## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Ratings:

1,040 pages18 hours

Magnetic Resonance Imaging (MRI) is among the most important medical imaging techniques available today. There is an installed base of approximately 15,000 MRI scanners worldwide. Each of these scanners is capable of running many different "pulse sequences", which are governed by physics and engineering principles, and implemented by software programs that control the MRI hardware. To utilize an MRI scanner to the fullest extent, a conceptual understanding of its pulse sequences is crucial. *Handbook of MRI Pulse Sequences* offers a complete guide that can help the scientists, engineers, clinicians, and technologists in the field of MRI understand and better employ their scanner.
Explains pulse sequences, their components, and the associated image reconstruction methods commonly used in MRI
Provides self-contained sections for individual techniques
Can be used as a quick reference guide or as a resource for deeper study
Includes both non-mathematical and mathematical descriptions
Contains numerous figures, tables, references, and worked example problems

Publisher: Academic PressReleased: Sep 21, 2004ISBN: 9780080533124Format: book

**Cover image **

**Title page **

**Copyright **

**Dedication **

**FOREWORD **

**FOREWORD **

**PREFACE **

**PART I: BACKGROUND **

**INTRODUCTION TO BACKGROUND **

**Chapter 1: TOOLS **

**PART II: RADIOFREQUENCY PULSES **

**INTRODUCTION TO RADIOFREQUENCY PULSES **

**Chapter 2: RADIOFREQUENCY PULSE SHAPES **

**Chapter 3: BASIC RADIOFREQUENCY PULSE FUNCTIONS **

**Chapter 4: SPECTRAL RADIOFREQUENCY PULSES **

**Chapter 5: SPATIAL RADIOFREQUENCY PULSES **

**Chapter 6: ADIABATIC RADIOFREQUENCY PULSES **

**PART III: GRADIENTS **

**INTRODUCTION TO GRADIENTS **

**Chapter 7: GRADIENT LOBE SHAPES **

**Chapter 8: IMAGING GRADIENTS **

**Chapter 9: MOTION-SENSITIZING GRADIENTS **

**Chapter 10: CORRECTION GRADIENTS **

**PART IV: DATA ACQUISITION, K-SPACE SAMPLING, AND IMAGE RECONSTRUCTION **

**INTRODUCTION TO DATA ACQUISITION, K-SPACE SAMPLING, AND IMAGE RECONSTRUCTION **

**Chapter 11: SIGNAL ACQUISITION AND K-SPACE SAMPLING **

**Chapter 12: BASICS OF PHYSIOLOGIC GATING, TRIGGERING, AND MONITORING **

**Chapter 13: COMMON IMAGE RECONSTRUCTION TECHNIQUES **

**PART V: PULSE SEQUENCES **

**INTRODUCTION TO PULSE SEQUENCES **

**Chapter 14: BASIC PULSE SEQUENCES **

**Chapter 15: ANGIOGRAPHIC PULSE SEQUENCES **

**Chapter 16: ECHO TRAIN PULSE SEQUENCES **

**Chapter 17: ADVANCED PULSE SEQUENCE TECHNIQUES **

**APPENDIX I: TABLE OF SYMBOLS **

**APPENDIX II: TABLE OF CONSTANTS AND CONVERSION FACTORS **

**APPENDIX III: COMMON ABBREVIATIONS **

**Index **

Nothing in this book implies endorsement by Mayo Foundation or University of Illinois of any medical equipment manufacturer.

Elsevier Academic Press

200 Wheeler Road, 6th Floor, Burlington, MA 01803, USA

525 B Street, Suite 1900, San Diego, California 92101-4495, USA

84 Theobald’s Road, London WCIX 8RR, UK

This book is printed on acid-flee paper. ∞

Copyright © 2004, Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher.

Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone: (+44) 1865 843830, fax: (+44) 1865 853333, e-mail: **permissions@elsevier.com.uk. You may also complete your request on-line via the Elsevier homepage (http://elsevier.com), by selecting Customer Support and then Obtaining Permissions. **

**Library of Congress Cataloging-in-Publication Data **

Application submitted

**British Library Cataloguing in Publication Data **

A catalogue record for this book is available from the British Library

ISBN-13: 978-0-12-092861-3

ISBN-10: 0-12-092861-2

For all information on all Elsevier Academic Press publications visit our Web site at **www.books.elsevier.com **

Printed in the United States of America

05 06 07 08 09 9 8 7 6 5 4 3 2

*To Rhoda, juliet, Sara, and Lee *

M.A.B.

