM, Mars M. Traumatic arteriovenous fistula: experience with 202 patients. Br J Surg. 1994;81: 1296–1299. 20. Lim RP, Storey P, Atanasova IP, et al. Three-dimensional electrocardiographically gated variable flip angle FSE imaging for MR angiography of the hands at 3.0 T: initial experience. Radiology. 2009;252:874–881. 21. Fayad LM, Hazirolan T, Bluemke D, Mitchell S. Vascular malformations in the extremities: emphasis on MR imaging features that guide treatment options. Skeletal Radiol. 2006;35:127–137. 22. Dayicioglu D, Martell EG, Ogilvie M, Gozu A, Panthaki ZJ, Armstrong MB. Vascular anomalies of the upper extremity in children. J Craniofac Surg. 2009;20:1025–1029. 23. van Rijswijk CS, van der Linden E, van der Woude HJ, van Baalen JM, Bloem JL. Value of dynamic contrast-enhanced MR imaging in diagnosing and classifying peripheral vascular malformations. AJR Am J Roentgenol. 2002;178:1181–1187. 24. Flinterman LE, Van Der Meer FJ, Rosendaal FR, Doggen CJ. Current perspective of venous thrombosis in the upper extremity. J Thromb Haemost. 2008;6:1262–1266. 25. Kahn SR, Elman EA, Bornais C, Blostein M, Wells PS. Postthrombotic syndrome, functional disability and quality of life after upper extremity deep venous thrombosis in adults. Thromb Haemost. 2005;93:499–502. 26. Sadowski EA, Bennett LK, Chan MR, et al. Nephrogenic systemic fibrosis: risk factors and incidence estimation. Radiology. 2007; 243:148–157.

317 27. Li W, Salanitri J, Tutton S, et al. Lower extremity deep venous thrombosis: evaluation with ferumoxytol-enhanced MR imaging and dual-contrast mechanism--preliminary experience. Radiology. 2007;242:873–881. 28. Mistretta CA. Undersampled radial MR acquisition and highly constrained back projection (HYPR) reconstruction: potential medical imaging applications in the post-Nyquist era. J Magn Reson Imaging. 2009;29:501–516. 29. Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magn Reson Med. 2007;58(6):1182–1195. 30. Wang J, Yarnykh VL, Molitor JA, et al. Micro magnetic resonance angiography of the finger in systemic sclerosis. Rheumatology (Oxford). 2008;47:1239–1243. 31. von Morze C, Xu D, Purcell DD, et al. Intracranial time-of-flight MR angiography at 7T with comparison to 3T. J Magn Reson Imaging. 2007;26:900–904. 32. Miyazaki M, Sugiura S, Tateishi F, Wada H, Kassai Y, Abe H. Noncontrast-enhanced MR angiography using 3D ECG-synchronized half-Fourier fast spin echo. J Magn Reson Imaging. 2000; 12:776–783. 33. Isogai J, Kobayashi Y, Ogawa Y, et al. A novel non-contrast MRA technique using Time-Spatial Labeling Inversion Pulse in combination with flow-spoiled FBI for the assessment of small arteries of the finger. Proc Int Soc Magn Reson Med. 15th Meeting 2007;Berlin:3152. 34. Sheehan JJ, Fan Z, Carr JC, Li D. Non contrast MRA of the hand in patients with Raynauds disease usng flow sensitized dephasing prepared SSFP. Proc Int Soc Magn Reson Med 17th Meeting 2009; Honolulu:423.

wwwwwwwwwwww

Lower Extremity Peripheral Arterial Disease

23

Jeremy D. Collins and Timothy Scanlon

Introduction Magnetic resonance angiography (MRA) is an accepted modality for arterial imaging of the lower extremities. Lower extremity MRA (LE-MRA) primarily relies on subtracted gadolinium-enhanced three-dimensional spoiled gradient echo sequences. The modality has evolved with the evolution of low-dose time-resolved gadolinium-enhanced angiography sequences. LE-MRA has also benefited from the continued evolution of noncontrast angiography techniques, enabling safe and accurate arterial imaging in patients with renal insufficiency. LE-MRA has several advantages over the competing modalities of ankle-brachial index assessment, Doppler ultrasonography, computed tomographic angiography (CTA), and catheter angiography. Lower extremity Doppler ultrasonography is a cost-effective screening tool for patients with suspected peripheral vascular disease. Advantages of the technique include the use of high-frequency sound waves to generate images, without significant energy absorption by the patient. However, artifacts from calcified plaque, the multifocal nature of peripheral vascular disease, limited Doppler assessment of lesions downstream from a significant stenosis, and dependence on operator expertise limits definitive assessment of entire lower extremity system. CT angiography, while a robust technique for visualizing the vasculature, has significant limitations in the assessment of lower extremity peripheral vascular disease. Significant limbto-limb differences in arterial transit times can compromise CT angiography performed with a single contrast-enhanced J.D. Collins, MD () Department of Radiology, Northwestern Memorial Hospital, Chicago, IL, USA e-mail: [email protected] T. Scanlon, MB, BCh, BAO Department of Cardiovascular Imaging, Radiology, Northwestern University Feinberg School of Medicine, Chicago, IL, USA

acquisition. Multiple-enhanced acquisitions are not routinely performed secondary to concerns regarding radiation exposure. The high prevalence of peripheral small vessel disease in the patient population also limits the utility of single-source CT acquisition; calcifications in small vessels are often hard to differentiate from diseased, but patent runoff vessels. Dualenergy techniques have been developed to facilitate semiautomated bone subtraction; however, robust techniques for automated removal of calcified plaque enabling differentiation from the contrast-opacified blood pool are lacking. Digital-subtraction catheter angiography (DSA) is still considered the gold standard for lower extremity imaging. However, the invasive nature of the technique, use of ionizing radiation, inability to assess the surrounding soft tissues, and limited contrast alternatives in patients with preexisting renal insufficiency relegate the technique to focused diagnostic studies and interventions. The principle advantage of DSA is the ability to perform an intervention at the time of a diagnostic study.

MRA Techniques LE-MRA can be performed with both contrast-enhanced and noncontrast techniques. Gadolinium-enhanced spoiled gradient echo techniques are the workhorse sequences for LE-MRA in the clinical setting [1]. All MRA sequences employ techniques to increase signal from the vasculature, while simultaneously suppressing signal from the adjacent nonvascular tissues, often maximizing visualization of the arteries over the veins.

Principles of Gadolinium-Enhanced MRA The principle technique employed for LE-MRA is threedimensional subtracted high-resolution gadolinium-enhanced angiography. High signal to noise images of the enhanced vasculature are generated by suppressing nonvascular signal,

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_23, © Springer Science+Business Media, LLC 2012

319

320

employing mask subtraction techniques, and shortening the T1 of blood by administering intravascular paramagnetic contrast agents. High-quality arterial imaging is dependent on acquisition during the arterial phase of contrast opacification with collection of the central lines of k-space during the peak of arterial enhancement. An understanding of the tradeoffs between spatial resolution and acquisition time is imperative to adapt this technique in the clinical setting. All LE-MRA gadolinium-enhanced sequences are acquired as three-dimensional acquisitions. This differentiates the technique from some noncontrast techniques and the majority of postcontrast soft tissue imaging which is acquired as single or multiple slices. Three-dimensional sequences have the advantages of increased signal to noise; however, the longer acquisition time requires optimization of contrast dynamics and instructing the patient to remain perfectly still throughout the acquisition. In three-dimensional acquisitions, motion artifacts and contrast mistiming impacts the entire dataset, rather than a few slices, and may result in a nondiagnostic examination. High-resolution LE-MRA utilizes a spoiled gradient echo technique, with a short TR and a low flip angle. The short TR suppresses signal in tissues with intermediate to long T1 relaxation times by preventing complete recovery of longitudinal relaxation between pulses. Appropriate choice of flip angle maximizes the signal from all tissues. The Ernst angle is the optimum flip angle, where by the greatest signal will be generated for a given TR. The angle is calculated by the equation arccos (TR/T1). Flip angles less than the Ernst angle do not yield enough transverse magnetization for optimum signal; larger flip angles reduce the longitudinal magnetization available for subsequent pulses. In isolation, the gradient echo readout described above generates low signal arteries and veins. Paramagnetic contrast agents are utilized to shorten the T1 of blood and greatly increase the vascular signal over that of the adjacent tissues. The T1 value of arterial blood is commonly cited at 1,200 ms; venous values are similar to those of the arteries at 1.5 Tesla (T) [2]. Exact T1 blood values are dependent on the body temperature, hematocrit, and oxygen saturation. A paramagnetic contrast agent can reduce the T1 of blood from 1,200 to 100 ms with a dispersed bolus and to as low as 50 ms with a tighter bolus profile, greatly increasing the recovery of longitudinal magnetization. Tissues with short T1 values remain relatively bright on spoiled gradient echo sequences. Lipidrich tissues have a T1 of 260 ms and demonstrate intermediate signal. The application of subtraction techniques maximizes the signal from gadolinium-enhanced vessels, while eliminating signal from any stationary tissue without arterial enhancement. An unenhanced acquisition is obtained with identical parameters to the subsequent postcontrast dataset. This unenhanced acquisition is referred to as the mask. Using identical

J.D. Collins and T. Scanlon

patient breath-holding and instructing the patient to remain completely still, subtracting mask data from the contrastenhanced data completely suppresses stationary, nonenhancing tissues. Combining image subtraction with pure arterial phase vascular opacification greatly improves arterial contrast to the surrounding tissues. The significantly greater signal from the vasculature and near complete suppression of nonvascular tissues makes subtracted three-dimensional datasets ideal for postprocessing with maximum intensity projection (MIP) and multiplanar reformatting (MPR) techniques. In addition to optimizing contrast resolution, tailoring spatial resolution to the imaged vascular territory is a requirement to obtain high-quality MRA images. Spatial resolution at MRA is described in terms of the volume pixel element, or voxel, but must be distinguished form the true spatial resolution. MR scanners report the voxel dimensions, which often incorporate k-space symmetry and interpolation methods (discussed subsequently) to increase the apparent spatial resolution without an increase in acquisition time. Reconstructions of three-dimensional data are best performed on isotropic volumes, whereby the spatial resolution is identical in all three directions. MRA datasets are acquired in the coronal or sagittal planes so that the readout, or frequency encoded, direction is in the z-axis. This is necessary to prevent phase wrap into the image and enables maximizing the imaging matrix in the direction requiring greatest coverage for LE-MRA without increasing the scan time. The x- (left-right) and y- (anteroposterior) axes are phaseencoded; increasing the resolution in the phase-encoded directions increases the acquisition time. The acquisition time for a spoiled gradient echo readout is determined by the TR and number of phase-encoded steps in both the x- (Nx) and y- (Ny) directions: Time Acq = TR × Nx × Ny. For this reason, MRA is often performed with an in-plane rectangularized field of view (i.e., 352 × 512). Further gains in acquisition time are achieved through the use of k-space undersampling. As mentioned above and discussed in detail subsequently, the central lines of k-space contribute to the contrast resolution and peripheral lines provide edge definition. Symmetries in k-space allow calculation of the undersampled peripheral data prior to the Fourier transform; alternatively this data can be quickly zero-filled. Either technique permits interpolation to a smaller voxel, improving image quality. It is important to note that interpolation techniques do not improve the actual spatial resolution; they simply improve the apparent resolution of the reconstructed image without a penalty in acquisition time. Interpolation schemes should be applied to achieve a nearly isotropic voxel in the reconstructed dataset. Nonisotropic data precludes careful assessment of vessels in orientations other than the acquisition slab. Despite the technical innovations to maximize image fidelity, LE-MRA image quality remains dependent on

23

Lower Extremity Peripheral Arterial Disease

Fig. 23.1 Poor image quality secondary to initiation of the MRA acquisition (a) before and (b) after the contrast bolus peaked in the thoracic aorta and the pelvis in two different patients. Both patients had to return to the imaging suite for repeat examinations

acquiring the central lines of k-space during the arterial phase of contrast opacification. The central aspect of k-space provides image contrast. Peripheral k-space data contributes to image sharpness, enhancing the perceived image resolution. Several k-space sampling schemas are available to re-order k-space filling with early collection of central data. Two well-described sampling methods are central and linear asymmetric. As the title implies, in central methods k-space is acquired starting at the center, filling the periphery by moving outwards. This strategy ensures that the central data is acquired before any peripheral data. Linear asymmetric sampling starts off-center, but fills in toward the center of k-space, such that the central k-space data is acquired approximately half-way through the acquisition. Contrast timing requirements are therefore dependent on the chosen k-space sampling method. Starting imaging acquisition too early results in band-like artifacts through the arteries; acquiring data after the arterial peak may result in excessive venous and soft tissue enhancement. Contrast-enhanced MRA datasets with problematic timing are presented in Fig. 23.1. Central MRA sampling schemes require scan initiation at the mid-point of the gadolinium bolus, during peak arterial opacification. This filling scheme is well suited to bolus timing. Asymmetric schemes fill the central lines of k-space approximately mid-way through the acquisition; bolus timing is optimized to the middle of the dataset acquisition. Bolus tracking methods, which enable near real-time visualization of contrast arrival in the vessels of interest, can be employed with automated or manual scan initiation and are often used with asymmetric k-space filling schemas. Filling the central lines of k-space in the first half of the acquisition also permits reduction of the gadolinium dose. The peripheral

321

lines can be acquired at the tail and after the dispersed bolus reaches the veins without substantially impacting image contrast. The development of parallel imaging technologies such as sensitivity encoding (SENSE) and generalized autocalibrating partially parallel acquisitions (GRAPPA) has been essential to the continued improvement in LE-MRA [3, 4]. Parallel imaging acquires reference lines at the start of the scan to partially encode spatial information based on the positioning of coil elements over the region of interest. This reduces the number of requisite phase encoding steps and allows k-space undersampling. Artifacts from k-space undersampling are therefore reduced. This allows acquisition of the center of k-space in a shorter period of time, reducing dependence on bolus timing, and enables further gains in spatial resolution without increasing acquisition time. SENSE factors of 2–3 are routinely used in clinical practice; higher parallel imaging factors have been employed with success in the research arena [5]. Gains in acquisition speed are directly proportional to the number of coil elements, with substantial parallel imaging factors theoretically achievable. Parallel imaging has been limited in clinical practice by reductions in signal to noise secondary to shorter acquisition times. Imaging with a more compact bolus profile by increasing the contrast injection rate partially compensates for the reduction in signal to noise. Coil sensitivity profiles are also not ideal in some regions of the body, limiting the achievable gains in acquisition time. MRA at higher field strengths, with associated greater signal to noise is well suited for parallel imaging techniques. Time-resolved or four-dimensional MRA generates a digital-subtraction angiogram-like dataset with dynamic flow information. The most commonly employed technique in time-resolved LE-MRA is k-space undersampling with an ultrashort TR. Parallel imaging is often employed to further reduce the scan time with resultant improvements in both temporal and spatial resolution. Contrast injection rates of 5–6 mL/s and smaller flip angles compensate for the lower signal to noise. Time-resolved imaging of contrast kinetics (TRICKS) and time-resolved imaging with stochastic trajectories (TWIST) are two common techniques employed clinically for dynamic calf LE-MRA [6]. Differing principally by the k-space undersampling mechanism, both techniques sample the central lines of k-space more frequently than those in the periphery. Individual datasets are reconstructed by sharing peripheral k-space data from adjacent temporal frames with acquired central k-space data. Zero-fill interpolation is used to generate an isotropic voxel in the reconstructed dataset. Additional modifications in k-space filling and image reconstruction have enabled further improvements in temporal resolution while maintaining spatial resolution. Projection reconstruction (PR) TRICKS utilizes a radially undersampled projection reconstruction algorithm to achieve

322

up to threefold gains in temporal resolution [7]. Linear artifacts may occur as a result of angular undersampling; however, these are seldom problematic with the high signal to noise ratio achieved with the technique. Highly constrained back-projection (HYPR) TRICKS improves spatial resolution by further undersampling peripheral k-space data during the dynamic arterial acquisition, acquiring this data as a high-resolution MRA during the venous phase of contrast opacification [8]. These two modifications can be combined in PR HYPR TRICKS with consequent improvements in both spatial and temporal resolution.

Gadolinium-Enhanced MRA Adaptations for Lower Extremity Angiography A requisite for LE-MRA is the ability to image the craniocaudal extent of the vasculature from the infrarenal aorta through the pedal vessels in a single examination. The variations of gadolinium-enhanced LE-MRA differ principally in the manner that they enable imaging beyond a single station field of view in the z-axis. Recent adaptations have employed time-resolved MRA to reduce the gadolinium dose and improve differentiation of arteries and veins. Early LE-MRA iterations optimized the z-axis coverage to divide the vasculature into three separate imaging stations, termed three-station LE-MRA. A separate contrast injection was performed for each station. Venous signal from prior injections was removed by acquiring the mask acquisition immediately before a subsequent contrast-enhanced dataset. Subtraction techniques removed remaining signal from the veins; however, this also reduced signal from the arteries, with a greater impact on the last acquired contrast-enhanced dataset. Consequently, contrast doses increased from station to station with a total contrast dose of 0.3–0.4 mmol/kg of gadolinium for a complete study. Macroscopic motion greatly reduced image quality, as venous signal was not adequately suppressed. Subsequently, researchers developed custom solutions to the scanner table for rapid z-axis translation. Initially termed “stepping-table” angiography, manufacturers increased the table translation speed to allow rapid patient repositioning between stations. This enabled bolus-chase imaging, where the bolus was timed for the pelvis and “chased” down the lower extremities with subsequent thigh and calf acquisitions. Contrast doses of 0.2–0.4 mmol/kg were administered for bolus-chase studies as the lengthy time of image acquisition required a long bolus. Mask acquisitions were performed prior to the administration of contrast. Consequently, image quality is often degraded by venous contamination during the acquisition of the calf station when the central lines of k-space are acquired when both the veins and arteries are opacified. This can be accentuated in patients with unilateral bypass graft occlusion. Improvements in gradient technology

J.D. Collins and T. Scanlon

permitted reductions in station acquisition time; however, limb-to-limb discrepancies in arteriovenous transit time remained problematic for distal stations. The challenge of venous contamination of distal stations was addressed through the addition of a separate calf acquisition to the bolus-chase LE-MRA technique. Termed Hybrid MRA, an additional calf acquisition was performed prior to the bolus-chase study with dedicated bolus timing to the popliteal arteries [9, 10]. Limb-to-limb discrepancies in contrast arrival were readily identified on the timing bolus and the subsequent two to three sequential calf acquisitions were optimized to assess the earliest enhancing calf. The subsequent bolus-chase study was performed as previously described, with subtraction techniques removing venous signal from the first bolus. Contrast doses for Hybrid MRA were divided into 0.1 and 0.2 mmol/kg for the calf and boluschase acquisitions, respectively. Improvements in gradient technology and the development of k-space undersampling schemes enabled timeresolved MRA with subsecond temporal resolution, as described above. The most recent iteration of contrastenhanced LE-MRA combines the high-spatial resolution of the bolus-chase technique with time-resolved imaging of the calves [11]. Adapting this technique to the anatomy of the calves lengthens the temporal acquisition to between 3.5 and 5 s. Time-resolved calf imaging readily identifies limb-tolimb discrepancies in contrast arrival, identifies retrograde arterial filling, and permits accurate differentiation between late arterial and early venous filling. The time-resolved technique does not require a test bolus and can be performed with less than 0.05 mmol/kg of gadolinium contrast. In clinical practice, the time-resolved calf acquisition is followed by the bolus-chase LE-MRA acquisition. As in Hybrid MRA, mask subtraction removes venous contamination from the datasets; the smaller gadolinium dose required for time-resolved calf imaging improves arterial signal on the bolus-chase acquisition compared to Hybrid MRA. Combined with an abdominal aortic timing bolus, contrast arrival time in the calf and pedal vessels on time-resolved imaging can provide the contrast transit information necessary to adapt a subsequent multistation LE-MRA data acquisition to minimize venous contamination [12]. Although a promising technique to improve image quality in the thighs and calves, the need for a highly trained technologist to modify the imaging protocol remains a limitation. Several investigators have developed techniques for continuous table motion (CTM) LE-MRA. This technique utilizes a single coronal or sagittal imaging slab, replacing the bolus-chase multistation LE-MRA with a single extended field-of-view acquisition. A precontrast mask acquisition is performed and subtraction techniques applied to increase the vascular signal to noise. At the start of scan, the table pauses to acquire data from the cranial field of view. Similarly, at the end of table movement, the acquisition continues for a few

23

Lower Extremity Peripheral Arterial Disease

Fig. 23.2 Differences in volume coverage between continuous table movement and stepping-table acquisitions. (a) Continuous table movement acquires a single slab of uniform thickness across the z-axis field of view. (b) Bolus chases stepping-table acquisitions require careful prescription of each imaging volume to optimize spatial resolution within the limits of acquisition time penalties

seconds to obtain the periphery of k-space data. Benefits of the technique are shorter scanning times without difficulties in positioning adjacent slabs to ensure adequate arterial overlap (Fig. 23.2). Holding parameters constant, scan time is shorter because the between-station table moving intervals are incorporated into continuous table movement with simultaneous image acquisition. The longer field of view is also free of field inhomogeneity artifacts at the edges of each station. Table velocity is dependent on the spatial resolution and the parallel imaging factor. A challenge for the technique is the requirement that the same y-axis slab thickness is utilized throughout the entire field of view, precluding optimization of spatial resolution by station. In patients with tortuous aortoiliac vessels, for example, the technique requires that the greater y-axis coverage needed in the pelvis is applied to the lower extremities. In addition, significant differences in arteriovenous transit from limb to limb may compromise image quality in the distal limbs. The continuous table movement acquisition has been combined with separate time-resolved calf acquisition to improve assessment of the calves.

Gadolinium Dosing The reports of nephrogenic systemic fibrosis and its association with acute and chronic renal insufficiency dramatically

323

changed gadolinium dosing at MRI and with LE-MRA in particular. As described above, early LE-MRA protocols utilized gadolinium doses in excess of 0.3 mmol/kg/1.73 m2. At our institution, we obtain point-of-care (iStat, Abbott Laboratories, Abbott Park, IL) eGFR calculations on the day of the scan. If the eGFR >60 mL/min/1.73 m2, a total of 0.20 mmol/kg gadopentetate dimeglumine (Magnevist, Bayer Healthcare Pharmaceuticals, Wayne, NJ) is administered. For patients with eGFR 30–60 mL/min/1.73 m2, a total of 0.15 mmol/kg gadobenate dimeglumine (Multihance, Bracco Diagnostics, Princeton, NJ). Gadobenate dimeglumine has a highly stable cyclic ring structure greatly reducing the amount of free gadolinium at equilibrium, as discussed subsequently. The gadolinium dose is divided approximately 1:3 between the time-resolved calf acquisition and the bolus-chase LE-MRA runoff. Noncontrast LE-MRA sequences are performed in patients with eGFR <30 mL/min/1.73 m2. In the rare event that noncontrast sequences are unable to provide the necessary diagnostic information and no other study is suitable, a study with low-dose gadobenate dimeglumine is considered. Informed consent is obtained from all such patients. At our institution, these patients undergo hemodialysis within 3 h of the examination and again at 24 h. Peritoneal dialysis is unable to efficiently remove the contrast agent; these patients also undergo temporary hemodialysis.

Angiographic Contrast Agents MRA contrast agents are paramagnetic polymers of the heavy rare earth element Gadolinium. These agents alter tissue contrast through enhancing the relaxivity of protons in proximity to the element. Although both T1 and T2 shortenings occur in the presence of gadolinium, the predominant effects are a consequence of T1 shortening. Angiographic gadolinium contrast agents in clinical use are primarily extracellular chelates employing a linear or uncharged cyclic ring chemical structure [13]. Cyclic ring agents are more stable with little Gadolinium existing as an unbound element in the soft tissues at histology. Although administered intravenously, these agents rapidly equilibrate with the interstitial space. The majority of these Gadolinium contrast agents are 0.5 M. A list of approved MR angiographic contrast agents appears in Table 23.1. A recently introduced contrast agent, Gadobutrol (Gadovist, Bayer Pharmaceuticals, Toronto, Ontario) has a 1.0 M concentration. The clinical utility of a 1 M contrast agent for LE-MRA is the subject of ongoing investigation [14]. The FDA has recently approved an intravascular contrast agent for use in MR angiography. Gadofosveset (Ablavar, Lantheus pharmaceuticals) is a Gadolinium-based contrast agent that binds tightly to albumin. Consequently, the volume of distribution is restricted to the intravascular space.

324

J.D. Collins and T. Scanlon Table 23.1 FDA approved contrast agents for MR angiography MR contrast agent Gadodiamide (Omniscan) Gadobentic acid (Multihance) Gadopentetic acid (Magnevist) Gadoteridol (Prohance) Gadofosveset (Ablavar) Gadoversetamide (Optimark) Gadobutrol (Gadovist)

Concentration 0.5 M 0.5 M 0.5 M 0.5 M 0.25 M 0.5 M 1M

This permits delayed vascular imaging without soft tissue enhancement and is well suited for venous imaging.

Noncontrast LE-MRA Techniques Initial enthusiasm for noncontrast techniques was tempered by the long acquisition times required for two-dimensional datasets over a lengthy z-axis field of view. The association between gadolinium-based contrast media and nephrogenic systemic fibrosis in patients with acute and chronic renal insufficiency has sparked renewed interest. In addition, a robust noncontrast technique can rescue a nondiagnostic gadolinium-enhanced LE-MRA examination, saving the patient a return visit to the MR suite. As with contrastenhanced MRA, noncontrast LE-MRA requires bright arterial signal in the vasculature, methods to suppress signal from stationary tissues, and techniques to remove signal from the adjacent veins. This section reviews several noncontrast techniques, highlighting those adapted for clinical applications.

Time-of-Flight Angiography An early iteration of noncontrast angiography, the technique relies on inflow of unsaturated protons to generate signal. Both two- and three-dimensional acquisitions are possible. Stationary tissues are exposed to repeated pulses and steadily lose longitudinal magnetization, with near complete suppression. A longer TR and a higher flip angle contribute to the suppression of stationary tissues. However, parameters that facilitate stationary tissue suppression also adversely impact the signal in flowing blood. For a given blood velocity, increases in the TR must be compensated for by decreasing the slice or slab thickness. In addition, the configuration of the vessels of interest relative to the imaging plane is an important consideration. Tortuous vessels with greater distances to traverse in the slice or volume may demonstrate saturation of signal; ideally the imaging volume or slice is positioned orthogonal to the imaging plane. For example, the

Chemical structure Uncharged, linear Charged, linear Charged, linear Uncharged, cyclic ring Charged, linear Uncharged, linear Uncharged, cyclic ring

Principle organ of elimination Renal Renal Renal Renal Renal >> Feces Renal Renal

complex anatomy of a tortuous aorto-iliac system, tortuous collateral vessels, and the horizontally oriented proximal anterior tibial artery could result in artifactual signal loss mimicking disease. Time-of-flight imaging has limited sensitivity to slow flow due to challenges differentiating an occluded segment from proton saturation secondary to slower than expected flow. Venous inflow is suppressed through the application of a saturation pulse below the imaging plane. Three-dimensional slab thickness is limited by arterial saturation deep into the imaging volume. Iterations of twodimensional imaging start at the exit of arterial blood from the field of view, moving opposite to the direction of arterial blood flow. This prevents saturation band associated signal loss in the arteries while increasing venous signal suppression. Specific iterations have been developed to overcome the deep imaging volume arterial blood saturation, including multiple overlapping thin slab acquisition (MOTSA) and titled optimized nonsaturated excitation (TONE). MOTSA covers a large z-axis dimension by splitting the volume into several thinner slabs. TONE applies larger flip angles deep into a slab to increase the signal from saturated blood protons. Flow-related dephasing is an important consideration in time-of-flight imaging. Both intravoxel and intervoxel dephasing can occur. The accumulated phase shift is dependent on the blood velocity, the amplitude of the applied gradient, and the square of the gradient duration. Gradient moment nulling has been applied to reduce flow-related dephasing. Rather than a single bipolar gradient, gradient moment nulling utilizes a pair of bipolar gradients of opposite polarity. Stationary tissues experience no net phase accumulation under both scenarios; however, moving protons accumulate a net phase shift with a single bipolar gradient. The application of a pair of bipolar gradients compensates for constant flow velocities with no net phase accumulation. Gradient moment nulling cannot compensate for higherorder flow patterns with accelerating protons or turbulent flow. In the presence of higher-order flows, minimizing the TE reduces the phase shift accumulation by shortening the period of time between the applied gradients and the echo. Higher-order flows are more sensitive to longer TEs than

23

Lower Extremity Peripheral Arterial Disease

constant velocity flow patterns. In addition, local field inhomogeneities experienced by the moving protons cannot be corrected for gradient moment nulling. Shorter TEs reduce the amount of phase shift accumulated; however, as in the case of higher-order flows, cannot completely compensate and artifacts result. Time-of-flight imaging is poorly suited to image the entire z-axis field of view required in LE-MRA due to the excessive imaging times required for three-dimensional imaging and proton inflow saturation effects. Selective application of twodimensional time of flight may be useful in the assessment of flow directionality. However, an appropriate suspicion is required, as the traveling saturation band suppresses retrograde arterial flow. When combined with contrast-enhanced LE-MRA, the technique is able to readily differentiate antegrade and retrograde flow patterns, as vessel patency is already known. Two-dimensional time of flight has demonstrated accuracy in assessment of pedal vessel stenosis and occlusion [15], although it has largely been replaced by novel noncontrast techniques and contrast-enhanced methods for this application.

Phase-Contrast Angiography Phase-contrast MR angiography (PC-MRA) generates images of the vasculature based on the absolute net phase shift acquired by protons moving through a time-varying magnetic field. PC-MRA is performed by the application of both flow compensated and flow-encoding bipolar gradients. As discussed previously, gradient moment nulling utilizes paired bipolar gradients of opposite polarity to correct for phase shifts accumulated by protons moving in the direction of the applied gradient. By also acquiring a velocity-encoded, uncompensated gradient acquisition, data regarding the net phase shift of the protons is obtained. Both acquisitions experience phase shifts secondary to field inhomogeneities; however, only the velocity-encoded uncompensated acquisition also contains velocity-related phase shift information. Phase shifts accumulated by moving protons after correction for field inhomogeneities are linearly related to the proton velocity. Net phase shift is converted to proton velocity by the estimation of the peak velocity experienced in a region of tissue. For accurate application, the user must accurately prescribe a velocity corresponding to 180° of accumulated phase shift. Voxels with an average net phase accumulation in one gradient direction of 180° would be assigned the peak velocity; however, if any voxels experienced greater than 180 or −180° of phase shift, aliasing would occur. For example, at a velocityencoding setting of 150 cm/s, a phase shift of 210, corresponding to an actual velocity of 210/180 × 150 = 175 cm/s would generate an aliased velocity corresponding to a phase shift

325

of −150° or −150/180 × 150 = −125 cm/s. Hence appropriate choice of velocity encoding is necessary. Although problematic at flow quantification, small amounts of aliasing at phase-contrast angiography do not introduce significant errors secondary to the vector sum calculation for average voxel velocity. Voxel signal intensity is equal to the square root of the sum of the squares of the corresponding velocity in each encoded direction; hence aliased negative velocities close to the peak forward velocity in magnitude introduce much smaller errors than seen with single direction flow quantification. The choice of the velocity-encoding gradient is important to consider in the context of an ECG-gated or ungated acquisition, reflecting the variation between diastolic and systolic peak versus average flow, respectively. PC-MRA can be performed with both two-dimensional and three-dimensional acquisitions. Although acquisition of a PC-MRA sequence at peak systole results in greater signal to noise ratio, imaging times may be prohibitive with threedimensional techniques. Consequently, applications of PC-MRA to the lower extremities have relied on ECG-gated two-dimensional acquisitions in the coronal plane, adjusting the velocity-encoding gradient to the expected average velocity across the cardiac cycle. Velocity encoding is much lower to reflect the high-resistance vascular bed supplied by the lower extremity vasculature at rest. Steffens et al. reported their experience imaging from aortic bifurcation to the tibioperoneal trunk with two-dimensional ECG-gated PC-MRA in 115 patients with atherosclerosis [16]. The authors employed three-dimensional flow encoding with velocityencoding gradients of 30 and 20 cm/s in the pelvis and thighs, respectively. The imaging volume was divided into three stations with an acquisition time of 4–7 min per station; two averages were performed with completion of imaging in 30 min. The lesion analysis demonstrated a sensitivity and specificity of 95% and 90% with positive and negative predictive values of 90% and 96%, respectively. PC-MRA overcomes the limitations associated with flowrelated enhancement at time-of-flight imaging by generating signal intensity proportional to proton velocity. Technical improvements with undersampling techniques have resulted in shorter imaging times [17]. However, clinically available PC-MRA iterations have imaging times impractical for routine LE-MRA applications. A related technique, two-dimensional flow quantification, discussed in detail elsewhere in this textbook, has been described for selective assessment of stenosis grading in the lower extremities. Similar to Doppler sonography, the velocity increase at a stenosis can be calculated based on the corrected phase shift; using the modified Bernoulli equation, 4 × Vmax2, the pressure gradient across the lesion can be estimated. Mohajer et al. showed success grading lesion severity in the superficial femoral arteries by assessing the peak velocity at and measuring the delay in systolic peak velocity beyond lesions, using MRA as standard

326

of reference. Although accurate for assessing a problematic individual lesion, this technique is impractical for assessing serial lesions in patients with multifocal disease.

3D MRA with Balanced Steady-State Free Precession Bright-blood balanced steady-state free precession (bSSFP) imaging generates robust images of the soft tissues and vessels, with signal intensity dependent on the ratio of T2/T1 [18]. Signal intensity is independent of inflow. Both arteries and the veins demonstrate high signal intensity; consequently most angiographic adaptations utilize additional preparatory pulses to suppress venous signal. Similar to other noncontrast techniques, signal loss may in result in regions of flow acceleration. Short TR times combined with high flip angles generate high signal to noise ratios, making these techniques well suited to parallel imaging. One technique to suppress fat tissue at bSSFP is the Dixon method. This utilizes the frequency shift of protons associated with lipids relative to those in water. At 1.5 T, there is a frequency offset of 220 Hz, with lipid protons processing at a slightly lower rate. Acquiring an echo with the fat and water protons in phase generates increased signal from voxels containing both lipid and water; acquisitions at an effective TE with out-of-phase lipid and water protons generate low signal intensity voxels. Tuning the center frequency of the magnet can also produce in- and out-of-phase images. Complex addition of these two datasets can generate wateror fat-only images. bSSFP sequences are highly susceptible to field inhomogeneities, with consequent off-resonance artifacts. These become increasingly problematic with longer TRs. Application in LE-MRA is particularly challenging due to the complex surface anatomy of the lower extremities. Venous signal notwithstanding, however, diagnostic images of the arteries can be obtained with this technique.

Arterial Spin Labeling Arterial spin labeling (ASL) as the name implies relies on the inflow of labeled or tagged spins into the imaging volume. A nonselective inversion pulse suppresses signal from stationary tissues. Earlier versions of technique were implemented with segmented turbo fast low angle shot gradient echo sequences and required two separate acquisitions, one with a nonselective inversion pulse and the other with selective inversion of tissue upstream from the region of interest [19]. The positioning of the upstream tagged slab is determined by the desired delay time between tagging and imaging. Subtracting the two acquisitions yields an angiographic image with suppressed background tissues.

J.D. Collins and T. Scanlon

Subsequent ASL iterations combined a spatially nonselective inversion pulse with a spatially selective upstream inversion pulse into a single acquisition. The delay time between the application of the nonselective inversion pulse and the initiation of imaging allows selective suppression of different tissues based on T1 relaxation. Acquiring an untagged imaging volume for subtraction enables better suppression of background tissue. Further improvements on the technique have incorporated bSSFP and partial Fourier FSE sequences. Combining ASL with bSSFP yields bright-blood images with excellent venous suppression and high signal to noise ratio [20]. Partial Fourier FSE sequences (discussed in detail subsequently) can be used with ASL and benefit from less susceptibility artifacts secondary to the spin echo readout. ECG gating is required for partial Fourier FSE readout to avoid spin-dephasing artifacts from systolic acquisition. Although an eloquent noncontrast MRA technique, ASL has limited application for imaging the lower extremities. This is primarily secondary to long transit times in the extremities. The inflow time for a peripheral imaging volume can approach the blood T1, negating differences between the tagged blood and stationary tissues in the imaging volume. Applying smaller imaging volumes is impractical with consequent increases in imaging time. In addition, ASL does not detect retrograde arterial flows. The technique assumes antegrade brisk arterial flow to generate adequate signal. Patent vessels with slow or retrograde flow are not well assessed with this technique and may appear occluded.

Emerging Noncontrast Techniques Several novel noncontrast techniques for lower extremity angiography deserve mention. The goal of this section is to introduce the reader to the concepts underpinning generation of arterial signal and venous suppression in each, while highlighting challenges and known pitfalls. Of these ECG-gated 3D partial Fourier FSE and quiescent interval single shot (QISS) are available commercially. The clinical utility and accuracy of these techniques is the subject of ongoing investigation.

ECG-Gated 3D Partial Fourier FSE This technique makes use of systolic arterial flow voids on T2-weighted fast spin echo (FSE) images to generate arterial images [21]. Systolic-triggered acquisitions demonstrate bright signal in the veins. Diastolic-triggered acquisitions demonstrate bright signal in both arteries and veins. Subtracting the two acquisitions yields arterial only images. Each partition in the three-dimensional dataset is acquired with a single-shot acquisition, lengthening the TR to equal two to three heartbeats to allow sufficient time for recovery

23

Lower Extremity Peripheral Arterial Disease

of longitudinal relaxation. The application of partial fast Fourier TSE readout yields acceptable acquisition times for clinical imaging, with 3–6 min reported for each station. Optimization of the systolic and diastolic trigger times can be achieved by obtaining preparatory ECG-triggered images at different trigger delays to choose the frames with greatest systolic suppression and adequate diastolic signal. Also, a preparation scan varying the spoiling gradients in the frequency direction can facilitate choice of dephasing gradient amplitude. A challenge applying this technique to LE-MRA is unwanted T2 blurring, which occurs in the phase encoding directions. In LE-MRA applications, the frequency encoding direction is oriented along the length of the vessel to facilitate spoiling of the systolic acquisition arterial signal. Consequently, the phase encoding directions are oriented orthogonal to the vessels of interest; blurring in these directions reduce the sharpness of the vessel walls. Single-shot FSE readouts acquire data over a range of approximately 350 ms. Solutions to reduce T2 blurring include a dual shot acquisition with consequent doubling of image acquisition time. Alternatively, centric ordering with rectilinear k-space sampling has been reported to reduce vessel blurring through less phase spread in the phase encoding directions. Although a promising technique, clinical application is challenging in several scenarios. Patients with irregular arrhythmias and tachyarrhythmias are not well suited to this technique due to differences in the length of diastole, precluding adequate sampling of the diastolic acquisition. Applying a dual shot acquisition may be helpful, but prolonged imaging times increases the probability of macroscopic patient motion. In addition, patients with significant peripheral vascular disease may demonstrate limb-to-limb and in-station segmental differences in the optimum delay time for the systolic arterial flow void. Optimizing flow to visualize collateral vessels may exaggerate the severity and extent of disease in adjacent native arteries. Despite these limitations, several studies have demonstrated promising results in clinical patients with atherosclerosis, particularly when the analysis is limited to those patients with good quality studies [22].

ECG-Gated Flow-Sensitive Dephasing bSSFP ECG-gated flow-sensitive dephasing (FSD) bSSFP acquires systolic- and diastolic-triggered datasets with a bSSFP bright-blood readout. Similar in principle to ECG-gated partial Fourier FSE, this technique reduces systolic arterial signal by applying a weak gradient moment in the frequency encoding direction, dephasing flowing signal [23]. Fast, laminar arterial flow patterns and weak gradient moments facilitate suppression of arterial signal in systole, while maintaining venous signal. Constant venous diastolic flow and slower

327

arterial diastolic flow facilitates bright signal in both vessels for the diastolic acquisition. Subtracting the two datasets results in pure arterial images. Acquisition for a single station is approximately 3 min. This technique has primarily been applied for imaging the calves alone. An ECG-gated phase-contrast imaging preparatory scan facilitates appropriate selection of the systolic and diastolic trigger delays. Similarly, use of a scout sequence applying different gradient moment strengths in the systolic phase optimizes arterial signal suppression [24]. A challenge to the technique is bright signal at systole in both arteries and veins with bSSFP. In clinical practice, this technique encounters similar difficulties to the ECG-gated partial Fourier FSE sequence, described above. Advantages of FSD bSSFP are shorter overall imaging time, greater diastolic arterial signal intensity free from short TR-associated saturation effects, and interleaved acquisition of systolic and diastolic phases.

Quiescent Interval Single-Shot MRA Quiescent interval single-shot (QISS) MRA is a novel ECGgated noncontrast technique that employs a two-dimensional bSSFP pulse sequence [25]. QISS MRA relies on combination of stationary tissue and fat signal suppression, venous signal saturation, and bright-blood bSSFP gated for acquisition during diastole. The technique utilizes a slice-selective radiofrequency pulse after a user specified delay time from the R wave to suppress stationary tissues within the imaging slice. A saturation pulse applied inferior to the imaging slice suppresses antegrade venous signal. Following a delay termed the quiescent interval, a chemical shift-selective fat suppression pulse is applied with a subsequent radiofrequency pulse forcing in-plane protons into the steady state. A two-dimensional bSSFP pulse sequence is subsequently acquired, triggered in diastole. The sequence is performed as stacks of two-dimensional slices starting caudally. Each station consists of sixty 3 mm slices with 0.6 mm overlap and takes approximately 45–70 s to acquire. A total of 10–13 stations are needed to cover the entire lower extremity field of view with adequate overlap between stations. Thinner slices can be acquired through diseased segments to better assess vascular stenosis. The technique benefits from parallel imaging with a factor of 2 at 1.5 T. Spatial resolution is limited in the slice direction by the slice thickness of 3 mm; however, 1 × 1 mm resolution is achieved in-plane. QISS MRA performed well in a feasibility study of eight volunteers. Preliminary analysis in a patient cohort referred for contrast-enhanced LE-MRA demonstrated high accuracy for QISS MRA using contrast-enhanced LE-MRA as the reference standard [26]. QISS MRA demonstrated 95% sensitivity and 92% specificity for detecting a stenosis ³50% using bolus-chase gadolinium-enhanced MRA as the reference

328

standard. The negative predictive value of QISS MRA was 96%; the positive predictive value was 90%. QISS MRA image quality assessed by Likert scores was comparable to contrast-enhanced LE-MRA. QISS MRA has several advantages compared to other noncontrast ECG-gated angiographic techniques. The sequence requires little patient-specific adjustments to generate robust images of the arteries with excellent soft tissue and venous signal suppression. Independence from systolic phase imaging improves the reliability in patients with peripheral vascular disease. As with other noncontrast ECGgated techniques, QISS MRA image quality may be reduced in patients with irregular arrhythmias or tachyarrhythmias. Finally, the two-dimensional acquisition technique reduces the effects of macroscopic patient motion.

J.D. Collins and T. Scanlon

Plaque Imaging As performed clinically, LE-MRA imaging protocols are optimized to grade stenoses, but lack the soft tissue contrast and resolution to characterize plaque components. Dedicated acquisitions with small fields of view, surface coils, and spinecho T1, T2, and proton density weighted imaging are necessary to achieve the requisite contrast and spatial resolution for plaque characterization. Even with optimized acquisitions, plaque characterization is limited to macroscopic plaque components in the femoral through popliteal arteries. The appearance of specific plaque components has been validated through histological correlation studies of carotid plaque specimens. MRI is currently used to assess changes in plaque burden with treatment in ongoing clinical research. Due to long imaging times and uncertain clinical utility, plaque imaging at MRI remains a research tool.

LE-MRA at 3 T The increasing availability of 3 T scanners has fostered adaptations of 1.5 T LE-MRA sequences for the higher field strength system. Despite the benefits from imaging at 3 T, several pitfalls deserve mention. Higher parallel imaging factors are achievable secondary to increases in signal. Specific absorption rate limits can be problematic at 3 T when 180° refocusing pulses or higher flip angles are utilized. Contrast-enhanced LE-MRA benefits from higher signal to noise ratio. Intravascular gadolinium contrast agents easily overcome the prolongation of T1 relaxation at 3 T. Greater signal to noise ratio enables higher parallel imaging factors, translating into gains in both spatial and temporal resolution at a fixed acquisition time. The development of total imaging matrix technology enables robust continuous moving table LE-MRA with a single data acquisition covering up to 128 cm with isotropic resolution [14]. Several noncontrast techniques deserve specific discussion at 3 T. Increased susceptibility artifacts adversely impact steady-state sequences. At 3 T, venous signal is less than arterial signal on bSSFP sequences due to the effect of reduced oxygenation in venous blood. In addition, longer TRs generate further venous signal suppression on bSSFP sequences. Stafford et al., who applied the Dixon technique at 3 T using a three-dimensional bSSFP sequence and a TR of 3.4 ms for noncontrast LE-MRA, demonstrated good quality arterial images with less venous signal [27]. The investigators employed a 50% overlap between coronal 3D bSSFP acquisitions to overcome inhomogeneity artifacts. The reduction in imaging time afforded by parallel imaging is particularly useful in ECG-gated partial Fourier FSE, with improvement in T2 blurring described above at 1.5 T. Finally, ASL benefits from the longer T1 relaxation times at 3 T. Preliminary work with QISS MRA suggests that the increased signal to noise ratio improves image quality in the tibial vessels and lower extremity branch vessels compared to 1.5 T.