*To Sherry and our son Christopher *

X.J.Z.

*To Beth *

K.F.K.

Magnetic resonance imaging is recognized as one of the most important medical advances of the century. It has opened new windows into the human body, revealing structure and function with a level of detail that would have been unimaginable only decades ago.

The award of the Nobel Prize in Medicine to Paul Lauterbur and Peter Mansfield is a wonderful recognition for their important contributions to the development of MRI. The award of the Nobel Prize in Medicine for work that is rooted in Physics and Engineering is also an important validation of these disciplines as important participants within the medical research community.

While the pioneering work of Lauterbur and Mansfield was critical to the early development of MRI, the amazingly wide range of its capabilities have resulted from the efforts of the many scientists and clinicians. Unlike other high-tech imaging modalities, a large proportion of the techniques currently used in state-of-the-art practice were *not *initially introduced by industry. Rather, these advances were made by investigators and physician specialists in academic centers and clinical imaging practices. In a very real way, MRImone of the most important tools of modem medical practice–is a technology that was *invented by its users! *

The *Handbook of MRI Pulse Sequences *is a comprehensive and highly readable reference targeted squarely at those who want to understand the current state-of-the-art, as well as those who are committed to continuing the process of innovation. The authors have segmented the vast subject of MRI technology into five logical sections: Background Concepts, Radiofrequency Pulses, Gradients, Data Acquisition, and Pulse Sequences. These sections contain carefully planned groups of chapters that systematically cover the field. The chapters use a modular building block approach that avoids repetition, allowing efficient presentation of a subject. Each of the main sections is selfcontained, allowing the reader to go directly to any section without the need to review previous sections. Consistent nomenclature is used throughout.

Such disciplined and accessible coverage of the field would have been difficult if the book followed the model of chapters written by dozens of authors. Instead the *Handbook of MRI Pulse Sequences *is the product of three highly dedicated authors. Drs. Bernstein, King, and Zhou have all made significant contributions to the MRI field and have collective experience spanning from industry to academia. All three are highly accomplished in teaching the nuts and bolts

of MRI technology.

Most chapters are accessible to readers with varying levels of expertise. The book should be useful to physicists and MRI radiologists alike. The chapters combine rigorous mathematical coverage of concepts to the level warranted by the subject matter, with qualitative descriptions wherever feasible.

This book is unique in the fact that it brings together in one reference many areas of MRI technology that are not well covered in other available references. Such areas include adiabatic fast passage radiofrequency pulses, stimulated echoes, ordered phase encoding, navigator echoes, and many others. The section on Gradients nicely clarifies the generally misunderstood terminology and purpose of the various types of correction gradients such as crushers, spoilers, and twisters.

Many scientists who are active in supervising MRI research by graduate and postdoctoral students have encountered, year after year, a relative paucity of comprehensive reference textbooks that can answer the recurring questions that arise. The *Handbook of MRl Pulse Sequences *will address this need in an outstanding fashion.

This book will become one of the classic texts in the field and will play a key role in helping the next generation of scientists and MRI clinicians to continue the process of invention.

**Richard L. Ehman, M.D. **

*Professor of Radiology, Mayo Clinic *

MRI has now been developing for over 30 years. The first experiments did not use any technique that would ordinarily be called a pulse sequence,

employing CW applications of radiofrequency (RF) fields and static main magnetic fields and gradients, with iterative back projection reconstruction and *T*1 discrimination by changing of RF intensity. It was soon realized, however, that time-dependent RF pulses and magnetic field gradients offered more flexibility in many applications, and the field of pulse sequence

design opened. This book is the best evidence of its subsequent proliferation, both for the sake of novelty and to tailor instrument responses to a variety of applications. An unfortunate side effect has been that such efforts have introduced innumerable acronyms or near-acronyms in an effort to distinguish the various sequences, or fix credit for their designs. This book is the most valiant and successful attempt yet to provide a useful description of this zoo,

and to relate and classify the various denizens in it. Such a taxonomy can never be complete, as the inhabitants keep multiplying and providing new hybrids which must be classified to facilitate discourse among physicists, engineers, physicians, and others, and to help pulse sequence designers avoid inadvertent duplication of efforts. I believe that no MRI developer or user can read this book without learning more about the field, as I have.