Imaging Processing Image postprocessing is an important trouble-shooting tool in the assessment of vessel pathology. Through more efficient display of three-dimensional data, postprocessing may reduce the amount of time required to review an imaging study. Both contrast-enhanced and noncontrast LE-MRA datasets are well suited to MPR postprocessing to better assess stenosis in the plane of a diseased vessel. MPR review is best performed on unsubtracted partition data. The higher signal to noise ratio and contrast to noise achieved with mask-subtracted contrast-enhanced datasets and noncontrast acquisitions with static tissue signal suppression are well suited to MIP algorithms. MIP postprocessing projects the voxel with the greatest signal intensity in a line through the entire dataset to generate an image in a given orientation. MIPs collapse an entire three-dimensional dataset into either a single frame or a smaller number of frames (sliding MIPS). The three-dimensional nature of the dataset can be partially preserved by generating a series of rotating images through small changes in the orientation of the dataset. A large dataset can be reviewed efficiently using a MIP algorithm. Artifacts that increase signal intensity of background tissues or cause misregistration between contrast-enhanced and mask datasets reduce the quality of MIP reconstructions. Reviewing postprocessed and subtracted datasets without considering the source data may lead to pitfalls in interpretation (Fig. 23.3); abnormalities on postprocessed images should be verified by reviewing the source data. Postprocessing greatly facilitates efficient review of timeresolved MRA datasets. As described above, time-resolved acquisitions generate multiple three-dimensional datasets with a single contrast injection. Reviewing a series of subtracted MIPs oriented in the coronal or sagittal planes organized by the time of acquisition enables efficient review of the

23

Lower Extremity Peripheral Arterial Disease

329

Atherosclerotic Disease

Fig. 23.3 MIP postprocessing artifacts. Gadolinium-enhanced LE-MRA performed in a 67-year-old male with buttock claudication. The patient has a right common iliac artery stent and a short segment nonflow limited dissection of the left common iliac artery. The thickslab MIP exaggerates the signal loss in the stent. The dissection flap is not well-seen secondary to bright contrast-enhanced blood on both sides of the intimal flap. Review of partition data clearly demonstrates flow through the stent and the intimal flap

entire time-resolved dataset. Correlation with partition data is recommended to confirm abnormal findings on MIP images.

Clinical Applications Contrast-enhanced LE-MRA is routinely performed to evaluate the vessels of the lower extremities. The lower extremity arterial system is divided by station into the inflow vessels (aorto-iliac), outflow vessels (femoral and popliteal arteries), runoff vessels (tibial), and pedal vessels (dorsalis pedis and medial plantar arteries).

Peripheral vascular disease is the third most common manifestation of atherosclerosis. The prevalence in the general population is estimated at 10%, increasing to 15% in individuals over the age of 70. Presenting symptoms are often indolent, with pain or muscle cramping with activity subsiding with rest. The effected muscle groups may indicate the level of arterial occlusive disease. Activity-associated claudication may progress to rest pain, nonhealing ulcers, and tissue gangrene. Risk factors include family history, longstanding diabetes mellitus, history of hypercholesterolemia, and smoking. Aorto-iliac occlusive disease manifests with intermittent buttock and thigh claudication. Impotence may also be a presenting symptom. Leriche’s syndrome is a constellation of symptoms and signs associated with infrarenal occlusion of the abdominal aorta (Fig. 23.4). Initially described in middle-aged males, patients complain of impotence and buttock claudication; on physical examination the common femoral arterial pulses are nonpalpable. Treatment is indicated for symptomatic claudication. The mainstay of therapy is surgical bypass, although endovascular treatment may successfully recanalize the native infrarenal aorta and iliac vessels in a minority of patients. External iliac and femoral arterial disease manifests with intermittent thigh claudication. Superficial femoral, popliteal, tibial, or peroneal artery disease presents with calf claudication (Fig. 23.5). Endovascular treatment with angioplasty and stenting is preferred in patients with suitable anatomy; surgical bypass may be performed in patients with distal targets. Acute occlusion may result from distal thromboembolization of plaque, as shown in Fig. 23.6. Without an upstream dissection, plaque donor site, or aneurysm with mural thrombus, a cardiac source should be considered and further assessment with echocardiography or cardiac MRI is indicated. Cardiac MRI can be performed at the same time as a thoracoabdominal angiogram. The presence of preexisting collateral vessels determines the acuity of intervention. Most patients with acute thromboembolism undergo operative thrombectomy with possible fasciotomy. Patients without limb threatening ischemia may be referred for catheter directed lysis. Atherosclerosis and aneurysm formation often coexists. The natural history of iliac artery aneurysms mirrors that of aneurysms in other locations. Iliac aneurysms are discovered incidentally in up to 65% of patients (Fig. 23.7). Symptoms are nonspecific in the remainder; the most common symptom is abdominal pain. The common iliac artery is considered aneurysmal if it is greater than 2.5 cm in diameter. Treatment is recommended if the aneurysm exceeds 3 cm in diameter. Aneurysm rupture is associated with a mortality of up to 80%. Other complications include thrombosis and thromboembolization. Treatment strategies depend on the aneurysm location and its association with an abdominal aor-

330

J.D. Collins and T. Scanlon

Fig. 23.4 Leriche’s Syndrome. 62-Year-old male presents with bilateral buttock claudication. (a) Pelvis, (b) thigh, and (c) calf MIPs from a subtracted contrast-enhanced runoff demonstrate infrarenal aortic

occlusion, with reconstitution of the common femoral arteries through mesenteric and abdominal wall collaterals

Fig. 23.5 Peripheral occlusive vascular disease. 59-Year-old male presents with left calf claudication. (a) Pelvis, (b) thigh, and (c) calf MIPS from a subtracted contrast-enhanced runoff demonstrate segmental

occlusion of the distal left superficial femoral artery in the region of the adductor canal, with two vessel runoff via the peroneal and posterior tibial arteries

tic aneurysm. Endovascular aneurysm repair is the preferred therapy. The popliteal artery is an uncommon location for aneurysm formation (Fig. 23.8). Popliteal aneurysms are commonly associated with aortic aneurysms (Fig. 23.8d). As with aneurysms in other vascular territories, complications are related to rupture, progressive thrombus accumulation with stenosis or occlusion, or thromboembolism. Preferred treatment methods repair the native vessel; saphenous vein interposition grafts are considered in those patients whose native vessels cannot be repaired.

artery [28]. This entity should be considered in a young male patient presenting with calf claudication and in older patients without risk factors for atherosclerotic disease. The average age at presentation is less than 30 years, although this entity has been diagnosed in the seventh decade of life. The male:female ratio is approximately 15:1. A rare congenital abnormality with a prevalence of <1%, this entity is not uncommonly bilateral with a range of 22–67% reported. Physical examination findings that suggest the diagnosis include reduced pedal pulses with active plantar flexion and occasionally with passive dorsiflexion. Six distinct classifications of popliteal entrapment have been described. The most common cause of popliteal artery entrapment is compression by the medial head of the gastrocnemius muscle. Type I involves medial deviation of the popliteal artery around the medial head of a normally inserting gastrocnemius muscle. In type II, the medial head of the

Popliteal Entrapment Popliteal entrapment is an uncommon cause of popliteal aneurysm, popliteal stenosis, acute thrombosis, tibial vessel thromboembolization, and chronic occlusion of the popliteal

23

Lower Extremity Peripheral Arterial Disease

Fig. 23.6 Distal thromboembolism. 58-Year-old female with a history of long standing diabetes mellitus and peripheral vascular disease presents with acute onset right lower extremity pain. Coronal MIP contrastenhanced subtraction LE-MRA of the (a) pelvis, (b) thighs, and (c) calves demonstrates abrupt occlusion of the tibial vessels in the mid to distal calf (arrows). A potential donor site is identified in the mid right common iliac artery (arrow). An early phase time-resolved coronal subtracted contrast-enhanced time-resolved MRA MIP (d) resolves a pure arterial phase of tibial vessel opacification. A high origin of the left profunda femoris artery is noted incidentally (thick arrow)

331

gastrocnemius arises laterally from the medial femoral condyle. The popliteal artery deviates medially around the muscle. In type III, an accessory lateral slip of the gastrocnemius muscle traverses between the popliteal artery and vein to attach on the medial aspect of the lateral femoral condyle. In type IV, the popliteus muscle may have an abnormal course, traversing between the popliteal artery and vein. The aberrant popliteus displaces the artery medially and anteriorly. A fibrous band is also considered a type IV lesion. Type V describes classifications I–IV, which also entrap the popliteal vein. Type VI is functional entrapment secondary to hypertrophy of the gastrocnemius and popliteus muscles, crowding the normally positioned popliteal artery and vein. MRA is well suited to assess the popliteal artery, popliteal vein, and surrounding soft tissues in a single examination (Fig. 23.9). Time-resolved MRA sequences assess dynamic changes in popliteal artery caliber in conditions of plantar and dorsiflexion. High-resolution MRA is also performed in the neutral position. Postcontrast T1-weighed sequences are useful to delineate the anatomic relationship between the popliteal artery, popliteal vein, medial head of the gastrocnemius muscle, and popliteus muscle. Medial deviation of the popliteal artery is suggestive of popliteal artery entrapment on bolus-chase LE-MRA. Adding a postcontrast axial T1 spin echo sequence through the popliteal fossa may demonstrate an abnormal course of the popliteal or gastrocnemius muscle better than can be appreciated on unsubtracted contrast-enhanced LE-MRA. Treatment is determined by the extent of injury to the popliteal artery. Endovascular approaches fail to address the underlying compression in the popliteal fossa and result in poor patency. In all cases, the structure entrapping the artery is released surgically, with further management dependent on the condition of the popliteal artery. Options include primary repair of a popliteal aneurysm, stenosis patch grafting, or interposition graft placement often using a reversed saphenous vein.

Cystic Adventitial Disease

Fig. 23.7 Iliac artery aneurysm in a 94-year-old female with acute aortic syndrome (data not shown). (a) Thick and (b) thin MIPs from the abdominal MRA study demonstrate a fusiform internal iliac artery aneurysm filling retrograde in this patient with a history of a repaired abdominal aortic aneurysm with an aorto-bifemoral bypass graft

A rare cause of popliteal artery stenosis, this entity is characterized by extrinsic arterial compression by adventitial cysts containing mucinous fluid [29]. The prevalence is estimated at <0.1%. Symptoms of claudication bring patients to clinical attention. Cystic adventitial disease is unilateral and occurs in men, presenting in the third to fifth decades. Although the most common site of involvement is the popliteal artery, this uncommon entity has been described in the external iliac and common femoral arteries, and rarely in the large arteries of the upper extremity. The etiology of the mucinous cysts is unclear. Four theories have been advanced. At pathology, cysts are associated

332

J.D. Collins and T. Scanlon

Fig. 23.8 Popliteal aneurysms. Contrast-enhanced LE-MRA in a 58-year-old male who presents with left calf claudication. (a) Subtracted MIP contrastenhanced MRA demonstrates ecstatic and irregular popliteal vessels bilaterally. (b) Axial and (c) coronal postcontrast T1-weighted fat-suppressed imaging delineates the size of the aneurysm sac and highlights the nonocclusive thrombus at the level of the knee joint. (d) The patient has a history of a repaired abdominal aortic aneurysm with a bifurcated aortic graft, with the excluded aneurysm sac identified on CE-MRA partition images

with both the adventitia and the media, raising the possibility that congenital rests of mesenchymal cells may produce the cysts. The proximity to joints suggests that the cysts may reflect joint capsule pathology in an atypical location with adventitial cysts resulting from dissecting ganglion or synovial cysts. Alternatively, the cysts could form as a response to repeated adventitial injury. Finally, cysts could be a local manifestation of a systemic myxoid degenerative disorder; however, the typical unifocality of vessel involvement makes this hypothesis less likely. MRA with T2-weighted sequences is an eloquent technique to diagnose this entity. Contrast-enhanced angiography sequences demonstrate extrinsic smooth crescentic mass effect on the popliteal artery with associated stenosis. T2 spin echo sequences preferentially performed in the axial plane correlate the regions of stenosis to well-circumscribed T2 hyperintense lesions contained within the vessel wall. If bilateral popliteal pathology is identified on MRA, primary consideration of an alterative diagnosis should be sought. Treatment is anecdotal due to the rarity of the condition. Simple cyst aspiration does not generate a durable result

as the cysts recur. The preferred treatment is surgical cyst evacuation; as the cysts are superficial the popliteal artery can usually be preserved. Otherwise, bypass is the treatment of choice. If patients present with arterial thrombosis, attempts at lysis or open thrombectomy are warranted to attempt to preserve the popliteal artery.

Thromboangiitis Obliterans (Buerger’s Disease) Buerger’s disease is a noninflammatory vasculitis involving medium- and small-sized vessels in the lower extremities [30]. Both arteries and veins can be involved and this entity can also involve the upper extremities. Thrombangiitis obliterans refers to the histological findings of highly cellular thrombus completely occluding vessel lumens. Patients, predominantly male, present in young to middle age with bilateral claudication, normal proximal pulses, and dampened or absent distal pulses. A strong history of smoking is the rule. The prognosis of Buerger’s disease is closely influenced by the patient’s smoking status. Patients who are able to successfully quit

23

Lower Extremity Peripheral Arterial Disease

333

Fig. 23.9 Popliteal artery entrapment, Type III. 51-Year-old male presents with claudication, associated with numbness and tingling after a spinning class. (a) High-resolution coronal MIP contrast-enhanced bolus-chase LE-MRA in the neutral position demonstrates segmental occlusion of the popliteal artery with well-developed collaterals. No significant atherosclerotic disease is present; the right popliteal artery is medially deviated with a mild focal stenosis of the above-knee vessel. MIP frames from subtracted time-resolved MRA datasets in active

(b) plantar flexion and (c) dorsiflexion demonstrate dynamic popliteal stenosis. (d) Axial T1-weighted postcontrast spin echo images showing an accessory slip of the medial gastrocnemius muscle coursing between the right popliteal artery and vein to insert on the lateral femoral condyle (arrow). Note the normal course of the popliteal muscle anterior to the popliteal artery and the normal configuration of the left popliteal artery and vein

smoking are unlikely to progress to ulceration, gangrene, and amputation. The etiology of this disorder is unclear. However, the strong association with cigarette smoking and improvement in symptoms following smoking cessation suggest a patient susceptibility to a causative environmental factor in cigarettes. Raynaud’s phenomenon is reported commonly; migratory thrombophlebitis of superficial veins is also associated. The angiographic appearance of small tortuous vessels along the course of focally or segmentally occluded tibial vessels, sparing the popliteal artery, is highly suggestive of

Buerger’s Disease in a young smoker (Fig. 23.10). The angiographic appearance of these collateral vessels has been likened to a corkscrew. Similar angiographic findings are seen in the arteries of the forearm and hands. The principal angiographic differential diagnosis involves other vasculitidies including Raynaud’s disease. Treatment strategies rely first and foremost on encouraging smoking cessation. Symptomatic vasospasm is treated with calcium channel blockers. Patients who present with acute thrombosis may benefit from lysis. Surgical bypass is challenging due to the segmental and distally predominant distribution of disease.

334

J.D. Collins and T. Scanlon

Fig. 23.10 Buerger’s disease. 31-Year-old male with a smoking history presents with claudication. Subtracted MIP contrast-enhanced LE-MRA of the (a) pelvis, (b) thighs, and (c) calves. Bilateral distal segmental occlusions are noted with numerous tortuous corkscrew-like

collateral vessels. (d) Tortuous tibial and peroneal collaterals are better appreciated on time-resolved MRA secondary to separation of arterial and venous phases of vascular enhancement

Table 23.2 Typical sequence parameters for MR angiography at 1.5 T, using a parallel imaging factor of 2 and 0.75 × 0.75 partial Fourier undersampling Parameters Acquisition time (s) Spatial resolution (mm3) Field of view (cm) Partial Fourier Rectangularized FoV TR (ms) TE (ms) Flip angle Matrix Partitions Bandwith (Hz/Pixel)

Time-resolved calf MRA 74 1.7 × 1.1 × 1.3 500 0.75 × 0.75 75% 2.28 0.85 25 448 60 740

Protocols LE-MRA Runoff At our institution, lower extremity anatomy is divided into three stations: pelvis, thighs, and calves. A hybrid technique is performed, initially acquiring a time-resolved acquisition of the calves, followed by bolus-chase subtraction contrastenhanced angiography of the pelvis, thighs, and calves (Table 23.2). • Single-shot steady-state free precession scout images of the pelvis, thighs, and calves. • Time-resolved calf angiography (0.05 mmol/kg, 5 cm3/s): Short TR with temporal resolution 3.5–5 s, 14 datasets. In-line subtraction of each dataset is performed, using the first as the mask acquisition.

CE-MRA pelvis 14 1.3 × 1.0 × 1.6 500 0.75 × 0.75 68.8% 3.02 1.05 20 512 72 440

CE-MRA thighs and calves 11 1.6 × 1.0 × 1.3 500 0.75 × 0.75 68.8% 2.49 0.91 20 512 72 610

• Three-dimensional three-station Bolus-Chase angiography, three stations: – Pelvis, thighs, and calves acquired in separate stations. – Mask slabs are acquired, optimizing volume coverage, and ensuring overlap at slab edges. – Fluoro-prep or contrast-bolus timing to the distal aorta. – Contrast-enhanced MRA of the pelvis. – Contrast-enhanced MRA of the thighs. – Contrast-enhanced MRA of the calves (first acquisition). – Contrast-enhanced MRA of the calves (second acquisition). – In-line contrast-enhanced mask subtraction for each dataset. • Optional: Postcontrast T1-weighted imaging of the popliteal fossa in patients with suspected popliteal artery entrapment.

23

Lower Extremity Peripheral Arterial Disease

Special Scenarios Popliteal Entrapment This protocol limits the volume of coverage to the distal thigh, popliteal fossa, and proximal calf. Time-resolved and high-resolution gadolinium-enhanced MRA is combined with pre- and postcontrast soft tissue sequences. • Single-shot steady-state free precession scout images of the pelvis, thighs, and calves. • Precontrast fat-suppressed T2-weighted imaging: axial and sagittal planes. • Precontrast fat-suppressed T1-weighted imaging: axial plane. • Time-resolved MRA in dorsiflexion (0.05 mmol/kg): Coronal acquisition, 14 measures, field of view centered on popliteal fossa. • Time-resolved MRA in plantarflexion (0.05 mmol/kg): Identical imaging parameters as in the previous acquisition. • High-resolution contrast-enhanced MRA in neutral position (0.10 mmol/kg): Coronal acquisition, same field of view as time-resolved imaging. • Postcontrast T1-weighted imaging, axial plane LE-MRA in Renal Insufficiency This protocol utilized noncontrast angiography sequences to evaluate the lower extremities. Use this protocol in patients with a GFR < 60 mL/min/1.73 m2. • Single-shot steady-state free precession scout images of the pelvis, thighs, and calves. • QISS MRA: Starting caudally, coverage from the pedal vessels through the infrarenal abdominal aorta. • Optional: Sagittal single-shot bSSFP sequences through the infrarenal aorta. • Alternative MRA sequence: 3D bSSFP, acquired in the coronal plane.

References 1. Prince MR. Gadolinium-enhanced MR aortography. Radiology. 1994;191:155–164. 2. Parker DL, Tsuruda JS, Goodrich KC, Alexander AL, Buswell HR. Contrast-enhanced magnetic resonance angiography of cerebral arteries. A review. Invest Radiol. 1998;33:560–572. 3. Zenge MO, Vogt FM, Brauck K, et al. High-resolution continuously acquired peripheral MR angiography featuring partial parallel imaging GRAPPA. Magn Reson Med. 2006;56:859–865. 4. Maki JH, Wilson GJ, Eubank WB, Hoogeveen RM. Utilizing SENSE to achieve lower station sub-millimeter isotropic resolution and minimal venous enhancement in peripheral MR angiography. J Magn Reson Imaging. 2002;15:484–491. 5. Koktzoglou I, Edelman RR. Ghost magnetic resonance angiography. Magnetic Resonance in Medicine. 2009;61:1515–1519. 6. Du J, Carroll TJ, Wagner HJ, et al. Time-resolved, undersampled projection reconstruction imaging for high-resolution CE-MRA of the distal runoff vessels. Magn Reson Med. 2002;48:516–522.

335 7. Vigen KK, Peters DC, Grist TM, Block WF, Mistretta CA. Undersampled projection-reconstruction imaging for time-resolved contrast-enhanced imaging. Magn Reson Med. 2000;43:170–176. 8. Huang Y, Wright GA. Time-resolved MR angiography with limited projections. Magn Reson Med. 2007;58:316–325. 9. Pereles FS, Collins JD, Carr JC, et al. Accuracy of stepping-table lower extremity MR angiography with dual-level bolus timing and separate calf acquisition: hybrid peripheral MR angiography. Radiology. 2006;240:283–290. 10. Morasch MD, Collins J, Pereles FS, et al. Lower extremity stepping-table magnetic resonance angiography with multilevel contrast timing and segmented contrast infusion. J Vasc Surg. 2003;37:62–71. 11. Collins J, Scanlon T, Hodnett P, Carr JC. Hybrid calf magnetic resonance angiography using a time resolved technique. Accepted for Presentation at the Proceedings of the 18th Annual Meeting of the International Society of Magnetic Resonance in Medicine, Stockholm, Sweden. 2010. 12. Potthast S, Wilson GJ, Wang MS, Maki JH. Peripheral movingtable contrast-enhanced magnetic resonance angiography (CE-MRA) using a prototype 18-channel peripheral vascular coil and scanning parameters optimized to the patient’s individual hemodynamics. J Magn Reson Imaging. 2009;29:1106–1115. 13. Bellin MF, Vasile M, Morel-Precetti S. Currently used non-specific extracellular MR contrast media. European Radiology. 2003;13:2688–2698. 14. Voth M, Haneder S, Huck K, Gutfleisch A, Schoenberg SO, Michaely HJ. Peripheral magnetic resonance angiography with continuous table movement in combination with high spatial and temporal resolution time-resolved MRA With a total single dose (0.1 mmol/kg) of gadobutrol at 3.0 T. Invest Radiol. 2009;44: 627–633. 15. McCauley TR, Monib A, Dickey KW, et al. Peripheral vascular occlusive disease: accuracy and reliability of time-of-flight MR angiography. Radiology. 1994;192:351–357. 16. Steffens J, Link J, Muller-Hulsbeck S, Freund M, Brinkmann G, Heller M. Cardiac-gated two-dimensional phase-contrast MR angiography of lower extremity occlusive disease. Am. J. Roentgenol. 1997;169:749–754. 17. Gu T, Korosec FR, Block WF, et al. PC VIPR: A high-speed 3D phase-contrast method for flow quantification and high-resolution angiography. AJNR Am J Neuroradiol. 2005;26:743–749. 18. Scheffler K, Lehnhardt S. Principles and applications of balanced SSFP techniques. Eur Radiol. 2003;13:2409–2418. 19. Nishimura DG, Macovski A, Pauly JM, Conolly SM. MR angiography by selective inversion recovery. Magn Reson Med. 1987;4: 193–202. 20. Spuentrup E, Manning WJ, B√∂rnert P, Kissinger KV, Botnar RM, Stuber M. Renal arteries: navigator-gated balanced fast field-echo projection MR angiography with aortic spin labeling: initial experience. Radiology. 2002;225:589–596. 21. Miyazaki M, Takai H, Sugiura S, Wada H, Kuwahara R, Urata J. Peripheral MR angiography: separation of arteries from veins with flow-spoiled gradient pulses in electrocardiography-triggered threedimensional half-Fourier fast spin-echo Imaging1. Radiology. 2003;227:890–896. 22. Nakamura K, Kuroki K, Akiyoshi Y, Hiramine A, Miyazaki M, Matsufuji Y. Fresh blood imaging (FBI) of peripheral arteries: comparison with 16-detector row CT angiography. Presented at the Proceedings of the 14th Annual Meeting of the International Society of Magnetic Resonance in Medicine, Seattle, WA, 2006. 23. Fan Z, Bi X, Sheehan J, Carr J, Renate J, Li D. 3D peripheral subtraction MRA using flow-spoiled ECG-triggered balanced SSFP. J Cardiovasc Magn Reson. 2009;11(Suppl 1):P288. 24. Fan Z, Bi X, Zhou X, Zuehlsdorff S, Carr J, Li D. 3D nongadolinium-enhanced mra using flow-sensitive dephasing (fsd) prepared

336 balanced ssfp: identification of the optimal first-order gradient moment. J Cardiovasc Magn Reson. 2010;12(Suppl 1):O65. 25. Edelman RR, Sheehan JJ, Dunkle E, Schindler N, Carr JC, Koktzoglou I. Quiescent interval single shot unenhanced magnetic resonance angiography of peripheral vascular disease: technical considerations and clinical feasibility. Magn Reson Med: 2010;63:951–958. 26. Hodnett P, Collins J, Scanlon T, Sheehan J, Carr J, Edelman RR. Quiescent interval single shot lower extremity angiography: comparison with bolus-chase LE-MRA. Presented at: Proceedings of the 18th Annual Meeting of the International Society of Magnetic Resonance in Medicine, Stockholm, Sweden. 2010.

J.D. Collins and T. Scanlon 27. Stafford RB, Sabati M, Mahallati H, Frayne R. 3D non-contrastenhanced MR angiography with balanced steady-state free precession dixon method. Magn Reson Med. 2008;59:430–433. 28. Henry MF, Wilkins DC, Lambert AW. Popliteal artery entrapment syndrome. Curr Treat Options Cardiovasc Med. 2004;6:113–120. 29. Elias DA, White LM, Rubenstein JD, Christakis M, Merchant N. Clinical Evaluation and MR imaging features of popliteal artery entrapment and cystic adventitial disease. AJR Am J Roentgenol. 2003;180:627–632. 30. Olin JW. Thromboangiitis obliterans (Buerger’s disease). N Engl J Med. 2000;343:864–869.

Noninvasive Imaging for Coronary Artery Disease

24

Reza Nezafat, Susie N. Hong, Peng Hu, Mehdi Hedjazi Moghari, and Warren J. Manning

Coronary Artery Disease Coronary artery disease (CAD) is the largest killer of Americans. Approximately every 25 s, an American will have a coronary event, and approximately every minute, someone will die of one [1]. In 2010, an estimated 785,000 Americans will have a myocardial infarction, and ~470,000 will have a recurrent infarction. An additional 195,000 silent first myocardial infarctions occur each year. Catheter-based, diagnostic invasive X-ray coronary angiography remains the clinical “gold standard” for the diagnosis of significant (³50% diameter stenosis) CAD with over a million catheterbased X-ray coronary angiograms performed annually in the USA and higher volume in Europe. Although numerous noninvasive tests are available to help discriminate among those with and without significant angiographic disease, over 35% of patients referred for their initial elective catheter-based X-ray coronary angiography are found to have no significant stenosis. Even without significant CAD, these individuals remain exposed to the cost, inconvenience, and potential morbidity (vascular complications, exposure to both ionizing radiation and iodinated contrast) of X-ray angiography. Data

R. Nezafat, PhD () Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA S.N. Hong, MD • M.H. Moghari, PhD Cardiovascular Division, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA P. Hu, PhD Department of Radiology, Ronald Reagan Medical Center, Los Angeles, CA, USA W.J. Manning, MD Cardiovascular Division, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA Department of Radiology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA

also suggest that in selected high risk populations such as those patients with aortic valve stenosis, the incidence of subclinical stroke associated with retrograde catheter crossing of the stenotic valve may exceed 20% [2]. Percutaneous coronary intervention in single vessel disease is commonly performed to relieve symptoms or decrease pharmaceutical use, but the greatest impact on mortality occurs with mechanical intervention among patients with left main (LM) and multivessel CAD. Thus, it would be desirable to have a noninvasive method that allowed direct visualization of the proximal/mid native coronary vessels for the accurate identification/exclusion of LM/multivessel CAD.

Noninvasive Coronary Imaging: MRI vs. MDCT Alternative noninvasive coronary artery imaging methods include multidetector computed tomography (MDCT) and coronary magnetic resonance imaging (MRI). Advantages of coronary MDCT include rapid image acquisition as well as superior isotropic spatial resolution. Advantages of coronary MRI include the lack of ionizing radiation or need for iodinated contrast (thereby facilitating repeated or follow-up scanning) and smaller artifacts related to epicardial calcium. The lack of iodinated contrast attribute of coronary MRI is of particular importance for the CAD population due to the common coexistence of renal dysfunction, while the lack of ionizing radiation is advantageous for younger patients with atypical chest pain or CAD. Both groups are likely to undergo follow-up studies for the development of new disease or to monitor disease progression. Independent of coronary anatomy, cardiac MRI also provides for superior assessment of myocardial anatomy, perfusion and viability, thereby providing a comprehensive cardiac assessment. Comparisons of coronary MRI with MDCT are few. Data suggest accuracy of coronary MRI to be superior to 4 slice [3] and similar with 16 slice MDCT [4]. Current data comparing whole-heart coronary MRI with 40 and 64 slice MDCT indicate coronary MRI has similar accuracy with

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_24, © Springer Science+Business Media, LLC 2012

337

338

R. Nezafat et al.

Fig. 24.1 Reformatted whole-heart coronary MRI (left), and MDCT (center), and corresponding X-ray coronary angiogram (right) of the right and left coronary artery system. (a) Normal right coronary artery. (b) Normal left coronary artery. (c) Serial stenoses of the proximal and distal RCA. (arrows) (adapted from Pouleur et al. [5])

MDCT if only interpretable segments are considered and inferior to MDCT if all segments are included due to superior ability of MDCT to image segments [5]. An example comparison between coronary MRI and MDCT is shown in Fig. 24.1. The presence of epicardial calcium, a common finding in patient at high risk of CAD, decreases the sensitivity and specificity of MDCT [6], whereas coronary MRI does not appear to be affected by epicardial calcium [7], resulting in superior MRI results for patients with calcified plaque >100 Hounsfield units [8]. Recently, advances in MDCT hardware such as 320-slice or dual-source MDCT have resulted in improved image acquisition at reduced radiation dose [9–24]. However, there is no current head-tohead comparison between coronary MDCT acquired with these new state-of-the-art MDCT systems with coronary MRI methods.

Coronary MRI: Current Clinical Applications Anomalous Coronary Arteries The incidence of anomalous coronary arteries is low and reported to be less than 1% of live births and 0.17% in

asymptomatic children and adolescents referred for echocardiography [25, 26]. However, anomalous coronaries are thought to be the second most common cause of sudden cardiac death (SCD) [27–30] among young athletes based on autopsy series. SCD due to anomalous coronary arteries are mostly attributed to high risk, inter-arterial anomalies, including those which acutely bend and course between the pulmonary artery and aorta (Fig. 24.2) [31, 32]. The mechanism of SCD is thought to be due to ischemia secondary to the sharp angle of the aberrant artery, which is exacerbated and possibly “kinked” with exercise as it courses between the engorged aorta and pulmonary artery [33]. Coronary MRI is ideal in the evaluation and screening of these younger patients in which anomalous coronary arteries are suspected. In addition to being noninvasive, coronary MRI lacks ionizing radiation and does not require contrast media, which are all advantages over coronary MDCT and conventional X-ray angiography. Additionally, several studies have confirmed the excellent accuracy of coronary MRI [34–37]. Owing to the advantages of coronary MRI and its diagnostic accuracy, coronary MRI is recommended and deemed appropriate in patients suspected of anomalous coronary artery disease by both the American College of Cardiology and American Heart Association [38, 39].

24

Noninvasive Imaging for Coronary Artery Disease

339

Fig. 24.2 Free-breathing targeted 3D coronary MRI demonstrating an anomalous left main coronary artery (LMCA) originating from the right coronary cusp, just proximal to the right coronary artery (RCA). Ao aorta

Coronary Vasculitides Although coronary aneurysms are uncommon, the vast majority are due to mucocutaneous lymph node syndrome (Kawasaki’s disease). Kawasaki’s disease is one of the most common vasculitides of childhood, often afflicting children younger than 5 years [40–43]. Aneurysms due to Kawasaki’s are a source of both short and long term morbidity and mortality [44]. Coronary MRI studies have confirmed the high accuracy of coronary CMR for both the identification and the characterization (diameter/length) of these aneurysms [45–47]. Similar data have been reported for ectatic coronary arteries and fistulas [48]. Coronary MRI may be ideal for the evaluation of patients suspected of coronary vasculitides, such as Takayasu arteritis and Behçet Syndrome (example shown in Fig. 24.3), given their tendencies to involve other large vessels (e.g. aorta, pulmonary vasculature) and the myocardium [49, 50]. Several cases using cardiac MR have already been reported in the characterization of Takayasu arteritis with coronary involvement in the literature [51–53].

Fig. 24.3 Steady-state free precession left ventricular outflow tract view demonstrating a large left anterior descending coronary aneurysm in a patient with Behçet Syndrome

Fig. 24.4 Steady-state free precession 4-chamber view demonstrating a large saphenous venous graft to RCA aneurysm with multiple thrombi

Coronary Artery Bypass Graft Assessment Internal mammary artery graft, reverse saphenous veins, and their complications can be imaged robustly using coronary MRI due to the larger graft diameter and relatively low respiratory and cardiac motion compared to naive coronary arteries [54–59]. Free-breathing or breath-hold 2D and 3D spin echo or gradient echo imaging sequences, with or without contrast media, have been reported to asses bypass graft patency. Figure 24.4 shows an example SSFP image demonstrating a

large saphenous vein graft to RCA aneurysm. Sensitivity of 86–100% and specificity of 59–100% have been reported [54–59]. The contrast-enhanced 3D approaches yield better sensitivity and specificity for assessing the patency of the graft. The limitations of coronary MRI bypass graft assessment include imaging artifacts due to metallic implants and inability to identify severely diseased yet patent graft. In addition to the coronary MRI, CMR perfusion and late gadolinium enhancement (LGE) have been advocated for guiding

340

R. Nezafat et al. Table 24.1 Single-center ECG-triggered, free-breathing, targeted 3D and whole-heart coronary MRI with and without contrast agents

Fig. 24.5 SVG aneurysm with first-pass perfusion demonstrating ischemia (broken arrows) in the inferior septal walls due to the large SVGto-RCA aneurysm (solid arrows)

revascularization for patients with bypass graft [60]. In the example first-pass perfusion image shown in Fig. 24.5, a large saphenous vein graft to RCA aneurysm results in ischemia in the inferior septal walls.

Study Noncontrast 3D targeted coronary MRI Bunce et al. [134] Sommer et al. [135] Bogaert et al. [136] Noncontrast 3D whole-heart coronary MRI Jahnke et al. [137] Sakuma [78] Sakuma [77] Pouleur [5] Contrast-enhanced 3D whole-heart coronary MRI Yang et al. [132]

# patients

Sensitivity

Specificity

46 107 21

50–89% 74–88% 85–92%

72–100% 63–91% 50–83%

21 39 131 77

79% 82% 82% 100%

91% 91% 90% 72%

62

94%

82%

Coronary MRI: Technical Advances and Impediments Over the past two decades, coronary MRI has made considerable technical improvements. The early approaches of 2D breath-hold electrocardiogram (ECG) triggered coronary MRI [61, 62] have largely been replaced by 3D free-breathing approaches to enable greater anatomical coverage and higher signal level. 3D coronary MRI can be performed using targeted or whole-heart approaches. In the targeted approach [63], a double-oblique 3D volume (~30-mm thick) aligned along the major axis of the left or right coronary artery is acquired [64–66]. Whole-heart coronary MRI [7, 67–80], an approach somewhat analogous to coronary MDCT in that an axial 3D volume encompassing the entire heart is sampled in a single acquisition, provides ease of imaging slab prescription and more complete anatomical coverage. However, it has not been shown to be superior for CAD assessment compared to the targeted approach based on single-center trials up to date (Table 24.1). Along with the transition from 2D to 3D is the introduction of the steady state free precession (SSFP) sequence for coronary MRI [81]. Compared to gradient recalled echo (GRE) sequence, SSFP provides intrinsically higher signal due to its balanced gradients and improved blood–myocardium contrast due to its partial T2 weighting. Both GRE and SSFP are used for targeted 3D coronary MRI and the diagnostic accuracy of both for CAD has been shown to be similar [82, 83]. For whole-heart noncontrast coronary MRI at 1.5 T, SSFP appears to be the sequence of choice due to its higher blood–myocardium contrast and superior inflow properties [84]. Despite these technical advances, coronary MRI is yet to be widely accepted for clinical use due to prolonged scan time and image quality issues due to coronary artery motion, suboptimal signal-to-noise (SNR) and

contrast-to-noise (CNR) ratios. Contributing factors to these impediments include the near constant motion of the coronary arteries during the cardiac and respiratory cycles, their small caliber (3–6 mm diameter), high level of tortuosity, and signal from adjacent epicardial fat and myocardium.

Coronary Motion The magnitude of motion from cardiac contraction, diaphragmatic/chest wall motion, and upper torso motion may greatly exceed the coronary artery diameter, thereby leading to blurring artifacts in the absence of motion-suppressive methods such as ECG triggering, navigator gating, or breathholding. Various cardiac and respiratory motion compensation techniques have been developed over the past two decades [85]. For cardiac motion, a patient-specific rest period during the cardiac cycle is commonly employed [86– 88], along with arrhythmia rejection [1]. For suppression of respiratory motion, breath-holding is the most commonly used method for short scans such as cine MRI. However, the short breath-holding duration limits spatial resolution and anatomic coverage. In addition, many patients are unable to adequately sustain a breath-hold. Respiratory navigators, proposed over a decade ago, have been refined to become the most widely used respiratory motion compensation technique for coronary MRI [65, 89–93]. In the most common implementation of this technique, a navigator echo, which tracks the motion of the diaphragm in the superior–inferior direction, is sampled in each cardiac cycle immediately before imaging data acquisition. An empirical superior–inferior tracking factor of 0.6 (i.e., the ratio between coronary and

24

Noninvasive Imaging for Coronary Artery Disease

341

Fig. 24.6 Example reformatted 3D coronary MRI of right coronary artery (a) acquired with no T2-Prep; (b) with T2-Prep; (c) with adiabatic T2-Prep. Arrows in (b) point to the artifacts resulting from T2-Prep

sequence; note the suppression of the banding artifacts in (c) and also the homogeneity of the signal

diaphragmatic motion) is commonly used [94–96]. Refinements of the navigator method include the use of multiple navigator locations, use of leading and trailing navigators, and navigators that provide guidance for affine transformations, i.e., 3D translations and rotations, of the slice prescription for each heartbeat [74, 91, 97, 98]. The affine transformation permits use of larger navigator windows, and hence higher navigator efficiency. It has also been proposed that the heart itself be tracked [72, 99–101]. For this reason, methods such as the fat navigator have been developed, in which epicardial fat is tracked to detect the heart position [102–105]. In addition to navigator gating, respiratory self-gating techniques have been investigated, a method that derive the respiratory position of the heart from the imaging data itself [71, 72, 79, 99, 100, 106, 107], thus avoiding certain issues with navigators such as subject-dependent tracking factor [96] and hysteresis effects [108]. Novel k-space trajectories and various image reconstruction based method such as cross-correlation of low resolution images have also been proposed for respiratory motion compensation [79, 109–111].

fields of 3 T and 7 T may be advantageous due to the intrinsic higher SNR available at higher magnetic field strengths. However, imaging at these higher field strengths is partially limited due to issues with higher B1 and B0 inhomogeneity and higher specific absorption rate. For example, coronary MRI at 3 T using SSFP sequences is challenging due to increased field inhomogeneity and high specific absorption rate. Therefore, localized shimming [119] and gradient echo imaging sequences, which are less sensitive to field inhomogeneity, have been advocated. Additionally improved preparation sequences such as adiabatic T2 magnetization preparation [115, 120] and adiabatic fat saturation have been used to reduce the impact of B1 inhomogeneity at the high field strengths. Figure 24.6 shows example coronary MR images acquired at 3 T using improved T2 magnetization preparation which suppresses the banding artifact resulting from conventional T2 magnetization preparation. Despite these technical improvements, there are currently no multicenter data on a head-to-head comparison between 3 T and 1.5 T coronary MRI for diagnosing CAD. Coronary MRI at even higher field strengths, such as 7 T [121], is even more challenging. Figure 24.7 shows an example coronary MRI from a healthy subject acquired at 7 T. Several technical issues, including coil design, motion compensation, and B0 and B1 field inhomogeneity, need to be addressed before clinical evaluation is possible. Whole-heart coronary MRI can potentially result in higher SNR but have long scan time. Several studies show excellent image quality of whole-heart coronary MRI [77, 80]. Figure 24.8 shows an example whole-heart coronary image from a single-center study [77]. However, the SNR gain is counteracted by the need for parallel imaging to reduce scan time, which carries an SNR penalty. Furthermore, whole-heart imaging suffers from saturation effects of the inflowing blood magnetization [84]. Coronary MRI SNR can also be increased by administration of vasodilators because the increased coronary blood flow secondary to vasodilatation reduces the inflow saturation effects [122, 123]. Figure 24.9 demonstrates impact of sublingual isosorbide dinitrate administration on 3D targeted coronary MRI up to 30 min after drug administration in

SNR and CNR The coronary arteries are surrounded by epicardial fat and myocardium. It is therefore desirable to suppress the fat and myocardium signal to improve coronary artery CNR. A traditional spectrally selective fat saturation prepulse is commonly performed immediately before the imaging data acquisition to suppress fat signal. Endogenous contrast preparation techniques, such as T2 magnetization preparation [112–115] or magnetization transfer [116, 117], are commonly used to improve the CNR between blood and myocardium at 1.5 T. Owing to the high spatial resolution required to adequately depict the coronary vascular tree, coronary MRI is limited by SNR. The SNR of coronary MRI can be enhanced by higher B0 field strength [118], larger 3D spatial coverage [80], vasodilator administration, and contrast agents based on gadolinium chelates. Noncontrast coronary MRI at high magnetic

342

R. Nezafat et al.

Fig. 24.7 7T coronary MRI of the right coronary artery acquired using a gradient echo imaging sequence. Proximal (a, c) and more distal (b) segments of the RCA are visualized (adapted from van Elderen [121])

Fig. 24.8 Whole-heart coronary MRI. (a) Curved multiplanar reconstruction image shows a stenosis in LAD (white arrow). (b) Volume-rendered image demonstrates three-dimensional view of LAD

with stenosis. (c) X-ray coronary angiography confirms proximal LAD stenosis (arrowhead) (adapted from Sakuma et al. [77])

Fig. 24.9 Targeted 3D coronary MRI of the RCA acquired before and after sublingual isosorbide dinitrate administration on two healthy subjects using a 3D free-breathing SSFP with 2.5 mg (top row) and 5 mg doses (bottom row) as a function of time. Improved RCA vasodilation

and signal enhancement can be observed in all postdrug images (arrows in top and bottom row). As a result of enhanced SNR, distal segments are visualized better with isosorbide dinitrate

terms of subjective image quality and objective SNR and vessel sharpness. An alternative flow-independent approach is administration of exogenous gadolinium contrast agents that shorten the T1 relaxation time. Both extracellular [68, 124, 125] and intravascular [126–131] MRI contrast agents have been reported. Conventional extracellular contrast agents,

i.e., gadopentetate dimeglumine (Gd-DTPA), diffuse rapidly into the interstitial space. Therefore, early contrast-enhanced coronary MRI studies focused on breath-hold coronary MRI to take advantage of the first-passage of these agents [125]. However, the short breath-hold and first-pass time limits spatial resolution and is not suitable for thick-slab whole-heart coronary acquisitions [80]. With availability of gadobenate

24

Noninvasive Imaging for Coronary Artery Disease

343

Fig. 24.11 An example (a) 3D volume rendered and (b) corresponding reformatted whole-heart SSFP (3.6/1.8; flip angle, 90°) coronary image acquired with a bolus injection of Gd-BOPTA. All three major coronary arteries and distal branches are clearly depicted. RCA right coronary artery, LAD left anterior descending, LCX left circumflex

Fig. 24.10 Contrast-enhanced 3D whole-heart coronary MRI using a slow infusion of Gd-BOPTA contrast agent in a patient with atypical chest pain. (a, b) Contrast-enhanced whole-heart CMRA maximum intensity projection images show a significant stenosis in the proximal LCX and a nonsignificant stenosis in the middle RCA (arrows), respectively. (c, d) The volume-rendered images have the same findings in LCX and RCA (arrows), which were consistent (e, f) with the findings (arrows) of conventional coronary angiography. AO aorta, OM obtuse marginal artery (adapted from Yang et al. [132])

dimeglumine (Gd-BOPTA; MultiHance; Bracco Imaging SpA, Milan, Italy), a high relaxivity extracellular contrast agent, improved whole-heart coronary MRI at 3 T is feasible using an inversion recovery (IR) GRE sequence with a slow infusion of this Gd-BOPTA [68]. Figure 24.10 shows an example contrast-enhanced whole-heart coronary MR image from a CAD patient and the corresponding X-ray angiogram, demonstrating the agreement between two modalities in detecting significant stenosis. When compared with X-ray angiography in a single center trial, 3 T whole-heart coronary MRA with slow infusion of Gd-BOPTA had 93% sensitivity,

89% specificity, and 90% accuracy for detecting >50% diameter stenosis on a per-vessel basis (and 94, 82, and 89% on a per-patient basis) [132]. Coronary MRI with a bolus infusion of Gd-BOPTA has been recently reported [133]. In the example shown in Fig. 24.11, whole-heart coronary images are acquired after a bolus injection of Gd-BOPTA and the three major coronary vessels are clearly depicted in the reformatted and 3D volume rendered images. A potential advantage of this approach is that its contrast injection method is compatible with LGE, making it possible to assess coronary artery stenosis and myocardium viability using a single bolus contrast injection. The bolus injection approach also simplifies the initiation time of coronary MRI acquisition compared to slow infusion. Despite the tremendous technical improvements in the last two decades, the sensitivity and specificity of coronary MRI for detection of CAD remain moderate based on singlecenter [5, 77, 78, 132, 134–137] (Table 24.1) and multicenter [63] studies. Coronary motion, SNR and CNR will remain major impediments to coronary MRI in the foreseeable future and these issues need to be addressed before clinical prime time for coronary MRI.