But what does this suggest about future editions or successors? New applications will continue to become important, and new opportunities will continue to suggest themselves to creative pulse-sequence designers. Will we soon reach the limit of our ability to absorb, retain, and understand such innovations, which may include aspects of high-resolution solid-state spectroscopy and quantum computing? There is some evidence that we already have, as much MRI seems even now driven by habit, custom, and instrumental constraints rather than by medical and other needs. In the future, the functions that this book serves may be accomplished by interactive database programs that understand

MRI physics, so that properly defined needs can be matched to the best sequences, and new sequences consistent with machine capabilities can be computer-developed as needed. I therefore look forward to the second edition of this book as a CD or equivalent, combined with a specialized search-engine. In the meantime, the first edition will well-repay careful study.

**Paul C. Lauterbur, Ph.D. **

*Center for Advanced Study, Professor of Chemistry, Biophysics and, Computational Biology, and in the Bioengineering Program, Research Professor of Medical Information Sciences, College of Medicine at Urbana-Champaign, and Distinguished Professor, College of Medicine, University of Illinois at Chicago *

Since its invention in the early 1970s, the development of magnetic resonance imaging (MRI) has been among the most active and exciting areas in science, technology, and medicine. Today, thousands of MRI scanners are operating throughout the world, providing crucial information in such diverse areas as material science, pharmaceutical development, and especially medical diagnosis.

MRI encompasses many areas of science and technology, including spin physics, biophysics, image reconstruction, and hardware design. Although each area performs a necessary function, it is the MRI pulse sequence that links many of the individual functions together and plays a pivotal role in data acquisition. Over the past 30 years, development of pulse sequences has been an extremely active area of MRI research. Many important concepts of MRI are realized through pulse sequence design, and virtually every imaging application is enabled by one or more pulse sequences. This book aims to help the reader gain a comprehensive understanding of many pulse sequences (and their associated techniques) in use today. The book is also intended to assist the reader to continue the innovation in pulse sequence design, and to carry MRI development into the future.

While this handbook focuses on MRI pulse sequences, it also covers the building blocks from which they are composed, as well as the related image reconstruction techniques and important mathematical tools that are helpful for understanding them. We have divided the book into five parts. Each part of the book is preceded by its own introduction, which provides an overview of its contents. Part I covers two basic mathematical tools: Fourier transforms and the rotating reference frame. Parts II through IV describe a variety of the basic building blocks and associated techniques of MRI pulse sequences: radiofrequency (RF) pulses, gradient waveforms, data acquisition strategies, and common image reconstruction methods. Part V describes how these components are integrated into some of the common pulse sequences. Selected applications of each pulse sequence are also discussed in Part V.

The five parts of the book are further subdivided into chapters. For example, Part III on Gradients contains chapters describing gradient lobe shapes, imaging gradients, motion-sensitizing gradients, and correction gradients. Each chapter in turn contains multiple sections that are usually arranged alphabetically. In all, there is a total of 65 sections. The sections in each chapter are independent, forming their own *self-contained *units. As such, the sections do not have to be read in any particular order. Each section contains an introduction to the subject and some qualitative description. Depending on the subject matter, mathematical analysis, practical implementation details, and/or descriptions of applications may also be included. At the end of each section, selected references are provided, and the related sections are listed.

It is also important to say what this book is *not*. This book is not meant to be an encyclopedia of pulse sequences. We have attempted to cover most of the MRI pulse sequences in common use today, but have made no attempt to describe every pulse sequence that has ever been published. We believe, however, that virtually every MRI pulse sequence that has ever been implemented (or even proposed) is composed, at least in part, of the building blocks described in detail in this handbook. We would also like to emphasize that exclusion of any particular sequence or technique is not meant to imply anything about its relative importance. The scope of this book encompasses MR imaging pulse sequences, but excludes purely spectroscopic pulse sequences, and only briefly touches on spectroscopic imaging. The physical description of spin dynamics (sometimes called spin gymnastics

) is limited to the classical picture, which is sufficient for the vast majority of the discussion throughout the book. Therefore, neither quantum mechanical formalism nor multiple-quantum phenomena are covered.