Coronary Vein MRI Knowledge of coronary vein anatomy is becoming increasingly important for diagnostic and interventional cardiac procedures including epicardial radiofrequency ablation [138, 139], retrograde perfusion therapy in high-risk or complicated coronary angioplasty [140], arrhythmia assessment [141, 142], stem cell delivery [143], coronary artery bypass surgery [144], and cardiac resynchronization therapy (CRT) [145, 146]. In patients with severe congestive heart failure, CRT has been proven as an adjuvant therapy to pharmacological treatment [147, 148]. In CRT, simultaneous pacing of the right ventricle and left ventricle (LV), or pacing the LV

344

R. Nezafat et al.

Fig. 24.12 The variations in the coronary venous anatomy can be observed in images from four healthy adult subjects acquired using magnetization transfer GRE during the systolic rest period. There are clear variations in the branching point, angle, and diameter of

different tributaries of coronary sinus, highlighting the potential for noninvasive assessment of the coronary venous anatomy in cardiac resynchronization therapy. For example, subject (a) has no visible lateral vein (LatV)

alone, results in hemodynamic improvement and restoration of a more physiological contraction pattern [149]. One of the technical difficulties of CRTs is achieving effective, safe and permanent pacing of the LV. Transvenous coronary sinus pacing is the most common technique as it has the least procedural risk, but it is associated with long procedure times, extensive radiation exposure from fluoroscopy, implantation failure and LV lead dislodgment. Two of the major difficulties of the transvenous approach are the small number of coronary vein branches adjacent to an appropriate LV wall and the great variability in coronary vein anatomy [146]. Ideally, coronary venous morphology should be assessed

noninvasively prior to CRT procedure, to determine whether epicardial or transvenous lead placement would be more appropriate. Coronary artery MRI techniques developed over the past two decades are applicable to imaging coronary veins except for magnetization preparation methods and optimal time window for imaging within the cardiac cycle. Magnetization transfer preparation sequences have been demonstrated as an alternative to T2 magnetization preparation, commonly used in coronary MRI, for both targeted [150] and whole-heart [151] approaches. Figure 24.12 shows example coronary vein MRI using a targeted approach with a magnetization

24

Noninvasive Imaging for Coronary Artery Disease

preparation sequence. Use of intravascular contrast agents such as gadocoletic acid trisodium salt [152] has also been reported to improve contrast in coronary vein MRI. With high relaxivity extracellular contrast agents such as Gd-BOPTA, coronary vein can also be easily visualized, similar to coronary arteries. While coronary artery MRI is commonly performed during mid-diastolic quiescent period, coronary vein MRI should be in the end-systolic quiescent period, as it coincides with the maximum size of the coronary veins [117].

References 1. Lloyd-Jones D, Adams RJ, Brown TM, et al. Heart disease and stroke statistics--2010 update: a report from the American Heart Association. Circulation. 2010;121:e46–e215. 2. Omran H, Schmidt H, Hackenbroch M, et al. Silent and apparent cerebral embolism after retrograde catheterisation of the aortic valve in valvular stenosis: a prospective, randomised study. Lancet. 2003;361:1241–1246. 3. Gerber BL, Coche E, Pasquet A, et al. Coronary artery stenosis: direct comparison of four-section multi-detector row CT and 3D navigator MR imaging for detection – initial results. Radiology. 2005;234:98. 4. Kefer J, Coche E, Legros G, et al. Head-to-head comparison of three-dimensional navigator-gated magnetic resonance imaging and 16-slice computed tomography to detect coronary artery stenosis in patients. J Am Coll Cardiol. 2005;46:92–100. 5. Pouleur AC, le Polain de Waroux JB, Kefer J, Pasquet A, Vanoverschelde JL, Gerber BL. Direct comparison of whole-heart navigator-gated magnetic resonance coronary angiography and 40and 64-slice multidetector row computed tomography to detect the coronary artery stenosis in patients scheduled for conventional coronary angiography. Circ Cardiovasc Imaging. 2008;1:114–121. 6. Raff GL, Gallagher MJ, O’Neill WW, Goldstein JA. Diagnostic accuracy of noninvasive coronary angiography using 64-slice spiral computed tomography. J Am Coll Cardiol. 2005;46:552–557. 7. Maintz D, Ozgun M, Hoffmeier A, et al. Whole-heart coronary magnetic resonance angiography: value for the detection of coronary artery stenoses in comparison to multislice computed tomography angiography. Acta Radiol. 2007;48:967–973. 8. Liu X, Zhao X, Huang J, et al. Comparison of 3D free-breathing coronary MR angiography and 64-MDCT angiography for detection of coronary stenosis in patients with high calcium scores. AJR Am J Roentgenol. 2007;189:1326–1332. 9. Choi SI, George RT, Schuleri KH, Chun EJ, Lima JA, Lardo AC. Recent developments in wide-detector cardiac computed tomography. Int J Cardiovasc Imaging. 2009;25(Suppl 1):23–29. 10. Hein PA, Romano VC, Lembcke A, May J, Rogalla P. Initial experience with a chest pain protocol using 320-slice volume MDCT. Eur Radiol. 2009;19:1148–1155. 11. Hoe J, Toh KH. First experience with 320-row multidetector CT coronary angiography scanning with prospective electrocardiogram gating to reduce radiation dose. J Cardiovasc Comput Tomogr. 2009;3:257–261. 12. Pasricha SS, Nandurkar D, Seneviratne SK, et al. Image quality of coronary 320-MDCT in patients with atrial fibrillation: initial experience. AJR Am J Roentgenol. 2009;193:1514–1521. 13. Rixe J, Conradi G, Rolf A, et al. Radiation dose exposure of computed tomography coronary angiography: comparison of dualsource, 16-slice and 64-slice CT. Heart. 2009;95:1337–1342.

345 14. Achenbach S, Anders K, Kalender WA. Dual-source cardiac computed tomography: image quality and dose considerations. Eur Radiol. 2008;18:1188–1198. 15. Achenbach S, Marwan M, Ropers D, et al. Coronary computed tomography angiography with a consistent dose below 1 mSv using prospectively electrocardiogram-triggered high-pitch spiral acquisition. Eur Heart J. 2010;31:340–346. 16. Achenbach S, Ropers D, Kuettner A, et al. Contrast-enhanced coronary artery visualization by dual-source computed tomography-initial experience. Eur J Radiol. 2006;57:331–335. 17. Achenbach S, Ropers U, Kuettner A, et al. Randomized comparison of 64-slice single- and dual-source computed tomography coronary angiography for the detection of coronary artery disease. JACC Cardiovasc Imaging. 2008;1:177–186. 18. Kuettner A, Gehann B, Spolnik J, et al. Strategies for dose-optimized imaging in pediatric cardiac dual source CT. Rofo. 2009; 181:339–348. 19. Lell M, Marwan M, Schepis T, et al. Prospectively ECG-triggered high-pitch spiral acquisition for coronary CT angiography using dual source CT: technique and initial experience. Eur Radiol. 2009;19:2576–2583. 20. Marwan M, Ropers D, Pflederer T, Daniel WG, Achenbach S. Clinical characteristics of patients with obstructive coronary lesions in the absence of coronary calcification: an evaluation by coronary CT angiography. Heart. 2009;95:1056–1060. 21. Pflederer T, Jakstat J, Marwan M, et al. Radiation exposure and image quality in staged low-dose protocols for coronary dualsource CT angiography: a randomized comparison. Eur Radiol. 2010;20:1197–1206. 22. Schepis T, Marwan M, Pflederer T, et al. Quantification of noncalcified coronary atherosclerotic plaques with dual source computed tomography: comparison to intravascular ultrasound. Heart. 2010;96: 610–615. 23. Blankstein R, Shah A, Pale R, et al. Radiation dose and image quality of prospective triggering with dual-source cardiac computed tomography. Am J Cardiol. 2009;103:1168–1173. 24. Kalra MK, Brady TJ. Current status and future directions in technical developments of cardiac computed tomography. J Cardiovasc Comput Tomogr. 2008;2:71–80. 25. Kimbiris D, Iskandrian AS, Segal BL, Bemis CE. Anomalous aortic origin of coronary arteries. Circulation. 1978;58:606–615. 26. Davis JA, Cecchin F, Jones TK, Portman MA. Major coronary artery anomalies in a pediatric population: incidence and clinical importance. J Am Coll Cardiol. 2001;37:593–597. 27. Maron BJ, Thompson PD, Ackerman MJ, et al. Recommendations and considerations related to preparticipation screening for cardiovascular abnormalities in competitive athletes: 2007 update: a scientific statement from the American Heart Association Council on Nutrition, Physical Activity, and Metabolism: endorsed by the American College of Cardiology Foundation. Circulation. 2007;115:1643–1455. 28. Maron BJ, Carney KP, Lever HM, et al. Relationship of race to sudden cardiac death in competitive athletes with hypertrophic cardiomyopathy. J Am Coll Cardiol. 2003;41:974–980. 29. Corrado D, Basso C, Schiavon M, Thiene G. Screening for hypertrophic cardiomyopathy in young athletes. N Engl J Med. 1998; 339:364–369. 30. Eckart RE, Scoville SL, Campbell CL, et al. Sudden death in young adults: a 25-year review of autopsies in military recruits. Ann Int Med. 2004;141:829–834. 31. Taylor AJ, Rogan KM, Virmani R. Sudden cardiac death associated with isolated congenital coronary artery anomalies. J Am Coll Cardiol. 1992;20:640–647. 32. Taylor AJ, Byers JP, Cheitlin MD, Virmani R. Anomalous right or left coronary artery from the contralateral coronary sinus: “highrisk” abnormalities in the initial coronary artery course and heterogeneous clinical outcomes. Am Heart J. 1997;133:428–435.

346 33. Cheitlin MD, De Castro CM, McAllister HA. Sudden death as a complication of anomalous left coronary origin from the anterior sinus of Valsalva, a not-so-minor congenital anomaly. Circulation. 1974;50:780–787. 34. McConnell MV, Ganz P, Selwyn AP, Li W, Edelman RR, Manning WJ. Identification of anomalous coronary arteries and their anatomic course by magnetic resonance coronary angiography. Circulation. 1995;92:3158–3162. 35. Post JC, van Rossum AC, Bronzwaer JG, et al. Magnetic resonance angiography of anomalous coronary arteries. A new gold standard for delineating the proximal course? Circulation. 1995;92:3163–3171. 36. Taylor AM, Thorne SA, Rubens MB, et al. Coronary artery imaging in grown up congenital heart disease: complementary role of magnetic resonance and x-ray coronary angiography. Circulation. 2000;101:1670–1678. 37. Bunce NH, Lorenz CH, Keegan J, et al. Coronary artery anomalies: assessment with free-breathing three-dimensional coronary MR angiography. Radiology. 2003;227:201–208. 38. Hendel RC, Patel MR, Kramer CM, et al. ACCF/ACR/SCCT/ SCMR/ASNC/NASCI/SCAI/SIR 2006 appropriateness criteria for cardiac computed tomography and cardiac magnetic resonance imaging: a report of the American College of Cardiology Foundation Quality Strategic Directions Committee Appropriateness Criteria Working Group, American College of Radiology, Society of Cardiovascular Computed Tomography, Society for Cardiovascular Magnetic Resonance, American Society of Nuclear Cardiology, North American Society for Cardiac Imaging, Society for Cardiovascular Angiography and Interventions, and Society of Interventional Radiology. J Am Coll Cardiol. 2006;48:1475–1497. 39. Hundley WG, Bluemke DA, Finn JP, et al. ACCF/ACR/AHA/ NASCI/SCMR 2010 expert consensus document on cardiovascular magnetic resonance: a report of the American College of Cardiology Foundation Task Force on Expert Consensus Documents. Circulation. 2010;55:2614–2662. 40. Burns JC, Glode MP. Kawasaki syndrome. Lancet. 2004;364: 533–544. 41. Holman RC, Curns AT, Belay ED, Steiner CA, Schonberger LB. Kawasaki syndrome hospitalizations in the United States, 1997 and 2000. Pediatrics. 2003;112(3 Pt 1):495–501. 42. Du ZD, Zhao D, Du J, et al. Epidemiologic study on Kawasaki disease in Beijing from 2000 through 2004. Pediatr Infect Dis J. 2007;26:449–451. 43. Burns JC, Cayan DR, Tong G, et al. Seasonality and temporal clustering of Kawasaki syndrome. Epidemiology 2005;16:220–225. 44. Akagi T, Rose V, Benson LN, Newman A, Freedom RM. Outcome of coronary artery aneurysms after Kawasaki disease. J Pediatr. 1992;121(5 Pt 1):689–694. 45. Greil GF, Stuber M, Botnar RM, et al. Coronary magnetic resonance angiography in adolescents and young adults with Kawasaki disease. Circulation. 2002;105:908–911. 46. Mavrogeni S, Papadopoulos G, Douskou M, et al. Magnetic resonance angiography, function and viability evaluation in patients with Kawasaki disease. J Cardiovasc Magn Reson. 2006;8:493–498. 47. Mavrogeni S, Papadopoulos G, Douskou M, et al. Magnetic resonance angiography is equivalent to X-ray coronary angiography for the evaluation of coronary arteries in Kawasaki disease. J Am Coll Cardiol. 2004;43(4):649–652. 48. Mavrogeni SI, Manginas A, Papadakis E, et al. Correlation between magnetic resonance angiography (MRA) and quantitative coronary angiography (QCA) in ectatic coronary vessels. J Cardiovasc Magn Reson. 2004;6:17–23. 49. Harrison A, Abolhoda A, Ahsan C. Cardiovascular complications in Behcet syndrome: acute myocardial infarction with late stent thrombosis and coronary, ventricular, and femoral pseudoaneurysms. Tex Heart Inst J. 2009;36:498–500.

R. Nezafat et al. 50. Yakut ZI, Odev K. Pulmonary and cardiac involvement in Behcet disease: 3 case reports. Clin Appl Thromb Hemost. 2007;13(3): 318–322. 51. Kanaan S, Baker C, Starnes V. Resection of giant coronary artery aneurysms in a Takayasu’s arteritis patient. Ann Thorac Surg. 2008;85:1795–1796. 52. Malik IS, Harare O, AL-Nahhas A, Beatt K, Mason J. Takayasu’s arteritis: management of left main stem stenosis. Heart. 2003;89:e9. 53. Khalaf HH, Arafah MR, Refaat AA, Ibrahim MF. Coronary artery bypass grafting for Takayasu arteritis with severe coronary, carotid, subclavian, and renal artery involvement and subsequent pregnancy. Interact Cardiovasc Thorac Surg. 2006;5:153–155. 54. White R, Caputo G, Mark A, Modin G, Higgins C. Coronary artery bypass graft patency: noninvasive evaluation with MR imaging. Radiology. 1987;164:681–686. 55. White R, Pflugfelder P, Lipton M, Higgins C. Coronary artery bypass grafts: evaluation of patency with cine MR imaging. AJR Am J Roentgenol. 1988;150:1271–1274. 56. Rubinstein R, Askenase A, Thickman D, Feldman M, Agarwal J, Helfant R. Magnetic resonance imaging to evaluate patency of aortocoronary bypass grafts. Circulation. 1987;76:786–791. 57. Wintersperger B, Engelmann M, Von Smekal A, et al. Patency of coronary bypass grafts: assessment with breath-hold contrastenhanced MR angiography--value of a non-electrocardiographically triggered technique. Radiology. 1998;208:345–351. 58. Engelmann M, Knez A, von Smekal A, et al. Non-invasive coronary bypass graft imaging after multivessel revascularisation. Int J Cardiol. 2000;76:65–74. 59. Vrachliotis T, Bis K, Aliabadi D, Shetty A, Safian R, Simonetti O. Contrast-enhanced breath-hold MR angiography for evaluating patency of coronary artery bypass grafts. AJR Am J Roentgenol. 1997;168:1073–1080. 60. Klein C, Nagel E, Gebker R, et al. Magnetic resonance adenosine perfusion imaging in patients after coronary artery bypass graft surgery. J Am Coll Cardiol Img. 2009;2:437–445. 61. Manning WJ, Edelman RR. Magnetic resonance coronary angiography. Magn Reson Q. 1993;9:131–151. 62. Manning WJ, Li W, Edelman RR. A preliminary report comparing magnetic resonance coronary angiography with conventional angiography. N Engl J Med. 1993;328:828–832. 63. Kim WY, Danias PG, Stuber M, et al. Coronary magnetic resonance angiography for the detection of coronary stenoses. N Engl J Med. 2001;345:1863–1869. 64. Stuber M, Botnar RM, Danias PG, Kissinger KV, Manning WJ. Breathhold three-dimensional coronary magnetic resonance angiography using real-time navigator technology. J Cardiovasc Magn Reson. 1999;1:233–238. 65. Stuber M, Botnar RM, Danias PG, Kissinger KV, Manning WJ. Submillimeter three-dimensional coronary MR angiography with real-time navigator correction: comparison of navigator locations. Radiology. 1999;212:579–587. 66. Stuber M, Botnar RM, Danias PG, et al. Double-oblique freebreathing high resolution three-dimensional coronary magnetic resonance angiography. J Am Coll Cardiol. 1999;34:524–531. 67. Bhat H, Zuehlsdorff S, Bi X, Li D. Whole-heart contrast-enhanced coronary magnetic resonance angiography using gradient echo interleaved EPI. Magn Reson Med. 2009;61:1388–1395. 68. Bi X, Carr JC, Li D. Whole-heart coronary magnetic resonance angiography at 3 Tesla in 5 minutes with slow infusion of Gd-BOPTA, a high-relaxivity clinical contrast agent. Magn Reson Med. 2007;58:1–7. 69. Bi X, Deshpande V, Carr J, Li D. Coronary artery magnetic resonance angiography (MRA): a comparison between the whole-heart and volume-targeted methods using a T2-prepared SSFP sequence. J Cardiovasc Magn Reson. 2006;8:703–707.

24

Noninvasive Imaging for Coronary Artery Disease

70. Kim YJ, Seo JS, Choi BW, Choe KO, Jang Y, Ko YG. Feasibility and diagnostic accuracy of whole heart coronary MR angiography using free-breathing 3D balanced turbo-field-echo with SENSE and the half-fourier acquisition technique. Korean J Radiol. 2006;7:235–242. 71. Lai P, Bi X, Jerecic R, Li D. A respiratory self-gating technique with 3D-translation compensation for free-breathing whole-heart coronary MRA. Magn Reson Med. 2009;62:731–738. 72. Lai P, Larson AC, Bi X, Jerecic R, Li D. A dual-projection respiratory self-gating technique for whole-heart coronary MRA. J Magn Reson Imaging. 2008;28:612–620. 73. Liu X, Bi X, Huang J, Jerecic R, Carr J, Li D. Contrast-enhanced whole-heart coronary magnetic resonance angiography at 3.0 T: comparison with steady-state free precession technique at 1.5 T. Invest Radiol. 2008;43:663–668. 74. Nehrke K, Bornert P, Mazurkewitz P, Winkelmann R, Grasslin I. Free-breathing whole-heart coronary MR angiography on a clinical scanner in four minutes. J Magn Reson Imaging. 2006;23:752–756. 75. Niendorf T, Hardy CJ, Giaquinto RO, et al. Toward single breathhold whole-heart coverage coronary MRA using highly accelerated parallel imaging with a 32-channel MR system. Magn Reson Med. 2006;56:167–176. 76. Okada T, Kanao S, Ninomiya A, et al. Whole-heart coronary magnetic resonance angiography with parallel imaging: comparison of acceleration in one-dimension vs. two-dimensions. Eur J Radiol. 2009;71:486–491. 77. Sakuma H, Ichikawa Y, Chino S, Hirano T, Makino K, Takeda K. Detection of coronary artery stenosis with whole-heart coronary magnetic resonance angiography. J Am Coll Cardiol. 2006;48: 1946–1950. 78. Sakuma H, Ichikawa Y, Suzawa N, et al. Assessment of coronary arteries with total study time of less than 30 minutes by using wholeheart coronary MR angiography. Radiology. 2005;237:316–321. 79. Stehning C, Bornert P, Nehrke K, Eggers H, Stuber M. Free-breathing whole-heart coronary MRA with 3D radial SSFP and self-navigated image reconstruction. Magn Reson Med. 2005;54:476–480. 80. Weber OM, Martin AJ, Higgins CB. Whole-heart steady-state free precession coronary artery magnetic resonance angiography. Magn Reson Med. 2003;50:1223–1228. 81. Deshpande VS, Shea SM, Laub G, Simonetti OP, Finn JP, Li D. 3D magnetization-prepared true-FISP: a new technique for imaging coronary arteries. Magn Reson Med. 2001;46:494–502. 82. Ozgun M, Hoffmeier A, Kouwenhoven M, et al. (2005) Comparison of 3D segmented gradient-echo and steady-state free precession coronary MRI sequences in patients with coronary artery disease. AJR Am. J Roentgenol. 185:103–109. 83. Maintz D, Aepfelbacher FC, Kissinger KV, et al. Coronary MR angiography: comparison of quantitative and qualitative data from four techniques. AJR Am J Roentgenol. 2004;182:515–521. 84. Nezafat R, Herzka D, Stehning C, Peters DC, Nehrke K, Manning WJ. Inflow quantification in three-dimensional cardiovascular MR imaging. J Magn Reson Imaging. 2008;28:1273–1279. 85. Scott A, Keegan J, Firmin D. Motion in cardiovascular MR imaging. Radiology. 2009;250:331–351. 86. Lu B, Mao S, Zhuang N, et al. Coronary artery motion during the cardiac cycle and optimal ECG triggering for coronary artery imaging. Invest Radiol. 2001;36:250–256. 87. Hofman M, Wickline S, Lorenz C. Quantification of in-plane motion of the coronary arteries during the cardiac cycle: implications for acquisition window duration for MR flow quantification. J Magn Res Imaging. 1998;8:568–576.

347 88. Wang Y, Vidan E, Bergman G. Cardiac motion of coronary arteries: variability in the rest period and implications for coronary MR angiography. Radiology. 1999;213:751–758. 89. Ehman R, Felmlee J. Adaptive technique for high-definition MR imaging of moving structures. Radiology. 1989;173:255–263. 90. Wang Y, Rossman P, Grimm R, Riederer S, Ehman R. Navigatorecho-based real-time respiratory gating and triggering for reduction of respiration effects in three-dimensional coronary MR angiography. Radiology. 1996;198:55–60. 91. McConnell M, Khasgiwala V, Savord B, et al. Comparison of respiratory suppression methods and navigator locations for MR coronary angiography. AJR Am J Roentgenol. 1997;168:1369–1375. 92. Taylor A, Jhooti P, Wiesmann F, Keegan J, Firmin D, Pennell D. MR navigator-echo monitoring of temporal changes in diaphragm position: implications for MR coronary angiography. J Magn Res Imaging. 1997;7:629–636. 93. Huber A, Nikolaou K, Gonschior P, Knez A, Stehling M, Reiser M. Navigator echo-based respiratory gating for three-dimensional MR coronary angiography: results from healthy volunteers and patients with proximal coronary artery stenoses. AJR Am J Roentgenol. 1999;173:95–101. 94. Wang Y, Riederer S, Ehman R. Respiratory motion of the heart: kinematics and the implications for the spatial resolution in coronary imaging. Magn Reson Med. 1995;33:713–719. 95. Danias PG, McConnell MV, Khasgiwala VC, Chuang ML, Edelman RR, Manning WJ. Prospective navigator correction of image position for coronary MR angiography. Radiology. 1997;203:733–736. 96. Danias PG, Stuber M, Botnar RM, Kissinger KV, Edelman RR, Manning WJ. Relationship between motion of coronary arteries and diaphragm during free breathing: lessons from real-time MR imaging. AJR Am J Roentgenol. 1999;172:1061–1065. 97. Nezafat R, Stehning C, Gharib AM, et al. Improved spatial-temporal resolution MR coronary blood flow imaging at 3 T. J Cardiovasc Magn Res. 2005;7:199. 98. Nehrke K, Bornert P. Prospective correction of affine motion for arbitrary MR sequences on a clinical scanner. Magn Reson Med. 2005;54(5). 99. Lai P, Larson AC, Park J, Carr JC, Li D. Respiratory self-gated four-dimensional coronary MR angiography: a feasibility study. Magn Reson Med. 2008;59:1378–1385. 100. Larson AC, Kellman P, Arai A, et al. Preliminary investigation of respiratory self-gating for free-breathing segmented cine MRI. Magn Reson Med. 2005;53:159–168. 101. Park J, Larson AC, Zhang Q, Simonetti O, Li D. 4D radial coronary artery imaging within a single breath-hold: cine angiography with phase-sensitive fat suppression (CAPS). Magn Reson Med. 2005;54:833–840. 102. Keegan J, Gatehouse P, Yang G, Firmin D. Non-model based correction of respiratory motion using beat-to-beat 3d spiral fat-selective imaging. J Magn Reson Imaging. 2007;26:624–629. 103. Nguyen T, Spincemaille P, Cham M, Weinsaft J, Prince M, Wang Y. Free-breathing 3D steady-state free precession coronary magnetic resonance angiography: comparison of diaphragm and cardiac fat navigators. J Magn Reson Imaging. 2008;28:509–514. 104. Nguyen T, Spincemaille P, Prince M, Wang Y. 2128 Free-breathing steady-state free precession 3D coronary MRA: comparison of diaphragm and cardiac fat navigator techniques. J Cardiovasc Magn Res. 2008;10(Suppl 1):A397. 105. Nguyen T, Spincemaille P, Cham M, Weinsaft J, Prince M, Wang Y. Free-breathing 3-dimensional steady-state free precession coronary magnetic resonance angiography: comparison of four navigator gating techniques. Magn Res Imaging. 2009;27:807–814.

348 106. Uribe S, Muthurangu V, Boubertakh R, et al. Whole-heart cine MRI using real-time respiratory self-gating. Magn Reson Med. 2007;57:606–613. 107. Brau AC, Brittain JH. Generalized self-navigated motion detection technique: Preliminary investigation in abdominal imaging. Magn Reson Med. 2006;55:263–270. 108. Nehrke K, Bornert P, Manke D, Bock JC. Free-breathing cardiac MR imaging: study of implications of respiratory motion--initial results. Radiology. 2001;220:810–815. 109. Gurney PT, Hargreaves BA, Nishimura DG. Design and analysis of a practical 3D cones trajectory. Magn Reson Med. 2006;55: 575–582. 110. Santos JM, Cunningham CH, Lustig M, et al. Single breath-hold whole-heart MRA using variable-density spirals at 3 T. Magn Reson Med. 2006;55:371–379. 111. Sussman M, Stainsby J, Robert N, Merchant N, Wright G. Variable-density adaptive imaging for high-resolution coronary artery MRI. Magn Reson Med. 2002;48:753–764. 112. Brittain JH, Hu BS, Wright GA, Meyer CH, Macovski A, Nishimura DG. Coronary angiography with magnetization-prepared T2 contrast. Magn Reson Med. 1995;33:689–696. 113. Botnar RM, Stuber M, Danias PG, Kissinger KV, Manning WJ. Improved coronary artery definition with T2-weighted, freebreathing, three-dimensional coronary MRA. Circulation. 1999;99: 3139–3148. 114. Nezafat R, Derbyshire J, Ouwerkerk R, Stuber M, McVeigh E. Spectrally selective B1 insensitive T2 preparation sequence for 3 T imaging. Proceedings of 14th Scientific Meeting, International Society for Magnetic Resonance in Medicine. 2006:596. 115. Nezafat R, Ouwerkerk R, Derbyshire AJ, Stuber M, McVeigh ER. Spectrally selective B1-insensitive T2 magnetization preparation sequence. Magn Reson Med. 2009;61(6):1326–1335. 116. Li D, Paschal CB, Haacke EM, Adler LP. Coronary arteries: threedimensional MR imaging with fat saturation and magnetization transfer contrast. Radiology. 1993;187:401–406. 117. Nezafat R, Han Y, Peters DC, et al. Coronary magnetic resonance vein imaging: imaging contrast, sequence, and timing. Magn Reson Med. 2007;58:1196–1206. 118. Stuber M, Botnar RM, Fischer SE, et al. Preliminary report on in vivo coronary MRA at 3 Tesla in humans. Magn Reson Med. 2002;48:425–429. 119. Schar M, Kozerke S, Fischer Se Boesiger P. Cardiac SSFP imaging at 3 Tesla. Magn Reson Med. 2004;51:799–806. 120. Nezafat R, Stuber M, Ouwerkerk R, Gharib AM, Desai MY, Pettigrew RI. B1-insensitive T2 preparation for improved coronary magnetic resonance angiography at 3 T. Magn Reson Med. 2006;55:858–864. 121. van Elderen SG, Versluis MJ, Webb AG, et al. Initial results on in vivo human coronary MR angiography at 7 T. Magn Reson Med. 2009;62:1379–1384. 122. Terashima M, Meyer CH, Keeffe BG, et al. Noninvasive assessment of coronary vasodilation using magnetic resonance angiography. J Am Coll Cardiol. 2005;45:104–110. 123. Hu P, Chuang ML, Ngo LH, et al. Coronary MR imaging: effect of timing and dose of isosorbide dinitrate administration. Radiology. 2010;254:401–409. 124. Zheng J, Bae KT, Woodard PK, Haacke EM, Li D. Efficacy of slow infusion of gadolinium contrast agent in three-dimensional MR coronary artery imaging. J Magn Reson Imaging. Nov 1999; 10:800–805. 125. Goldfarb JW, Edelman RR. Coronary arteries: breath-hold, gadolinium-enhanced, three-dimensional MR angiography. Radiology. 1998;206:830–834.

R. Nezafat et al. 126. Knuesel PR, Nanz D, Wolfensberger U, et al. Multislice breathhold spiral magnetic resonance coronary angiography in patients with coronary artery disease: effect of intravascular contrast medium. J Magn Reson Imaging. 2002;16:660–667. 127. Herborn CU, Barkhausen J, Paetsch I, et al. Coronary arteries: contrast-enhanced MR imaging with SH L 643A--experience in 12 volunteers. Radiology. 2003;229:217–223. 128. Prompona M, Cyran C, Nikolaou K, Bauner K, Reiser M, Huber A. Contrast-enhanced whole-heart MR coronary angiography at 3.0 T using the intravascular contrast agent gadofosveset. Invest Radiol. 2009;44:369–374. 129. Kelle S, Thouet T, Tangcharoen T, et al. Whole-heart coronary magnetic resonance angiography with MS-325 (Gadofosveset). Med Sci Monit. 2007;13:CR469-CR474. 130. de Haen C, Anelli PL, Lorusso V, et al. Gadocoletic acid trisodium salt (b22956/1): a new blood pool magnetic resonance contrast agent with application in coronary angiography. Invest Radiol. 2006;41:279–291. 131. Tang L, Merkle N, Schar M, et al. Volume-targeted and whole-heart coronary magnetic resonance angiography using an intravascular contrast agent. J Magn Reson Imaging. 2009;30:1191–1196. 132. Yang Q, Li K, Liu X, et al. Contrast-enhanced whole-heart coronary magnetic resonance angiography at 3.0-T: a comparative study with X-ray angiography in a single center. J Am Coll Cardiol. 2009;54:69–76. 133. Hu P, Chan J, Smink J, et al. Contrast-enhanced whole-heart coronary MRI with a bolus infusion of gadobenate dimeglumine at 1.5 T. Magn Reson Imaging. 2011;66:392–398. 134. Bunce NH, Jhooti P, Keegan J, et al. Evaluation of free-breathing three-dimensional magnetic resonance coronary angiography with hybrid ordered phase encoding (HOPE) for the detection of proximal coronary artery stenosis. J Magn Reson Imaging. 2001; 14:677–684. 135. Sommer T, Hofer U, Hackenbroch M, et al. [Submillimeter 3D coronary MR angiography with real-time navigator correction in 107 patients with suspected coronary artery disease]. Rofo. 2002;174:459–466. 136. Bogaert J, Kuzo R, Dymarkowski S, Beckers R, Piessens J, Rademakers FE. Coronary artery imaging with real-time navigator three-dimensional turbo-field-echo MR coronary angiography: initial experience. Radiology. 2003;226:707–716. 137. Jahnke C, Paetsch I, Schnackenburg B, et al. Coronary MR angiography with steady-state free precession: individually adapted breath-hold technique versus free-breathing technique. Radiology. 2004;232:669–676. 138. Kusano KF, Morita H, Fujimoto Y, Hirose E, Ohe T. Catheter ablation of an epicardial accessory pathway via the middle cardiac vein guided by monophasic action potential recordings. Europace. 2001;3:164–167. 139. Haissaguerre M, Gaita F, Fischer B, Egloff P, Lemetayer P, Warin JF. Radiofrequency catheter ablation of left lateral accessory pathways via the coronary sinus. Circulation. 1992;86:1464–1468. 140. Kar S, Nordlander R. Coronary veins: an alternate route to ischemic myocardium. Heart Lung. 1992;21:148–157. 141. Stellbrink C, Diem B, Schauerte P, Ziegert K, Hanrath P. Transcoronary venous radiofrequency catheter ablation of ventricular tachycardia. J Cardiovasc Electrophysiol. 1997;8: 916–921. 142. Butter C, Meisel E, Engelmann L, et al. Human experience with transvenous biventricular defibrillation using an electrode in a left ventricular vein. Pacing Clin Electrophysiol. 2002;25:324–331. 143. Thompson CA, Nasseri BA, Makower J, et al. Percutaneous transvenous cellular cardiomyoplasty. A novel nonsurgical approach

24

Noninvasive Imaging for Coronary Artery Disease

for myocardial cell transplantation. J Am Coll Cardiol. 2003;41: 1964–1971. 144. Oesterle SN, Reifart N, Hauptmann E, Hayase M, Yeung AC. Percutaneous in situ coronary venous arterialization: report of the first human catheter-based coronary artery bypass. Circulation. 2001;103:2539–2543. 145. de Paola AA, Melo WD, Tavora MZ, Martinez EE. Angiographic and electrophysiological substrates for ventricular tachycardia mapping through the coronary veins. Heart. 1998;79:59–63. 146. Singh JP, Houser S, Heist EK, Ruskin JN. The coronary venous anatomy: a segmental approach to aid cardiac resynchronization therapy. J Am Coll Cardiol. 2005;46:68–74. 147. Bax JJ, Abraham T, Barold SS, et al. Cardiac resynchronization therapy: Part 1--issues before device implantation. J Am Coll Cardiol. 2005;46:2153–2167.

349 148. Bax JJ, Abraham T, Barold SS, et al. Cardiac resynchronization therapy: Part 2--issues during and after device implantation and unresolved questions. J Am Coll Cardiol. 2005;46:2168–2182. 149. Abraham WT, Fisher WG, Smith AL, et al. Cardiac resynchronization in chronic heart failure. N Engl J Med. 2002;346:1845–1853. 150. Nezafat R, Han Y, Peters DC, et al. Magnetic resonance coronary vein imaging: sequence, contrast and timing. Magn Reson Med. 2007;58:1196–1206. 151. Stoeck CT, Han Y, Peters DC, et al. Whole heart magnetizationprepared steady-state free precession coronary vein MRI. J Magn Reson Imaging. 2009;29:1293–1299. 152. Rasche V, Binner L, Cavagna F, et al. Whole-heart coronary vein imaging: a comparison between non-contrast-agent-and contrastagent-enhanced visualization of the coronary venous system. Magn Reson Med. 2007;57:1019–1026.

wwwwwwwwwwww

Venous Imaging: Techniques, Protocols, and Clinical Applications

25

Amir H. Davarpanah, Philip Hodnett, Jeremy D. Collins, James C. Carr, and Tim Scanlon

Introduction Magnetic resonance imaging (MRI) is intrinsically sensitive to flowing blood. The earliest investigators of MRI recognized that blood flow altered the intraluminal magnetic resonance (MR) signal. MRI techniques in imaging blood flow have been principally directed toward interrogation of the arterial system [1]. However, the same techniques that have made magnetic resonance angiography (MRA) such a useful clinical tool have also been applied to the venous system [2]. MR angiographic techniques can be divided into two categories: contrast material-enhanced and nonenhanced MRA [3]. Since its introduction in 1994 by Prince [1], first-pass contrast-enhanced (CE) MRA with gadolinium-based contrast material has seen widespread acceptance. The approach has benefited from many technical advances, including strategies to synchronize arrival of the bolus of contrast material with MR acquisition, moving bed technology for multistation studies, such as peripheral MRA, shortened acquisition times with parallel imaging, and k-space sharing methods, such as time-resolved imaging of contrast kinetics or TRICKS, for “time-resolved” MR angiographic acquisitions [1–3].

A.H. Davarpanah, MD () Department of Radiology, Yale School of Medicine, Yale University, New Haven, Connecticut, USA e-mail: [email protected] P. Hodnett, MD Department of Radiology, New York University, NY J.D. Collins, MD Department of Radiology, Northwestern Memorial Hospital and Northwestern University Feinberg School of Medicine, Chicago, IL, USA J.C. Carr, MD Northwestern University, Feinberg School of Medicine, Chicago, IL, USA T. Scanlon, MD Consultant radiologist, Limerick regional Hospital, Ireland

Several factors contribute to a renewed interest in nonenhanced MR angiographic methods [1]. Improvements in MR hardware and software, including the widespread availability of parallel imaging, have helped reduce acquisition times and made some methods clinically practical. Moreover, the recent association between high doses of gadolinium-based contrast material for MRA and the debilitating and occasionally life-threatening entity, nephrogenic systemic fibrosis, originally known as nephrogenic fibrosing dermopathy, has made it imperative that patients with moderate to severe renal insufficiency and vascular disease have nonenhanced alternatives for angiography/venography [4–6]. The purpose of this chapter is to review the techniques, clinical applications, and useful protocols of magnetic resonance venography (MRV), both noncontrast and contrastenhanced, in relation to the chest, abdomen, pelvis, and lower limbs, to aid in everyday practice.

MRV Techniques Multiple techniques, both noncontrast and contrast-enhanced pulse sequences, have been used to assess venous anatomy and pathology [2, 3]. Both categories of techniques include different strategies, including time-of-flight (TOF) imaging, phase-contrast imaging, MR direct thrombus imaging (MRDTI), gadolinium-enhanced MRV, and black blood techniques [2]. Contrast-enhanced MR venography (CEMRV) remains the mainstay technique for imaging the venous system. Noncontrast (NC) MRV is employed when there are contraindications to gadolinium contrast administration.

Noncontrast-Enhanced MR Venography Given the frequency of renal functional impairment in patients with peripheral vascular disease (PVD) and concerns about nephrogenic systemic fibrosis, over the last few years, several unenhanced MRA techniques have been introduced [7]. Three-dimensional fast spin echo (FSE) sequences with ECG

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_25, © Springer Science+Business Media, LLC 2012

351

352

A.H. Davarpanah et al.

Table 25.1 Protocol for noncontrast MR venography (NCMRV) Technique 1. 2D single-shot SSFP (axial and coronal) 2. T1 GRE FS (axial and coronal) 3. 3D SSFP (axial/LAO)

Comment Free breathing; ECG gated to diastole Noncontrast Free breathing; respiratory gated with navigator

gating to generate systolic and diastolic images can be modified to obtain venous information. The ability to generate 3-dimensional reconstruction of venous anatomy without contrast and without arterial contamination is very appealing; however, the efficacy of these techniques in body MRV has only begun to be evaluated. In a recent study, it was demonstrated that portal vein and its branches can be well-delineated with unenhanced MRA performed with a single-breath-hold, coronal, in-plane, ECG-synchronized, 3D half-Fourier, FSE sequence [8]. Similarly, in another study, a combination of 3D half-Fourier, FSE, flow-refocused fresh-blood imaging (FR-FBI) and the swap phase-encode arterial double-subtraction elimination (SPADE) technique images demonstrated to be highly accurate and reproducible for detecting deep vein thrombosis (DVT) without interference from implant susceptibility artifacts, especially in patients with nonmagnetizing metal implants during orthopedic surgery [9]. NCMRV protocol is detailed in Table 25.1.

Time of Flight TOF imaging employs a gradient-recalled echo pulse sequence [10]. In contradistinction to spin echo imaging, the signal from flowing blood is consistently replenished by the inflow of unsaturated (excited) spins producing increased intraluminal signal, resulting in “bright blood” images. The acquisitions may be either two dimensional (slice by slice) or three dimensional (a volume is acquired and then partitioned). A major drawback of TOF is the long acquisition time, which may result in image degradation due to motion artifact from patient movement. Additionally, in regions of slow flow, turbulence, or significant in-plane coursing of the vessel, there is loss of the intraluminal signal. Portions of the vessel lumen then appear darker and may mimic intraluminal pathology [10]. TOF MRV has generally been superseded by other NCMRV techniques due to the reasons mentioned above; however, the technique may be of use at higher field strength due to the increased signal-to-noise ratio and resultant improved image quality. TOF imaging has been used for imaging the intracranial veins [11, 12] and portal venous system [13]. Steady-State Free Precession Steady-state free precession (SSFP) is a bright blood imaging technique that is used extensively for cine MRI of the heart [14] and has been employed successfully as a noncon-

trast MRA technique for imaging the arterial system in several different vascular territories [3, 15]. Since, SSFP produces bright signal from both arteries and veins, it can be used as a method for NCMRV. The technique is implemented in 2D, as a stack of images, or in 3D, where respiratory gating needs to be utilized if used in the thorax or abdomen. ECG gating, where image acquisition occurs during diastole, improves image quality, particularly in the thorax. A major advantage of SSFP is its short scan time resulting in acquisition times significantly shorter than TOF imaging. SSFP has been used as an NCMRV technique in the lower extremity venous system [16] and also in the thorax.

Signal Targeting Alternative Radiofrequency and Flow-Independent Relaxation Enhancement A novel noncontrast MRV technique, designed by Koktzoglou and Edelman, is Signal Targeting Alternative Radiofrequency and Flow-Independent Relaxation Enhancement (STARFIRE) [17]. STARFIRE uses a bSSFP pulse sequence for readout of the MR signal to make blood vessels appear bright. Although bSSFP can be effective for imaging large vessels, such as the abdominal aorta and inferior vena cava (IVC), it can be problematic to use this technique for making projection angiograms of smaller vessels, such as the peripheral arteries and veins. With bSSFP, the smaller vessels tend to be obscured in MIP images by using the signal intensities of fat and muscle. Although chemical shift-selective methods for fat suppression can be applied with bSSFP, a substantial drawback is that they produce nonuniform signal suppression over large fields of view owing to static magnetic field inhomogeneity, a problem which is exacerbated by the sensitivity of peripheral array coils to the high signal intensity of subcutaneous fat. Moreover, the effect of the fat suppression radiofrequency (RF) pulse may be attenuated during the course of the bSSFP echo train. The goal of the STARFIRE technique is to suppress fat and muscle signals while maintaining the high signal intensity of the vasculature. Different contrast mechanisms are involved for fat and muscle signal suppression. With STARFIRE, fat suppression is performed predominantly on the basis of a T1-dependent mechanism, relying on the fact that fat has a much shorter T1 relaxation time than blood. The main rationale for using a T1-based, rather than chemical shift-based, mechanism for fat suppression is that the T1 relaxation times of the tissues are only negligibly dependent on static magnetic field inhomogeneity [17]. STARFIRE produces flow-independent images of all blood vessels (i.e., both arteries and veins). The presaturation pulses equalize arterial signal on the tagged and untagged image sets and suppress arterial signal on subtraction images. STARFIRE enables imaging of the vasculature over large fields of view as a result of the uniform suppression of both fat and muscle signals [17].

25

Venous Imaging: Techniques, Protocols, and Clinical Applications

3D Fast Spin Echo Recently, unenhanced MRA, using 3D half-Fourier FSE has been shown effective for visualization of the coronary arteries, renal arteries, and peripheral vessels [2, 3]. The 3D half-Fourier FSE method is one of the new MRA techniques that allows selective visualization of both arteries and veins. In general, an FSE sequence uses a long TE value and shows rapid blood flow as signal void; however, reduced echo train spacing enables compact echo sampling and reduces this phenomenon, especially when the flow velocity is low, as in the vein [3]. FSE images can also be acquired during systole, where arterial signal is suppressed due to rapid flow, and during systole, where both arteries and veins are visible together. Subtraction of both acquisitions has been used for arterial imaging, such as in the thoracic aorta [18]. However, images from either systole or diastole depict veins and can therefore be used for NCMRV imaging. MR-Directed Thrombus Imaging MRDTI relies on a gradient echo sequence which visualizes thrombus directly against a suppressed background [19]. In this instance, the sequence is a T1-weighted, magnetizationprepared, three dimensional gradient echo volume acquisition. The sequence employs a water-only excitation radiofrequency pulse that effectively eliminates any signal from fat. Unlike simple TOF GRE sequences which produce high signal from flowing blood, an inversion pulse is incorporated into this acquisition and is timed to effectively eliminate blood signal. It is known that acute venous thrombus contains significant methemoglobin [2, 3]. Methemoglobin causes T1 shortening within the thrombus causing it to appear bright. By suppressing the surrounding tissue and blood signal, the conspicuity of the thrombus becomes more apparent. MRDTI has been used to detect lower limb DVT, avoiding some of the flow related artifacts associated with other NCMRV techniques. Susceptibility-Weighted Imaging Susceptibility-weighted imaging (SWI) is a means of enhancing contrast in MRI. It is complementary to conventional spindensity, T1- and T2-weighted imaging methods. SWI is particularly suited for imaging venous blood, as it is very sensitive to deoxyhemoglobin, making it useful for imaging hemorrhages from trauma, where blood products can be visualized. The iron in deoxyhemoglobin in venous blood acts as an intrinsic contrast agent, causing T2*-related losses in the magnitude image and a shift in the phase image caused by susceptibility differences. The oxygen in oxyhemoglobin shields the iron so that T2* and susceptibility effects are only seen in venous blood. This provides a natural separation of venous and arterial blood, and allows for venographic images without any arterial contamination. Veins are dark because of T2* losses, whereas the arteries are bright from TOF inflow enhancement. In addition, it is possible to increase the contrast in the arteries without overly degrading the venographic images by using a slightly higher flip angle, short TR, and a

353

Fig. 25.1 Schematic drawing illustrating the range of each sequence. The parenchyma sequence is more sensitive to lower concentrations while the angiographic phase is more sensitive to higher concentrations. An overlap exists, making both sequences usable in a given range

thin slab [2, 20]. This is particularly effective at high and ultrahigh field because shorter echoes can be used to reduce flow dephasing in the arteries without affecting the venous contrast. SWI has been used to image the intracranial venous system.

Contrast-Enhanced MR Venography CEMRV is similar to CEMRA of the arterial system, except that the contrast injection is timed to the venous phase (Fig. 25.1). A spoiled gradient echo pulse sequence is used, as with CEMRA, and gadolinium contrast agent is injected intravenously. Minimum TR and parallel imaging are sued to shorten the acquisition time. The short scan time allows rapidly flowing blood to become saturated, thereby losing its bright signal. The intravenous administration of gadolinium shortens the T1 relaxation time of blood to such an extent that recovery occurs after each RF pulse and thus blood appears white. The surrounding tissues appear dark as they have become saturated and produce little signal. CEMRV can be implemented as a conventional “timed” MRV or as time-resolved MRV. Table 25.2 shows sequence protocol for CEMRV.

Conventional CEMRV Conventional CEMRV is a timed examination of the vessel of interest, similar to CEMRA. CEMRV may be performed using a direct or indirect approach. With the indirect approach (Fig. 25.2), nondiluted contrast is injected in a nontargeted peripheral vein and imaging takes place during the venous phase of the vessel of interest. Because considerable dilution of contrast occurs by the time it arrives to the area of interest, images are acquired in the early equilibrium phase to avoid redistribution. In general, it is recommended that larger doses of contrast agent are used, i.e., 0.2 mmol/kg, in order to fully fill the large capacitance venous system. The transit time can be estimated by adding 10–20 s to the arterial transit time, measured using a timing

354

A.H. Davarpanah et al.

bolus technique. Bolus tracking is less successful for venous imaging due to contrast dilution during the venous phase. Three to four postcontrast 3D images of the venous system are acquired to ensure adequate venous enhancement. Alternatively, the direct approach uses significantly less contrast because of targeted administration proximal to the area of interest. The direct technique involves injection of dilute contrast in a peripheral vein and imaging of the draining venous system. Thus, it is analogous to conventional X-ray venography. Dilution of gadolinium with saline prior to injection is necessary to avoid T2* effect, resulting in signal loss. Contrast

is injected into a peripheral vein and images are acquired before gadolinium reaches the heart. To ensure optimal opacification, bilateral injection may be of benefit for imaging the central veins in the chest. Bolus triggering is optional but not necessary, since the injection rate is relatively low and the contrast reaches the region of interest quickly. A set of 3D images is obtained, with mask-mode subtraction, at different time points during venous filling in the same manner as conventional X-ray venography. Postprocessing includes MIPs with and without subtraction. First-pass venography can be used for visualization of the subclavian veins and the superior vena cava, using bilateral injection (Fig. 25.3), and of the forearm veins, using unilateral injection.

Table 25.2 Protocol for contrast-enhanced MR Venography (CEMRV) Technique 1. 2D single-shot SSFP (axial and coronal) 2. T1 GRE FS (axial and coronal) 3. VIBE (axial and coronal) Inject contrast 4. TRMRV (LAO/ coronal)

Comment Free breathing; ECG gated to diastole Pre and post contrast Pre and post contrast

4 cc GAD at 4 cc/s; this can be used for timing; alternatively, this can be repeated several times at different temporal resolution, if required 5. CEMRV (coronal) Breath-hold, if needed; 3–4 postcontrast 3D measurements 6. High-spatial-resolution If blood pool agent used, increase MRV spatial resolution to 0.8 mm3 7. High-spatial-resolution If blood pool agent used, increase VIBE spatial resolution to 0.8 mm3

Fig. 25.2 Contrast-enhanced MR venography at equilibrium. Preparation of pure contrast flushed with saline. Masks are obtained before injection for further subtraction. Contrast is injected at a high rate. MRA sequence is peformed once, followed by the parenchyma sequence which is repeated twice

Time-Resolved MRV A variant of conventional CEMRV is to perform timeresolved MRV (TRMRV). With this approach, the acquisition speed is considerably increased using acceleration strategies, such as echo sharing (i.e., TWIST, TRICKS) or parallel imaging. Multiple 3D images are acquired in rapid succession at a time resolution of 3–6 s per frame so that filling of the venous system can be observed in real time. In-line mask-mode subtraction and MIP calculation can be carried out so that venous images are displayed as a cine series showing filling of arteries and veins over time. Since speed of acquisition is the priority with TRMRV, spatial resolution needs to be sacrificed resulting in lower image quality compared to conventional CEMRV. TRMRV is particularly advantageous in regions, where arteriovenous transit is rapid, such as for pulmonary venous imaging (Fig. 25.4). Another advantage of TRMRV is the ability to use smaller doses of

25

Venous Imaging: Techniques, Protocols, and Clinical Applications

355

Fig. 25.3 Contrast-enhanced MR venography at first pass. Preparation of dilute contrast media. A mask is obtained before injection for further subtraction. Contrast is injected at a slow rate. MRA sequence is repeated multiple times during injection

Fig. 25.4 Coronal dynamic TWIST MIP showing left inferior pulmonary vein stenosis post ablation

contrast. Although a conventional CEMRV acquisition may require 20–40 ml of contrast, a temporally resolved acquisition may only use as little as 3–4 ml.

Contrast Agents Extracellular Gadolinium Contrast Agents Standard extracellular contrast media are currently used for almost all CEMRA and CEMRV applications. Angiography during the first pass provides strong and selective enhancement of the vessel of interest. Extracellular agents are welltolerated with low incidence of side effects. Extracellular contrast agents are characterized by rapid extravasation of the agent from the vascular space into the interstitium, resulting in decreasing vascular enhancement over time and increasing background signal. Thus, precise bolus timing is mandatory so that contrast arrival is timed to the venous system. There is a narrow window for imaging with extracellular agents. Blood Pool Contrast Agents Blood pool contrast agents are characterized by prolonged intravascular half-life and low rate of extravasation of contrast from the vascular to the extravascular space. This feature makes blood pool agents ideally suited for venous imaging

since they reside within the venous system for prolonged periods of time, allowing high spatial resolution images to be acquired. Gadofosveset trisodium is currently the only contrast media with predominant blood pool intravascular distribution that is approved for use in patients [21]. Gadofosveset is noncovalently bound to albumin in human plasma and is primarily excreted renally. In addition to its prolonged intravascular half-life, the agent has significantly higher relaxivity compared to conventional extracellular agents, allowing it to be administered at lower doses. In plasma, gadofosveset exhibits a relaxivity at 0.5 T that is approximately six to ten times that of gadopentetate dimeglumine [22].

Iron-Based Contrast Agents Iron-based blood pool agents are coated ultrasmall superparamagnetic iron oxide (USPIO) particles with a strong T1 and T2 shortening effect. Such compounds are retained within the intravascular space in a prolonged fashion. Because of the T1 and T2 shortening effects, a dual-contrast approach can be used for imaging. First-pass and equilibrium images can be acquired using T1 gradient echo approach, similar to conventional CEMRV, and T2-weighted TSE can be employed to exploit the T2 shortening effect. Promising results from USPIO at first-pass and steady-state angiography have been published, but no USPIO is approved yet [21]. Iron-based agents have also been used to assess deep venous thrombosis [23].

Clinical Applications MRV has applications in many clinical settings. In most instances, it complements other venous imaging modalities. In some cases, it is the primary diagnostic modality. If renal

356

A.H. Davarpanah et al.

Fig. 25.5 Axial T2 proton density sequence showing DVT with hematocrit effect

function or a history of contrast allergy precludes iodinated contrast administration and body habitus prevents adequate assessment by ultrasound, MRI can be performed.

Abdominal and Lower Extremity Deep Vein Thrombosis Assessment Although considered an underestimation, the reported incidence of acute DVT is approximately 5 cases per 10,000 per year. Complications include postphlebitic syndrome, pulmonary embolism, and possibly death. As many as 300,000 deaths per year in the USA are caused by pulmonary embolism [24–26]. For diagnosing DVT, physicians have heavily relied on TOF imaging. [25–27]. This basic protocol is supplemented with fat-suppressed T2-weighted FSE or fast IR images and, where appropriate, phase-contrast images. When using these sequences, nonturbulent flowing blood is bright (white). Thrombus appears dark if not black, especially when acute. When present, acute thrombus typically appears to occlude the entire lumen, although not infrequently the lumen is only partially occluded by thrombus (Fig. 25.5). Using TOF imaging, Evans et al. [28] reported 100% sensitivity and specificity for the detection of DVT in the thigh in 61 patients when compared with venography. In the same study, the authors reported a sensitivity of 87% with a specificity of 97% for the assessment of the calves when compared with venography. Fraser et al., in a study of 101 patients with suspected DVT, compared MRDTI with venography and reported sensitivities of 97 and 100% for femoropopliteal and iliofemoral DVT, respectively [19]. Preliminary investigations are ongoing using new contrast agents in the hope of facilitating the assessment of thrombus.

Fig. 25.6 Contrast-enhanced (CE) MRA pelvis thick MIP showing right persistent sciatic vein and IVC thrombus

Li et al. [23] demonstrated the potential utility of ferumoxytol for detecting lower extremity thrombus. This iron-based compound was able to detect thrombus using both black blood (FSE) and white blood (3-D GRE) techniques. The authors suggest that although the anatomy is better seen using the white blood technique the precise extent of thrombus was more readily appreciated on the black blood images [25, 26, 29]. Spuentrup et al. [30] have demonstrated the feasibility of using a fibrin-specific contrast agent (EP-2104R) to detect thrombus within the arterial system. Such approaches clearly have applicability in the venous system as well. The traditional reference standard for assessing venous thrombus is ascending venography. The reported sensitivity and specificity for venography have been as high as 89 and 97%, respectively [30]. However, X-ray venography is associated with complications, including pain, inflammation, extravasation, induced DVT, allergy, and renal failure [24, 26]. Accordingly, there are some situations, where MRI should be primarily considered. These include the following: (a) clinical suspicion of central (pelvic) venous thrombosis (Fig. 25.6), (b) pelvic trauma, (c) ovarian vein thrombosis, (d) cryptogenic stroke, (e) pelvic congestion syndrome (Fig. 25.7), (f) hepatic and portal venous thrombosis (Fig. 25.8), (g) venous anomalies. In addition, if ultrasound is unclear or if there is potentially a large differential diagnosis, MRI is the diagnostic tool of choice. Septic thrombophlebitis occurs primarily in women who have recently undergone a cesarean section [31, 32]. Although

25

Venous Imaging: Techniques, Protocols, and Clinical Applications

Fig. 25.7 3D coronal CEMRA/V showing enlarged left gonadal vein and pelvic varicosities, suggesting pelvic congestion syndrome

Fig. 25.8 3D CE FAT-SAT MRA/V showing right gonadal (ovarian) vein thrombus and main portal vein thrombus

the left ovarian vein may be thrombosed, it is the right ovarian vein (Fig. 25.8) that is principally affected. MRI is considered more sensitive than either computed tomography (CT) or ultrasound for making this diagnosis. Cryptogenic stroke is defined as brain ischemia with no apparent etiology [33]. Studies suggest that cryptogenic stroke may occur in up to 40% of stroke patients. This number appears to be higher in younger patients. A patent foramen

357

ovale is more common in patients with cryptogenic stroke when compared with stroke of determined origin or normal controls, suggesting that right-to-left shunting of emboli is a potential cause of stroke in this population. In a small study of 16 patients with cryptogenic stroke, all had intra-atrial communication: a patent foramen ovale was present in 14, whereas 2 subjects had atrial septal defects. In four cases, MRI showed pelvic thrombus. In an additional seven cases, MRI was suspicious for prior DVT. In each instance, ultrasound was normal [33]. MRI combined with MRV efficiently clarifies cases, where multiple diagnoses are simultaneously being entertained. For example, for a patient presenting with acute calf pain with swelling and possible mass, a typical differential might include DVT, ruptured Baker’s cyst, compartment syndrome, infection (myositis), and muscle injury. Whereas ultrasound may diagnose or exclude several of the aforementioned entities, MRI/MRV can readily distinguish and diagnose each entity. MRI is potentially able to distinguish between acute and chronic thrombus. Erdman et al. [34] demonstrated that perivascular edema seen on T2-weighted images is highly predictive of acute DVT. As discussed above, acute thrombus appears bulky filling much if not all of the involved lumen. In addition, the thrombus appears dark in the TOF images. Over time, the thrombus becomes less bulky with higher signal intensity due to recanalization and lysis of the clot. In MRDTI images, with age, methemoglobin is lost resulting in decreased signal over time [35]. The sequelae of prior DVT include partial filling defects and/or Webs, vessel narrowing, thickened vessel walls, and the development of collateral circulation. MRI is capable of identifying these changes, in addition to venous anomalies. In a fashion similar to the pelvic veins, MRI/MRV is useful in the assessment of the IVC, hepatic, portal, mesenteric, and renal veins. A detailed map of the portal and hepatic venous anatomy is an important tool for surgeons to plan hepatic resection and living-related donor liver transplantation or to aid an interventional radiologist plan a TIPS procedure. Other clinical applications include arterio-portal fistula, cavernous transformation of the portal vein, Budd-Chiari, hepatic and/or portal vein thrombosis, and congenital anomalies. In the past, conventional angiography was the standard method for visualization of the portal venous anatomy. 3D contrast-enhanced MR portography has been shown to be as effective as digital subtraction angiography for assessing the portal vein [36]. Half-Fourier FSE is an unenhanced MRA technique [2, 3], which allows coronal acquisition, which is not possible with TOF imaging, especially for body MRA, and thus enables shorter 3D acquisition time [3]. In addition, this sequence is T2 weighted and the liver parenchyma becomes relatively low

358

signal intensity, resulting in good contrast between the portal vein and liver parenchyma; however, the long TE decreases the signal at the portal confluence, where higher flow velocity might cause flow void in comparison with images acquired with true SSFP [3]. Recently, Ono et al. demonstrated high accuracy and reproducibility of noncontrast- enhanced MRV using both the FR-FBI and the SPADE techniques for detecting DVT, as compared using conventional X-ray venography as the reference standard. Forty-one legs of 32 consecutive patients in 32 consecutive patients [9].

Klippel-Trénaunay Syndrome and Parkes Weber Syndrome Klippel-Trénaunay syndrome is a congenital disorder classically characterized by three findings: a port-wine stain (nevus flammeus), abnormal venous structures (such as varicosities and venous malformations), and osseous and soft-tissue hypertrophy. In 1907, Frederick Parkes Weber noted similar findings in association with arteriovenous malformations. This entity is referred to as Parkes Weber or KlippelTrénaunay–Weber syndrome. The diagnosis of KlippelTrénaunay syndrome can be made when any two of the three features are present [37, 38]. Complications associated with Klippel-Trénaunay syndrome are most often related to the vascular system. Such complications include stasis dermatitis, thrombophlebitis, and cellulitis. More serious complications include deep venous thrombosis, pulmonary embolism, coagulopathy, and congestive heart failure (in patients with associated arteriovenous malformations). Various imaging techniques can be used in the diagnosis of suspected Klippel-Trénaunay syndrome. Both plain radiography and CT have been used to depict phleboliths that suggest the presence of a venous malformation and disease-related complications, such as deep venous thrombosis. Sonography with Doppler capabilities can be used to assess the condition of the venous system within an affected extremity. MRI can be used to evaluate extremity hypertrophy and vascular malformations in these patients. MR arteriography and MRV can be used to define both the type and extent of vascular malformations in Klippel-Trénaunay syndrome [39]. Specifically, MRI allows differentiation of low-flow (venous) from high-flow (arteriovenous) malformations. The venous malformations typically associated with Klippel-Trénaunay syndrome are hyperintense on T2-weighted images and lack flow voids. The arteriovenous malformations associated with Parkes Weber syndrome are high flow because they are fed by the arterial system and therefore typically have flow voids. Occasionally, however, conventional angiography or venography is needed to define the vascular lesions associated with these conditions [39, 40].

A.H. Davarpanah et al.

Pelvic Congestion Syndrome Chronic pelvic pain with associated ovarian vein varicosities is called pelvic congestion syndrome (PCS). It can present with atypical and intractable pelvic pain usually in women of childbearing age. The etiology of PCS, while poorly understood, may be from incompetent valves in ovarian or pelvic veins, with associated hormonal factors. Ovarian vein dilatation is seen in 10% of women, up to 60% of whom may develop PCS [41]. Noninvasive methods, such as MRI, contrast-enhanced CT, and duplex ultrasound, may be used to depict dilated veins. Catheter venography, however, remains the gold standard in accurately depicting flow dynamics, but is an invasive procedure associated with radiation exposure. Conventional MRI and computed tomography yield good cross-sectional information; however, they fail to demonstrate the flow dynamics. MRV has been used to study ovarian vein incompetence, but lacks sufficient temporal resolution to demonstrate retrograde ovarian vein filling. In a study of 23 female patients with signs and symptoms of pelvic venous congestion comparing MRV with phlebography as gold standard, Asciutto et al. found that MRV was highly sensitive (88%), however demonstrated lower specificity (67%) in detecting congested ovarian veins [42]. In this study, the presence of retrograde ovarian vein filling was demonstrated in only 66% cases by MRV when compared to gold standard phlebography. Time-resolved MRA (TR-MRA) has been proven to be a quick and noninvasive technique that allows visualization of the physiologic blood flow dynamics and is predicted to be helpful for the detection of PCS because of its presumed ability to accurately determine whether anterograde or retrograde flow in the ovarian vein is present. In two recent studies, TR-MRA has shown utility in demonstrating ovarian vein reflux and diagnosing PCS [43, 44]. Pandey et al. used 4D time-resolved angiography with central keyhole (TRAK) acquisition MR technique as a noninvasive alternate modality for diagnosing pelvic congestion syndrome. The technique achieves a high temporal resolution by sharing k-space and acquiring only the central lines (keyhole) repeatedly over a short period of time. Faster image acquisition allows for improved visualization of flow dynamics and vascularity. This technique is extremely useful for depicting early reversal of venous flow and incompetence of the ovarian vein. It also has the capability to better resolve the arterial phase helping in the detection of arterial feeders to the lesion and any underlying arteriovenous malformation [43]. Nutcracker Syndrome Nutcracker syndrome refers to compression of the left renal vein (LRV) by the superior mesenteric artery and aorta. Patients typically present with left flank pain and associated symptoms of pelvic congestion. Hematuria is frequently

25

Venous Imaging: Techniques, Protocols, and Clinical Applications

present, and vulvar or lower extremity varices are seen in a subset of patients. Diagnosing nutcracker phenomenon can be done by anatomic assessment, with finding LRV compression between the aorta and SMA, a ratio between the distended and narrowed portions of the LRV of >1.5 for the anteroposterior diameter and a sharp rather than 90° branching angle of the SMA from the aorta, all indicating the phenomenon. The standard (although invasive) method of physiologic assessment is measuring the reno-caval pressure gradient during phlebography. If the pressure gradient is >3 mmHg, the patient is considered to have venous hypertension. A noninvasive alternative is to measure the peak velocities in the distended and narrowed segments of the LRV using Doppler ultrasonography, and then calculating a velocity ratio. However, quantitative analysis of the venous flow by Doppler sonography has potential pitfalls. It is difficult to obtain a spectral Doppler sampling from an entrapped segment of the LRV in the aortomesenteric space. TR-MRA findings compatible with nutcracker syndrome include the angle of the superior mesenteric artery (SMA) to the aorta (normal > 60°), LRV diameter on the basis of coronal maximum-intensity-projection images, and continuity of flow from the LRV into the IVC [44]. Wong et al. showed the usefulness of FSE T2-weighted MRI in diagnosing nutcracker syndrome. Hyperintense LRV on MR FSE T2WI indicated marked flow stagnation, and this inferred the presence of venous hypertension and suggested a diagnosis of nutcracker syndrome [45].

May-Thurner Syndrome Iliac vein compression syndrome (IVCS), also termed as May-Thurner syndrome or Cockett syndrome, is caused by compression of the left common iliac vein between the right common iliac artery and overlying vertebrae. Pulsatile wall compression of the vein induces replacement of normal intima and media of vein is largely replaced by well-organized connective tissue covered with endothelium that could cause DVT or venous hypertension without thrombosis in the left lower extremity. The syndrome most commonly presents as DVT; however, patients also can present with left-sided leg pain, swelling, and venous insufficiency without a thrombosis, but these occur less frequently [46, 47]. Conventional venography is the gold standard for IVCS diagnosis; however, different modalities have been shown to demonstrate the compression successfully as well. Different MRV types are available for evaluation of lower extremity venous system with their limitations. TOF technique is susceptible to flow artifacts and saturation and long acquisition times are needed. Basically, two different contrast-enhanced techniques using subtraction of arterial phase from the venous-arterial equilibrium phase and direct

359

contrast-enhanced venous phase have been investigated. Capillary passage of extracellular contrast during the arterial phase reduces venous return resulting in a reduced venous signal. Furthermore, a selective venous enhancement without concomitant arterial display is often difficult to obtain in peripheral vascular segments in this technique. Direct contrast-injection 3D MRV seems to overcome these problems [48]. Ruehm et al. reported that direct CEMRV has sensitivity and specificity values over 90% compared to conventional venography in the evaluation of varicosities, postthrombotic changes of the lower extremity venous system [49]. Direct CEMRV performed with diluted and long-lasting automated injection of the extracellular contrast agent from the pedal veins seems to be an improved imaging technique of lower extremity deep venous system. In terms of equipment and application, this technique is simple and feasible at any center where lower extremity contrast-enhanced 3D MRA is being performed in routine clinical practice while providing a practical approach with better image quality.

Renal Vein Thrombosis The term renal vein thrombosis (RVT) is used to describe the presence of thrombus in the major renal veins or their tributaries. This condition may either present with acute symptoms or go unnoticed because of lack of symptoms until a complication like pulmonary embolism or worsening renal function draws attention to it. The etiology of RVT is variable, but can be extrinsic or intrinsic. The intrinsic form is triggered by an intrarenal thrombotic event precipitated by acidosis, hemoconcentration, or arteriolar constriction. In adults, this process is typically the result of an underlying renal neoplasm. Additional intrinsic causes include membranous glomerulonephritis, pyelonephritis, amyloidosis, polyarteritis nodosa, sickle cell anemia, cardiac disease/low flow states, trauma, diabetic nephropathy, lupus nephropathy, coagulopathy, dehydration, or trauma. Extrinsic processes include umbilical vein catheterization, extension of IVC thrombosis, pancreatitis, retroperitoneal fibrosis, metastasis, and pancreatic tail carcinoma. MRI characterizes renal vein and IVC involvement by RCC with a higher accuracy for staging than computed tomography [50]. Standard sequences supplemented with 3D gadolinium-enhanced images provide high contrast and spatial resolution. In one study, MRA could delineate the entire course of the renal vessels in 88% of cases compared to 58% with Color Doppler ultrasound and 43% on spin-echo MRI. Similarly, the anatomic variants, vessel displacement, collateral circulation, and neoplastic vessel infiltration were demonstrated more accurately by MRA [51]. Laissy et al. study the performance of gadolinium-enhanced TOF MRV in 26 patients with RCC and tumor thrombus. For detection of venous thrombus, the sensitivity and specificity are

360

A.H. Davarpanah et al.

100 and 96%, respectively [52]. Using gadolinium-based contrast for evaluation of bland versus neoplastic thrombus, sensitivity and specificity of 89 and 96%, respectively, are achieved [52].

Portal Vein Obstruction Portal vein occlusion may occur with tumor or bland thrombus or it may occur secondary to extrinsic compression. Infectious, inflammatory, and malignant conditions of the abdomen are the most commonly implicated local risk factors for the development of portal vein thrombosis. In adults, acute appendicitis, cholecystitis, acute necrotizing pancreatitis, cholangitis, diverticulitis, and perforated peptic ulcers all have been reported to cause portal vein thrombosis. Extrinsic compression most commonly is due to metastatic liver disease but also may occur with benign masses, such as hemangiomas. Lobar or segmental portal vein obstruction by tumor may cause discrete wedge-shaped regions of increased intensity on T2-weighted and Gd-enhanced images [53]. Tumor and bland thrombus can be distinguished from each other by the observation that tumor thrombus is higher in signal on T2-weighted images, is intermediate in signal intensity on TOF, and enhances with Gd. In comparison, bland thrombus is low in signal intensity on both T2-weighted and TOF images and does not enhance with Gd. Portal vein thrombosis can be demonstrated using bright blood techniques (e.g., TOF and three-dimensional GRE sequences and black blood techniques, e.g., spin echo with superior and inferior saturation pulses). After administration of Gd, parenchymal manifestations of portal vein compromise are seen on MRI as transient areas of increased enhancement during the arterial phase that correspond to areas of decreased portal perfusion during the portal venous phase. This phenomenon results from the compensatory autoregulatory mechanism geared toward maintaining the net parenchymal perfusion by augmenting hepatic arterial blood supply [54]. Several authors have demonstrated that contrast-enhanced threedimensional MRA is as accurate as digital subtraction angiography (DSA) in assessing the portal venous system and determining surgical respectability in patients with pancreatobiliary tumors [53, 55].

Upper Extremity and Central Vein Evaluation Primary indications for MRV/MRI of the chest and upper extremity include defining SVC and central venous obstruction and invasion, venous thrombosis, stenosis, occlusion, and venous access planning. In patients presenting with abrupt swelling of the arms or facial swelling suggesting SVC syndrome, both noncontrast- and contrast-enhanced MRV are essential diagnostic tools.

Fig. 25.9 Coronal SHARP post contrast showing right internal jugular vein thrombus

Central venous occlusion often results in congestion, edema, and venous hypertension. The underlying cause varies, but previous radiation therapy, extrinsic mass/compression, or inflammation frequently is present. Common central venous occlusive conditions include superior vena cava (SVC) and IVC syndrome. SVC syndrome may be the result of a complete or partial occlusion of the SVC or its tributaries. Eighty to ninety percent of secondary obstructions are neoplastic in origin. Common offending neoplasms are bronchogenic carcinoma (greater than 50%), lymphoma, and mediastinal tumors. Granulomatous diseases, aneurysms, constrictive pericarditis, and substernal goiter are among the non-neoplastic causes. Catheter-induced occlusion or stenosis has become a more frequent cause of SVC syndrome. IVC syndrome can have intrinsic or extrinsic etiologies. Intrinsic caval occlusion typically has a neoplastic etiology (leiomyoma, leiomyosarcoma, and endothelioma), but may be non-neoplastic (congenital membrane). Extrinsic obstruction often occurs at the mid-IVC as a result of enlarged lymph nodes or an adjacent retroperitoneal, renal, pancreatic, or hepatic mass. Functional obstruction can result from a pregnant uterus, valsalva maneuver, or supine positioning with a large abdominal mass. Whereas ultrasound readily detects acute thrombus within the arms, MRV more readily visualizes more central thrombus (Figs. 25.9 and 25.10). Many patients with end-stage renal disease undergo both short-term and long-term dialysis via indwelling catheters. Initially, a multitude of suitable access sites are available for catheter placement. However, over time, the number and quality of accessible sites diminish due to multiple stenoses and thrombosis associated with long-standing catheterization. In such patients, MRV readily evaluates potential access sites.

25

Venous Imaging: Techniques, Protocols, and Clinical Applications

Fig. 25.10 Partition CEMRA/V showing SVC occlusion secondary to thrombus

Fig. 25.11 STARFIRE sequence showing DVT

Finn et al. [56], using only TOF imaging [10] in three orthogonal planes, reported excellent correlation between MRV and contrast venography. The authors acknowledged that collateral vessels were better appreciated with contrast venography; otherwise, there was essentially complete agreement between the two imaging modalities. The major disadvantage of TOF acquisitions is the relatively long acquisition time. In addition, these techniques suffer from numerous artifacts, including decreased signal intensity in the vessel due to slow flow, loss of signal secondary to turbulent flow, and pulsation artifacts, especially in the SVC. Koktzoglou’s noncontrast STARFIRE technique [17] has been utilized for both lower limb DVT and central venous mapping. Initial results look promising (Fig. 25.11). Temporally resolved MRA/MRV may prove to be a useful alternative to standard gadolinium-enhanced MRV in patients with significant renal impairment as significant dose reductions of gadolinium are possible [4–6]. Studies concerning

361

NSF suggest that in addition to using a more stable gadoliniumcontaining compound, dose reduction also reduces the probability of acquiring NSF, suggesting that it may be possible to safely image patients with poor renal function [4]. Kim et al., in a retrospective analysis of 27 consecutive patients undergoing both conventional gadolinium-enhanced MRV and temporally resolved MRV, demonstrated that the addition of the time-resolved images improved specificity in the detection of venous occlusions and improved reader confidence while actually reducing image interpretation time [56– 58]. Unfortunately, the time-resolved imaging, while sensitive for occlusion (95%), was only moderately specific (56%). More recently, several investigators have reported findings with SSFP imaging as a means to assess the central veins. SSFP has several potential advantages over traditional TOF imaging [10], including (a) faster acquisition times due to the shortened repetition time, (b) less flow artifacts due to the incorporation of balanced gradients in the pulse sequence, and (c) better contrast to noise between flowing blood and the adjacent stationary tissue. Like TOF imaging, SSFP imaging does not require intravenous contrast. A primary cause for the reduced sensitivity of the SSFP pulse sequence appeared to be the variable signal intensity of thrombus over time [10]. Acute thrombus displayed increased signal making it relatively isointense to blood. Older thrombus was more readily detected presumably due to its longer T1 value. If gadoliniumbased contrast agents are unable to be administered, we currently use a combination of axial and sagittal, true FISP, black blood (e.g., Haste or double-inversion IR), and STARFIRE images [17]. In addition, cardiosynchronous acquisitions are usually obtained in the axial, saggital, and coronal planes through the SVC. Although these studies require longer acquisition and interpretation times, clinically useful information is routinely obtained noninvasively.

Pulmonary Vein Evaluation MRA has become an established modality for pulmonary vein (PV) evaluation prior to and following RF ablation in patients with atrial fibrillation. Contrast material-enhanced MR angiographic techniques are primarily used for PV evaluation because of their high spatial resolution and reliability [59]. In patients who are pregnant or at risk of NSF, the use of gadolinium-based contrast material is contraindicated. For these patients, unenhanced MR angiographic techniques must be of equal quality and reliability as the contrastenhanced MR angiographic techniques. Balanced SSFP imaging techniques have been used to evaluate the thoracic vasculature, including the coronary arteries and thoracic aorta, because of their inherent high signal-to-noise and contrast-to-noise ratios. While singleshot techniques are rapid and useful for the urgent evaluation

362

of patients with suspected acute aortic syndromes or patients incapable of extended breath holding, 3D techniques with near-isotropic resolution are preferred for evaluation of smaller vessels, such as PVs. Free-breathing, navigator-gated 3D SSFP techniques have been developed to evaluate the coronary and renal arteries. Francois et al. [59] demonstrated that unenhanced freebreathing, T2-segmented 3D SSFP imaging sequence for PV assessment is comparable with time-resolved 3D CEMRA in patients in the assessment of patients prior to and following RF ablation for atrial fibrillation. Qualitatively, the images obtained were not different between the techniques. PV variants detected by using CEMRA were identified at 3D SSFP imaging as well. No significant difference in PV measurements was found between the 3D SSFP and contrastenhanced MR angiograms, including assessment of PV stenosis (Fig. 25.4). While both CT and MR angiography are commonly used for PV imaging [59], CT angiography has substantial drawbacks, including the use of ionizing radiation and nephrotoxic [4–6] contrast agents. Because patients with atrial fibrillation return for follow-up examinations at frequent short-term intervals prior to and following RF ablation, MRA is frequently preferred. Advances in MR sequences have made MRA an excellent noninvasive method of evaluating the vasculature. MRA is particularly suited for evaluating the PVs, especially with the use of parallel imaging and time-resolved techniques to offset the effects of cardiac motion. MRA, when combined with other nonangiographic MRI techniques, can be used as the primary or sole imaging modality in the evaluation of patients with atrial fibrillation prior to and following RF ablation. In particular, TRMRV can produce purely pulmonary venous-phase images, free from overlap of other vascular structures, such as the pulmonary arteries or aorta. An important point is that pulmonary vein ostia have to be measured accurately using multiplanar reformatting to generate orthogonal dimensions. In addition to providing morphologic information on the PVs, cardiac MR can be used to assess cardiac function and help detect evidence of myocardial scar. Because of the risk of causing atrioesophageal fistulas during RF ablation, it is also important to define the relationship between the atrium and esophagus with cross-sectional imaging prior to therapy. Because the contrast-enhanced technique used in this study is a subtraction technique, to optimize visualization of the vasculature, extravascular structures are rarely noted. However, the 3D SSFP technique permits evaluation of the vasculature in addition to extravascular anatomy, including the detection of extracardiac pathologic anomalies, such as cardiac masses, foregut cysts, lung nodules, lymphadenopathy, and hiatal hernias.

A.H. Davarpanah et al.

Importantly, secondary complications of the ablation procedure can be readily assessed with both noncontrast and contrast-enhanced MR PV mapping, including left atrial appendage thrombus.

Future Directions Imaging at field strengths higher than the conventional 1.5 T, such as with the increasingly available 3-T systems, has different benefits and drawbacks for the various nonenhanced MR techniques [3]. The intrinsically higher signal-to-noise ratio at 3 T does lend itself well to the use of parallel imaging. Implementation of parallel imaging can be particularly beneficial to nonenhanced MR angiographic methods by reducing acquisition times and consequently decreasing undesirable blurring and motion artifacts. Most applications of parallel imaging suffer a trade-off of reduced signal for shorter scan times; however, partial Fourier FSE with parallel imaging allows for compensating benefits that include a reduction in T2 blurring. With shorter acquisition times, multistation imaging with nonenhanced methods becomes feasible.

Conclusion Using a myriad of techniques, MRV has proven to be a useful tool in the assessment of venous abnormalities. Developments on the horizon include techniques that provide time-resolved imaging for assessment of flow dynamics by using nonenhanced approaches.

References 1. Prince MR. Gadolinium-enhanced MR aortography. Radiology. 1994;191(1):155–64. 2. Spritzer CE. Progress in MR imaging of the venous system. Perspect Vasc Surg Endovasc Ther. 2009;21(2):105–16. 3. Miyazaki M, Lee VS. Nonenhanced MR angiography. Radiology. 2008;248(1):20–43. 4. Prince MR, Arnoldus C, Frisoli JK. Nephrotoxicity of high-dose gadolinium compared with iodinated contrast. J Magn Reson Imaging. 1996;6(1):162–6. 5. Marckmann P, Skov L, Rossen K, Heaf JG, Thomsen HS. Casecontrol study of gadodiamide-related nephrogenic systemic fibrosis. Nephrol Dial Transplant. 2007;22(11):3174–8. 6. Cowper SE, Robin HS, Steinberg SM, Su LD, Gupta S, LeBoit PE. Scleromyxoedema-like cutaneous diseases in renal-dialysis patients. Lancet. 2000;356(9234):1000–1. 7. Guerrero A, Montes R, Munoz-Terol J, et al. Peripheral arterial disease in patients with stages IV and V chronic renal failure. Nephrol Dial Transplant. 2006;21(12):3525–31. 8. Ito K, Koike S, Shimizu A, et al. Portal venous system: evaluation with unenhanced MR angiography with a single-breath-hold

25

9.

10.

11.

12. 13. 14.

15.

16.

17.

18.

19.

20. 21. 22.

23.

24.

25.

26.

27.

28.

Venous Imaging: Techniques, Protocols, and Clinical Applications ECG-synchronized 3D half-Fourier fast spin-echo sequence. Ajr. 2008;191(2):550–4. Ono A, Murase K, Taniguchi T, et al. Deep venous thrombosis: diagnostic value of non-contrast-enhanced MR venography using electrocardiography-triggered three-dimensional half-Fourier FSE. Magn Reson Med. 2010;64(1):88–97. Wehrli FW, Shimakawa A, Gullberg GT, MacFall JR. Time-offlight MR flow imaging: selective saturation recovery with gradient refocusing. Radiology. 1986;160(3):781–5. Kirchhof K, Welzel T, Jansen O, Sartor K. More reliable noninvasive visualization of the cerebral veins and dural sinuses: comparison of three MR angiographic techniques. Radiology. 2002;224(3): 804–10. Liauw L, van Buchem MA, Spilt A, et al. MR angiography of the intracranial venous system. Radiology. 2000;214(3):678–82. Laissy JP, Trillaud H, Douek P. MR angiography: noninvasive vascular imaging of the abdomen. Abdom Imaging. 2002;27(5):488–506. Carr JC, Simonetti O, Bundy J, Li D, Pereles S, Finn JP. Cine MR angiography of the heart with segmented true fast imaging with steady-state precession. Radiology. 2001;219(3):828–34. Liu X, Berg N, Sheehan J, et al. Renal transplant: nonenhanced renal MR angiography with magnetization-prepared steady-state free precession. Radiology. 2009;251(2):535–42. Stafford RB, Sabati M, Mahallati H, Frayne R. 3D non-contrastenhanced MR angiography with balanced steady-state free precession Dixon method. Magn Reson Med. 2008;59(2):430–3. Edelman RR, Koktzoglou I. Unenhanced flow-independent MR venography by using signal targeting alternative radiofrequency and flow-independent relaxation enhancement. Radiology. 2009;250(1):236–45. Miyazaki M, Sugiura S, Tateishi F, Wada H, Kassai Y, Abe H. Noncontrast-enhanced MR angiography using 3D ECG-synchronized half-Fourier fast spin echo. J Magn Reson Imaging. 2000;12(5): 776–83. Fraser DG, Moody AR, Morgan PS, Martel AL, Davidson I. Diagnosis of lower-limb deep venous thrombosis: a prospective blinded study of magnetic resonance direct thrombus imaging. Annals of internal medicine. 2002;136(2):89–98. Spritzer CE, Blinder RA. Practical aspects of vascular imaging using MRI. Crit Rev Diagn Imaging. 1990;31(2):145–85. Bremerich J, Bilecen D, Reimer P. MR angiography with blood pool contrast agents. European radiology. 2007;17(12):3017–24. Baxter AB, Melnikoff S, Stites DP, Brasch RC. AUR Memorial Award 1991. Immunogenicity of gadolinium-based contrast agents for magnetic resonance imaging. Induction and characterization of antibodies in animals. Investigative radiology. 1991;26(12):1035–40. Li W, Salanitri J, Tutton S, et al. Lower extremity deep venous thrombosis: evaluation with ferumoxytol-enhanced MR imaging and dual-contrast mechanism--preliminary experience. Radiology. 2007;242(3):873–81. Fowkes FJ, Price JF, Fowkes FG. Incidence of diagnosed deep vein thrombosis in the general population: systematic review. Eur J Vasc Endovasc Surg. 2003;25(1):1–5. Spritzer CE, Sostman HD, Wilkes DC, Coleman RE. Deep venous thrombosis: experience with gradient-echo MR imaging in 66 patients. Radiology. 1990;177(1):235–41. Spritzer CE, Norconk JJ, Jr., Sostman HD, Coleman RE. Detection of deep venous thrombosis by magnetic resonance imaging. Chest. 1993;104(1):54–60. Sampson FC, Goodacre SW, Thomas SM, van Beek EJ. The accuracy of MRI in diagnosis of suspected deep vein thrombosis: systematic review and meta-analysis. European radiology. 2007; 17(1):175–81. Evans AJ, Sostman HD, Knelson MH, et al. 1992 ARRS Executive Council Award. Detection of deep venous thrombosis: prospective

363

29.

30.

31.

32.

33.

34. 35.

36.

37.

38. 39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

comparison of MR imaging with contrast venography. Ajr. 1993; 161(1):131–9. Erdman WA, Jayson HT, Redman HC, Miller GL, Parkey RW, Peshock RW. Deep venous thrombosis of extremities: role of MR imaging in the diagnosis. Radiology. 1990;174(2):425–31. Spuentrup E, Botnar RM, Wiethoff AJ, et al. MR imaging of thrombi using EP-2104R, a fibrin-specific contrast agent: initial results in patients. European radiology. 2008;18(9):1995–2005. Bertsch NM, Mastrobattista JM, Kawashima A, Kramer LA. Antepartum bilateral ovarian vein thrombosis: magnetic resonance imaging diagnosis. Am J Perinatol. 1997;14(10):597–9. Twickler DM, Setiawan AT, Evans RS, et al. Imaging of puerperal septic thrombophlebitis: prospective comparison of MR imaging, CT, and sonography. Ajr. 1997;169(4):1039–43. Rordorf G, Koroshetz WJ, Copen WA, et al. Regional ischemia and ischemic injury in patients with acute middle cerebral artery stroke as defined by early diffusion-weighted and perfusion-weighted MRI. Stroke. 1998;29(5):939–43. Erdman WA, Parkey RW. MR imaging of deep venous thrombosis. Ajr. 1990;155(4):897. Kubik-Huch RA, Hebisch G, Huch R, Hilfiker P, Debatin JF, Krestin GP. Role of duplex color Doppler ultrasound, computed tomography, and MR angiography in the diagnosis of septic puerperal ovarian vein thrombosis. Abdom Imaging. 1999;24(1):85–91. Shimada K, Isoda H, Okada T, et al. Non-contrast-enhanced MR portography with time-spatial labeling inversion pulses: comparison of imaging with three-dimensional half-fourier fast spin-echo and true steady-state free-precession sequences. J Magn Reson Imaging. 2009;29(5):1140–6. Jacob AG, Driscoll DJ, Shaughnessy WJ, Stanson AW, Clay RP, Gloviczki P. Klippel-Trenaunay syndrome: spectrum and management. Mayo Clin Proc. 1998;73(1):28–36. Cha SH, Romeo MA, Neutze JA. Visceral manifestations of Klippel-Trenaunay syndrome. Radiographics. 2005;25(6):1694–7. Vilanova JC, Barcelo J, Smirniotopoulos JG, et al. Hemangioma from head to toe: MR imaging with pathologic correlation. Radiographics. 2004;24(2):367–85. Elsayes KM, Menias CO, Dillman JR, Platt JF, Willatt JM, Heiken JP. Vascular malformation and hemangiomatosis syndromes: spectrum of imaging manifestations. Ajr. 2008;190(5):1291–9. Ganeshan A, Upponi S, Hon LQ, Uthappa MC, Warakaulle DR, Uberoi R. Chronic pelvic pain due to pelvic congestion syndrome: the role of diagnostic and interventional radiology. Cardiovasc Intervent Radiol. 2007;30(6):1105–11. Asciutto G, Mumme A, Marpe B, Koster O, Asciutto KC, Geier B. MR venography in the detection of pelvic venous congestion. Eur J Vasc Endovasc Surg. 2008;36(4):491–6. Pandey T, Shaikh R, Viswamitra S, Jambhekar K. Use of time resolved magnetic resonance imaging in the diagnosis of pelvic congestion syndrome. J Magn Reson Imaging. 2010;32(3):700–4. Kim CY, Miller MJ, Jr., Merkle EM. Time-resolved MR angiography as a useful sequence for assessment of ovarian vein reflux. Ajr. 2009;193(5):W458-63. Wong HI, Chen MC, Wu CS, et al. The usefulness of fast-spin-echo T2-weighted MR imaging in Nutcracker syndrome: a case report. Korean J Radiol. 2010;11(3):373–7. Oguzkurt L, Ozkan U, Tercan F, Koc Z. Ultrasonographic diagnosis of iliac vein compression (May-Thurner) syndrome. Diagnostic and interventional radiology (Ankara, Turkey). 2007;13(3):152–5. Wolpert LM, Rahmani O, Stein B, Gallagher JJ, Drezner AD. Magnetic resonance venography in the diagnosis and management of May-Thurner syndrome. Vasc Endovascular Surg. 2002;36(1):51–7. Gurel K, Gurel S, Karavas E, Buharalioglu Y, Daglar B. Direct contrast-enhanced MR venography in the diagnosis of May-Thurner Syndrome. Eur J Radiol. 2010.