We make no rigorous attempt to provide an historical record of who invented or first published any particular pulse sequence or technique. Given the activity in the field of MRI, a thorough study of that subject could probably fill a large volume itself. The selected references at the end of each section are only intended to provide the reader with additional material to understand the subject matter, and do not necessarily correspond to the earliest work on that subject. The references are a collection of papers, review articles, conference abstracts, books, book chapters, and patents selected from those we have found instructive during our own study. The reader may also notice that some of the sections in Part V contain many more references than the earlier sections. Perhaps this is an expression that there are many more ways to combine and permute the building blocks, compared to the number of building blocks themselves. Because of the breadth and depth of activity in MRI research, we realize that we have inevitably omitted some important references. We are eager to receive feedback about the book, so feel free to contact us if you would like to suggest additional references that should be included in future editions.

This book is written primarily for scientists and engineers working in the field of MRI. It can also serve as a supplementary textbook for a graduate course on medical imaging, medical physics, or biomedical engineering. Radiologists and other medical professionals interested in an in-depth understanding of various MRI pulse sequences will also benefit from the book. To cover the diverse background of the readership, both non-mathematical and mathematical descriptions are provided. Readers not interested in the mathematical description of a particular topic can skip over those portions of a section that contain complicated equations. It is our aim to provide those readers with sufficient qualitative description in text and figures to convey the gist of each section. The level of mathematical sophistication varies widely from section to section. This is because we have tried to tailor the mathematical level to the specific subject material, rather than to a specific readership. For example the mathematical level used in the sections describing SLR pulses (§ 2.3) gridding (§ 13.2) and parallel imaging reconstruction (§ 13.3) is considerably higher than that describing real-time imaging (§ 11.4). In general, however, the level of mathematical sophistication is at the first-year university level, with use of vectors, basic calculus, and linear algebra. A notable exception is the use of Fourier transforms, which are relied on extensively throughout the book. To help keep the book self-contained, § 1.1 is entirely devoted to the properties of continuous and discrete Fourier transforms.

In writing this book, we have assumed that the reader has already acquired some basic knowledge of MRI physics. Although we provide discussion of a few essential physical concepts, such as the B loch equations, gradients, and rotating reference frame, we encourage the reader to consult any of the excellent introductory MRI physics books, should there be a background concept that is unfamiliar. Similar to the level of mathematics, the level of physics is also tailored to individual sections. For example, w 12.1 on cardiac triggering requires little background on spin physics, while to understand the adiabatic pulses described in **Chapter 6 requires a somewhat deeper background in MRI physics. Many of the physical concepts used to understand adiabatic pulses, however, are described in that chapter, or in other sections of the book. **

The idea of writing a handbook on MRI pulse sequences, with selfcontained sections, was initiated by one of the authors (X.J.Z.) in the spring of 1999, but it was only through the effort of all three authors working in concert that the book took its current shape. Between 1999 and 2000, the three authors held numerous meetings and conference calls to define the scope. We focused on designing a book with three primary features, all aimed at helping the reader extract the maximal information with the minimal effort. First, as mentioned earlier, each section is self-contained so the reader can go straight to the material of interest without reading the preceding sections in sequential order. Second, also mentioned earlier, each subject in a section is discussed both qualitatively and quantitatively to serve readers with various backgrounds. Third, special attention is paid to maximizing consistency across all the sections. This frees the reader from having to make the mental adjustment for inconsistent notation, conventions, and terminology usually required when reading a set of original research papers, or an edited book that has many contributing authors. Many of the mathematical symbols, constants, abbreviations, and acronyms used across multiple sections of the book are listed in the appendices.

This handbook was written during a period from 2000 to early 2004. It is the product of three authors who *contributed equally*. The author names in various places are listed alphabetically and the order does not imply relative contribution to the book. Each step in the process was designed to ensure maximal consistency throughout while maintaining the breadth of coverage that three authors can provide. In this electronic era, being based in separate cities did not hamper this effort. A primary author drafted each section, and then sent it to one of the other authors for review. Based on the comments and additions received, the primary author revised the section and sent it to the third author for further review. The process was iterated as many times as necessary (in some cases, seven or eight iterations) until all authors were satisfied with the manuscript. To make the handbook as seamless as possible, the three authors participated in telephone conference calls at regular intervals. As a result, each author made substantial contributions to every section. In spite of our efforts, mistakes undoubtedly still exist. We encourage the reader to contact us, should she/he find any errors, omitted material, or any explanations that are not clear. Correspondence about a particular section can be addressed to the primary section author who is listed in the table of contents, and who will continue to manage the revisions of that section.