364 49. Ruehm SG, Wiesner W, Debatin JF. Pelvic and lower extremity veins: contrast-enhanced three-dimensional MR venography with a dedicated vascular coil-initial experience. Radiology. 2000;215(2): 421–7. 50. Heiss SG, Shifrin RY, Sommer FG. Contrast-enhanced three-dimensional fast spoiled gradient-echo renal MR imaging: evaluation of vascular and nonvascular disease. Radiographics. 2000;20(5):1341–52; discussion 53–4. 51. Pfluger T, Czekalla R, Hundt C, et al. MR angiography versus color Doppler sonography in the evaluation of renal vessels and the inferior vena cava in abdominal masses of pediatric patients. Ajr. 1999;173(1):103–8. 52. Laissy JP, Menegazzo D, Debray MP, et al. Renal carcinoma: diagnosis of venous invasion with Gd-enhanced MR venography. European radiology. 2000;10(7):1138–43. 53. McFarland EG, Kaufman JA, Saini S, et al. Preoperative staging of cancer of the pancreas: value of MR angiography versus conventional angiography in detecting portal venous invasion. Ajr. 1996; 166(1):37–43.

A.H. Davarpanah et al. 54. Morrin MM, Pedrosa I, Rofsky NM. Magnetic resonance imaging for disorders of liver vasculature. Top Magn Reson Imaging. 2002;13(3):177–90. 55. Smedby O, Riesenfeld V, Karlson B, et al. Magnetic resonance angiography in the resectability assessment of suspected pancreatic tumours. European radiology. 1997;7(5):649–53. 56. Finn JP, Zisk JH, Edelman RR, et al. Central venous occlusion: MR angiography. Radiology. 1993;187(1):245–51. 57. Tanju S, Sancak T, Dusunceli E, Yagmurlu B, Erden I, Sanlidilek U. Direct contrast-enhanced 3D MR venography evaluation of upper extremity deep venous system. Diagnostic and interventional radiology (Ankara, Turkey). 2006;12(2):74–9. 58. Thornton MJ, Ryan R, Varghese JC, Farrell MA, Lucey B, Lee MJ. A three-dimensional gadolinium-enhanced MR venography technique for imaging central veins. Ajr. 1999;173(4):999–1003. 59. Francois CJ, Tuite D, Deshpande V, Jerecic R, Weale P, Carr JC. Pulmonary vein imaging with unenhanced three-dimensional balanced steady-state free precession MR angiography: initial clinical evaluation. Radiology. 2009;250(3):932–9.

Pediatric MR Angiography: Principles and Applications

26

Bharathi D. Jagadeesan and David N. Loy

Introduction Magnetic resonance angiography (MRA) is perhaps the most valuable in children when compared to all other patient groups. While CT angiography has made vast strides in imaging the neurovascular system in adults, these advances do not lend themselves to easy application in the pediatric patient population primarily due to the risk from the considerable radiation exposure that CT angiographic procedures entail. Likewise, conventional catheter angiography procedures also result in radiation exposure. While many CTA and conventional digital subtraction angiography (DSA) procedures have been shown to result in radiation exposure which is less than the threshold for deterministic effects such as epilation or erythema [1–3], they nevertheless result in an increased risk for malignancy secondary to stochastic effects. Although the risk from any single procedure may be low, children with neurovascular disorders are often subject to multiple CTAs or angiograms in a year, resulting in large cumulative radiation doses. The estimated incidence of new cancers resulting from these procedures was 890.6/100,000 exposed males and 1,222.5/100,000 exposed females [4]. Swoboda et al. [4] found that the average 9-year-old patient with a neurovascular accumulates radiation exposure equivalent to 235 frontal and 177 lateral skull radiographs over the year from CT and DSA procedures. Radiation risks are higher for the youngest children and for females. Radiation risks for newborn males and females per unit dose of radiation are approximately three and six times higher than for adult patients, respectively [5, 6]. Most CT angiography studies are now performed on multidetector CT scanners. Likewise, multiphase CT imaging also results in increased radiation doses [7]. Improvements in

B.D. Jagadeesan, MD () • D.N. Loy, MD, PhD Washington University School of Medicine, Barnes-Jewish Hospital, Mallinckrodt Institute of Radiology, Saint Louis, MO, USA e-mail: [email protected]

image quality with these techniques come at a cost of increased radiation dose. Recent efforts by vendors to develop better radiation dose estimates in children have been an important step in support of the “Image Gently” campaign [8]. MRA should play an increasingly crucial role in the diagnosis, classification, prognostication, and follow-up of pediatric neurovascular disorders to reduce radiation exposure in children. Numerous unique challenges confront MRA in this developing patient population. These include the small size of intracranial vascular structures, the hyper-dynamic circulation in children with various multisystem disorders, and the immature or compromised renal function that is encountered in certain pediatric patient populations. Pediatric MRA techniques should possess high spatial and temporal resolution and should also be feasible without the administration of intravenous contrast when possible. In this chapter, we briefly review the basic principles of MRA as they specifically pertain to pediatric neuroimaging. We can then proceed to classifying and reviewing the common pediatric neurovascular disorders and the role of MRA techniques in each.

Principles of Magnetic Resonance Angiography in Children All MRA techniques strive to accentuate the contrast ratio between the intraluminal blood and the surrounding background tissue. They aim to achieve this by accentuating the signal originating from the flowing blood and by attenuating the signal originating from the stationary tissues. Some techniques rely on a combination of both these endeavors. Hence, most neurovascular imaging can be considered to be “white blood imaging” where the flowing blood is hyperintense. Black blood imaging which is popular elsewhere in the body for vascular imaging plays almost no role in routine neurovascular imaging with the notable exception of MR bold venography, which is useful in select scenarios.

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_26, © Springer Science+Business Media, LLC 2012

365

366

Techniques in Pediatric Magnetic Resonance Imaging Noncontrast Magnetic Resonance Angiography of the Head and Neck The basic noncontrast MRA techniques are time-of-flight MRA, phase contrast MRA, and BOLD MRI venography or susceptibility weighted imaging (SWI).

Time-of-Flight MRA Time-of-flight (TOF) angiography relies on the enhancement produced by imaging the inflow of fresh unsaturated protons in the blood stream into a slice or slab of tissue that has been presaturated with repetitive RF pulses with short TRs [9]. When compared to 2D TOF MRA, 3D TOF MRA offers much higher spatial resolution but it is relatively insensitive to slow flow. Modifications to 3D TOF imaging such as Multiple Overlapping Thin Slice Acquisition (MOTSA), addition of magnetization transfer, and the use of ramped flip angles through the imaging volume can considerably improve the signal in slow flow lesions [10–12]. Additionally, a few authors have also advocated the intravenous administration of a small dose (0.5 mL) of gadolinium, a T1 shortening agent, in order to improve contrast in distal intracranial branches [13]. Nevertheless, TOF MRA suffers from inherently poor sensitivity to turbulent blood flow in vascular beds with complex anatomy. Therefore, time-of-flight imaging is of limited use in complex head and neck vascular malformations or for imaging the carotid arteries. TOF MRA can also be misleading in patients with the evaluation of thrombosed vessels since thrombi with short T1 signal will appear hyperintense similar to flowing blood, making it impossible to distinguish between the two [14]. Despite these caveats, the lack of a need for intravenous contrast administration and the excellent spatial resolution with 3D techniques makes TOF angiography an attractive MRA technique for imaging the circle of Willis in children. Unlike techniques dependent on intravenous contrast administration, TOF can also be safely performed in children with renal dysfunction. TOF MRA studies can also be repeated multiple times in the same session or the same day (i.e., preand postoperatively) when indicated. This is invaluable for the study of children, especially those who cannot be sedated or those with contraindications to intravenous contrast including sickle cell disease. The more widespread availability of 3 Tesla scanners and the increase in signal-to-noise that is inherent to higher field strengths has significantly improved the quality of TOF MRA in children. Recently, several investigators have also described techniques for the simultaneous acquisition of TOF MR arteriographic images and susceptibility weighted MR venographic

B.D. Jagadeesan and D.N. Loy

imaging data with a multi-echo sequence [15, 16]. This exciting development allows for the acquisition of instantly co-registered distinct arterial and venous maps of the cerebral vascular system without additional imaging time when compared to routine TOF imaging. These techniques are likely to have a major role in the evaluation of complex brain vascular malformations such as pial or dural arteriovenous malformations in the near future. Another recent development is hybrid opposite contrast MR angiography, which combines TOF and flow sensitive black blood contrasts [17]. This technique combines the excellent sensitivity of TOF techniques for high flow vessels with the excellent sensitivity of black blood techniques for slow flow vessels without increasing the imaging time. This development may increase the ability of TOF MRA to diagnose cerebral vasculopathies involving medium and small vessels but DSA comparison studies will be essential for validation.

Phase Contrast MRA Phase contrast imaging technique utilizes the difference in phase accumulation between stationary and moving protons, which are subjected to magnetic gradients. In order to exploit this difference, two opposing magnetic gradients, which are similar in magnitude and duration, are applied to a selected imaging slice in rapid succession. In the case of stationary protons within the excited tissue, position dependent phase differences, which accumulate along the axis of the first gradient are promptly neutralized after the application of the second gradient. However, in the case of protons, which are not stationary along the axis of these gradients, e.g. protons within a blood vessel oriented along the axis of the gradient, the phase accumulation is not reversed by application of the second gradient. This leads to the accumulation of a net positive or negative phase in mobile protons in flowing blood. The magnitude of this net phase accumulation depends upon the velocity of blood flow, whereas the sign of the net phase accumulation depends on the direction of blood flow. Needless to say phase differences are also similarly produced by higher order motion such as acceleration and jerk. The magnitude and the duration of the gradients can be adjusted to determine the range of velocities that can be encoded between the phase angles of −P and +P degrees. This step is called velocity encoding (VENC). In order to obtain technically adequate phase contrast images, it is essential to set an appropriate VENC. A VENC that is too high may result in poor signal from slow flowing vessels, whereas a VENC that is too low will result in signal aliasing in vessels with rapid flow. In the end, this technique results in the generation of a phase image, which shows the direction of blood flow and a magnitude image, the signal in which is proportional to the overall flow rate within the interrogated blood vessel [18]. Additionally, phase contrast imaging can also be used to generate flow velocity curves by continuously interrogating a

26 Pediatric MR Angiography: Principles and Applications

blood vessel during the entire cardiac cycle. Unlike time of flight imaging, in phase contrast imaging, it is very easy to distinguish between clotted blood with short T1 signal and flowing blood. Therefore phase contrast imaging is especially useful in imaging for cerebral venous sinus thrombosis in children, a not uncommon situation in neonates. Additionally, it can be used in other populations of children such as those with sickle cell disease in whom intravenous contrast administration is risky or contra-indicated. The sensitivity of phase contrast MRA (PCMRA) to the direction of blood flow is also very useful in evaluating arteriovenous malformations and collateral vessels. Both 3-D and 2-D PCMRA techniques are available. The relatively long acquisition times constitute a major drawback with PC-MRA techniques [19]. Recently, accelerated 4D methods for phase contrast imaging have been described which also enable in vivo demonstration hemodynamics as vascular flow streamlines [20]. Such depictions are likely to increase fundamental understanding of the evolution of pediatric neurovascular diseases including vessel wall shear stresses and collateral recruitment after large vessel occlusions. Phase contrast imaging has also been used in highly constrained projection reconstruction techniques (HYPR) to increase the fidelity of time-resolved MRA images [21]. The relatively long acquisition times constitute a major pitfall for PC-MRA techniques. However, parallel imaging techniques, improved coil performance, and 3-T field strengths are rendering these techniques faster and more clinically feasible.

Time-Resolved Contrast-Enhanced MRA in Children Contrast-enhanced MR angiography methods have grown rapidly in the past decade. In order to gain the maximum information from a contrast-enhanced MRA study, it is essential to distinctly image the arterial and venous phases during the first pass of the intravenously injected T1 shortening agent through the cerebral vasculature [22]. These techniques are often also referred to magnetic resonance digital subtraction angiography or MRDSA since they rely on subtraction of an initial non-contrast mask image of the area in question from the sequential phases of the contrast-enhanced study or with continuous subtractions of stagnate signal in sliding window techniques. Image acquisition during arterial and venous transit of contrast is traditionally timed using bolus tracking methods. A test bolus of contrast agent is injected prior to image acquisition. The MRA acquisition is triggered using a fixed delay after contrast injection (calculated from the test bolus injection) or the bolus is monitored and triggered in real time using MR fluoroscopy. Bolus timing methods are often successful in adults. However, in children, the test bolus dose (approximately 1 mL of contrast agent) is likely to constitute a larger fraction of the total amount of contrast that can be

367

used for imaging based on the child’s weight. Furthermore, greater variability in the heart rates and shorter arteriovenous circulation times in children make test bolus injections MRA techniques more challenging and often unfeasible [23]. Therefore, time-resolved MRA techniques with rapid continuous dynamic image acquisition are more commonly used in pediatric MRA studies. Broadly speaking, these techniques rely on centric or elliptical reordering of k space acquisition, partial acquisition and refreshment of k space, parallel imaging methods, radial sliding window reconstruction techniques and sliding mask subtraction algorithms [24–29]. These approaches have been used in various different combinations to increase the temporal resolution of these techniques while maintaining spatial resolution. Currently, sub-second acquisition of individual MRA frames is possible. Chooi et al. [30] studied 15 children with cerebrovascular malformations using MRDSA after hand injection of 1 or more doses of gadolinium contrast agent and found that MRDSA contributed significantly to avoiding catheter angiography (in 33% of patients), and justified catheter angiography in 27% of their patients. They also reported that MRDSA could be safely performed in neonates. However, MRDSA was limited in the evaluation of high flow vascular malformations of the brain and it could not be reliably used for treatment planning. Continuous improvement in acquisition techniques is ongoing. Most recently, time of arrival mapping techniques have been applied to three-dimensional time-resolved MRDSA with promising results [31]. However, MRDSA studies have not to date replaced conventional catheter-based DSA techniques in the evaluation of complex cerebral and cervical arteriovenous vascular malformations, especially for the evaluation of small residual shunts after treatment. In the following sections, we shall briefly review the role of these MRA techniques in the evaluation of specific neurovascular disorders in children. Brain vascular malformations in children: Pediatric cerebrovascular malformations consist of both diseases that are unique to the pediatric patient population such as vein of Galen malformations and Sturge–Weber syndrome as well as malformations that occur both in children and adults such as pial arteriovenous malformations (AVMs), dural arteriovenous malformations, dural arteriovenous fistulas (DAVFs), and developmental venous anomalies with or without associated cavernomas. The following sections consist of a brief review of the latter type of malformations followed by a more detailed review of the appropriate roles for MRA.

Arteriovenous Malformations Pediatric AVMs differ from their adult counterparts demonstrating more immature architecture [32, 33], a more diffuse pattern of AV shunting [34], and perhaps a higher cumulative

368

B.D. Jagadeesan and D.N. Loy

Fig. 26.1 Pial AVM in a 6-year-old child with seizures. Saggital (a), coronal (b) and axial (c) maximum intensity projections from a precontrast time of flight MR angiography showed a small periventricular arteriovenous malformation nidus (white arrow) supplied by the right pericallosal arteries and draining into the deep venous system. Postcontrast time-of-flight MR angiography in the same patient with identical saggital (d), coronal (e), and axial (f) maximum intensity projections shows improved contrast in the images but also significantly

increased venous contamination. Axial SWI image (g) shows hyperintense nidus (white arrow). Note the hyperintense blood products within the atrium of the right lateral ventricle (Asterisks, G, see A also). Anteroposterior (h) and oblique (i) DSA obtained during injection of the right vertebral artery demonstrates the nidus of the arteriovenous malformation, which is also supplied by posterior pericallosal branches of the right posterior cerebral artery

lifetime risk for hemorrhage [35]. Intraparenchymal hemorrhage constitutes the commonest presenting symptom in children with AVMs [35]. In a study of 26 children with nontraumatic intracerebral hemorrhage as the presenting event, Papadias et al. [36] found that AVMs represented the underlying etiology in seven of the patients. MRA was utilized to study these malformations and the authors opined that the overall performance of MRA was satisfactory in the detection of AVMs but did suffer from lack of temporal information and poor spatial resolution. Fasulakis et al. [37] directly compared the performance of 3D TOF MRA and DSA in nine children with brain AVMs. They reported agreement between these two modalities in four patients and discordant results in one patient. DSA demonstrated an additional supplying vessel not visible on MRA in one case. On the other

hand, MRA demonstrated an additional supplying vessel not described on DSA in two cases. Although there are few studies which compare MRA with DSA specifically in the pediatric population, it is likely that the overall performance of MRA in the detection of AVMs in children is similar to MRA in adults (Fig. 26.1). DSA remains the gold standard because of the superior temporal and spatial resolution. Aneurysms in Children: Aneurysms are much less common in children than in adults accounting for less than 5% of all detected aneurysms. In most cases, the etiology of these aneurysms is unknown. In the remaining patients, trauma, infectious disease, or connective tissue disorders may constitute the underlying etiology. These children usually present with subarachnoid hemorrhage or symptoms related to mass effect.

26 Pediatric MR Angiography: Principles and Applications

369

Fig. 26.2 Cerebral aneurysm. Nine-month-old child who presented to the hospital with nausea, vomiting, and poor appetite, her clinical examination had revealed a left VIth nerve palsy. Subsequent MRI examination of the head with time-of-flight MR angiography (a), susceptibility weighted imaging (b), and gadolinium-enhanced T1-weighted imaging (c) of the brain showed a large saccular aneurysm arising from the left internal carotid artery. There is thrombus layering in a dependent

manner within the aneurysm. Note: The thrombus is hyperintense on the time of flight maximum intensity projection image (a) and maybe confused for blood flow but is clearly identifiable as thrombus based on its hypointensity on the SWI image and the postcontrast MR angiogram. Digital subtraction angiography confirmed the large saccular aneurysm arising from the left internal carotid artery (d, e)

Saccular aneurysms represent the commonest morphologic subtype but giant aneurysms are considerably more common in children when compared to adults. With respect to infants, Buis et al. [38] conducted a retrospective review of 110 published articles describing a total of 131 aneurysms in children less than 1 year of age and found that most aneurysms occurred in the anterior circulation with the MCA being the most common vessel involved. They also reported a mean size of 1.8 ± 1.4 cm. Patients who presented with hemorrhage tended to be younger and had smaller aneurysms. They reported no gender bias. Traumatic pseudoaneurysms account for 14–39% of intracranial aneurysms in children [39] and they can occur with closed head injury, especially in the pericallosal or callosomarginal branches of the anterior cerebral arteries where these branches impact against the falx cerebri [40]. Most literature comparing MRA with DSA in adults generally suggest that MRA is excellent at the detection of aneurysms larger than 5 mm. It is likely that the performance of MRA in the detection of aneurysms in children is similar. Additionally, given that giant aneurysms constitute 16–29% of all aneurysms in children as compared to just 2% of

aneurysms in adults, MRA may be expected to a higher overall percentage of aneurysms in children compared to adults (Fig. 26.2). Allison et al. [41] reported that false-negative MRA studies occurred in the presence of spasm, slow flow or small aneurysm size in a study comparing MRA with DSA in children.

Vein of Galen Malformation In normal adults, the vein of Galen is a deep venous structure that drains the internal cerebral veins and the basal veins of Rosenthal. It empties into the straight sinus. The term, vein of Galen malformation, can be broadly used to describe any condition in which the vein is abnormally large or has abnormal morphology. Perhaps the best way to approach these malformations is to consider them as either high flow or low flow malformations. High flow malformations are the result of abnormal arteriovenous shunting, whereas low flow malformations are either the result of venous egress failure of the vein of Galen or from excessive venous inflow from developmental venous anomalies.

370

B.D. Jagadeesan and D.N. Loy

High Flow Vein of Galen Malformations Vein of Galen Aneurysmal Malformation By the 11th–12th week of gestation, a transient venous structure develops in the roof of the diencephalon, which receives drainage from the choroid plexus of the telencephalon. This structure is called the median vein of the prosencephalon or the vein of Markowski. With the development of the basal ganglia, the choroidal venous drainage shifts to the paired internal cerebral veins and the median vein of the prosencephalon regresses, except for its caudal extreme, which is joined by the internal cerebral veins and thalamostriate to form the vein of Galen. When this pattern is not established and there is arteriovenous shunting to the persistent embryonic median vein of the prosencephalon, the resultant malformation is called a vein of Galen aneurysmal malformation (VGAM) [42]. The underlying arteriovenous communications take place in the cistern of the velum interpositum and the quadrigeminal cistern. There is resultant ballooning of the median prosencephalic vein, which is thin walled and unsupported [42–44]. These VGAMs were subdivided into choroidal and mural subtypes [45, 46]. In the choroidal type (Fig. 26.3), there is a nidus supplied by the choroidal arteries, which drains into the primitive vein of the mesencephalon. In the mural type, arteriovenous fistula(s) are located in the inferolateral wall of the median vein of the prosencephalon. The communications occur in the anterior end of the median prosencephalic vein. Arterial supply to the malformation arises from those arteries that normally feed the structures in the vicinity, namely the arteries that feed the tela choroidea and the quadrigeminal plate. They may belong to either the anterior or prosencephalic group (the anterior cerebral, anterior choroidal, middle cerebral, and the posterolateral choroidal arteries), or the posterior or mesencephalic group (the posteromedial choroidal, posterior thalamoperforating, quadrigeminal, and superior cerebellar arteries) [42, 43]. The arterial supply that is recruited from the subependymal and thalamo-perforator arteries arises secondarily from a sump effect. These secondary arterial-feeders usually tend to disappear following treatment of the main feeders to the shunt. Persistence of a primitive limbic arterial arch, which connects the anterior and posterior choroidal vessels, may also be noted [47]. The presence of a high-flow arteriovenous shunt also results in the persistence of fetal patterns of venous drainage from the deeper structures of the brain. The thalamostriate veins in a VGAM no longer drain afferent into the median vein of the prosencephalon but may drain into the posterior and inferior thalamic veins and eventually drain in to a subtemporal vein, or the lateral mesencephalic vein and into the superior petrosal sinuses. This pattern results in the typical

Fig. 26.3 Vein of Galen malformation in a 22-month-old child. Saggital (a) and axial (b) maximum intensity projection images from a noncontrast time-of-flight MRA study show an enlarged median prosencephalic vein fed by arterial branches from both the anterior and posterior cerebral arteries. The arterial feeders can be seen to complete a primitive limbic circle (a). The posterior choroidal feeders are depicted (b). Postcontrast saggital T1-weighted image (c) and sagittal reconstruction of a postcontrast susceptibility weighted image (d) from the same study show multiple enlarged draining vessels around the midbrain with the formation of multiple large venous aneurysms. Interestingly, the axial pre- (e) and postcontrast SWI (f) images show not only bright signal in the enlarged veins with arterialized flow from arteriovenous shunting but also dark passively congested cortical veins, suggesting venous hypertension. Venous hypertension is commonly associated with ventriculomegaly in these patients

epsilon shape (“epsilon vein”) on the lateral views in conventional catheter angiograms [48, 49]. Efferent venous drainage from a VGAM is usually into the straight sinus. However, the VGAM nidus may also drain to some extent into the pontomesencephalic vein via the choroidal veins. The pontomesencephalic vein in turn usually drains into the falcine sinus with or without a stenosis at the sinovenous junction. This results in persistence of the falcine sinus [45]. The presence of a connection between the VGAM nidus and the pontomesencephalic vein may also result in pial venous drainage due to communications between the choroidal and striate venous systems. Documentation of

26 Pediatric MR Angiography: Principles and Applications

the drainage pattern is crucial since the presence of pial venous drainage excludes a patient from transvenous endovascular embolization [47].

Vein of Galen Aneurysmal Dilatation In contrast to vein of Galen arteriovenous malformations, the vein of Galen aneurysmal dilatation (VGAD) consists of a pial AVM, which drains into a normally formed vein of Galen. The pial AVM may be supratentorial or infratentorial. The clinical manifestations depend on the location of the AVM and the extent of outflow obstruction at the drainage of the Vein of Galen into the straight sinus. The VGAMs and VGAD are also classified according to the Yasargil system [50]. In the Yasargil classification, VGADs constitute the type IV malformations, whereas the VGAMs are classified as types I–III according to the arterial feeders. Dural Vein of Galen Dilatations: Vein of Galen dilatation can also occur due to development of dural arteriovenous shunting after straight sinus thrombosis in adults or thrombosis of the falcine vein or vein of Galen in children [51, 52].

Low Flow Vein of Galen Dilatations There can also be dilatation of a normally formed vein of Galen secondary to congestive heart failure or from drainage of a venous angioma known as a vein of Galen Varix [47].

Role of MRA in Vein of Galen Malformations In the light of the above discussion, it becomes clear that in order for any angiographic technique to be useful in the evaluation of vein of Galen malformations, the technique must be capable of differentiating between high flow and low flow vein of Galen malformations, and it should also be able to provide exquisite information regarding the arterial supply and the venous drainage pattern associated with vein of Galen malformations. Time-resolved dynamic contrastenhanced MRA technique is likely to offer the greatest detail if performed with high temporal and spatial resolution. Physiological immaturity in renal function often precludes intravenous administration of gadolinium agents in neonates. Hyperdynamic circulations and congestive cardiac failure associated with VGAMs also makes time-resolved MRA challenging. Recently, Chooi et al. [30] studied a set of 15 pediatric patients with MRDSA and found that the extremely high flow rates associated with VGAMs resulted in poor visualization of the surrounding feeding cerebral vessels.

371

They also opined that the MRDSA results could not be used for treatment planning in these children. In this setting, a high quality time-of-flight MRA may be the study of choice. Newer rapid phase contrast imaging techniques may also be useful. Time-of-flight imaging is excellent in the depiction of the arterial supply but may or may not reveal the venous drainage pattern in its entirety. In this context, we have also found SWI to be a useful tool, since the BOLD technique offers exquisite representation of the deep venous system of the cerebral hemispheres [53]. By reviewing both the TOF and SWI images, it may be possible to adequately evaluate vein of Galen malformations but the technique has not been applied enough in VGAMs to make conclusions at this time. Segmental neurovascular malformations in children: The segmental neurovascular malformations are a group of rare disorders that share several common features. According to Krings et al. [54], these features consist of their segmental distribution pattern, their link to the concept of neural crest development, the metameric origin and cephalic migration of the cells concerned, their evolution with growth of the child, and their varying expressivity. These malformations include several eponymous syndromes such as Sturge–Weber syndrome, Wyburn-Mason syndrome, Klippel–Trenaunay syndrome, Cobbs syndrome, and PHACES [55–58]. Of these, Wybrun-Mason or Dechaume–Blanc syndromes represent the occurrence of brain arteriovenous malformations, orbital arteriovenous malformations and maxillofacial vascular abnormalities in the same patient. Extracranial vascular malformations of the head and neck are also a part of these segmental malformations. Midline brain vascular malformations seem to be associated with midline facial vascular malformations and lateral malformations are associated with maxillary malformations. The role of MRA in the detection of the orbital components of these malformations is clearly evident but in the presence of cerebral and facial malformations on MRA, the orbital anomalies can usually be studied more thoroughly with ophthalmologic evaluations. The cerebral AVMs found in these malformations differ from isolated AVMs in their multifocality, recruitment of dural vessels, and their low shunt volumes [54, 59]. Sturge–Weber (SWS) syndrome: This syndrome also known as encephalo-trigeminal angiomatosis consists of a large unilateral leptomeningeal venous malformation which undergoes progressive calcification with accompanying atrophy of the involved cerebral hemisphere and an associated dermal venous malformation or facial port wine stain over the ipsilateral V1 division territory. Children with this syndrome typically present with the cosmetic deformity and epilepsy. There is also enlargement of the choroid plexus in the ipsilateral cerebral hemisphere and accompanying vascular malformations in the orbit. In the majority of affected children, the

372

Fig. 26.4 Sturge–Weber Syndrome in a 2-year-old child with seizures. Noncontrast axial head CT (a) shows cortical calcification involving the left frontoparietal convexity. There is corresponding signal loss on an axial SWI image (b) and leptomeningeal enhancement on a postcontrast T1-weighted image (c) from the same patient suggesting a diagnosis

diagnosis can be made on the basis of dermatological and neurological findings. However, occasionally, the neurologic findings are mild and the dermatologic findings may be absent. In such cases, neuroimaging can play an important role. Recently, MRI/MRA studies have become the neuroimaging study of choice. On these studies, the diagnostic imaging features are gyral contrast enhancement, leptomeningeal vessel enhancement, enlarged deep medullary veins, and an enlarged ipsilateral choroid plexus (Fig. 26.4). Among these features, enhancement of the leptomeningeal vessels is usually considered to be the most important feature [60, 61]. MR contrast venography plays a useful role in this setting by demonstrating thrombosed cortical veins and enlarged deep draining medullary veins. Recently, Hu et al. [62] evaluated the role of SWI or MR bold venography in imaging patient with SWS. In their series of 12 patients, they found that SWI provide complimentary information to contrast-enhanced MRI. Specifically, SWI showed superior depiction of abnormal transmedullary veins, periventricular veins, cortical gyriform hypointensities, and gray–white junction abnormalities. However, conventional contrast-enhanced MRI was reportedly better at demonstrating abnormal leptomeningeal enhancement. Extracranial vascular malformations in children: These malformations can also be broadly classified as low-flow and high-flow lesions. Among the low-flow lesions, hemangiomas form a distinct class when compared to other venous, lymphatic or venolymphatic malformations of the head and neck (Fig. 26.5). The high-flow vascular malformations are essentially arteriovenous malformations. The goal of imaging studies in head and neck vascular malformations is to classify the malformations as high- or low-flow, evaluate their anatomy with respect to feeding and draining vessels, and to evaluate their effect on other tissues in the head and neck. Routine MR imaging by itself is often enough to

B.D. Jagadeesan and D.N. Loy

of Sturge–Weber Syndrome. Note: The extent of leptomeningeal enhancement in image C exceeds the boundaries of the calcified region in image A suggesting increased sensitivity of MRI in the detection of meningeal vascular abnormalities in these children

Fig. 26.5 Cystic hygroma. Two-year-old child with neck swelling. Axial T2-weighted image (a) shows a multilobulated mass which is predominantly T2 hyperintense but also shows a few locules with fluid levels and T2 hypointensity. Axial source image from a time of flight angiogram of the neck vessels (b) in the same patient shows no evidence for arterial supply to this mass. Coronal dynamic contrastenhanced MRA image obtained during the venous phase after injection of intravenous gadolinium (c) also shows no evidence for abnormal blood vessels associated with this mass. These imaging findings are characteristic of a cystic hygroma or a low flow veno-lymphatic malformation of the neck

address the questions regarding the anatomy of the vascular malformations with high spatial resolution and to document the relationship of the vascular malformations to the surrounding structures. The major role of MRA is simply to distinguish between high-flow and low-flow vascular malformations.

26 Pediatric MR Angiography: Principles and Applications

Thus, the role of MRA in the extracranial malformations is somewhat different from its role in intracranial malformations, where both high spatial and temporal resolution are required, this is simply due to the fact that extracranial vascular malformations tend to be much larger than their intracranial counterparts and are often distinctly situated away from normal vascular structures. Therefore, MRA techniques with high temporal resolution and a large field of view are preferred to evaluate the extracranial vascular malformations at the expense of spatial resolution. Ziyeh et al. [63] performed MR projectional angiography in eight patients with head and neck vascular malformations and compared the result with DSA. They found that MRPA was adequate in the differentiation of high-flow from low-flow vascular malformations. Cerebral venous thrombosis and magnetic resonance venography: The incidence of cerebral venous thrombosis in the neonatal population may be as high as 40.7 per 100,000 per year. In the Canadian Pediatric Ischemic Stroke Registry, the incidence cerebral venous sinus thrombosis was 0.6 per 100,000 children per year aged term birth to 18 years. Some of the many predisposing causes include disorders such as leukemia, trauma, dehydration, infection, prothrombotic disorders, congenital heart disease, and prematurity. In the majority of children, the superior saggital sinus and the transverse sinus are involved. Unlike adults, the presenting symptoms in children can be atypical or subtle. Hence neuroimaging and particularly magnetic resonance venography (MRV) plays a major role in the detection of pediatric cerebral sinovenous thrombosis. Time of flight MRV is unreliable in this setting because of the T1 shortening effect of thrombus which can mimic flow in a thrombosed vascular segment and the high incidence of false-positive findings. Rollins et al. [64] directly compared the performance of 2D time-of-fight techniques with 3D contrast-enhanced MR venograms of the cerebral venous system in 37 children and found that the results of the techniques were comparable in only 17 of the patients. In 19 children, various vascular anomalies such as flow gaps and segmental atresia were suggested by 2D TOF which were found to be absent on the postgadolinium studies. In the neonatal population where the accurate diagnosis of cerebral venous thrombosis is most crucial, Widjaja et al. [65] found that TOF MRV showed spurious flow gaps in 31 of 51 neonates who had normal venous anatomy on CT venography. When postcontrast imaging is performed without bolus timing or subtraction techniques, especially with a longer acquisition time, the organized thrombus itself may enhance resulting in a false-negative study. MRV studies with bolus timing methods and MR fluoroscopic monitoring of contrast bolus arrival in the superior sagittal sinus followed by centric

373

re-ordering of echo acquisition may be required. However, these studies may suffer from poor spatial resolution, which may be a significant drawback in very young children. Recently, Meckel et al. [66] evaluated a combination of these two techniques for the evaluation of cerebral venous thrombosis, and to differentiate between acute thrombus and chronic enhancing organized thrombus. They found that this combination of 4D MR venography outperformed other modalities such as TOF MRV and gradient echo imaging in the detection of dural venous sinus thrombosis. This method had the highest sensitivity in the detection of subacute venous sinus thrombosis. However, in common with other studies, their study too showed that the sensitivity of MRV techniques for the detection of cortical venous thrombosis was lower than that of traditional gradient recalled echo (GRE) MR imaging. Isolated cortical venous thrombosis, in the absence of dural venous sinus thrombosis is a rare phenomenon and it is possible that SWI bold venography will play an increasing role in its evaluation in the future. Phase contrast imaging is also commonly used in the evaluation of the cerebral venous system; its main advantages are the lack of need for intravenous contrast, the ability to differentiate between thrombus and flow, and it ability to provide quantitative hemodynamic information.

MRA in the Evaluation of Acute Ischemic Stroke in Children Acute ischemic stroke is a rare disorder in children thought to affect 1.2–3.3/100,000 children/year [67–71]. The majority of the cases of childhood stroke occur in the absence of any known risk factors. However, in many cases, a specific etiology can still be identified. The most common etiologies include vasculopathies, cardiac disease, and pro-thrombotic disorders. The vasculopathies, including transient cerebral arteriopathy (TCA), fibromuscular dysplasia, childhood primary angiitis of the central nervous system, and Moyamoya disease, can be identified in 18–80% of children with acute ischemic strokes depending on the population studied. Arterial dissection can also be found in 7.5–20% of cases of acute ischemic stroke in children [70–72]. Transient cerebral vasculopathy: TCA is the most common among these conditions. In one series, TCA accounted for approximately 26% of sample of children suffering from stroke. TCA is an inflammatory vasculopathy, which has been reported to be triggered by prior varicella zoster infection in approximately 44% of cases and by other bacterial or viral infections in the rest. It usually manifests as unilateral, often multiple, focal or elongated segmental narrowings of the terminal internal carotid artery and the proximal middle

374

B.D. Jagadeesan and D.N. Loy

Fig. 26.6 Moyamoya disease. Maximum intensity projection images (a) and volume rendered images (b) of the circle of Willis from a time-of-flight MR angiography study in an 8-year-old child demonstrate severe stenoses of the terminal internal carotid arteries and the proximal anterior and middle cerebral arteries with prominent deep lenticulostriate collateral vessels. The M2 and M3 segments of the middle cerebral arteries fill (retrograde on DSA) from dural vessels which are the result of a prior encephaloduromyosynangiosis procedure (a–d). Serial maximal intensity projection images of the vertebrobasilar circulation from time of flight studies obtained at initial presentation (e) and 2 years later (f, g) in the same child with progressive stenosis of the P1 segment of the right posterior cerebral artery

and anterior cerebral arteries with resolution on follow-up imaging [73]. Moyamoya disease and Moyamoya syndrome: In contrast to transient cerebral vasculopathy, bilateral progressive narrowing and eventual occlusion of the terminal ICA and the proximal MCAs and ACAs characterize Moyamoya disease and syndrome. Among the common underlying etiologies, sickle cell disease, hemolytic uremic syndrome, neurofibromatosis, Down’s syndrome, and radiation are the most widespread [74–76]. There is development of an abnormal network of collaterals including enlarged lenticulostriate arteries, which results in the puff of smoke appearance on conventional angiograms as well as development of extensive, dural to pial collaterals. Although ischemic stroke is more common in children with Moyamoya disease, hemorrhagic stroke may also occur and it is believed that abnormal dilatation of the anterior choroidal artery may predict an increased risk for hemorrhagic stroke [77]. It has also been shown that increasing

prominence of collateral vessels predicts an increased risk of stroke. Several studies have shown that MRA using time-offlight technique compares favorably with DSA in evaluating the cranial circulation in children with untreated sickle cell disease. It has also been shown that imaging at 3 T provides better information on the extent of collateralization compared to 1.5 T. This is simply because of the improved spatial resolution and signal-to-noise ratio of higher field strengths [78]. MRA can also be used in the follow up of patients with sickle cell disease who are treated with encephalo duro arterio synangiosis or encephalo duro arterio myosynagiosis (EDAS or EDAMS, respectively). Following EDAS, there is an increase in size of the distal MCA branches (underlying the graft) and a decrease in collaterals at the base of the brain, which imparts a gradual but significant decrease in the risk for stroke (Fig. 26.6). Yoon et al. [79] suggested that these findings could be adequately visualized with routine TOF-MRA.

26 Pediatric MR Angiography: Principles and Applications

It is unlikely that contrast-enhanced MRA will have any appreciable role in the evaluation of sickle cell disease in the near future given the decreased renal function in most children with sickle cell disease. However, new dynamic phase contrast studies may have a role in understanding the hemodynamics of treated and untreated sickle cell disease. Neff et al. [80] used 2D phase contrast MRA for flow quantification and found that there was 50% or more reduction in blood flow volumes in the internal carotid arteries of subjects with Moyamoya disease when compared with controls whereas flow in the basilar arteries increased substantially. It remains to be seen if such flow quantification studies can predict which subjects are at an increased risk for intracranial hemorrhage. For now, O15 PET studies remain the gold standard for quantification of oxygen extraction fractions and stroke risk in Moyamoya patients. Vascular dissection and stroke in children: Vascular dissections are often under diagnosed and may account for a significant proportion of strokes in children. Vascular dissections may represent the underlying etiology for arterial ischemic stroke in 7.5–20% of children with acute ischemic stroke. Dissections in children differ from those in adults in both clinical presentation and distribution. Unlike adults, most dissections in children present with infarcts and dissections are more common among male children. Likewise a greater proportion of dissections tend to be intracranial when they involve the anterior circulation although extracranial dissections of the vertebral arteries at the C1–C2 level are still the most common. Dissections can result in luminal narrowing of the affected vessel (when the dissection plane is subintimal) or pseudoaneurysm formation (when the dissection plane is subadventitial). Subarachnoid hemorrhages can also be the presenting manifestation of a dissection especially when there is pseudoaneurysm formation. Dissections can occur in the absence of any known trigger (i.e., spontaneous dissections) or may occur in the setting of trauma or a known vessel wall disorder such as fibromuscular dysplasia. When imaging the vessels in a child with suspected dissection of the craniocervical arteries, it is not only important to image the vessel lumen but also to image the vessel wall in order to detect the intramural hematoma. Axial high-resolution fatsaturated T1- and T2-weighted images are crucial in the detection of intramural hematomas. Childhood vasculitis: Aviv et al. [81] compared the performance of catheter angiography and MR angiography in the selection of childhood vasculitis. Children with a clinical diagnosis of primary angiitis of childhood were included in the study and those with other causes for childhood vasculopathy such as varicella or sickle cell disease were excluded. They studied a total of 25 patients and found that 3D time-offlight MRA identified only 45 of the 64 lesions detected on

375

catheter angiography and there was only a modest agreement between the two modalities (k = 0.4). However, when the performance of MRA for making the diagnosis of vasculitis was evaluated rather than the identification of individual lesions, MRA had a sensitivity of 95% and a positive predictive value of 83%. They reported that unilateral disease was common and that proximal MCA and ACA lesions accounted for most stenoses. These lesions lead most commonly to basal ganglia infarctions. The authors also concluded that MRA performs well in quantifying the degree of stenosis in individual affected segments when the stenosis is greater than 75% of the luminal diameter but overestimates the degree of stenosis in patients with 50–75% stenosis. Eleftheriou et al. [82] also compared the performance of MRA and DSA in childhood central nervous system vasculitis. They reported that MRA correctly detected only 28 of 44 lesions identified by catheter angiography. MRA was 63% sensitive and 89% specific with a PPV of 62% and an NPV of 88%. They found that MRA performed particularly poorly in the posterior circulation. The 3D TOF technique used in both these studies also did not offer any dynamic flow information. In conclusion, at present a positive MRA study may prove to be reliable indicator for the presence of cerebral vasculitis but a negative study does not exclude the presence of childhood vasculitis. Unlike adults where 20–40% of patient with cerebral vasculitis do not have angiographically detectable changes, most children with vasculitis do show changes on DSA. Therefore, a catheter angiogram is still necessary for diagnosis in most cases of suspected pediatric vasculopathy.

Spinal Vascular Malformations The most widely used classification for spinal vascular lesions was devised by Anson and Spetlzler and classifies them into type 1, spinal dural arteriovenous fistulas (dorsal intradural fistulas); type II, intradural glomus arteriovenous malformations (these may be compact or diffuse with an extramedullary component); type III, juvenile or metameric AVMs; and type IV, intradural perimedullary AVFs (ventral intradural fistulas). The incidence of pediatric spinal vascular malformations is not entirely known although it is thought that they constitute approximately 10–20% of all central nervous system vascular shunts. In younger children, spinal vascular malformations with arteriovenous shunting can be associated with hereditary hemorrhagic telengiectasia or Osler–Rendu–Weber syndrome. They may also be found in association with Klippel–Trenaunay syndrome, Parkes–Weber syndrome, or neurofibromatosis. Alternatively, they may form part of a spinal arteriovenous metameric syndrome. Most pediatric

376

spinal cord vascular lesions are isolated, spontaneous anomalies. The commonest of the spinal vascular malformations in adults, namely the type 1 malformations (dural AV fistula) are thought to be acquired in nature and rarely occur in children. Pediatric patients with spinal vascular malformations most commonly present with complications from hemorrhage or venous congestion. Hence, most undergo an initial MR imaging study where enlarged flow voids are usually evident on T2-weighted images. MRA is reported to be most useful for the diagnosis of type 1 spinal vascular malformations. MRA may play a role in identifying the approximate levels of arterial feeders in other types of spinal AVMs but is not helpful for characterization of the types of AV shunting. Identification of hypertrophied segmental feeders with MRA can be useful to direct spinal catheter angiography, thus reducing iodinated contrast doses in small children. MRA is insensitive for the detection of rare type IV malformations. The spinal vasculature presents many unique challenges for MRA. These arise from the unique architecture of the spinal vasculature, with paired ventral arteries and veins, which can only be distinguished from each other by dynamic imaging; the extremely small caliber of the vessels concerned; the extreme variability in the origin and course of crucial vessels such as the artery of Admakewicz, and the affliction of nervous tissue remote from the neurovascular abnormality. Therefore, any MRA technique for imaging the spinal vasculature has must exhibit high spatial resolution, high temporal resolution, and a relatively large field of view. TOF and PC MRA are not useful in the spine given the slow flow in spinal vessels and their small size. However, some authors have successfully used these techniques for the diagnosis and follow up of spinal vascular lesions in adult patients [83, 84]. Therefore, contrast-enhanced MRA methods are most commonly utilized for spinal MR angiography. The usual arteriovenous circulation time of blood in the spinal cord ranges between 9 and 12 s in adults and is likely shorter in children, especially those with hyperdynamic circulation from other vascular anomalies. In order to be useful, dynamic MRA techniques should ideally be able to acquire images of the entire spinal vasculature within the order of this time frame, which will allow the reader to differentiate between the arterial and venous systems. Hence unlike in the brain where subsecond temporal resolutions present the major

B.D. Jagadeesan and D.N. Loy

technical challenge for MRA, in the spine the challenge is to maximize the field of view and achieve high spatial resolution while maintaining a temporal resolution of several seconds. In small children, the field of view that is required may be smaller, but the vessels that have to be imaged are also smaller, making spinal vascular imaging even more challenging. Even with acquisition times that are longer than the arteriovenous circulation time, adequate differentiation between arteries and veins can be achieved by bolus timing the arrival of contrast in the aorta and acquiring the center of k space during the time window when the arteriovenous contrast difference is greatest. Typically dual-phase imaging is performed. The first phase is acquired during the maximal contrast difference between arteries and veins and the second phase is obtained immediately thereafter to study the venous system. Backes et al. [85] speculate that the second phase may be more focused with the use of a smaller field of view and higher resolution. They also suggest the use of intravascular contrast agents which stay in the blood pool for an extended time. Using dynamic bolus timed MRA, Mull et al. [86] identified the main feeding artery in 10 out of 11 patients with spinal intradural arteriovenous malformations and were also able to differentiate between fistulous and nidal malformations in four out of six patients. Jaspers et al. [87] have successfully used a key hole imaging technique to image the spinal vasculature with a temporal resolution of 6–8 s. However, as elsewhere, bolus-timing methods are not optimal for the study of spinal vessels in children. Some authors have suggested using a higher gadolinium contrast dose in the range of 0.2 mmol/kg to improve sensitivity [85], but this may not be feasible in children with immature or poor renal function. Therefore, dynamic subtraction imaging methods, which do not rely on timing may be required to optimally image the spinal vessels (Fig. 26.7). Hyodoh et al. [88] performed a dynamic study of the spinal vessels using a 3D fast spoiled gradient recalled acquisition in steady state (GRASS) method, and imaged five consecutive phases of the transit of contrast bolus through the spinal vasculature. They then performed a double subtraction of the maximum intensity projection (MIP) images to delineate the artery of Adamkiewicz. A similar technique can be used to improve the hemodynamic evaluation of spinal arteriovenous malformations in children.

26 Pediatric MR Angiography: Principles and Applications

Fig. 26.7 Type III metameric spinal AVM in a 6-year-old child who presented with neck pain and swelling. MRI and MRA examinations after conventional radiographic images showed lytic lesions involving her C1 to C4 vertebrae. T2-weighted images in the saggital (a) and (b) axial planes showed multiple hypointense vascular flow voids involving the C1 to C4 veretbrae, with intradural extension. There was also abnormal T2 hyperintensity within the cervical spinal cord parenchyma.

References 1. Swoboda NA, Armstrong DG, Smith J, Charkot E, Connolly BL. Pediatric patient surface doses in neuroangiography. Pediatr Radiol. 2005;35:859–866. 2. Glennie D, Connolly BL, Gordon C. Entrance skin dose measured with MOSFETs in children undergoing interventional radiology procedures. Pediatr Radiol. 2008;38:1180–1187. 3. Hollingsworth CL, Yoshizumi TT, Frush DP, et al. Pediatric cardiac-gated CT angiography: assessment of radiation dose. AJR Am J Roentgenol. 2007;189:12–18. 4. Raelson CA, Kanal KM, Vavilala MS, et al. Radiation dose and excess risk of cancer in children undergoing neuroangiography. AJR Am J Roentgenol. 2009;193:1621–1628. 5. Pierce DA, Shimizu Y, Preston DL, Vaeth M, Mabuchi K. Studies of the mortality of atomic bomb survivors. Report 12, Part I. Cancer: 1950–1990. Radiat Res. 1996;146:1–27. 6. Preston DL, Shimizu Y, Pierce DA, Suyama A, Mabuchi K. Studies of mortality of atomic bomb survivors Report 13: Solid cancer and noncancer disease mortality: 1950–1997. Radiat. Res. 2003;160: 381–407. 7. Yang CY, Chen YF, Lee CW, Huang A, Shen Y, Wei C, Liu HM. Multiphase CT angiography versus single-phase CT angiography: comparison of image quality and radiation dose. Am J Neuroradiol. 2008;29:1288–1295.

377

Maximum intensity projection images obtained during the arterial phase of a dynamic contrast-enhanced MR angiography examination of the neck show a large arteriovenous malformation in the upper cervical spine region supplied by both vertebral arteries (c), large venous channels can be seen draining the arteriovenous malformation to the internal jugular veins (d). Spinal DSA (e)

8. Strauss KJ, Goske MJ, Fursh DP, Butler PF, Morrison G. Image Gently Vendor Summit: working together for better estimates of pediatric radiation dose from CT. AJR Am J Roentgenol. 2009; 192:1169–1175. 9. Bosmans H, Wilms G, Dymarkowski S, Marchal G. Basic principles of MRA. Eur J Radiol. 2001;38:2–9. 10. Haacke EM, Brown RW, Thompson MR, Venkatesan R. Magnetic Resonance Imaging—Physical Principles and Sequence Design. New York: Wiley-Liss; 1999:10. 11. Blatter DD, Parker DL, Robison RO (1991) Cerebral MR angiography with multiple overlapping thin slab acquisition. Part I. Quantitative analysis of vessel visibility. Radiology. 179:805–811. 12. Atkinson D, Brant-Zawadzki M, Purdy D, Laub G. Improved MR angiography: magnetization transfer suppression with variable flip angle excitation and increased resolution. Radiology. 1994;190:890–894. 13. Ozsarlak O, Van Goethem JW, Parizel PM. 3D time-of-flight MR angiography of the intracranial vessels: optimization of the technique with water excitation, parallel acquisition, eight-channel phased-array head coil and low-dose contrast administration. Eur Radiol. 2004;14:2411–2444. 14. Ozsarlak O, Goethem JWV, Maes M, Parizel PM. MR angiography of the intracranial vessels: technical aspects and clinical applications. Neuroradiology. 2004;46:955–972. 15. Du YP, Jin Z, Hu Y, Tanabe J. Multi-echo acquisition of MR angiography and venography of the brain at 3 Tesla. J Magn Reson Imaging. 2009;30:449–454.

378 16. Deitsung A, Dittrich E, Sedalick J, Rauscher A, Reichenbach JR. ToF-SWI: simultaneous time of flight and fully flow compensated susceptibility weighted imaging. J Magn Reson Imaging. 2009; 29:1478–1484. 17. Tsuchiya K, Yoshida M, Imai M, Nitatori T, Kimura T, Ikedo M, Takemoto S. Hybrid of opposite-contrast magnetic resonance angiography of the brain by combining time-of-flight and black blood sequences: its value in Moyamoya disease. J Comput Assist Tomogr. 2010;34:242–246. 18. Marks MP, Pelc NJ, Ross MR, Enzmann DR. Determination of cerebral blood flow with a phase-contrast cine MR imaging technique: evaluation of normal subjects and patients with arteriovenous malformations. Radiology. 1992:182:467–476. 19. Markl M, Chan FP, Alley MT, et al. Time-resolved three-dimensional phase-contrast MRI. J Magn Reson Imaging. 2003:17:499–506. 20. Yamashita S, Isoda H, Hirano M, et al. Visualization of hemodynamics in intracranial arteries using time-resolved three-dimensional phase-contrast MRI. J Magn Reson Imaging. 2007;25:473–478. 21. Velikina JV, Johnson KM, Wu Y, Samsonov AA, Turski P, MIstretta CA. PC HYPR flow: a technique for rapid imaging of contrast dynamics. J Magn Reson Imaging. 2010;31:447–456. 22. Korosec FR, Frayne R, Grist TM, et al. Time-resolved contrastenhanced 3D MR angiography. Magn Reson Med. 1996;36: 345–351. 23. Muthupillai R, Vick GW, Flamm SD and Chung T. Time-resolved contrast-enhanced magnetic resonance angiography in pediatric patients using sensitivity encoding. J Magn Reson Imaging. 2003; 17:55–564. 24. Pinto C, Hickey R, Carroll TJ, et al. Time-resolved MR angiography with generalized autocalibrating partially parallel acquisition and time-resolved echosharing angiographic technique for hemodialysis arteriovenous fistulas and grafts. J Vasc Interv Radiol. 2006;17:1003–1009. 25. Lim RP, Shapiro M, Wang EY. 3D time-resolved MR angiography (MRA) of the carotid arteries with time-resolved imaging with stochastic trajectories: comparison with 3D contrast-enhanced boluschase MRA and 3D time-of-flight MRA. AJNR Am J Neuroradiol. 2008;29:1847–1854. 26. Jeong HY, Cashen TA, Hurley MC, Eddleman C, Getch C, Batjer H, Carroll TJ. Radial sliding-window magnetic resonance angiography (MRA) with highly-constrained projection reconstruction (HYPR). Magn Reson Med. 2009;61:1103–1113. 27. Eddleman CS, Jeong HY, Hurley MC, Zuehlsdorff S, Dabus G, Getch H, Batjer H, Bendok BR, Carroll TJ. 4D radial acquisition contrast-enhanced MR angiography and intracranial arteriovenous malformations: quickly approaching digital subtraction angiography. Stroke. 2009;40;2749–2753. 28. Willinek WA, Hadizadeh DR, von Falkenhausen M, et al. 4D timeresolved MR angiography with keyhole (4D-TRAK): more than 60 times accelerated MRA using a combination of CENTRA, keyhole, and SENSE at 3.0T. J Magn Reson Imaging. 2008;27:1455–1460. 29. van Vaals JJ, Brummer ME, Dixon WT, et al. Keyhole method for accelerating imaging of contrast agent uptake. J Magn Reson Imaging. 1993;3:671–675. 30. Chooi WK, Connolly DJ, Coley SC, Griffiths PD. Assessment of blood supply to intracranial pathologies in children using MR digital subtraction angiography. Pediatr Radiol. 2006;36:1057–1062. 31. Riederer SJ, HAider CR and Borisch E. Time-of-arrival mapping at three-dimensional time-resolved contrast-enhanced MR angiography. Radiology. 2009;253:532–542. 32. Klimo P, Jr, Rao G, Brockmeyer D. Pediatric arteriovenous malformations: a 15-year experience with an emphasis on residual and recurrent lesions. Childs Nerv Syst. 2007;23:31–37. 33. Gabriel EM, Sampson JH, Wilkins RH. Recurrence of a cerebral arteriovenous malformation after surgical excision. Case report. J Neurosurg. 1996;84:879–882.

B.D. Jagadeesan and D.N. Loy 34. Luessenhop AJ. Natural history of cerebral arteriovenous malformations. In: Wilson CB, Stein MB, eds. Intracranial Arteriovenous Malformations. Baltimore: Williams & Wilkins; 1984:12–23. 35. Humphreys RP. Brain vascular malformations. In: Choux M, Di Rocco C, Hockley A, Walker M, eds. Paediatric Neurosurgery. Edinburgh: Churchill Livingstone; 1999:665–677. 36. Papadias A, Taha A, Sgouros S, Walsh AR, Hockley AD. Incidence of vascular malformations in spontaneous intra-cerebral haemorrhage in children. Childs Nerv Syst. 2007;23:881–886. 37. Fasulakis S, Andronikou S. Comparison of MR angiography and conventional angiography in the investigation of intracranial arteriovenous malformations and aneurysms in children. Pediatr Radiol. 2003;33:378–378. 38. Buis DR, van Ouwerkerk WJ, Takahata H, Vandertop WP. Intracranial aneurysms in children under 1 year of age: a systematic review of the literature. Childs Nerv Syst. 2006;22:1395–1409. 39. Ventureyra EC, Higgins MJ. Traumatic intracranial aneurysms in childhood and adolescence. Case reports and review of literature. Childs Nerv Syst. 1994;10:361–369. 40. Nakstad P, Nornes H, Hauge HN. Traumatic aneurysms of the pericallosal arteries. Neuroradiology. 1986;28:335–338. 41. Allison JW, Davis PC, Sato Y, Jamees CA, HAque SS, Angtuaco EJ, Glasier CM. Intracranial aneurysms in infants and children. Pediatr Radiol. 1998;28:223–229. 42. Raybaud CA, Strother CM, Hald JK. Aneurysms of the vein of Galen: embryonic considerations and anatomical features relating to the pathogenesis of the malformation. Neuroradiology. 1989;31:19–28. 43. Casasco A, Lylyk P, Hodes JE, Kohan G, Aymard A, Merland JJ. Percutaneous transvenous catheterization and embolization of vein of Galen aneurysms. Neurosurgery. 1991;28:260–266. 44. Gupta AK, Varma DR. Vein of Galen malformations: review. Neurol India. 2004;52:43–53. 45. Berenstein A, Lasjaunias P. Arteriovenous fistulas of the brain. In: Surgical Neuroangiography: Endovascular Treatment of Cerebral Lesions. Berlin: Springer-Verlag; 1992:267–317. 46. Garcia-Monaco R, Lasjaunias P, Berenstein A. Therapeutic management of vein of Galen aneurysmal malformations. In: Vinuela F, Halbach VV, Dion JE, eds. Interventional Neuroradiology: Endovascular Therapy of the Central Nervous System. New York: Raven Press; 1992:113–127. 47. Alvarez H, Garcia Monaco R, Rodesch G, Sachet M, Krings T, Lasjaunias P. Vein of Galen aneurysmal malformations. Neuroimaging Clin N Am. 2007;17:189–206. 48. Lasjaunias P. Vascular Diseases in Neonates, Infants and Children: Interventional Neuroradiology Management. Berlin: Springer; 1997. 49. Lasjaunias P. Vascular Diseases in Neonates, Infants and Children: Interventional Neuroradiology Management. 2nd edition. Berlin: Springer; in press. 50. Yasargil MG. Microneurosurgery IIIB. New York: Thieme Medical Publishers; 1988:323–357. 51. Fournier D, Rodesch G, Ter Brugge K, et al. Acquired mural (dural) arteriovenous shunt of the vein of Galen: report of 4 cases. Neuroradiology. 1991;33:185–192. 52. Lasjaunias P, Ter Brugge K, Lopez Ibor, et al. The role of dural anomalies in vein of Galen aneurysms: report of six cases and review of the literature. AJNR Am J Neuroradiol. 1987;8:185–192. 53. Jagadeesan BD, Delgado Almandoz JE, Moran CJ, Benzinger TL. Accuracy of susceptibility-weighted imaging for the detection oa arteriovenous shunting in vascular malformations of the brain. Stroke. 2011;42:87–92. 54. Krings T. Geibprasert S, Luo CB, Bhattacharya JJ, Alvarez H, Lasjaunias P. Segmental neurovascular syndromes in children. Neuroimaging Clin N Am. 2007;17;245–258. 55. Bergstrand H, Olivecrona H, Tonnis W. Gefa ss- missbildungen und Gefaessgeschwulste des Ge- hirns. Leipzig (Germany): Thieme; 1936.

26 Pediatric MR Angiography: Principles and Applications 56. Bonnet P, Dechaume JP, Blanc E (1937) Lanevrisme cir-soide de la retine. Ses relations avec lanevrysme cirsoide du cerveau [in French]. Journal de Me decine de Lyon. 18:165–178. 57. Wyburn-Mason R. Arteriovenous aneurysm of midbrain and retina, facial naevi and mental changes. Brain. 1943;66:163–203. 58. Bhattacharya JJ, Luo CB, Alvarez H, et al. PHACES syndrome: a review of eight previously unreported cases with late arterial occlusions. Neuroradiology. 2004;46:227–33. 59. Lasjaunias P, Ter Brugge K, Berenstein A. Surgical Neuroangiography. Berlin: Springer; 2006. 60. Griffiths PD. Sturge-Weber syndrome revisited: the role of neuroradiology. Neuropediatrics. 1996;27:284–294. 61. Truhan AP, Filipek PA. Magnetic resonance imag- ing. Its role in the neuroradiologic evaluation of neurofibromatosis, tuberous sclerosis, and Sturge-Weber syndrome. Arch Dermatol. 1993;129: 219–226. 62. Hu J, Yu Y, Juhasz C, et al. MR susceptibility weighted imaging (SWI) complements conventional contrast enhanced T1 weighted MRI in characterizing brain abnormalities of Sturge-Weber Syndrome. J Magn Reson Imaging. 2008;28:300–307. 63. Ziyeh S, chumacher M, Streker R, Rossler J, Hochmuth A, Klisch J. Head and neck vascular malformations: time-resolved MR projection angiography. Neuroradiology. 2003;45:681–686. 64. Rollins N, Ison C, Reyes T, Chia J. Cerebral MR venography in children: comparison of 2D time-o-flight and gadolinium enhanced 3D gradient-echo techniques. Radiology. 2005;235:1011–1017. 65. Widjaja E, Shroff M, Blaser S, Laughlin S, Raybaud C. 2D time-offlight MR venography in neonates: anatomy and pitfalls. AJNR Am J Neuroradiol. 2006;27:1913–1918. 66. Meckel S, Reisinger C, Bremerich J, et al. Cerebral venous thrombosis: diagnostic accuracy of combined, dynamic and static, contrast-enhanced 4D MR venography. AJNR Am J Neuroradiol. 2010;31:527–535. 67. Schoenberg BS, Mellinger JF, Schoenberg DG. Cerebrovascular disease in infants and children: a study of incidence, clinical features, and survival. Neurology. 1978;28:763–768. 68. Broderick J, Talbot GT, Prenger E, et al. Stroke in children within a major metropolitan area: the surprising importance of intracerebral hemorrhage. J Child Neurol. 1993;8:250–255. 69. Giroud M, Lemesle M, Gouyon JB, et al. Cerebrovascular disease in children under 16 years of age in the city of Dijon, Franc: a study of incidence and clinical features from 1983 to 1993. J Clin Epidemiol. 1995;48:1343–1348. 70. deVeber G. Stroke and the child’s brain: an overview of epidemiology, syndromes and risk factors. Curr Opin Neurol. 2002;15:133–138. 71. Fullerton HJ, Wu YW, Sidney S et al. (2007) Risk of recurrent childhood arterial ischemic stroke in a population-based cohort: the importance of cerebrovascular imaging. Pediatrics. 119(3); 495–501. 72. Kirkham F, Sebire G, Steinlin M, et al. Arterial ischemic stroke in children. Thromb Haemost. 2004;92:697–706. 73. Braun KP, Bulder MM, Chabrier S, Kirkham FJ, Uiterwaal CS, Tardieu M, Sébire G. The course and outcome of unilateral

379

74. 75.

76. 77.

78.

79.

80.

81.

82.

83.

84.

85. 86.

87.

88.

intracranial arteriopathy in 79 children with ischaemic stroke. Brain. 2009;132(Pt 2):544–557. Scott RM, Smith ER. Moyamoya disease and Moyamoya syndrome. N Engl J Med. 2009; 360:1226–1237. Singla M, John E, Hidalgo G, et al. Moyamoya vasculopathy in a child after hemolytic uremic syndrome: a possible etiopathogenesis. Neuropediatrics. 2008;39:128–130. Rea D, Brandsema J, Armstrong D, et al. Cerebral arteriopathy in children with neurofibromatosis type 1. Pediatrics. 2009;124:e476-e483. Morioka M, Hamada JI, Kawano T, Todaka T, Yano S, Kai Y, Ushio Y. Angiographic dilatation and branch extension of the anterior choroidal and posterior communicating arteries are predictors of hemorrhage in adult Moyamoya patients. Stroke. 2003;34;90–95. Fushimi Y, Miki Y, Kikuta K, et al. Comparison of 3.0- and 1.5-T Three-dimensional time-of-flight MR angiography in Moyamoya disease: preliminary experience. Radiology. 2006;239:232–239. Yoon HK, Shin HJ, Lee M, Byun HS, Na DG, Han BK. MR angiography of Moyamoya disease before and after Encephaloduroarteriosynangiosis. AJR Am J Roentgenol. 2000;174:195–2000. Neff W, Horn P, Schmiedek P, Duber C, Dinter DJ. 2D Cine phasecontrast MRI for volume flow evaluation of the brain-supplying circulation in Moyamoya disease. AJR Am J Roentgenol. 2006; 187:W107-W116. Aviv RI, Benseler SM, DeVeber G, Silverman ED, Tyrrell PN, Tsang LM, Armstrong D. Angiography of primary central nervous system angiitis of childhood: conventional angiography versus magnetic resonance angiography at presentation. Am J Neuroradiol. 2007;28:9–15. Eleftheriou D, Cox T, Saunders D, Klein NJ, Brogan PA, Ganesan V. Investigation of childhood central nervous system vasculitis: magnetic resonance angiography versus catheter cerebral angiography. Dev Med Child Neurol. 2010;52:863–867. Mascalchi M, Quilici N, Ferrito G, et al. Identification of the feeding arteries of spinal vascular lesions via phase-contrast MR angiography with three-dimensional acquisition and phase display. AJNR Am J Neuroradiol. 1997;18:351–358. Saraf-Lavi E, Bowen BC, Quencer RM, et al. Detection of spinal dural arterio-venous fistulae with MR imaging and contrastenhanced MR angiography: sensitivity, specificity, and prediction of vertebral level. AJNR Am J Neuroradiol. 2002;23:858–867. Backes WH, Nijenhuis RJ. Advances in spinal cord MR angiography. Am J Neuroradiol. 2008;29:619–631. Mull M, Nijenhuis RJ, Backes WH, Krings T, Wilmink JT, Thron A. Value and limitations of contrast-enhanced MR angiography in spinal arteriovenous malformations and dural arteriovenous fistulas. AJNR Am J Neuroradiol. 2007;28:1249–1258. Jaspers K, Nijenhuis RJ, Backes WH. Differentiation of spinal cord arteries and veins by time-resolved MR angiography. J Magn Reson Imaging. 2007;26:31–40. Hyodoh H, Shirase R, Akiba H, et al. Double-subtraction maximum intensity projection MR angiography for detecting the artery of Adamkiewicz and differentiating it from the drainage vein. J Magn Reson Imaging. 2007;26:359–365.

wwwwwwwwwwww

Contrast Agents for MR Angiography Christoph U. Herborn

Recent years have seen the rapid development of techniques and applications for contrast-enhanced magnetic resonance angiography (CE-MRA) and the growing acceptance of the method in clinical routine. In addition to advances in hardware and software design, especially the development in the field of contrast media for CE-MRA has triggered the modality to increasingly become the diagnostic standard of reference for vascular imaging. Despite the majority of MR contrast agents is still lacking the direct approval for CE-MRA, this chapter reviews the properties and characteristics of both currently available agents and those which have been developed recently and are now on their way to the clinical market. Two major groups of MR contrast agents for CE-MRA can be distributed: On the one hand, paramagnetic agents that mostly rely on gadolinium and on the other hand superparamagnetic agents that are based on iron oxide particles. Paramagnetic agents can be further divided into a group with weak or strong interaction with serum proteins and a group without such interaction [1–6]. In addition, there are macromolecular agents that tend to stay longer in the blood pool. Superparamagnetic agents, on the contrary, can be divided upon their particle size and coating into groups of ultrasmall size (ultrasmall particles of iron oxide, USPIO) or groups coated with dextran or starch [7–12]. Figure 27.1 and Table 27.1 summarize the various agents and their brand names and physical properties, respectively.

Paramagnetic Contrast Agents At present, up to nine gadolinium-based contrast agents are approved worldwide and might be used for CE-MRA. Six of these agents do not show any interaction with blood serum C.U. Herborn, MD, MBA () University Medical Center Hamburg-Eppendorf, Martinistrasse 52, D-20251 Hamburg, Germany e-mail: [email protected]

27

and are therefore mainly used for first pass CE-MRA, i.e., acquisition of data during the first pass of the agents through the arterial bed after peripheral venous injection. A rather new agent, Gadobutrol (Gadovist), possesses a higher molarity than the other agents (1.0 M as compared to 0.5 M) and might have advantages over the other agents due to a smaller bolus size [7]. Two other agents, Gadobenate Dimeglumine (MultiHance) and Gadofosveset trisodium (Ablavar), have either a weak and transient or a strong interaction to serum in the blood, respectively. Thereby, these two agents remain longer in the blood pool than any of the aforementioned agents, which tend to diffuse into the interstitium after the first transit through the arterial and especially the venous bed [13–21]. Chemical structures of the agents are shown in Fig. 27.2.

Non-protein-Binding MR Contrast Agents The agents without any interactions with serum proteins are considered nonspecific and extracellular; they are excreted through the kidneys by glomerular filtration. With regard to T1 relaxation rates, a strong indicator for signal intensities in CE-MRA, these agents are comparable with values ranging between 4.3 and 5.6 L/mmol/s. Therefore, all these agents show more or less equivocal results in their vascular imaging performance with good quality and diagnostic performance. The decision for or against a certain dye has been made on reports about adverse events or comparatively subjective results at various institutions. The vast majority of scientific evaluations of CE-MRA has been made with these nonprotein-binding chelates and – as a matter of fact – these agents still are mostly used for CE-MRA examinations in routine practice. As mentioned before, Gadobutrol (Gadovist) has several advantages over the other non-protein-binding agents due to its higher molarity, the lower viscosity of the dye and finally the compact and concentrated bolus that achieves higher intravascular signal.

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_27, © Springer Science+Business Media, LLC 2012

381

382

C.U. Herborn

Fig. 27.1 Paramagnetic and superparamagnetic agents

Table 27.1 Brand names and physical properties of paramagnetic and superparamagnetic agents Molarity (mol/L) Molecular structure Thermodynamic stability constant (log Keq) Osmolarity (Osm/kg) Viscosity (mPas @ 37°C) T1 relaxivity (L/mmol/s), plasma

Magnevist Dotarem ProHance Omniscan Gadovist OptiMARK 0.5 0.5 0.5 0.5 1.0 0.5 Linear, ionic Cyclic, ionic Cyclic, non-ionic Linear, non-ionic Cyclic, non-ionic Linear, non-ionic

MultiHance 0.5 Linear, ionic

22.1

25.8

23.8

16.9

21.8

16.6

22.6

1.96

1.35

0.63

0.65

1.6

1.11

1.97

2.9

2.0

1.3

1.4

4.96

2.0

5.3

4.9

4.3

4.6

4.8

5.6

N/A

9.7

While early CE-MRA examinations were focused on single stations in the pelvis or the lower leg, the introduction of multistation CE-MRA with moving tables and bolus-chase techniques, have pushed the application toward the coverage of large vascular territories, including whole-body MRA from head to toe. The typical dose used for these examinations ranges between 0.1 and 0.3 mmol/kg bodyweight. Recent guidelines reflecting the risk associated with higher doses of gadolinium-based agents (see below) recommend

doses between 0.1 and 0.2 mmol/kg bodyweight, which still permits good image quality. All non-protein-binding agents have been used for various CE-MRA indications and have produced comparable image quality to that achieved with selective DSA. However, rapid loss of intravascular signal due to leakage of the agents into interstitial tissue has lead to both refinement of patient preparation with cuffs for femoral or even pelvic compression and to the development of contrast agents with protein interaction.

27

Contrast Agents for MR Angiography

383

Ablavar was previously known as vasovist. Both agents present with a significantly higher T1 relaxivity after their binding to albumin than compared to values achieved by the non-protein-binding chelates. This finding translates into more signal from the vascular bed which is potentially helpful for the assessment of small peripheral vessels. In addition, smaller doses of gadolinium seem to be applicable for virtually all vascular territories from the carotids to the feet [22–29]. Gadofosveset causes an extended signal increase from the vascular bed due to a strong and enduring noncovalent tie with albumin in the blood thereby permitting CE-MRA during a more prolonged timeframe after the initial first arteriovenous transit. Due to this binding, a gadolinium dose of merely 0.03 mmol/kg bodyweight is sufficient for diagnostic quality comparable to DSA examinations of the lower limb arteries. The phase of equilibrium distribution of contrast material in both arteries and veins is called steady-state and might open new windows for artery display with high spatial resolution. Furthermore, the agent might be used for the ever promising approach of CE-MRA toward the coronary arteries [30, 31].

Macomolecular Blood Pool Agents All aforementioned paramagnetic agents have a similar small size which allows for a rapid diffusion out of the vascular bed in case of lacking interaction with serum protein. Some gadolinium-based agents, however, have a macromolecular structure that prevents a fast leakage into the interstitium. Representatives of this group are P792 and gadomer-17 with molecular weights between 6.5 and 35 kDa, respectively, which is vastly bigger than the size of the other agents that range between 0.56 and 1.0 kDa. Nevertheless, the macromolecular agents are filtered through the glomerulus and excreted unmetabolized. These agents are still in a preclinical phase of evaluation, very first results hint at a potential use for CE-MRA of the coronary arteries.

Fig. 27.2 Chemical structures of 6 paramagnetic contrast agents. (a) Gadopentate. (b) Gadobutrol. (c) Gadodiamide. (d) Gadoteridol. (e) Gd-DOTA. (f) Gd-DTPA

Protein-Binding MR Contrast Agents Generally, Gadobenate Dimeglumine (MultiHance) with a transient and weak interaction with serum protein must be discriminated from Gadofosveset trisodium (ablavar), which shows a relatively strong binding interaction with protein.

Risk Profile of Paramagnetic Contrast Agents for MRA Gadolinium-based contrast agents have proved to be among the safest available for clinical imaging. In particular, with the doses used for CE-MRA, there are low rates of occurrence of nephrotoxicity and allergic reactions compared with those rates for iodinated contrast agents used for DSA or computed tomography angiography (CTA). Until recently, gadolinium-based paramagnetic contrast agents were used especially in patients with renal failure and patients undergoing

384

dialysis. In these patients, doses above 0.3 mmol/kg bodyweight have been injected. Interestingly, it was demonstrated that some paramagnetic contrast agents, namely gadodiamide and gadoversetamide interfere with the colorimetric test for serum calcium, resulting in spurious hypocalcemia in routine laboratory testing. This false measure was caused by free gadolinium that binds with calcium. As calcium metabolism becomes more important in renal failure, this observation is critical in this subgroup of frequently severely ill patients. In addition, this finding was shown to be a greater problem with higher doses of gadolinium, e.g., as given for CE-MRA of the renal arteries. Stability of gadolinium chelates also matters when dealing with potential tetention of the ion in the body following possible transmetallation and the release of a free gadolinium3+ ion. The recent discovery of an association between the administration of gadolinium-based contrast agents and nephrogenic systemic fibrosis (NSF) has also changed the way that gadolinium is used. High doses above 0.2 mmol/kg bodyweight are now rare because most applications have been adjusted to a standard dose of 0.1 mmol/kg, especially in patients undergoing dialysis or patients with a glomerular filtration rate (GFR) of less than 30 mL/min. When patients are undergoing dialysis, CE-MRA with gadolinium is ideally scheduled for just before the next dialysis appointment to facilitate prompt clearance. NSF is a rare fibrosing condition occurring in patients with profound renal failure or patients undergoing dialysis. In NSF, patches of skin become thickened and tethered to the underlying tissue, reducing range of motion and leading to contractures. The fibrosing process can also involve internal organs, including the lungs, heart, and muscles. Although many cases are mild and limited to dermatologic manifestations, an estimated 5% of cases have a more fulminant course resulting in death. Because treatment options are limited, an emphasis on prevention has been under way, limiting gadolinium exposure in patients with severe renal failure (estimated GFR <30 mL/min) and dialysis patients. A virtual absence of new cases with onset in 2008 or 2009 suggests that these preventive measures have been successful and are now allowing safe judicious use of gadolinium-based contrast agents in patients with renal failure and dialysis patients when clinically necessary [32–34].

Superparamagnetic Contrast Agents for MRA So far, several USPIO have been preclinically evaluated for CE-MRA: SH U 555C, AMI227 and NC100150 are the most popular representatives of the group, the latter agent has undergone the greatest early development. However, currently there are no official dedicated MRA studies performed with neither of these agents despite their potential advantages,

C.U. Herborn

including a long intravascular half-life with minimal leakage to the interstitial space and high interavascular signal. Yet, the complex signal characteristics of USPIO with strong T2* effects and relatively mild T1 effects have prevented a broad use of the agents for CE-MRA. In addition, a longer excretion and deposition of the dyes in reticuloendothelial cells influencing future MR imaging has so far limited their application. While many questions in clinical routine can nowadays be accurately answered with CE-MRA using mostly paramagnetic contrast agents as outlined in this chapter, the future will see further developments and progress of new contrast agents, e.g., for vessel wall imaging or atherosclerotic plaque composition characterization, only to mention a few. The search for the perfect dye for CE-MRA is still ongoing.

References 1. Cavagna FM, Maggioni F, Castelli PM, et al. Gadolinium chelates with weak binding to serum proteins. A new class of high-efficiency, general purpose contrast agents for magnetic resonance imaging. Invest Radiol. 1997;32:780–796. 2. Hany TF, Schmidt M, Hilfiker PR, Steiner P, Bachmann U, Debatin JF. Optimization of contrast dosage for gadolinium-enhanced 3D MRA of the pulmonary and renal arteries. Magn Reson Imaging. 1998;16:901–906. 3. Lauffer RB, Parmelee DJ, Dunham SU, et al. MS-325: albumintargeted contrast agent for MR angiography. Radiology. 1998; 207:529–538. 4. Grist TM, Korosec FR, Peters DC, et al. Steady-state and dynamic MR angiography with MS-325: initial experience in humans. Radiology. 1998; 207:539–544. 5. Runge VM, Knopp MV. Off-label use and reimbursement of contrast media in MR. J Magn Reson Imaging.1999;10:489–495. 6. Knopp MV, von Tengg-Kobligk H, Floemer F, Schoenberg SO. Contrast agents for MRA: future directions. J Magn Reson Imaging. 1999;10:314–316. 7. Tombach B, Reimer P, Prumer B, et al. Does a higher concentration of gadolinium chelates improve first-pass cardiac signal changes? J Magn Reson Imaging. 1999;10:806–812. 8. Reimer P, Allkemper T, Matuszewski L, Balzer T. Contrastenhanced 3D-MRA of the upper abdomen with a bolus-injectable SPIO (SH U 555 A). J Magn Reson Imaging. 1999;10:65–71. 9. Goyen M, Ruehm SG, Debatin JF. MR-angiography: the role of contrast agents. Eur J Radiol. 2000;34:247–256. 10. Port M, Corot C, Raynal I, et al. Physicochemical and biological evaluation of P792, a rapid-clearance blood-pool agent for magnetic resonance imaging. Invest Radiol. 2001;36:445–454. 11. Port M, Corot C, Rousseaux O, et al. P792: a rapid clearance blood pool agent for magnetic resonance imaging: preliminary results. Magma. 2001;12:121–127. 12. Knopp MV, Schoenberg SO, Rehm C, et al. Assessment of gadobenate dimeglumine for magnetic resonance angiography: phase I studies. Invest Radiol. 2002;37:706–715. 13. Gerber BL, Bluemke DA, Chin BB, et al. Single-vessel coronary artery stenosis: myocardial perfusion imaging with Gadomer-17 first-pass MR imaging in a swine model of comparison with gadopentetate dimeglumine. Radiology. 2002; 225:104–112. 14. Kraitchman DL, Chin BB, Heldman AW, Solaiyappan M, Bluemke DA. MRI detection of myocardial perfusion defects due to coronary

27

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Contrast Agents for MR Angiography artery stenosis with MS-325. J Magn Reson Imaging. 2002; 15:149–158. Bachmann R, Conrad R, Kreft B, et al. Evaluation of a new ultrasmall superparamagnetic iron oxide contrast agent Clariscan, (NC100150) for MRI of renal perfusion: experimental study in an animal model. J Magn Reson Imaging. 2002; 16:190–195. Hentsch A, Aschauer MA, Balzer JO, et al. Gadobutrol-enhanced moving-table magnetic resonance angiography in patients with peripheral vascular disease: a prospective, multi-centre blinded comparison with digital subtraction angiography. Eur Radiol. 2003;13:2103–2114. Balzer JO, Loewe C, Davis K, et al. Safety of contrast-enhanced MR angiography employing gadobutrol 1.0 M as contrast material. Eur Radiol. 2003;13:2067–2074. Herborn CU, Goyen M, Lauenstein TC, Debatin JF, Ruehm SG, Kroger K. Comprehensive time-resolved MRI of peripheral vascular malformations. AJR Am J Roentgenol. 2003;181:729–735. Herborn CU, Lauenstein TC, Ruehm SG, Bosk S, Debatin JF, Goyen M. Intraindividual comparison of gadopentetate dimeglumine, gadobenate dimeglumine, and gadobutrol for pelvic 3D magnetic resonance angiography. Invest Radiol.2003;38:27–33. Wyttenbach R, Gianella S, Alerci M, Braghetti A, Cozzi L, Gallino A. Prospective blinded evaluation of Gd-DOTA- versus Gd-BOPTAenhanced peripheral MR angiography, as compared with digital subtraction angiography. Radiology. 2003;227:261–269. Goyen M, Herborn CU, Kroger K, Lauenstein TC, Debatin JF, Ruehm SG. Detection of atherosclerosis: systemic imaging for systemic disease with whole-body three-dimensional MR angiography – initial experience. Radiology. 2003;227:277–282. Herborn CU, Barkhausen J, Paetsch I, et al. Coronary arteries: contrast-enhanced MR imaging with SH L 643A – experience in 12 volunteers. Radiology. 2003;229:217–223. Huber ME, Paetsch I, Schnackenburg B, et al. Performance of a new gadolinium-based intravascular contrast agent in free-breathing inversion-recovery 3D coronary MRA. Magn Reson Med. 2003;49:115–121. Herborn CU, Ajaj W, Goyen M, Massing S, Ruehm SG, Debatin JF. Peripheral vasculature: whole-body MR angiography with midfemoral

385

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

venous compression – initial experience. Radiology. 2004;230: 872–878. Huppertz A, Balzer T, Blakeborough A, et al. Improved detection of focal liver lesions at MR imaging: multicenter comparison of gadoxetic acid-enhanced MR images with intraoperative findings. Radiology. 2004;230:266–275. Paetsch I, Huber ME, Bornstedt A, et al. Improved three-dimensional free-breathing coronary magnetic resonance angiography using gadocoletic acid (B-22956) for intravascular contrast enhancement. J Magn Reson Imaging. 2004;20:288–293. Prokop M, Schneider G, Vanzulli A, et al. Contrast-enhanced MR Angiography of the renal arteries: blinded multicenter crossover comparison of gadobenate dimeglumine and gadopentetate dimeglumine. Radiology. 2005;234:399–408. Goyen M, Edelman M, Perreault P, et al. MR angiography of aortoiliac occlusive disease: a phase III study of the safety and effectiveness of the blood-pool contrast agent MS-325. Radiology. 2005;236:825–833. Fink C, Goyen M, Lotz J. Magnetic resonance angiography with blood-pool contrast agents: future applications. Eur Radiol. 2007;17(Suppl 2):B38-B44. Meaney JF, Goyen M. Recent advances in contrast-enhanced magnetic resonance angiography. Eur Radiol. 2007;17(Suppl 2): B2-6. Vogt FM, Herborn CU, Parsons EC, Kroger K, Barkhausen J, Goyen M. [Diagnostic performance of contrast-enhanced MR angiography of the aortoiliac arteries with the blood pool agent Vasovist: initial results in comparison to intra-arterial DSA]. Rofo. 2007;179:412–420. Prince MR, Zhang H, Morris M, et al. Incidence of nephrogenic systemic fibrosis at two large medical centers. Radiology. 2008;248:807–816. Prince MR, Zhang HL, Prowda JC, Grossman ME, Silvers DN. Nephrogenic systemic fibrosis and its impact on abdominal imaging. Radiographics. 2009;29:1565–1574. Prince MR, Zhang HL, Roditi GH, Leiner T, Kucharczyk W. Risk factors for NSF: a literature review. J Magn Reson Imaging. 2009; 30:1298–1308.

wwwwwwwwwwww

CE-MRA in the Age of Nephrogenic Systemic Fibrosis

28

Aditya Bharatha, Sean P. Symons, and Walter Kucharczyk

Introduction The purpose of this chapter is to review the current knowledge concerning nephrogenic systemic fibrosis (NSF) and discuss its implications for contrast-enhanced MR angiography (CE-MRA). NSF is a rare but serious systemic fibrosing condition occurring in patients with profound renal impairment. An increasing body of literature has implicated gadolinium (Gd)-based MR contrast agents (GBCAs) as the causative factor in the development of NSF. We discuss the clinical aspects of the disease, review the literature supporting an etiologic role of GBCAs, discuss risk factors for the disease, and cover current recommendations and prevention strategies with particular emphasis on MRA applications.

What Is NSF? Clinical Aspects of the Disease NSF is a rare, potentially debilitating, fibrosing illness that most commonly affects the skin, but has now been recognized to affect multiple organ systems. The disease was first recognized in 1997 in renal-transplant patients who developed extensive cutaneous indurations. Most of the affected patients had failed or failing renal transplants [1]. It was apparent that the condition was clinically distinct from scleroderma/morphea and appeared to resemble scleromyxedema at histological analysis, leading to an initial working hypothesis of a scleromyxedema related to transplantation. However, the condition was also noted in patients without a history of transplant, but who were on dialysis [2]. In addition, A. Bharatha, MD, FRCPC () • W. Kucharczyk, MD, FRCPC Department of Medical Imaging, University Health Network, Rm 1C552, Toronto General Site, 585 University Avenue, Toronto, ON, Canada M5G 2N2 e-mail: [email protected] S.P. Symons, MD, FRCPC Sunnybrook Health Sciences Centre, 2075 Bayview Avenue, AG31D, Toronto, ON, Canada M4N 3M5

it became clear that the condition was distinct from scleromyxedma in that the distribution of disease favored the extremities and spared the face (the opposite of scleromyxedema) and the typically seen serum paraprotein (IgG lambda) was not detected in these patients. This led to the terms fibrosing dermopathy of dialysis and nephrogenic fibrosing dermopathy. With recognition of the systemic manifestation of fibrosis of the internal organs and suspected role of circulating fibrocytes, the currently accepted term, “NSF,” was subsequently adopted [3]. NSF occurs equally in both genders without race predilection. Cases have been reported in patients ranging from pediatric to elderly, with a median age of 46 years. The majority of reported cases have been from the USA; however, cases have been reported in a multitude of other countries worldwide. NSF lesions are typically symmetric, involving more frequently the lower than upper extremities and extending from the hands to the mid-upper arms and from the feet to the mid-thighs. There is typically a distal to proximal progression of the disease. The trunk is less commonly involved and the face is typically spared. The typical cutaneous lesion is an indurated, firm, skin-colored to erythematous papule that may coalesce into thickened plaques (Fig. 28.1). The skin may develop a “peau d’orange” texture and the lesions may be associated with mild to moderate edema. The plaques characteristically spare the antecubital and popliteal fossae. The skin ultimately thickens and develops a woody texture. As the lesions cross joints, contractures may develop which can limit mobility; some patients become wheelchair bound. Patient may report pain and pruritis in the affected areas [3]. The presence of extracutaneous lesions in NSF is now well-established. The involvement of skeletal muscle, fascia, tendons, joints, and periarticular tissues has been demonstrated both clinically and histopathologically [3, 4]. More recently, multiorgan involvement with involvement of myocardium, lungs, kidneys, testes, and dura mater has been reported [5]. However, the literature concerning deep organ involvement is somewhat less reliable with many presumptive

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_28, © Springer Science+Business Media, LLC 2012

387

388

A. Bharatha et al.

Fig. 28.2 Coronal T1-weighted MR image through the thighs in a patient with NSF shows subcutaneous edema, skin thickening, and deep dermal stranding from thickened fibrous bands. Reprinted from Prince MR et al. [6] with permission

Fig. 28.1 Photographs of skin lesions in NSF; note skin thickening with erythematous plaques (a) and thickened skin with hyperpigmented plaques (b) (reprinted from Prince MR et al. [6] with permission)

cases based on functional testing or imaging or with nonspecific pathological findings. The diagnosis of NSF can be challenging and requires a thorough investigation of the patient’s medical history. Initial presentation, pattern and progression of skin lesions, renal function, and history of GBCA exposure must be determined. Other similar conditions, such as scleromyxedema, scleroderma, amyloidosis, calciphylaxis, and other fibrosing illnesses, must be ruled out on clinical grounds. As noted above, a chief differential diagnostic consideration is scleromyxedema; however, the variable distribution of lesions and absence of serum paraprotein in NSF usually allow its differentiation. Serological markers of scleroderma, such as antiScl-70 and antinuclear and anticentromere antibodies, are usually absent. A deep skin biopsy containing representative epidermis, dermis, subcutaneous fat, and, where possible, muscle and fascia from affected areas should be performed. Histologically, one may see thickened dermis involved with thickened collagen fibers and infiltrated with dermal spindle cells (fibrocytes) positive for CD-34 and procollagen-I. The thick collagen bundles are often broadened and separated from one another by prominent clefts. CD68, factor XIIIapositive dendritic cell proliferation, is seen [3, 5]. Although nonspecific, imaging findings include diffuse skin thickening and subcutaneous stranding on cross-sectional

imaging modalities (Fig. 28.2), diffuse soft tissue tracer uptake on bone scans, and skin and muscle activity on PET imaging [6]. Ultimately, both supportive clinical features, history of GBCA exposure, and suggestive biopsy findings are needed to make the diagnosis. The prognosis of the disease is generally guarded. In some patients, a fulminant aggressive course has been reported, as have several deaths. However, because of associated comorbities, it has not been possible to determine the increased mortality caused by the disease. Without improvement in renal function, progressive disease is often seen and complete spontaneous remission has not been seen (although a few patients have reported small improvement in symptoms over time). Improvement of renal function [i.e., renal transplant or resolution of acute renal failure (ARF)] has terminated the progression of NSF in some patients and even allowed a gradual resolution in some. However, in others, transplant has not led to any appreciable change leading some to speculate that early transplantation prior to the fibrosing process becoming well-established may be needed. Apart from improving renal function, there is no established therapy for NSF. Physiotherapy to limit progression of contractures seems prudent. A myriad of other treatments have been reported in the literature, but evidence is lacking. These include extracorporeal photophoresis, plasmapheresis, steroids, methotrexate, thalidomide, photodynamic therapy, pentoxifylline therapy, IVIG, and others. The role of immunomodulating therapies remains unclear [3, 5].

28

CE-MRA in the Age of Nephrogenic Systemic Fibrosis

Etiologic Role of Gadolinium-Based Contrast Agents The fact that NSF was a “new” disease that had not been recognized before the late 1990s strongly implicated a new pharmacologic agent or medical technique, infectious or toxic agent, as the causative factor. The first article that postulated GBCAs as an etiologic cause for NSF was published by Grobner in 2006 [7]. He reported five cases of NSF in nine patients with ESRD whom he had referred for GBCAenhanced MRI. Although initially met with skepticism, an increasing body of literature has emerged which has strongly implicated GBCAs as the etiologic agent responsible for NSF, beginning with reports of Gd in the biopsy specimens of patients with NSF [8, 9] and culminating with reports of hundreds of cases of NSF following GBCA administration [6]. However, it is very important to keep in mind that NSF remains an exceedingly rare disease. After an estimated 200 million administrations of GBCAs, there are only around 500 cases of NSF reported in the peer-reviewed literature. Although nearly all cases of NSF have been reported following administration of GBCAs, a tiny number of cases have been reported, where the history of GBCA administration was not established. Nevertheless, as discussed later, the link with GBCAs is extremely convincing and is now widely accepted. It is also interesting to note that the number of new cases of NSF has dropped markedly since 2007, coinciding with widespread restriction of the use of GBCAs in patients with severe renal dysfunction, further indirect support to the etiologic role of GBCAs in the development of the disease [10]. NSF has been reported to develop on average approximately 2 months following administration of GBCAs, but ranging from days to many months. In order to better understand the putative mechanisms for the causative role played by GBCA in NSF, a brief discussion of the physical and chemical properties of Gd is required. Gd is a rare earth element in the lanthanide family that was discovered in 1880 by Swiss chemist Jean-Charles Galissard de Marignac. The most common oxidation state for free Gd is +3. The high number of unpaired electrons in Gd3+ (seven) and the very slow electronic relaxation time explain its highly paramagnetic property, which is the basis for its use as an MR imaging contrast agent. However, the ionic radius for Gd3+ is almost equal to that of divalent calcium (Ca2+); hence, Gd3+ can compete for calcium in most biological systems (such as enzymes) that require Ca2+ and it binds to them with greater affinity due to its trivalent charge. This can alter the enzyme function changing the kinetics of enzyme-dependent biological processes. Furthermore, it is known that even at very low concentrations (nano- to micromolar range) Gd3+ inhibits many voltage-gated calcium channels which are critical to the function of smooth and skeletal muscle, cardiac, and neural

389

tissues. Gd can also have the opposite effect, acting as an agonist in certain calcium-sensitive receptors. For these reasons, free Gd is classically recognized as being toxic [11]. As a result of its inherent toxicity, Gd must be chelated or bound to a molecular scaffold (termed the ligand) which serves to insulate the biological system from exposure to free Gd and makes it water soluble and suitable for use as an IV-administered contrast agent. Structurally, there are two types of Gd chelates. “Macrocyclic” chelates consist of circular “closed-off” ligands, where the Gd3+ is contained within the cavity of the ligand. “Linear” chelates are open-chain ligands. The Gd chelate can also be either nonionic or ionic. Following IV administration of standard nonspecific, extracellularly distributed GBCAs, the chelate is rapidly distributed in the extracellular space and then excreted by passive glomerular filtration. The presence of a lipophilic aromatic moiety within the ligand of certain GBCAs (e.g., gadobenate dimeglumine and gadoxetic acid) results in protein binding and hepatobiliary excretion in addition to renal clearance. Gadofosveset contains a lipophilic group which binds reversibly to albumin, thereby acting as a blood pool agent with MR angiography applications. It is mainly renally excreted. Hence, the elimination half-life of these agents is increased in patients with renal disease [11, 12]. Much attention has been focused in the literature toward the differences in chemical properties between different Gd chelates. This is because the majority of “unconfounded” NSF cases reported in the literature (i.e., those cases associated with administration of a single GBCA) have been associated with the use of linear GBCAs, particularly gadodiamide (80–90% of cases), but unconfounded cases have also been reported with gadopentetate dimeglumine and gadoverstamide. While there have been scattered (confounded) cases of NSF associated with the other agents, no unconfounded cases have been reported (confounded meaning two different contrast agents were administered prior to the development of NSF) [13]. The apparent higher risk of these agents persists even when account is taken of differences in the total number of worldwide administrations of these contrast agents; however, since NSF was first reported with gadodiamide, an ascertainment bias cannot be completely excluded. Nevertheless, much attention has been paid to differences in stability of the different Gd chelates as a possible explanation for the apparent differences in their predilection to trigger NSF. The solution stability of a chelate is described as an equilibrium between the free metal, the free ligand, and the chelate. More stable chelates have the equilibrium shifted toward the metal–ligand complex, resulting in lower concentrations of the free metal. The stability of a chelate is critically dependent on the chemical configuration of the ligand, but is also dependent on local conditions, such as temperature and pH. All of the commercially available GBCA chelates are very

390

A. Bharatha et al.

Table 28.1 Nomenclature, chemical properties and kinetic stability of commercially available GBCAs

GBCA acronym/generic/ trade name Gd-DTPA/Gadopentetate dimeglumine/Magnevist Gd-EOB-DTPA/Gadoxetic acid disodium salt/Primovist Gd-BOPTA/Gadobenate dimeglumine/MultiHance MS325/Gadofosveset Trisodium salt/Ablavar Gd-DTPA-BMA/ Gadodiamide/Omniscan Gd-DTPA-BMEA/ Gadoversetamide/OptiMARK Gd-DOTA/Gadoterate meglumine/Dotarem Gd-HP-DO3A/Gadoteridol/ ProHance Gd-BT-DO3A/Gadobutrol

Manufacturer Bayer-Schering

Charge Di-ionic

Chemical structure Open chain

Kinetic stability from Idée JM et al. [11] and defined as per Laurent S et al. [25] as time index of dissociation <0.3 (low); 0.3–0.95 (medium); >0.95 (high) Low

Bayer-Schering

Di-ionic

Open chain

Medium

Bracco

Di-ionic

Open chain

Medium

Lantheus

Tri-ionic

Open chain

Medium

GE Healthcare

Nonionic

Open chain

Low

Covidien

Nonionic

Open chain

Low

Guerbet

Ionic

Macrocyclic

High

Bracco

Nonionic

Macrocyclic

High

Bayer-Schering

Nonionic

Macrocyclic

High

Modified from Idée JM et al. [11]

stable under physiological conditions; however, their stabilities do vary. One measure of this is the so-called kinetic stability, which describes the kinetics of dechelation of GBCAs. In vivo, it is hypothesized that the dechelation may be accompanied by transmetallation, whereby an endogenous metal replaces the Gd3+ in the chelate and the Gd3+ is free to precipitate. Various experimental methods have been used to measure the stability of gadolinium preparations. In general, macrocyclic chelates are more stable than ionic open-chain chelates, which in turn are more stable than nonionic, openchain chelates. Table 28.1 summarizes the naming, chemical structure, and kinetic stability of a variety of commercially available GBCAs. It should be noted that another way to reduce the concentration of free metal is to place excess free ligand in the chemical preparation of the GBCA to scavenge free Gd. This is used in the less-stable preparations (GD-DTPA-BMA and -BMEA) to try to ensure the absence of free Gd3+ cations in the contrast preparation for the duration of its shelf life. The reason for the emphasis on thermodynamic/chemical stability of gadolinium chelates is that free Gd has been implicated as the specific causative agent in the development of NSF. The hypothesis that dechelated Gd may cause NSF is based on the fact that in renal impairment the elimination half-life of GBCAs is substantially increased and that free Gd release can therefore occur to a greater extent than in patients with normal renal function. The released Gd presumably precipitates and is phagocytosed by tissue macrophages (as noted above, Gd has been detected in the tissues

of affected patients, and in some studies it has been possible to localize the Gd to the intracellular space). In experimental models of NSF, rats injected with very high serum doses of Gadodiamide, but not several other marketed GBCAs, developed a fibrosing dermopathy similar to NSF [14]. The rat lesions showed increased dermal cellularity with dendriticlike cells and spindle-shaped fibrocytes. In several studies, increased cellularity and infiltration of CD34-positive fibroblasts has been demonstrated, similar to lesions seen in humans. To investigate the possible role of free Gd, a weak Gd chelate, Gd-EDTA, which is constantly releasing free Gd3+, has been injected into rodent models and has been associated with the development of NSF-like lesions. By contrast, the propensity to induce skin lesions has been negatively correlated with high concentrations of excess ligand, and excess ligand is associated with lower tissue concentrations of Gd3+, lending further indirect support to the free Gd hypothesis [15]. Moreover, it has been shown that free unchelated Gd has a proliferative effect on human dermal fibroblasts in vitro [16]. However, there is also experimental evidence suggesting that chelated Gd is not as inert as previously believed, and that gadolinium chelates themselves may exert a biological action which triggers or contributes to the development of NSF. Based on studies in which peripheral blood monocytes were exposed to GBCAs, it has been shown that chelated Gd (Gadodiamide and Gadopentetate dimeglumine) can directly stimulate human monocytes and macrophages to express and release cytokines and growth factors that can stimulate tissue fibrosis.

28

CE-MRA in the Age of Nephrogenic Systemic Fibrosis

A direct effect of chelated Gd on fibroblasts has also been shown. Gadodiamide added to culture medium causes human fibroblast growth and increases hyaluronan and collagen synthesis (a feature also noted in fibroblasts cultured from NSF patients). There is also experimental evidence to suggest that receptor-mediated cellular uptake of GBCAs may be the mechanism by which they exert a direct profibrotic action [17]. Another factor which has been consistently associated with NSF is increased GBCA dose. Among the cases where dose information is available, 90% of patients received greater than the standard (0.1 mM/kg) dose. The mean dose in patients with NSF, where dose information is available, was just over double dose. Since single-dose exams are obviously much more common, high Gd dose is clearly a significant risk factor. Many high-dose exams were undoubtedly performed as MRA studies. MR systems commonplace in the mid-1990s often required high doses of GBCA to obtain consistent high-quality MRA studies, and since there were no known dangers of GBCAs at the time use of double- and even higher dose studies became standard. Fortunately, technical advances have allowed substantial improvements in the diagnostic quality of low-dose CEMRA and noncontrast MRA using today’s MR imaging systems [18].

NSF and Renal Dysfunction NSF appears to occur exclusively in patients with severe renal dysfunction, with the vast majority of patients on dialysis at the time of exposure. Patients with stage 4 (estimated glomerular filtration rate [eGFR] 15–29 mL/min/1.73 m2) and stage 5 (GFR <15 mL/min/1.73 m2) disease are at greatest risk, and risk appears to be greatest in patients with worse renal function. No well-documented cases have been published of NSF in patients with stage 3 (GFR 30–59 mL/ min/1.73 m2) chronic kidney disease, although a few cases have been verbally reported at the 2009 Yale 3rd Annual Scientific Symposium on NSF [10]. In a prospective study of 168 patients with stage 3 disease administered GBCA, no cases of NSF were reported [19]. Estimates of the risk of NSF following GBCA administration in patients with severe renal disease have varied widely in the literature, ranging from less than 4% to more than 10%; however, such estimates must be interpreted with caution as they are based on retrospective series with variability in their size and quality of clinical and laboratory data [3]. Patients with acute renal failure appear to be at increased risk of NSF. Among a subgroup of 126 patients in which it was possible to determine whether their renal failure was acute (ARF) or chronic, 44% had an element of ARF. The odds ratio of ARF for NSF has been estimated at over 13. It is important to recognize that a small number of patients

391

with NSF were in ARF but had estimated GFR > 30 at the time of GBCA administration – but in this context, estimates of GFR based on serum Cr overestimate renal function. Recovery or improvement may be seen in up to 50% of patients with ARF with recovery of renal function. Although based on a limited number of cases, there is evidence to suggest that waiting until renal function has begun to recover can reduce the risk in the setting of ARF [19].

Other Risk Factors In the worldwide literature, approximately 30% of patients are reported to have had renal transplants around the time of GBCA administration and predating the onset of NSF. It appears, however, that in most of these cases the graft was failing (hence renal function was compromised) and moreover many studies were likely being performed as CEMRA studies to assess transplant vasculature. Therefore, it is likely that renal transplant was a marker for impaired renal function (and possibly also high-dose GBCA CEMRA study) rather than necessarily an independent risk factor for NSF [19]. A “proinflammatory event,” such as recent surgery, acute vessel thrombosis, infection, recent myocardial infarction, and active collagen vascular disease, has been noted in a relatively large number of patients with NSF. In a review of the world literature, in 82 cases in which it was possible to determine whether there was indeed a “proinflammatory event” at the time of GBCA administration, over 50% had such a history. Erythropoetin (Epo) is a medication commonly given to renal failure patients to boost hematocrit. It has been reported to possibly have a proinflammatory effect. It acts through stimulation of red cell production in bone marrow, but other bone marrow proliferation is also stimulated, including fibrocytes. Where medication history was published, EPO use was reported in a high percentage (80%) of patients with NSF. It is hypothesized that these events may lead to activation of circulating inflammatory mediators which may in turn render individual patients more susceptible to fibrocyte activation and differentiation, which may be precursors to NSF [19]. Acidosis is commonly seen in patients with renal failure and is a suspected risk factor for NSF. All five of the affected patients in Grobner’s original series had acidosis at the time of GBCA administration [7]. One possible explanatory mechanism is that at lower pH the extra positively charged protons may compete for Gd-binding sites on the chelator, weakening the Gd–chelate bond and resulting in greater free Gd via transmetallation. For similar reasons, hyperphosphatemia is a suspected risk factor for NSF. Hyperphosphatemia is common in renal failure patients and also common in patients with NSF. Hyperphosphatemia presumably increases the risk of phosphate binding and release of free Gd. Increased

392

free Gd release from GBCA in the setting of increased phosphate concentration has also been demonstrated in experimental work [19]. Many other endogenous ions, such as calcium, zinc, iron, and copper, are also elevated in renal failure patients and may contribute to risk via transmetallation. Recently, attention has been drawn to possible additional risk of NSF conferred by liver failure and particularly patients undergoing liver transplantation. In a recent review of the world literature, 47 NSF patients with liver disease at the time of GBCA administration were identified. Seventy-one percent had cirrhosis. Sixty-six percent underwent liver transplant, a median of about 2 months prior to development of NSF. In the patients in whom the gadolinium exposure could be definitively established, the majority of cases were with Gadodiamide; one case was with Gadobenate dimeglumine. Among the 35 patients in whom renal status was reported, 34 had severe renal insufficiency at the time of GBCA administration, and 33 were on dialysis. The sole patient with mild–moderate renal failure was suffering from ARF on the basis of hepatorenal syndrome following hepatic artery thrombosis of a recent liver transplant, with an eGFR declining from 70 to 35 during which he received 4 doubledose exams. The leading cause of renal failure in these patients was hepatorenal syndrome (28%) and glomerular disease (28%), with ARF in 41%. These data suggest that severe liver disease and liver transplantation per se may not confer additional risk for NSF; however, these patients, particularly those in the peritransplant period, are at high risk for development of renal insufficiency. The nature of their illness often necessitated CEMR (i.e., HCC screening) and CEMRA examinations. Moreover, eGFR based on serum creatinine is known to be inaccurate in patients with liver failure, as the loss of lean muscle mass in liver disease is not incorporated into the formulae used for its calculation. This can lead to an overestimation of renal function. Therefore, the use of GBCAs in these patients should be undertaken with caution and with careful attention to their renal function.

Prevention Suggestions Regulatory bodies and institutions in North America [20–22], Europe [23], and Asia (for example, 24) have published guidelines concerning the use of GBCAs in the era of NSF. It is strongly recommended that radiologists familiarize themselves with the specific regulations and recommendations pertinent to their jurisdictions and develop appropriate institutional policies which are in compliance. In this section, we discuss the issues and general approach used in our practice, which we believe is keeping with typical North American practice. Given the debilitating nature, chronic course, and paucity of effective treatments for NSF, in this case, prevention is

A. Bharatha et al.

clearly the best medicine. In all cases, consideration should be given to whether contrast is needed in the particular clinical context, and indeed whether CEMR is the best imaging test, weighing the pros and cons of other modalities and their respective contrast agents/tracers. The next critical issue is the patient’s renal function. Multiple sources have established that the risk of NSF is greatest in patients with severe renal failure defined as eGFR < 30 mL/min/1.73 m2. Aside from patients with ARF, no well-documented cases have been published in patients with eGFR > 30 mL/min/1.73 m2. Therefore, it is critical to identify patients with severe CRF and those in ARF prior to administering GBCAs. In otherwise healthy and stable patients, as applies to most outpatients and some inpatients who are medically stable, routine serum creatinine screening is probably not warranted and is not recommended by either the Food and Drug Administration (FDA) [21] or the Canadian Association of Radiologists (CAR) [23]. However, a detailed screening questionnaire to identify potential risk factors for CRF should be obtained in all patients. If a risk factor (i.e., history of renal disease, dialysis, solitary kidney, transplant; age > 60; hypertension; diabetes; ischemic heart disease; stroke; vascular disease; organ transplant; myeloma; chemotherapy) is identified, then screening with serum creatinine and calculation of eGFR using standard formulae (such as a Cockroft-Gault) should be performed. If the eGFR is >30 mL/min/1.73 m2 in this group, the risk of NSF is either absent or extremely low and it is generally felt that it is appropriate to proceed with GBCA, and detailed patient counseling regarding risk is probably not needed. If a highdose GBCA exam is going to be performed, consideration should be given to routine screening. In the case of inpatients, particularly unstable patients, consideration should be given to screening even if the questionnaire is negative. Finally, in any patient where there is suspicion of acute renal injury, serum Cr should be drawn and a nephrology consultation should be strongly considered keeping in mind that the eGFR may be unreliable. In patients at high risk for NSF (eGFR < 30 and those with ARF), it must be first determined whether GBCA-enhanced exam is really necessary and whether other imaging tests may provide the needed information. Specifically pertaining to vascular MR imaging, there has been significant improvement in noncontrast MRA techniques, including time-offlight MRA, phase-contrast imaging, arterial spin labeling, and balanced steady-state free precession techniques. In some cases, Doppler ultrasound or functional imaging with scintigraphy may provide adequate information. Consideration should also be given to CT angiography which is performed using iodinated contrast agents. In patients with chronic endstage (anuric) renal failure on dialysis, CT angiography is usually a viable alternative; however, in patients with CRF with residual renal function, the risk of iodinated contrast

28

CE-MRA in the Age of Nephrogenic Systemic Fibrosis

media-induced nephropathy must be weighed against the risks posed by GBCAs. Finally, in some instances (i.e., where intervention is highly likely), proceeding directly to DSA with iodinated contrast agent may also be appropriate. The technique of CO2 angiography is also occasionally used, but has many limitations. However, if the information from CEMRA is deemed necessary, i.e., the risks associated with CEMRA are outweighed by the benefit of the information that is obtained, at-risk patients should not be simply denied necessary imaging. At our institution, a nephrology consultation is obtained and the decision to proceed is based on collaborative decision making among referring team, nephrology, and radiology, taking account of the risks and benefits of the procedure. A detailed consent discussion with the patient is mandatory. Once the decision to proceed has been made, it is suggested that Gadodiamide, Gadopentetate dimeglumine, and Gadoversetamide be avoided as these are the agents that have been convincingly associated with NSF. The lowest possible dose of GBCA providing a diagnostic study should be used and repeated CEMR studies in short time intervals ought to be avoided. GBCA should not be used for CT or DSA studies as an alternative to iodinated contrast in patients with renal dysfunction, as was occasionally done in the past when the risk of NSF was not known. Patients should be instructed of the signs and symptoms of NSF and to seek medical attention should they develop. Published data suggests that prompt hemodialysis enhances GBCA elimination (over 98% free Gd is removed after three sessions) while it appears that peritoneal dialysis is not very effective in removing GBCA. It is, therefore, strongly recommended that prompt hemodialysis following administration of GBCAs be considered, although it remains unproven whether this reduces the risk of development of NSF. It is also unclear whether longer dialysis sessions would be of benefit. At our institution, GBCA exams in at-risk patients are coordinated with the nephrology team to facilitate hemodialysis immediately following the exam whenever possible. There is currently no data to support the use of renal protective protocols, such as bicarbonate administration or aggressive hydration in at-risk patients [13].

Conclusion In this chapter, we have reviewed the clinical features of NSF, risk factors for its development, and its relationship to GBCAs and severe renal dysfunction. Current recommendations and prevention strategies have been discussed with particular attention to how this relates to MRA applications. Although the emergence of NSF has increased the complexity of decision making in CEMRA imaging, it is reassuring that measures taken to reduce the use of GBCA in at-risk

393

patients seem to have stemmed the tide of the disease, evidenced by the dwindling number of cases reported in the literature. The NSF story serves as an important reminder, however, that even the most seemingly safe and widely used medical/imaging procedures are accompanied by risk, some that may be unknown – and therefore should be used judiciously.

References 1. LeBoit PE. What nephrogenic fibrosing dermopathy might be. Arch Dermatol. 2003;139:928–930. 2. Cowper SE, Robin HS, Steinberg SM, Su LD, Gupta S, LeBoit PE. Scleromyxoedema-like cutaneous diseases in renal-dialysis patients. Lancet. 2000;356:1000–1001. 3. Mayr M, Burkhalter F, Bongartz G. Nephrogenic systemic fibrosis: clinical spectrum of disease. J Magn Reson Imaging. 2009;30: 1289–1297. 4. Jimenez SA, Artlett CM, Sandorfi N, et al. Dialysis associated systemic fibrosis (nephrogenic fibrosing dermopathy): study of inflammatory cells and transforming growth factor beta1 expression in affected skin. Arthritis Rheum. 2004;50:2660–2666. 5. Juluru K, Vogel-Claussen J, Macura KJ, Kamel IR, Steever A, Bluemke DA. MR imaging in patients at risk for developing nephrogenic systemic fibrosis: protocols, practices, and imaging techniques to maximize patient safety. Radiographics. 2009;29:9–22. 6. Prince MR, Zhang HL, Prowda JC, Grossman ME, Silvers DN. Nephrogenic systemic fibrosis and its impact on abdominal imaging. Radiographics. 2009;29:1565–1574. 7. Grobner T. Gadolinium – a specific trigger for the development of nephrogenic fibrosing dermopathy and nephrogenic systemic fibrosis? Nephrol Dial Transplant. 2006;21:1104–1108. 8. High WA, Ayers RA, Chandler J, Zito G, Cowper SE. Gadolinium is detectable within the tissue of patients with nephrogenic systemic fibrosis. J Am Acad Dermatol. 2007;56:21–26. 9. Wiginton CD, Kelly B, Oto A, et al. Gadoliniumbased contrast exposure, nephrogenic systemic fibrosis, and gadolinium detection in tissue. AJR Am J Roentgenol. 2008;190:1060–1068. 10. Weinreb JC, Abu-Alfa AK. Gadolinium-based contrast agents and nephrogenic systemic fibrosis: why did it happen and what have we learned? J Magn Reson Imaging. 2009;30:1236–1239. 11. Idée JM, Port M, Robic C, Medina C, Sabatou M, Corot C.Role of thermodynamic and kinetic parameters in gadolinium chelate stability. J Magn Reson Imaging. 2009;30:1249–1258. 12. Aime S, Caravan P. Biodistribution of Gadolinium-based contrast agents, including gadolinium deposition. J Magn Reson Imaging. 2009;30:1259–1267. 13. Leiner T, Kucharczyk W. NSF prevention in clinical practice: summary of recommendations and guidelines in the United States, Canada, and Europe. J Magn Reson Imaging. 2009;30:1357–1363. 14. Sieber MA, Lengsfeld P, Frenzel T, Golfier S, Schmitt-Willich H, Siegmund F, Walter J, Weinmann HJ, Pietsch H. Preclinical investigation to compare different gadolinium-based contrast agents regarding their propensity to release gadolinium in vivo and to trigger nephrogenic systemic fibrosis-like lesions. Eur Radiol. 2008;18:2164–2173. 15. Sieber MA, Steger-Hartmann T, Lengsfeld P, Pietsch H. Gadoliniumbased contrast agents and NSF: evidence from animal experience. J Magn Reson Imaging. 2009;30:1268–1276. 16. Varani J, DaSilva M, Warner RL, Deming MO, Barron AG, Johnson KJ, Swartz RD. Effects of gadolinium-based magnetic resonance imaging contrast agents on human skin in organ culture and human skin fibroblasts. Invest Radiol. 2009;44:74–81.

394 17. Newon BB, Jimenez SA. Mechanism of NSF: New evidence challenging the prevailing theory. J Magn Reson Imaging. 2009; 30:1277–1283. 18. Prince MR, Zhang HL, Roditi GH et al. Risk factors of NSF: a literature review. J Magn Reson Imaging. 2009;30:1298–1308. 19. Bryant BJ 2nd, Im K, Broome DR. Evaluation of the incidence of nephrogenic systemic fibrosis in patients with moderate renal insufficiency administered gadobenate dimeglumine for MRI. Clin Radiol. 2009;64:706–713. 20. United States Food and Drug Administration. Information for Healthcare Professionals Gadolinium-Based Contrast Agents for Magnetic Resonance Imaging (marketed as Magnevist, MultiHance, Omniscan, OptiMARK, ProHance). Washington, D.C. http://www. fda.gov/Drugs/DrugSafety/PostmarketDrugSafetyInformationfor PatientsandProviders/ucm142884.htm. Accessed March 22, 2010. 21. American College of Radiology (ACR). Manual on Contrast Media, 6.0 edition. American College of Radiology; 2008:54–56. www.acr.

A. Bharatha et al.

22.

23.

24.

25.

org/SecondaryMainMenuCategories/quality_safety/contrast_manual. aspx. Accessed March 22, 2010. Padilla-Thornton A, Rafat Zand K, Barrett B, Stein L, Andrew G, Forster BB. Canadian Association of Radiologists national advisory on gadolinium administration and nephrogenic systemic fibrosis. Can Assoc Radiol J. 2008;59:237–240. European Society of Urogenital Radiology. ESUR guideline: gadolinium-based contrast media and nephrogenic systemic fibrosis. Vienna, Austria. http://www.esur.org/Nephrogenic_Fibrosis.39.0. html. Accessed March 22, 2010. Miki Y, Isoda H, Togashi K. Guideline to use gadolinium-based contrast agents at Kyoto University Hospital. J Magn Reson Imaging. 2009;30:1364–1365. Laurent S, Elst LV, Muller RN. Comparative study of the physicochemical properties of six clinical low molecular weight gadolinium contrast agents. Contrast Media Mol Imaging. 2006;1: 128–137.

Emerging Interventional MR Applications

29

Clifford R. Weiss, Aravindan Kolandaivelu, Jeff Bulte, and Aravind Arepally

Introduction The desire to help patients without doing harm has driven medicine to develop minimally invasive methods for diagnosing and treating disease. It is not surprising, then, that over the past three decades medicine has evolved an increasing emphasis on image-guided intervention. Traditionally, these interventions have been performed using fluoroscopy, ultrasound, and computed tomography. Most recently, however, radiologists’ interventional skills and trends toward minimally invasive surgery have converged to create a burgeoning interest in the use of magnetic resonance imaging for guidance in interventional procedures, including the delivery of cellular therapeutics [1]. Initial clinical applications were seen for neurosurgery, where intraoperative magnetic resonance imaging (MRI) systems with movable magnets provided unique glimpses into the operative field [2, 3]. With the advent of 3D gradient echo techniques, further applications for vascular and nonvascular frontiers have appeared [4]. The feasibility of angioplasty of the aorta and iliac arteries with MR tracking has been demonstrated in animal models [5, 6]. C.R. Weiss, MD () Division of Cardiovascular and Interventional Radiology, The Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins Hospital, Blalock 544, 600 N. Wolfe St., Baltimore, MD 21287, USA A. Kolandaivelu, MD Division of Cardiology, Cardiac Arrhythmia Service, Johns Hopkins Hospital, Carnegie 568, 600 N. Wolfe St., Baltimore, MD 21287, USA J. Bulte, PhD Division of MR Research, The Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins Medical Institutes, Blalock 544, 600 N. Wolfe Street, Baltimore, MD 21287, USA A. Arepally, MD Division of Interventional Radiology, Piedmont Healthcare, 1984 Peachtree Road, Suite 505, Atlanta, GA 30309, USA

However, despite the significant technological leaps with this modality, one main criticism has been the lack of any major clinical implementation. Initial efforts to replicate prior fluoroscopic procedures have failed and have not resulted in any clinical impact. More recently, there have been advancements in vascular interventional MRI for emerging clinical applications that previously were not feasible or were limited when performing conventional X-ray imaging. This chapter discusses two emerging vascular applications for interventional vascular MRI: MR-guided atrial fibrillation (AF) ablation and MR-guided cellular therapy.

Emerging Clinical Applications MR-Guided Atrial Fibrillation Ablations Atrial Fibrillation Background Atrial fibrillation is the most common clinically relevant arrhythmia, affecting 1 in 200 people in the general population and 1 in 10 people over the age of 75. The principal morbidities associated with AF are stroke due to embolization of atrial thrombus and symptoms related to poor heart rate regulation with resting heart rates commonly over 110 beats per minute. In 1998, Haissaguerre and colleagues [7] reported that triggering foci for AF typically arise from one or more pulmonary veins. Circumferential ablation around the pulmonary vein connections to the atrium can block the exit of these triggers and has emerged as a primary goal of atrial fibrillation catheter ablation [8]. AF ablation can achieve success rates of 80% in patients with intermittent AF and an otherwise normal heart. However, multiple procedures are commonly required to achieve this and the success rate drops to 50% or less for the more chronic forms of AF associated with age and ischemic, hypertensive, and valvular heart disease [9]. There also remains a 4% risk of significant complications, including cardiac perforation, pulmonary vein stenosis, stroke, and the rare but potentially lethal risk of atrial-esophageal fistula formation [10].

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0_29, © Springer Science+Business Media, LLC 2012

395

396

Intraprocedural MRI Applications Performing electrophysiology procedures within the MRI scanner has a number of potential advantages toward the goal of improving ablation efficacy and safety. First is the ability to visualize ablation lesions with high spatial resolution. The ability to identify regions of incomplete ablation could permit targeting of additional treatment during the procedure. Second, MRI permits catheter position and contact to be visualized relative to soft tissue structures. Catheter– tissue contact is an important factor for efficient lesion formation that is poorly assessed by X-ray fluoroscopy [11]. The ability to provide continuous visualization of soft tissue anatomy without ionizing radiation exposure may also permit more precise and safe catheter manipulation. Third, the 3D MRI anatomy and myocardial scar images used to guide ablation procedures can be acquired at the time of the procedure in the same coordinate system as the “real-time” images used for catheter guidance. This avoids the errors and time delay currently introduced by registering preacquired 3D image data to a separate electrospatial mapping (ESM) catheter tracking system. Over the past 15 years, the basic techniques to enable fully MRI-guided EP procedures have been developed. Real-Time Imaging Lardo and colleagues [12] introduced the potential of using “real-time” MRI (rtMRI) for guiding EP procedures in 2000. Continuous MRI at 1 fps was used to guide a nonferromagnetic EP catheter from an internal jugular vein to selected locations in the right atrium and right ventricle. They also demonstrated the ability to perform and monitor creation of an ablation lesion in the MRI scanner. Delivery of RF ablation energy during imaging caused significant MR image degradation. However, this noise could be adequately suppressed by low-pass filtering of the ablation energy source. After ablation, imaging showed the lesion position and extent using both DEMRI and T2-weighted imaging techniques. Nazarian et al. [13] subsequently demonstrated the ability to use rtMRI to direct a catheter to standard electrophysiology study locations and record satisfactory intracardiac electrograms during rtMRI scanning. Importantly, this was the initial report of performing rtMRI-guided electrophysiology procedure in patients. Recently, Hoffmann and colleagues [14] reported the feasibility of performing a full ablation protocol under five image per second rtMRI guidance. The cavotricuspid isthmus (CTI) is a region between the tricuspid valve and inferior vena cava that can be ablated to cure the typical atrial flutter arrhythmia. Using a nonferromagnetic ablation catheter, guidance to and ablation of the entire CTI region were performed. Completion of CTI ablation was guided by T2-weighted imaging. After MRI-guided ablation, the current clinical end point of conduction block across the CTI was demonstrated in 15 of 18 procedures.

C.R. Weiss et al.

Lesion Imaging These promising but imperfect results allude to the need for better intraprocedure MR lesion assessment techniques. Acute interstitial edema likely leads to the hyperintense region on T2-weighted MRI that corresponds to acute RF ablation damage; however, this region may overestimate the region of ablation-induced tissue necrosis [15]. RF ablation lesions may also be visualized by noncontrast T1-weighted imaging, but the lesion contrast-to-noise ratio is poor compared with T2-weighted imaging or DEMRI [15]. Gadolinium contrast DEMRI appears capable of providing good lesion visualization; however, the limit on total gadolinium administration limits its use for serial lesion monitoring during the procedure [16]. MR thermography is a promising technique that utilizes the decrease in proton resonance frequency with increasing temperature to estimate the region of heatinginduced tissue necrosis [17]. This technique has been used to follow tumor ablation in the uterus, liver, prostate, and brain using diverse energy sources, including RF, but is sensitive to motion. Its use for following RF ablation in the beating heart is being investigated. Device Tracking Another area of active investigation is improved techniques for rtMRI guidance of complex arrhythmia ablation. While fluoroscopy provides projection images where the entire catheter body and tip are easily visualized, standard 2D rtMRI only depicts a slice through the body that is 5–10-mm thick. Curved devices, such as catheters, may pass in and out of the MR image plane and lead to misinterpretation of the device position. One approach to this problem is to track the catheter tip position to provide similar information to that of current ESM systems. The currently favored method was first described by Dumoulin et al. [18] and uses rapid 1D projection rtMRI to identify the 3D position of a small receiver coil located in the catheter tip. Using this method, multicoil catheter designs have been used for MRI catheter guidance in the atrium and ventricles to target ablation and obtain surface maps of myocardial electrical properties [19, 20]. This position tracking technique has also been interleaved with rtMRI sequences to automatically move the image plane position to the catheter tip location during device manipulation. This technique likely represents the next step in rtMRI catheter guidance. Safety/Devices MR-guided intravascular procedures raise a number of safety concerns beyond the standard SAR tissue heating concerns associated with exposure to scanning-associated electromagnetic radiation [21]. In addition to avoiding ferromagnetic materials that could experience significant forces when brought close to the scanner, catheters must be designed to avoid RF transmission-induced current that can lead to significant

29

Emerging Interventional MR Applications

heating [22]. A number of techniques have been developed to avoid MRI-induced heating of interventional EP devices, including the use of nonmetallic polymer materials for structural elements, high-resistance alloys, RF chokes, and transformer transmission lines [23–26]. Development of devices for safe MRI-guided EP procedures requires cooperation among academic centers, imaging companies, and regulatory agencies.

MR-Guided Cellular Therapy Overview of Cellular Therapy Analogous to MR-guided ablation of atrial fibrillation, image-guided cellular therapies necessitate visualization of the delivery system and also the agents being delivered. Physicians have used image guidance for delivery of various therapeutic agents for over three decades. The main goal of all local delivery is to increase the concentration of a specific therapeutic agent in a target tissue with minimal nontarget distribution. Compared to systemic therapy, local delivery provides a high level of therapeutic efficacy with minimal systemic effects. In a similar manner, cellular therapies can take advantage of these techniques to enhance the delivery process. Although some cellular therapies are adequately managed with systemic delivery, certain organs and conditions require a targeted approach for maximum affect. In fact due to the anatomical constraints of the vascularity of specific organs, systemic therapy is precluded and image-guided techniques are necessary to overcome some of these barriers. For example, the vasculature of the brain, liver, and pancreas has been shown to have unique anatomical boundaries that create a barrier to conventional therapies and therefore do not respond to conventional systemic therapies. With the brain, the blood–brain barrier acts as a physiologic barrier that prevents the migration and the transport of agents from the systemic vasculature. With both the liver and pancreas, there is a separate anatomic venous vascular supply, termed portomesenteric system, that is completely isolated from systemic circulation. Therefore, with disorders involving these organs (i.e., stroke, cirrhosis, and diabetes), accessing these secluded organs is critical to the success of cellular therapy. Finally, another relevant opportunity with image-guided therapy is the ability to not only deliver to target organs, but also to administer only into injured tissues. As demonstrated by multiple studies, substantial mobilization of stem cells has been demonstrated after myocardial infarction and also with liver injury. Clearly, the homing of cells to injured tissues is a major mechanism of regeneration and there has been extensive work in utilizing image-guided therapy to help foster this process by providing targeted delivery into injured tissue.

397

Cell Tracking Using MRI The clinical use of novel experimental immune and stem cell therapies calls for suitable methods that can monitor the cellular biodistribution noninvasively following administration. Among the different clinically used imaging techniques, MRI has superior spatial resolution with excellent soft tissue contrast. In order for exogenous therapeutic cells to be detected, they need to have a different contrast from endogenous cells. There are several different approaches to endow cells with MR-visible properties [27]. The most sensitive and widely used MR labels today are the superparamagnetic iron oxide nanoparticles or SPIOs. SPIOs are clinically approved and create strong, local magnetic field disturbances that spoil the MR signal leading to hypointense contrast. As of 2009, at least four clinical MRI cell tracking studies have been performed, reviewed in detail elsewhere [1]. Cells can be labeled with SPIOs, notably Feridex®, using simple incubation [28, 29], following coating with transfection agents [30] or by magnetoelectroporation [31]. The optimal labeling technique depends on the cell type (i.e., is the cell phagocytic or very hard to transfect) and its application. Paramagnetic gadolinium chelates can also be used [32], although the induced contrast can be ambigious – the positive contrast decreases at higher fields while negative, susceptibilitybased contrast takes over. It has not been used clinically and there are serious concerns about its safety for cell tracking – gadolinium is toxic once dechelated, which may occur following prolonged retention by cells in acidic compartments (i.e., lysosomes). In addition, the NSF issue has dampened enthusiasm to initiate preliminary clinical studies. Conventional approaches require the cells to be prelabeled with contrast agent before injection. An example of homing of bone marrow (BM) stem cells is shown in Fig. 29.1. Hematopoietic bone marrow cells participate in the formation of atherosclerosis. Following transplantation of Feridexlabeled BM stem cells, large MR signal voids of the aorta walls were seen that were attributed to the “blooming” effect of migrated Feridex-BM stem cells in the plaques. Homing of Feridex-labeled BM stem cells has also been demonstrated for injured arteries [33]. In the recipient mice, the left femoral arteries were injured using a cuff-constriction or endothelium-damage approach while the right femoral arteries were uninjured to serve as controls. MRI showed larger regions of hypointensity with Feridex-labeled cells at the sites of the injured arteries as compared to control arteries (p < 0.01) (see Fig. 29.2). Both studies clearly demonstrate evidence that supports the potential use of MRI to detect homing of intravenously injected BM cells to vascular abnormalities. For MRI cell tracking, an emerging application is to perform 1H MRI only for anatomical information in conjunction with the use of 19F as a tracer molecule. As there is no endogenous background signal, “hot-spot” images of the tracer can

Fig. 29.1 (a–c) Cross-sectional view of representative in vivo 4.7-T MR images of aortas from atherosclerotic ApoE mice fed a high-cholestrol diet. Insets outline the ascending aorta, showing large MR signal void of the aortic wall in animal c following Feridex-labeled bone marrow (BM) stem cell transplantation while the aortae in animals a and b appear as bright rings. (d–f) Magnification of insets of a–c. In d and e (controls, treated with no BM stem cells and unlabeled BM stem cells), the thickened aortic walls due to atherosclerotic plaques (arrows) are visualized, with the aortic walls appearing as bright rings. In f (treated

with Feridex-labeled BM stem cells), larger signal voids (open arrows) of the aortic wall are seen. (g–i) Histochemical staining for Feridex with Prussian blue (scale bar = 50 mm) and (j–l) histochemical staining for LacZ with X-gal (scale bar = 20 mm). Numerous Feridex- and LacZpositive cells (blue color, arrows) are detected in aortic tissues of mice receiving BM stem cell transplants (in k, i, and l), which are not visualized in the control aortic tissues (in g, j and h) (reproduced, with permission, from Qiu B et al. [43])

Fig. 29.2 (a–d) (upper panel) Cross-sectional view of representative in vivo 4.7-T MR images of different arteries with various treatments. Images a and c are taken from the right legs of mice, which are vertically flipped for convenient comparison. Insets outline the femoral artery areas, showing larger hypointense areas at the site of the injured femoral artery in animal d with both Feridex-labeled BM stem cell transplantation and the cuff placement while no such findings appear at the sites of the

uninjured femoral arteries in animals a and c. In animal b, the cuff itself with circulated blood creates a small hypointense circle. (e–f) (middle panel) Histochemical staining for iron with Prussian blue. (i–l) (lower panel) Histochemical staining for LacZ with X-gal. Numerous ironpositive cells (blue color in h) and LacZ-positive cells (blue color in l) are detected, which are not visualized in the control tissues (e–g and i–k). 40× (reproduced, with permission, from Gao F et al. [33])

29

Emerging Interventional MR Applications

be obtained and superimposed on the anatomical 1H images [34–36]. The strategies for intracellular labeling is similar to that used for labeling with SPIO particles: in this case, 19F particles can be simply phagocytized [37], mixed with transfection agents [35], or electroporated into cells. Finally, novel semipermeable microcapsules have been developed that immunoprotect cells and are visible with multiple modalities. Initially, X-ray-visible alginate capsules were developed that contained barium or bismuth (“X caps”) [38], but capsules containing SPIO (“magnetocapsules”) have now also been described [39]. In principle, any type of contrast agent can be coencapsulated allowing multimodality tracking. Thus, depending on the specific cellular imaging application, a quite complete set of tools exists for MRI cell tracking.

Device Tracking for Cell Therapy A promising and exciting new development for cell and drug delivery is the fusion of transvascular and percutaneous approaches for a new procedure that is best described as a hybrid technique. In this method, the vascular system (arterial or venous) is used as a conduit to perform punctures of the target tissues for local drug/cellular delivery. Compared to the previously discussed methods, this technique is technically more invasive, requires sophisticated imaging support/ devices, and experience of the operator in performing endovascular procedures. However, the main advantages are that it is less invasive than current surgical options, provides access to certain tissues and organs that are difficult to reach, and minimizes potential systemic side effects by delivering only to pathological tissues. The advantage of this technique is best seen for procedures, such as direct intramyocardial injections, which currently require surgical exposure. Due to the results of phase I clinical trials that have shown that in vivo tissue engineering of the myocardium is feasible with local surgical intramyocardial delivery, there has been tremendous research dedicated to finding a percutaneous option. The challenge with performing this procedure is to deliver the therapeutic agent only to the damaged myocardium. MRI with its inherent ability to provide real-time visualization of the myocardium without radiation or iodinated contrast has become the modality of choice for this procedure. The initial feasibility of MR-guided hybrid procedures was demonstrated by Lederman et al. [40], where a commercially available injection catheter (Stiletto™; Boston Scientific, Natick, MA) was modified for real-time, MR-guided intramyocardial injections. Using commercial real-time imaging software and a 1.5-T MR scanner, the modified Stileto™ system was readily visible during advancement, successfully oriented in the left ventricle followed by delivery of dilute gadolinium–DTPA into the myocardium of swine. Additionally, Dick et al. [41] further

399

modified the Stiletto™ injection catheter system so that the guide catheters were arranged as one RF antenna. The second RF antenna, a microcoil, was built into the distal tip of the injection needle system which created a high-intensity signal at the distal tip in order to enhance positioning before myocardial injections. Based upon this work, several other groups have also demonstrated the accuracy of the hybrid technique. Saeed et al. [42] used an XMR system along with a modified clinical catheter for myocardial delivery; in their study, X-ray was used for 2D guidance into the left ventricle from the femoral artery and 3D MR fluoroscopy was utilized for injections. Additionally, they were able to demonstrate the fundamental benefit of MRI by delivering gadolinium chelate agents only to an artificially created target of 1.5–2 cm in the myocardium. To further demonstrate this accuracy, other authors have successfully delivered iron particles (which have T1/T2 effects) and dysprosium-based contrast agents (T2/T2* effects) only into the infracted myocardium of animal models. Despite the ability of these investigators to modify commercially available catheter/injection systems, a key limitation is that current drug delivery devices are designed for fluoroscopy and not fully optimized for MR delivery. Due to those concerns, novel catheter designs for hybrid delivery under MR have now been developed. At our institution, we have developed a steerable intramyocardial injection catheter with a defl ectable distal section, which can be actively tracked and used to deliver therapeutics to target tissue under MR guidance. The components of the catheter are arranged to form a “loopless-antenna” RF receiver coil that provides a region of high signal along the length of the coil to enable active tracking. The distal tip of the catheter was modified to create a “coiled tip” that provides high-intensity signal at the distal tip. Therefore, the position of the distal tip as it apposes the target tissue can be visualized before the needle is advanced. Using this steerable myocardial injection catheter, successful targeted delivery of gadolinium contrast and ironlabeled mesenchymal stromal cells to myocardial infarct border targets was performed.

Summary As demonstrated, recently, there has been significant improvement in both imaging and hardware design that now allows for highly targeted delivery of agents. Cardiac applications, such as ablation of the atrium or delivery of stem cells to the myocardium, require a complex imaging platform to ensure proper placement of the therapeutic modality. In addition to precise delivery, MRI provides multiparametric imaging that yields anatomic, physiologic, and functional data that is currently not feasible with other imaging modalities.

400

Furthermore, with real-time tracking and feedback mechanisms from the target tissue, MRI technology significantly impacts and enables such disruptive technologies. As research continues to further integrate MRI with novel clinical procedures, including the advent of cellular therapeutics, the use of interventional vascular MRI will continue to grow.

References 1. Bulte JW. In vivo MRI cell tracking: clinical studies. AJR Am J Roentgenol. 2009;193:314–325. 2. Tronnier VM, Wirtz CR, Knauth M, et al. Intraoperative diagnostic and interventional magnetic resonance imaging in neurosurgery. Neurosurgery. 1997;40:891–900; discussion 900–892. 3. Maurer CR, Jr., Hill DL, Martin AJ, et al. Investigation of intraoperative brain deformation using a 1.5-T interventional MR system: preliminary results. IEEE Trans Med Imaging. 1998;17:817–825. 4. Ladd ME, Quick HH, Debatin JF. Interventional MRA and intravascular imaging. J Magn Reson Imaging. 2000;12:534–546. 5. Godart F, Beregi JP, Nicol L, et al. MR-guided balloon angioplasty of stenosed aorta: in vivo evaluation using near-standard instruments and a passive tracking technique. J Magn Reson Imaging. 2000;12:639–644. 6. Becker GJ. 2000 RSNA annual oration in diagnostic radiology: the future of interventional radiology. Radiology. 2001;220:281–292. 7. Haissaguerre M, Jais P, Shah DC, et al. Spontaneous initiation of atrial fibrillation by ectopic beats originating in the pulmonary veins. N Engl J Med. 1998;339:659–666. 8. Pappone C, Rosanio S, Oreto G, et al. Circumferential radiofrequency ablation of pulmonary vein ostia: A new anatomic approach for curing atrial fibrillation. Circulation. 2000;102:2619–2628. 9. Calkins H. Further insight into the technique and outcomes of “curative” catheter ablation of atrial fibrillation. Circ Arrhythm Electrophysiol. 2008;1:238–239. 10. Dong J, Calkins H. Technology insight: catheter ablation of the pulmonary veins in the treatment of atrial fibrillation. Nat Clin Pract Cardiovasc Med. 2005;2:159–166. 11. Wittkampf FH, Nakagawa H. RF catheter ablation: Lessons on lesions. Pacing Clin Electrophysiol. 2006;29:1285–1297. 12. Lardo AC, McVeigh ER, Jumrussirikul P, et al. Visualization and temporal/spatial characterization of cardiac radiofrequency ablation lesions using magnetic resonance imaging. Circulation. 2000; 102:698–705. 13. Nazarian S, Kolandaivelu A, Zviman MM, et al. Feasibility of realtime magnetic resonance imaging for catheter guidance in electrophysiology studies. Circulation. 2008;118:223–229. 14. Hoffmann BA, Koops A, Rostock T, et al. Interactive real-time mapping and catheter ablation of the cavotricuspid isthmus guided by magnetic resonance imaging in a porcine model. Eur Heart J. 2010;31:450–456. 15. Dickfeld T, Kato R, Zviman M, et al. Characterization of acute and subacute radiofrequency ablation lesions with nonenhanced magnetic resonance imaging. Heart Rhythm. 2007;4:208–214. 16. Dickfeld T, Kato R, Zviman M, et al. Characterization of radiofrequency ablation lesions with gadolinium-enhanced cardiovascular magnetic resonance imaging. J Am Coll Cardiol. 2006;47: 370–378. 17. Rieke V, Butts Pauly K. MR thermometry. J Magn Reson Imaging. 2008;27:376–390.

C.R. Weiss et al. 18. Dumoulin CL, Souza SP, Darrow RD. Real-time position monitoring of invasive devices using magnetic resonance. Magn Reson Med. 1993;29:411–415. 19. Dukkipati SR, Mallozzi R, Schmidt EJ, et al. Electroanatomic mapping of the left ventricle in a porcine model of chronic myocardial infarction with magnetic resonance-based catheter tracking. Circulation. 2008;118:853–862. 20. Schmidt EJ, Mallozzi RP, Thiagalingam A, et al. Electroanatomic mapping and radiofrequency ablation of porcine left atria and atrioventricular nodes using magnetic resonance catheter tracking. Circ Arrhythm Electrophysiol. 2009;2:695–704. 21. Shellock FG, Crues JV. MR procedures: biologic effects, safety, and patient care. Radiology. 2004;232:635–652. 22. Nitz WR, Oppelt A, Renz W, Manke C, Lenhart M, Link J. On the heating of linear conductive structures as guide wires and catheters in interventional MRI. J Magn Reson Imaging. 2001;13:105–114. 23. Weiss S, Vernickel P, Schaeffter T, Schulz V, Gleich B. Transmission line for improved RF safety of interventional devices. Magn Reson Med. 2005;54:182–189. 24. Krueger S, Schmitz S, Weiss S, et al. An MR guidewire based on micropultruded fiber-reinforced material. Magn Reson Med. 2008; 60:1190–1196. 25. Wirtz D, Lips O, David B, Krueger S, Wiess S. Diagnostic MR-electrophysiology catheter with highly resistive wires for reduction of RF-heating. In: Joint Annual Meeting ISMRM-ESMRMB. Berlin, 2007. 26. Ladd ME, Quick HH. Reduction of resonant RF heating in intravascular catheters using coaxial chokes. Magn Reson Med. 2000;43: 615–619. 27. Long CM, Bulte JW. In vivo tracking of cellular therapeutics using magnetic resonance imaging. Expert Opin Biol Ther. 2009;9: 293–306. 28. de Vries IJ, Lesterhuis WJ, Barentsz JO, et al. Magnetic resonance tracking of dendritic cells in melanoma patients for monitoring of cellular therapy. Nat Biotechnol. 2005;23:1407–1413. 29. Long CM, van Laarhoven HW, Bulte JW, Levitsky HI. Magnetovaccination as a novel method to assess and quantify dendritic cell tumor antigen capture and delivery to lymph nodes. Cancer Res. 2009;69:3180–3187. 30. Frank JA, Miller BR, Arbab AS, et al. Clinically applicable labeling of mammalian and stem cells by combining superparamagnetic iron oxides and transfection agents. Radiology. 2003;228:480–487. 31. Walczak P, Kedziorek DA, Gilad AA, Lin S, Bulte JW. Instant MR labeling of stem cells using magnetoelectroporation. Magn Reson Med. 2005;54:769–774. 32. Anderson SA, Lee KK, Frank JA. Gadolinium-fullerenol as a paramagnetic contrast agent for cellular imaging. Invest Radiol. 2006;41:332–338. 33. Gao F, Kar S, Zhang J, et al. MRI of intravenously injected bone marrow cells homing to the site of injured arteries. NMR Biomed. 2007;20:673–681. 34. Bulte JW. Hot spot MRI emerges from the background. Nat Biotechnol. 2005;23:945–946. 35. Ahrens ET, Flores R, Xu H, Morel PA. In vivo imaging platform for tracking immunotherapeutic cells. Nat Biotechnol. 2005;23: 983–987. 36. Partlow KC, Chen J, Brant JA, et al. 19 F magnetic resonance imaging for stem/progenitor cell tracking with multiple unique perfluorocarbon nanobeacons. Faseb J. 2007;21:1647–1654. 37. Ruiz-Cabello J, Walczak P, Kedziorek DA, et al. In vivo “hot spot” MR imaging of neural stem cells using fluorinated nanoparticles. Magn Reson Med. 2008;60:1506–1511.

29

Emerging Interventional MR Applications

38. Barnett BP, Kraitchman DL, Lauzon C, et al. Radiopaque alginate microcapsules for X-ray visualization and immunoprotection of cellular therapeutics. Mol Pharm. 2006;3:531–538. 39. Barnett BP, Arepally A, Karmarkar PV, et al. Magnetic resonanceguided, real-time targeted delivery and imaging of magnetocapsules immunoprotecting pancreatic islet cells. Nat Med. 2007;13: 986–991. 40. Lederman RJ, Guttman MA, Peters DC, et al. Catheter-based endomyocardial injection with real-time magnetic resonance imaging. Circulation. 2002;105:1282–1284.

401 41. Dick AJ, Guttman MA, Raman VK, et al. Magnetic resonance fluoroscopy allows targeted delivery of mesenchymal stem cells to infarct borders in swine. Circulation. 2003;108:2899–2904. 42. Saeed M, Lee R, Martin A, et al. Transendocardial delivery of extracellular myocardial markers by using combination X-ray/MR fluoroscopic guidance: feasibility study in dogs. Radiology. 2004;231:689–696. 43. Qiu B, Gao F, Walczak P, et al. In vivo MR imaging of bone marrow cells trafficking to atherosclerotic plaques. J Magn Reson Imaging. 2007;26:339–343.

wwwwwwwwwwww

Index

A Abdominal aorta and mesenteric vessels anatomy abdominal aorta, 273 celiac artery, 273 inferior mesenteric artery, 273–274 mesenteric arterial system, 273 superior mesenteric artery, 273 black blood techniques, 272 contrast-enhanced MRA contrast administration and bolus timing, 271–272 hardware considerations, 270 pulse sequence, 270–271 mesenteric arterial system, MRA acute mesenteric arterial thrombosis, 277 acute mesenteric ischemia, 275–276 acute SMA embolism, 276 aortic dissection, 277 atherosclerotic CMI, 277–279 chronic mesenteric ischemia, 277 mesenteric venous thrombosis, 277 nonatherosclerotic vascular pathology, 279 non-occlusive mesenteric ischemia, 277 MRA, abdominal aorta abdominal aortic aneurysm, 274 aortic dissections, 275 aortic occlusive disease, 274–275 aortitis, 275 congenital anomalies, 275 MRA at 3 T, 272–273 noncontrast MRA, 270 phase-contrast (PC) MRA, 269 steady-state free precession, 269–270 time-of-flight MRA, 269 time-resolved MRA, 272 transplant surgery, MRA, 279–280 T1-weighted gradient-echo fat-saturated imaging, 272 Acceleration errors, 63 Acute ischemic stroke, pediatric MRA moyamoya disease, 374–375 transient cerebral vasculopathy, 373–374 vascular dissection and stroke, 375 vasculitis, 375 Acute mesenteric arterial thrombosis, 277 Allison, J.W., 369 Alsaid, H., 205 Aneurysm, MRA techniques, 217–219 Anomalous venous drainage, 265

Anson, B.J., 298 Aortic stenoses, 248 Arepally, A., 395 Arterial disease, MRA techniques aneurysm, 217–219 atherosclerotic disease, 214–215 dissection, 216–217 moyamoya disease, 216 vasculitis, 215–216 Arterial spin labeling (ASL) noncontrast coronary artery imaging, 135–136 noncontrast LE-MRA techniques, 326 renal vascular diseases, 288 Arteriovenous fistulas, 309 Atherosclerotic CMI, 277–279 Atherosclerotic disease, 214–215 Atherosclerotic plaques, 199 Atrial fibrillation clinical applications, MRI device tracking, 396 electrophysiology procedures, 396 lesion imaging, 396 real-time imaging, 396 safety/devices, 396–397 pulmonary vascular imaging, 264–265 Aviv, R.I., 375 Axel, L., 40

B Backes, W.H., 376 Balanced steady-state free precession (bSSFP) lower extremity MRA, 326 renal vascular diseases, 287 thoracic aorta, 245 Balu, N., 113 Barnes, S., 157 Beddy, P., 253 Beilvert, A., 202 Bernstein, M.A., 61 Bharatha, A., 387 Black-blood techniques, 113 abdominal aorta and mesenteric vessels, 272 coronary MRA, 134 thoracic aorta, 241 Blatter, D.D., 177 Block, W., 169 Block, W.F., 172

J.C. Carr and T.J. Carroll (eds.), Magnetic Resonance Angiography: Principles and Applications, DOI 10.1007/978-1-4419-1686-0, © Springer Science+Business Media, LLC 2012

403

404 Blood pool contrast agents, 315, 355 Boada, F.E., 175 Bornert, P., 172 Botnar, R., 203 Briley-Saebo, K.C., 125, 202 Bronchial arteries imaging, 262 Buerger’s disease, 332–333 Buis, D.R., 369 Bulte, J., 395

C CAD visualization anomalous coronary arteries and coronary aneurysms, 133 coronary artery bypass graft assessment, 133 coronary flow imaging, 134 coronary stenosis identification, 132–133 coronary vessel wall imaging, 134 Cardiac motion suppression, 129–130 Carotid and vertebral circulation, clinical applications contrast-enhanced 3D Fourier transform MRA efficacy, 229 limitations, 229 technique, 228 3 T scanners, 229 craniocervical dissections imaging findings, 231 MRA technique, 231 posttraumatic and spontaneous cases, 230–231 extracranial carotid atherosclerosis carotid endarterectomy, 225 carotid stenosis measurement, 226 CASANOVA, 226 contrast-enhanced MRA, 229 stroke, 225 TOF MRA, 228 facial/trigeminal nerve, vascular compression, 235–236 fibromuscular dysplasia, 232–233 head and neck neoplasms, 236–237 intracranial carotid atherosclerosis, 233–234 moyamoya disease and moyamoya syndrome, 234–235 MRA techniques, 226 sickle cell disease, 235 subclavian steal syndrome, 233 Time-of-flight MRA application, 226–227 efficacy, 228 limitations, 227–228 vertebral artery atherosclerosis, 229–230 Carr, J., 239, 351 Cartesian methods, CE-MRA k-space sampling, 76–78 MRI data acquisition, 83 parallel acquisition, 78–79 parallel imaging, angiography, 187–188 receiver coils, 80–81 sampling, 76 signal enhancement effect, 81–82 temporal fidelity, 82 temporal footprint analysis, 79–80 in vivo results, 83–84 Catheter angiography, 257 Celiac artery, anatomy, 273 Cellular therapy, MRI

Index cell tracking, 397–399 device tracking, 399 CE-MRA. See Contrast-enhanced MR angiography (CE-MRA) Cerebral arteriovenous malformations (cAVM), 219–220 Chooi, W.K., 367, 371 Cho, Z.H., 151 Cine SSFP, 240 Coleman, S.S., 298 Collins, J., 351 Collins, J.D., 319 Common Carotid (CC) method, 226 Contrast agents MRA. See also Targeted contrast agents, molecular imaging gadolinium-based agents, 383–384 lower extremity MRA, 323–324 macromolecular agents, 383 non-protein-binding agents, 381–382 paramagnetic agents, 381 protein-binding agents, 383 superparamagnetic agents, 384 upper extremity and hand vessels, MRA, 315 Contrast-enhanced 3D Fourier transform MRA carotid and vertebral circulation efficacy, 229 limitations, 229 technique, 228 3 T scanners, 229 Contrast-enhanced MR angiography (CE-MRA). See also Low-dose contrast-enhanced MR angiography abdominal aorta and mesenteric vessels contrast administration and bolus timing, 271–272 hardware considerations, 270 pulse sequence, 270–271 best guess technique, 70 bolus timing considerations, 69–70 Cartesian methods (See Cartesian methods, CE-MRA) contrast agents, 67 contrast dose and injection rate, 68 coronary arteries (See Coronary artery, CE-MRA) Fourier ( k-space) considerations, 66–67 hand vs. power injection, 69 intracranial arterial and venous disease, 213–214 MR fluoroscopy, 70–71 nephrogenic systemic fibrosis (See Nephrogenic systemic fibrosis (NSF), CE-MRA) patient preparation, 68–69 postproccessing and display, 71–72 pulmonary embolism, 257 pulse sequence, 66 test bolus technique, 70 theory, 65 thoracic aorta, 241–245 triggering, automatic, 70 whole-body MRA, 72 Contrast media, gadolinium-based, 285, 287 Contrast-to-noise (CNR), 341–343 Conventional CE-MRA, 107–108, 242–243 Coronary artery, CE-MRA blood vs. surrounding tissue, contrast, 142 contrast agent administration, 143 contrast agents benefits, 142 contrast-enhanced EPI whole-heart technique, 143 contrast-enhanced whole-heart CMRA, 144 high-field coronary imaging, 144 k-space trajectories, 143 motion compensation, 141

Index parallel imaging benefits, 143–144 spatial resolution, 141–142 SSFP whole-heart CMRA, 143 whole-heart coronary MRA, 3T clinical applications, 145–147 navigator efficiency, 145 patient training, 144–145 survey scanning, 145 vector electrocardiogram, 145 Coronary artery disease, noninvasive imaging clinical applications anomalous coronary arteries, 338 bypass graft assessment, 339–340 vasculitides, 339 coronary vein MRI, 343–345 MRI vs. MDCT, 337–338 technical advances and impediments coronary motion, 340–341 SNR and CNR, 341–343 Coronary vessel wall imaging, 124 Craniocervical dissections imaging findings, 231 MRA technique, 231 posttraumatic and spontaneous cases, 230–231 Cross-linked iron oxides (CLIO), 201 Cystic adventitial disease, 331–332

D Dale, B., 176 Davarpanah, A.H., 351 DeBakey, M.E., 245, 246 Debrey, S.M., 229 Deistung, A., 160 3D fast spin echo (FSE) imaging thoracic aorta, 245 upper extremity and hand vessels, MRA, 300–301 Dick, A.J., 399 Diffusion-weighted imaging (DWI), 214 Double inversion recovery (DIR), 114–116 2D TOF MRA, 227 3D TOF MRA, 227 Dumoulin, C.L., 396 Dural AV fistula, 220–221 Duyn, J.H., 177, 178, 180 Du, Y.P., 160 Dynamic contrast-enhanced MRI, 288–290

E ECST method, 226 Eddy currents, 61–62 Eleftheriou, D., 375 Embolic disease, 306–307 Erdman, W.A., 357 Evangelista, A., 243 Evans, A.J., 356 Extracranial carotid atherosclerosis MRA technique, clinical uses carotid endarterectomy, 225 carotid stenosis measurement, 226 CASANOVA, 226 contrast-enhanced MRA, 229 stroke, 225 TOF MRA, 228

405 F Facial/trigeminal nerve, vascular compression, 235–236 Fasulakis, S., 368 Fat suppression, 119 Fenchel, M., 287 Ferumoxytol, 315 Fibromuscular dysplasia, 232–233, 279 Filtered back projection (FBP), 175 Finn, J.P., 107, 361 Flow compensation, 42 Flow-dependent noncontrast MR angiography in abdomen, 93 2D/3D partial-Fourier FSE flow-in technique, 98–99 flow-out technique, 99 tag-on and tag-off alternate subtraction, 99–100 fresh blood imaging, 91–92 noncontrast MR venography, 96–97 peripheral MRA FBI-Navi, 94 MIP procession, 95 readout (RO)/frequency direction, 93 STIR pulse, 96 systolic and diastolic triggering, 93 phase contrast, 91 principles, 92–93 SSFP Imaging b noncontrast MRA, 101 subtractive methods, 101–102 quiescent interval single-shot MRA, 102–104 time-of-flight (TOF) MOTSA, 89 SORS pulse, 90 TONE, 89 venography, 96–98 Flow enhancement effect (FRE), 41, 42 Flow quantification, 56. See also Phase-contrast MRI Flow-sensitive 4D MRI, 58–60 Fourier ( k-space) considerations, 66 Frahm, J., 172 Francois, C.J., 362 Fraser, D.G., 356 Frias, J., 202 First-order gradient moment nulling, 42 Fushimi, Y., 235

G Gadobenate dimeglumine (Gd-BOPTA), 315 Gadobenate dimeglumine (multihance), 381, 383 Gadobutrol (gadovist), 381, 383 Gadofosveset trisodium (ablavar), 381, 383 Gadolinium-based contrast agents (GBCA), 383–384 nephrogenic systemic fibrosis, CE-MRA chelate, solution stability of, 389 chemical properties and kinetic stability, 390 dosing limit, 391 history, 389 inherent toxicity, 389 nomenclature, 390 putative mechanisms, 389 thermodynamic/chemical stability, 390 Gadolinium chelates, 201 Gadolinium contrast agents, 355 Gadolinium-enhanced MRA lower extremity MRA

406 Gadolinium-enhanced MRA (continued) adaptations, 322–323 dosing, 323 high-resolution, 320 k-space, 321 parallel imaging technologies, 321 spatial resolution, 320 three-dimensional acquisitions, 320 time-resolved, 321 Gadolinium-enhanced MRV, 303, 304 Galen vein aneurysmal dilatation, 371 aneurysmal malformation, 370–371 malformation, 369 MRA role, malformation, 371–373 Garovic, V.D., 286 Generalized autocalibrating partially parallel acquisitions (GRAPPA), 189–191 Glover, G.H., 172 Griswold, M., 185 Gullberg, G.T., 39

H Haacke, E.M., 157 Haage, P., 259 Hadizadeh, D.R., 213 Hagspiel, K.D., 269 Haider, C.R., 75, 80 Hand vessels, MRA. See Upper extremity and hand vessels, MRA Hand vs. power injection, 69 Hays, A., 129 Head and neck neoplasms, 236–237 Herborn, C.U., 381 Heverhagen, J.T., 271 Hinshaw, W.S., 39 Hodnett, P., 351 Hoffmann, B.A., 396 Hong, S.N., 337 Hu, J., 372 Hunt and Hess classification, intracranial aneurysm, 218 Hu, P., 337 Hurst, D.R., 259 Hypothenar hammer syndrome, 308–309

I Ichinose, N., 89 Inferior mesenteric artery, anatomy, 273–274 Intracerebral hemorrhage, 221–222 Intracerebral vasculitis, 215 Intracranial arterial and venous disease arterial disease, stenoocclusive disease/stroke aneurysm, 217–219 atherosclerotic disease, 214–215 dissection, 216–217 moyamoya disease, 216 vasculitis, 215–216 cerebral arteriovenous malformations, 219–220 dural AV fistula, 220–221 intracerebral hemorrhage, differential diagnosis, 221–222 MRA techniques clinical recommendations, 214 contrast-enhanced, 213–214 phase-contrast, 213 time-of-flight, 213

Index venous disease thrombosis, 219 Intracranial carotid atherosclerosis, 233–234 Intramural hematoma (IMH), 246 Intravascular ultrasound (IVUS), 199–200 Invasive angiography, cardiac catheterization, 199 Iron-based blood pool agents, 355 Ito, K. 100

J Jaarsveld, B.C., 291 Jagadeesan, B.D., 365 Jaspers, K., 376 Johnson, C.P., 75, 80, 87 Joseph, P.M., 175 Jung, B., 51

K Kanazawa, H., 89, 98, 99 Kassai, Y., 89 Kelly, K., 205 Kim, C.Y., 361 Kim, S.-E., 39 King, K.F., 172 Klippel-Trénaunay syndrome, 358 Kluge, A., 260 Kolandaivelu, A., 395 Koopmans, P.J., 154 Kramer, U., 68, 272 Kreitner, K.F., 261 Krings, T., 371 Krum, H., 292 Kucharczyk, W., 387

L Ladd, M.E., 149 Laissy, J.P., 359 Lancelot, E., 205 Lardo, A.C., 396 Larmor equation, 51 Laub, G., 107 Lederman, R.J., 399 Lee, V.S., 297 Leiner, T., 283 Leriche’s syndrome, 330 Li, D., 141, 143 Li, F., 121 Lim, R.P., 297 Linear Eddy current correction, 176–177 Li, R., 113 Liu, J., 179 Li, W., 315, 356 Lohan, D.G., 108 Low-dose contrast-enhanced MR angiography abdomen–pelvis, 110 chest, 110 contrast agent type, 108–109 contrast timing and injection protocol, 109 conventional CE-MRA, 107–108 fast imaging tools, 108 gadolinium association, 107 head and neck, 109–110 lower extremity runoff, 110–111

Index Lower extremity MRA (LE-MRA) arterial spin labeling (ASL), 326 balanced steady-state free precession (bSSFP), 326 clinical applications, 329–330 contrast agents, 323–324 cystic adventitial disease, 331–332 ECG-gated 3D partial Fourier FSE, 326–327 ECG-gated flow-sensitive dephasing (FSD) bSSFP, 327 gadolinium-enhanced MRA adaptations, 322–323 dosing, 323 high-resolution, 320 k-space, 321 parallel imaging technologies, 321 spatial resolution, 320 three-dimensional acquisitions, 320 time-resolved, 321 imaging processing, 328–329 noncontrast techniques, 324 phase-contrast angiography, 325–326 plaque imaging, 328 popliteal entrapment, 330–331, 335 protocols, 334 quiescent interval single-shot (QISS) MRA, 327–328 renal insufficiency, 335 thromboangiitis obliterans/Buerger’s disease, 332–333 time-of-flight angiography, 324–325 3 T scanners, 328 Loy, D.N., 365 Lumenal stenosis, 199 Lustig, J., 181

M Macromolecular contrast agents, 383 Magnetic resonance angiography (MRA) bSSFP, 34–36 CE-MRA, 21 contrast materials, 21–22 data acquisition order, 24–25 imaging sequence, 22–24, 26–27 mask subtraction, 25 multistation exams, 26 phase-contrast, 29–32 processing methods, 32–33 scan synchronization, 24 three-dimensional TOF MRA, 28–29 time-of-flight MRA, 26 two-dimensional TOF MRA, 27–28 velocity encoding and velocity aliasing, 34 Magnetic resonance imaging (MRI) atrial fibrillation ablations device tracking, 396 electrophysiology procedures, 396 lesion imaging, 396 real-time imaging, 396 safety/devices, 396–397 Bloch equation, 10 bulk magnetization, 5–6 cellular therapy cell tracking, 397–399 device tracking, 399 contrast mechanisms T1 relaxation time, 8–9 T2 relaxation time, 6–8 T1 vs. T2 relaxation rates, 9–10

407 description, 3 frequency-encoding, 14–16 imaging parameters, 10–11 mechanism, 5 MR data ( k-space), 17–20 MR image formation, 11–13 parallel imaging, 21 phase-encoding, 16–17 signal-to-noise ratio (SNR), 20–21 slice selection, 13–14 types, 3, 4 Manning, W.J., 337 Markl, M., 51 Maxwell terms, 60–61 May-Thurner Syndrome, 359 McAteer, M., 206 Meade, T. J., 199 Meaney, J.F.M., 253 Meckel, S., 373 Median arcuate ligament syndrome, 279 Mesenteric aneurysms, 279 Mesenteric arterial system, 273 Mesenteric arterial system, MRA acute mesenteric arterial thrombosis, 277 acute mesenteric ischemia, 275–276 acute SMA embolism, 276 aortic dissection, 277 atherosclerotic CMI, 277–279 chronic mesenteric ischemia, 277 mesenteric venous thrombosis, 277 nonatherosclerotic vascular pathogies, 279 nonocclusive mesenteric ischemia, 277 Mesenteric ischemia, 275–277 Mesenteric venous thrombosis, 277 Michaely, H., 283 Micron-sized particles of iron oxide (MPIO), 200, 201 Mittal, S., 164 Miyazaki, M., 89, 92, 316 Moghari, M.H., 337 Mohajer, 321 Moody, A.R., 261 Morawski, A., 203 Mostardi, P.M., 75, 78, 82 Motion artifact reduction, 119–120 Moyamoya disease MRA technique, clinical uses, 234–235 MRA techniques, 216 MR angiography and high field strength benefits, 151–153 contrast-enhanced MRA, 7 T, 153–154 high-field MR, 154–155 7 T non-neuro MRA techniques, 155 MR fluoroscopy, 70–71 MRV techniques contrast agents, 355 contrast-enhanced MRV, 353–354 3D fast spin echo, 352–353 MR-directed thrombus imaging, 353 noncontrast-enhanced MRV, 351–352 STARFIRE, 352 steady-state free precession, 352 susceptibility-weighted imaging, 353 time of flight, 352 time-resolved MRV, 354 Mukherjee, S., 225 Mulder, W.J., 202

408 Mull, M., 376 Multicontrast techniques, 116–117 Multiple overlapping thin 3D slab acquisition (MOTSA) technique, 39, 89, 227

N Nael, K., 107 NASCET method carotid stenosis measurement, 226 Nayler, G.L., 51 Nazarian, S., 396 Neff, W., 375 Nephrogenic systemic fibrosis (NSF), CE-MRA clinical findings, 387 diagnosis, 388 gadolinium-based contrast agents chelate, solution stability of, 389 chemical properties and kinetic stability, 390 dosing limit, 391 history, 389 inherent toxicity, 389 nomenclature, 390 putative mechanisms, 389 thermodynamic/chemical stability, 390 prevention measurement, 392–393 renal dysfunction, 391 risk factors, 391–392 treatment, 388 Nezafat, R., 337 Nishimura, D.G., 39 Nonatherosclerotic vascular pathogies, 279 Non-cartesian MR angiography asymmetric FOVs, 171 field of view, 170 flow sensitivity, 171 gradient spoiling, 170 off resonance, 170–171 parallel imaging, angiography, 191–192 projection imaging 2D, 172–173 3D radial, 173–175 undersampled, 173 reconstruction filtered back projection (FBP), 175 gridding, 175–176 image degradation, k-space sampling errors, 176 linear Eddy current correction, 176–177 off-axis imaging, 177–178 steady-state free precession (SSFP), 178–179 sampling density, 171 sampling region, 170 spiral trajectory design, 171–172 SSFP Imaging, 101 b temporal processing, 180–181 time-resolved MRA quantitative velocity imaging, 180 temporal processing, 179–180 trajectory design, 170 Noncontrast coronary artery imaging arterial spin labeling, 135–136 CAD visualization anomalous coronary arteries and coronary aneurysms, 133 coronary artery bypass graft assessment, 133 coronary flow imaging, 134 coronary stenosis identification, 132–133

Index coronary vessel wall imaging, 134 cardiac motion suppression, 129–130 high field imaging, 136–137 parallel imaging, 136 radial and spiral imaging techniques, 135 respiratory motion compensation breath-hold technique, 130 contrast enhancement, 131 free-breathing technique, 130–131 whole-heart technique, 131–132 steady-state free-precession, 136 Noncontrast-enhanced MRA technique, 286–287 Non-contrast MRA (NC-MRA). See also Flow-dependent noncontrast MR angiography abdominal aorta and mesenteric vessels, 270 pediatric MRA, 366 thoracic aorta, 244–245 Nonocclusive mesenteric ischemia, 277 Non-protein-binding contrast agents, 381–382 Norton, P.T., 269 Nutcracker syndrome, 358–359

O Ohno, Y., 257, 259 O’ Keefe, 258 Ono, A., 97, 358 Oudkerk, M., 257

P Paget-Schroetter syndrome, 314 Pancreatic transplantation, 280 Pandey, T., 358 Papadias, A., 368 Parallel imaging, angiography Cartesian method, 187–188 conjugate gradient SENSE, 192–193 3D, 191 dynamic methods, 195 GRAPPA, 189–191 non-Cartesian, 191–192 PILS, 188 radial GRAPPA, 193–195 remarks, 187 sensitivity encoding, 188–189 SNR losses, 187 Parallel imaging with localized sensitivities (PILS), 188 Paramagnetic contrast agents, 381 Parker, D.L., 39 Parkes Weber syndrome, 358 PC angiography, 57–58 Pediatric MRA acute ischemic stroke moyamoya disease, 374–375 transient cerebral vasculopathy, 373–374 vascular dissection and stroke, 375 vasculitis, 375 arteriovenous malformations, 367–369 Galen vein aneurysmal dilatation, 371 aneurysmal malformation, 370–371 malformation, 369 MRA role, malformation, 371–373 noncontrast MRA, 366 phase contrast MRA, 366–367

Index principles, 365 spinal vascular malformations, 375–377 time-of-flight MRA, 366 time-resolved contrast-enhanced MRA, 367 Pelvic congestion syndrome (PCS), 358 Penetrating atherosclerotic ulcer (PAU), 247 Perfusion-weighted imaging (PWI) atherosclerotic disease, 214 Peters, D.C., 173 Phase contrast MRA (PC-MRA) abdominal aorta and mesenteric vessels, 269 intracranial arterial and venous disease, 213 lower extremity MRA, 325–326 pediatric MRA, 366–367 renal vascular diseases, 288 thoracic aorta, 244 upper extremity and hand vessels, MRA, 301 Phase-contrast MRI (PC-MRI) 2D CINE, applications, 56–57 flow-sensitive 4D MRI, 58–60 implementation and clinical protocols, 55–56 PC angiography, 57–58 principle bipolar gradient, 53 image reconstruction, 52 Larmor equation, 51 signal-to-noise ratio (SNR), 53 temporal footprint (TE), 52 velocity nois, 54 pulse sequences, 54–55 sources of errors acceleration errors, 63 Eddy currents, 61–62 gradient field nonlinearities, 61 Maxwell terms, 60–61 PIOPED III study, 258–259 Pipe, J.G., 175, 176 Plaque imaging, 121–122 Prince, M.R., 65, 351 Protein-binding contrast agents, 383 Pruessmann, K.P., 193 Pseudostenosis, 305 Pulmonary arteriovenous malformation (AVM), 263 Pulmonary artery sling, 263 Pulmonary embolism (PE) animal studies, MRA, 259 applied pulmonary artery anatomy, 253 blood-pool imaging, 260 catheter angiography, 254–255 CE-MRA and catheter angiography, 257 clinical considerations, 253–254 cross-sectional imaging, 255–256 direct thrombus imaging, 260–261 MRA, 256–257 vs. CTA, 257–258, 261 non-contrast MRA, 260 perfusion imaging, 259 PIOPED III study, 258–259 time-resolved MRA, 259 Pulmonary hypertension, 261 Pulmonary sequestration, 263 Pulmonary vascular imaging anomalous venous drainage, 265 atrial fibrillation, 264–265 bronchial arteries, 262 malignancies imaging, 262–263

409 pulmonary arteriovenous malformation (AVM), 263 pulmonary artery sling, 263 pulmonary embolism animal studies, MRA, 259 applied pulmonary artery anatomy, 253 blood-pool imaging, 260 catheter angiography, 254–255 CE-MRA and catheter angiography, 257 clinical considerations, 253–254 cross-sectional imaging, 255–256 direct thrombus imaging, 260–261 MRA, 256–257 MRA vs. CTA, 257–258, 261 non-contrast MRA, 260 perfusion imaging, 259 PIOPED III study, 258–259 time-resolved MRA, 259 pulmonary hypertension, 261 pulmonary sequestration, 263 pulmonary vein, 263–264 Pulmonary vein, 263–264

Q Qanadli, S.D., 254 Quick, H.H., 149 Quiescent interval single-shot MRA, 102–104

R Radial and spiral imaging techniques, 135 Rauscher, A, 154 Raynaud’s phenomenon, 309–310 Renal vascular diseases anatomical considerations, 283–284 ASL techniques, 288 ASTRAL trial, 291 dynamic contrast-enhanced MRI, 288–290 functional renal imaging, 290–291 gadolinium-based contrast media, 287 MR imaging, 283 noncontrast-enhanced MRA technique, 286–287 phase-contrast MRA, 288 renal artery stenosis functional significance assessment, 287–288 MR angiography, 284–286 Renal vein thrombosis, 359–360 Respiratory motion compensation breath-hold technique, 130 contrast enhancement, 131 free-breathing technique, 130–131 whole-heart technique, 131–132 Riederer, S.J., 75, 81 Ritt, M., 290 Rollins, N., 373 Ruehm, S., 201 Ruehm, S.G., 359

S Saam, T., 124 Saeed, M., 399 Saleh, R., 107 Scanlon, T., 319, 351 Seiberlich, N., 185 Seo, J.B., 259

410 Sheehan, J.J., 316 Shimizu, K., 173 Sickle cell disease, 235 Signal targeting alternative radiofrequency and flow-independent relaxation enhancement (STARFIRE), 352 Signal-to-noise (SNR), 341–343 Singh, N., 122 Single-shot two-dimensional SSFP, 240 Slice-selective off-resonance sinc (SORS) pulse, 90 SMA embolism, 276 SNR optimization, 118 Sodickson, D.K., 185 Soulez, G., 285 SPGRE pulse sequence, 41 Spuentrup, E., 287, 356 SSFP Imaging b noncontrast MRA, 101 subtractive methods, 101–102 Stafford, R.B., 328 Steady-state free precession abdominal aorta and mesenteric vessels, 269–270 thoracic aorta, 239–240 Steffens, J., 325 Steinman, D.A., 125 Steno-occlusive disease, 305–306 Stenoocclusive disease/stroke. See Arterial disease, MRA techniques Stuber, M., 129 Sturge-Weber (SWS) syndrome, 371–372 Subclavian steal syndrome, 233 Subclavian steal-syndrome, 215 Sugiura, S., 89 Superior mesenteric artery, anatomy, 273 Superparamagnetic contrast agents, 384 Superparamagnetic iron oxide nanoparticles (SPIO), 201 Suryan, G., 39 Susceptibility weighted imaging (SWI) blood properties, different field strengths, 162 caffeine and acetazolmide role, 163–164 cerebral microbleeds, 164–165 concepts, 157–159 flow compensated gradient echo sequence, 159–160 mapping and oxygen saturation measurement, 159 single echo approach, 160–162 T1 and T2* role, 162–163 Swoboda, N.A., 365 Symons, S.P., 387 Systemic-bladder drainage (SBD), 280

T Takayasu’s arteritis, 249–250 Targeted contrast agents, molecular imaging adhesion molecules, 205 cytokines and integrins, 205 integrins and angiogenesis, 207–208 lipid-rich plaque regions and lipoproteins Gd(III) chelates, 202 lipid-targeting method, 202 micelles, 202, 203 oxidized low-density lipoprotein (OxLDL), 202 macrophages, 201–202 matrix metalloproteinases, 205 MRI contrast agent classes iron oxide nanoparticles, 200 lipid-based nanostructure, 201 myeloperoxidase, 208–209

Index P-selectin and vascular cell adhesion molecule, 205–207 thrombus, plaque-associated activated platelets, 204 acute and subacute, 203 EP-1873, 203 fibrin, 203–204 human carotid plaques, 204 LIBS-MPIO, 204 Test bolus technique, 70 Thoracic aorta aneurysms, 247–248 aortic dissection, 245–246 computed tomography, 239 congenital abnormalities, 248–249 contrast-enhanced MRA balanced steady-state free precession, 245 conventional CE-MRA, 242–243 3D FSE, 245 non-contrast MRA, 244–245 phase contrast MRA, 244 time-resolved MRA, 241–242 giant cell arteritis, 250 inflammatory conditions, 250 intramural hematoma, 246 MRI techniques black-blood techniques, 241 cine SSFP, 240 contrast-enhanced MRA, 241–245 single-shot two-dimensional SSFP, 240 steady-state free precession, 239–240 three-dimensional SSFP, 240–241 T1-weighted gradient echo fat-saturated imaging, 241 penetrating atherosclerotic ulcer, 247 pseudoaneurysms, 247 stenoses, 248 Takayasu’s arteritis, 249–250 transesophageal echocardiography, 239 vasculitis, 249 Thoracic aortic aneurysm (TAA), 247–248 Thoracic outlet syndrome, 311–313 Three-dimensional SSFP, 240–241 Thromboangiitis obliterans/Buerger’s disease, 332–333 Tilted optimized non-saturating excitation (TONE) pulses, 46, 89 Time-of-flight (TOF) angiography 2D TOF, 44–45 3D TOF, 45–47 high magnetic fields, 49 k-space sampling strategies, 47 lower extremity MRA, 324–325 magnetization transfer, 47–48 MR signal flow sensitivity, 39 goal, 40 maximum intensity projection (MIP), 39 MOTSA acquisitions, 39 spoiled gradient echo (SPGRE) sequence, 39 multiple overlapping thin 3D slab acquisition, 48–49 phase dispersion and flow compensation 3D SPGR pulse sequence, 42, 43 first-order gradient moment nulling, 42 phase dispersion, 42 pulse sequence, 43 second-order gradient motion rephasing, 44 signal intensity, 42 velocity-dependent phase dispersion, 43

Index quantification arterial velocity, 42 Bloch equations, 40 flow enhancement effect (FRE), 41, 42 maximum transverse magnetization, 41 RF pulses, 40 speed definition, 40 SPGRE pulse sequence, 41 spoiled gradient sequence, 41 velocity, 41 Time-of-flight MR-angiography (TOF-MRA) abdominal aorta and mesenteric vessels, 269 carotid and vertebral circulation, clinical applications application, 226–227 efficacy, 228 limitations, 227–228 intracranial arterial and venous disease, 213 pediatric MRA, 366 upper extremity and hand vessels, MRA, 301 Time-resolved contrast-enhanced MRA, 367 Time-resolved 3D CE-MRA. See Cartesian methods, CE-MRA Time-resolved MRA abdominal aorta and mesenteric vessels, 272 pulmonary embolism, 259 thoracic aorta, 241–242 Transesophageal echocardiography, 239 Trattnig, S., 154 T1-weighted gradient-echo fat-saturated imaging abdominal aorta and mesenteric vessels, 272 thoracic aorta, 241

U Ultrasmall iron oxide nanoparticle (USPIO), 200 Underhill, H, 123 Upper extremity and hand vessels, MRA bright blood localizer, 299 clinical indications arteriovenous fistulas, 309 embolic disease, 306–307 hypothenar hammer syndrome, 308–309 Paget-Schroetter syndrome, 314 Raynaud’s phenomenon, 309–310 steno-occlusive disease, 305–306 thoracic outlet syndrome, 311–313 trauma, 307–308 vascular malformations, 311 vasculitis, 310–311 venous thrombosis, 313–314 contrast agents, 315 2D fast spin echo imaging, 300–301 extraluminal pathology, 305 flow-related artifacts, 305 gadolinium-enhanced MRA, 301 acquisition timing, 301–302 contrast dosage, 301 sequence parameters, 302 time-resolved, 302 gadolinium-enhanced MRV, 303, 304 high-field imaging, 315 image acceleration, 315 inaccurate timing, 303, 305 motion artifact, 305 noncontrast-enhanced techniques, 315–316 phase-contrast and time-of-flight MRA, 301 pseudostenosis, 305

411 scanning coil selection, 299 patient positioning, 299 vascular anatomy arterial anatomy, 297–298 venous anatomy, 298–299 vascular mimics, 305 Urata, J., 95 Urbach, H., 213

V Vasbinder, G.B., 285 Vasculitis, 215–216, 249, 310–311 Vector electrocardiogram, 145 Venkataraman, S., 275 Venous disease MRA techniques, 219 Venous imaging clinical applications DVT assessment, 356–358 Klippel-Trénaunay syndrome, 358 May-Thurner Syndrome, 359 nutcracker syndrome, 358–359 Parkes Weber syndrome, 358 pelvic congestion syndrome (PCS), 358 portal vein occlusion, 360 pulmonary vein evaluation, 361–362 renal vein thrombosis, 359–360 upper extremity and central vein evaluation, 360–361 MRV techniques contrast agents, 355 contrast-enhanced MRV, 353–354 3D fast spin echo, 353 MR-directed thrombus imaging, 353 noncontrast-enhanced MRV, 351–352 STARFIRE, 352 steady-state free precession, 352 susceptibility-weighted imaging, 353 time of flight, 352 time-resolved MRV, 354 Venous thrombosis, 313–314 Vertebral artery atherosclerosis, 229–230 Vessel wall imaging techniques applications natural history studies, 122–123 vessel morphology, 120–121 vulnerable plaque imaging, 121–122 black-blood techniques, 113 blood suppression, spin echo, 114 clinical trials, 124 contrast weightings contrast enhancement techniques, 117–118 multicontrast techniques, 116–117 coronary vessel wall imaging, 124 diffusion preparation, 116 double inversion recovery (DIR), 114–116 flow and relaxation properties, blood, 113–114 hemodynamic study, 125 higher resolution, 125 molecular imaging, 124–125 practical considerations fat suppression, 119 field strength, 118–119 image processing, 120 localization, 120

412 Vessel wall imaging techniques (continued) motion artifact reduction, 119–120 SNR optimization, 118 saturation band, 114 Von zur Muhlen, C., 204 Vulnerable plaque, 199

W Wang, J., 315 Ward, E., 239 Waters, E.A., 199 Weiss, C.R., 395 Weiss, R.G., 129 Whole-body MRA, 72 Whole-heart coronary MRA, 3T clinical applications, 145–147 navigator efficiency, 145 patient training, 144–145 SSFP, 143 survey scanning, 145 vector electrocardiogram, 145

Index Widjaja, W., 373 Wieben, O., 169 Willinek, W. A., 213 Willoteaux, S., 285 Wilson, G.J., 287 Windkessel effect, 56 Wintermark, M., 225 Winter, P., 207 Wittram, C., 254 Wong, H.I., 359

Y Yamada, I., 235 Yang, Q., 141 Yoon, H.K., 374 Yuan, C., 113, 121

Z Zhang, H., 65 Zhang, W., 65 Ziyeh, S., 373