In addition to our own review process, we also asked a number of experts to review several sections of the manuscript. We are indebted to these guest reviewers who improved the quality of those sections with their insightful comments. Stephen Riederer, Ph.D. reviewed § 11.4 on real-time imaging, and § 15.3 on time-of-flight and contrast-enhanced MR angiography. James Glockner, M.D., Ph.D. reviewed § 12.1 on cardiac triggering, as did Kiaran McGee, Ph.D. Heidi Ward, Ph.D. reviewed § on navigator echoes, and § 14.1 on gradient echoes. Elisabeth Angelos, Ph.D. reviewed § 13.3 on parallel imaging. Yihong Yang, Ph.D. reviewed § 17.1 on arterial spin tagging. Sarah Patch, Ph.D. reviewed § 17.5 on projection acquisition. We greatly appreciate all of their help. Naturally, any errors remaining in those sections are solely ours.

We also thank the many contributors (Drs. Kimberly Amrami, Walter Block, Reed Busse, Kim Butts, Norbert Campeau, Bruce Daniel, J. Kevin DeMarco, Kevin Glaser, E. Mark Haacke, Romhild M. Hoogeveen, John Huston III, Emanuel Kanal, Chen Lin, Kiaran McGee, Paul McGough, Gary Miller, Koichi Oshio, Scott Reeder, Pr. Regent, Larry Tanenbaum, Heidi Ward, Qing-San Xiang, Yihong Yang, and Frank Q. Ye) who provided images for this handbook. Their contributions are individually acknowledged again in the figure captions.

We would also like to thank Linda Greene for her review of some of the page proofs, David Thomasson, Ph.D. for providing a glossary of MR terms used by Siemens Medical Solutions, and Tim Hiller for providing a list of acronyms used by Philips Medical Systems and other vendors. Our appreciation is also extended to Dr. K. Noelle Gracy, Anne Russum, Marcy Barnes-Henrie, Paul Gottehrer, and others at Elsevier who aided us during various stages of planning and writing the book.

Many other individuals also helped us in various ways to make this book possible. We would like to individually express our whole-hearted gratitude to them:

*MAB: *I am indebted to all my teachers and mentors. In the field of physics, the guidance from Drs. Bill Friedman, Gary Glover, and Norbert Pelc has been especially valued. Drs. John Huston III, Paul McGough, Patrick Turski, and many others have patiently taught me some of the clinical aspects of MRI. Larry Ploetz introduced me to the subtleties of pulse sequence programming. I am also indebted to my colleagues on the clinical MRI Physics team at the Mayo Clinic in Rochester, including Drs. Joel Felmlee, Chen Lin, and Kiaran McGee, as well as Diana Lanners and Renee Jonsgaard. Of course without the patience and support of my wife Rhoda Lichy, my contribution to this book would not have been possible.

*KFK: *Thanks to Paul Moran for introducing me to medical imaging and teaching me about magnetic resonance. Richard Kinsinger gave me career opportunities for which I will always be very grateful. The work of Gary Glover and Norbert Pelc has provided a standard by which I measure all of my own work. Thanks to the many colleagues at GE with whom I have had the privilege and pleasure to work. You have instructed, challenged, and inspired me. Special thanks to Carl Crawford, Alex Ganin, and Lisa Angelos. Without your collaboration, the work would not have been nearly as much fun.

*XJZ: *I would like to thank Professor Paul C. Lauterbur who introduced me to the field of MRI, shared his remarkable insight into MRI physics, and encouraged the idea of writing this book. I would also like to thank Drs. G. Allen Johnson, Keith R. Thulborn, and Richard E. Kinsinger who guided me, in various stages of my career, to acquire knowledge of pulse sequence design. My gratitude is owed to Drs. Norman E. Leeds, Edward F. Jackson, Haesun Choi, and Srikanth Mahankali for teaching me a great deal on the clinical aspects of MRI. Many colleagues, friends, and students also helped me in developing the materials that are included in the book. In particular, I am indebted to Dr. Fernando E. Boada, Gary P. Cofer, Dr. John D. Hazle, Dr. Christof Karmonik, Dr. Zhi-Pei Liang, Dr. James R. MacFall, Joseph K. Maier, Dr. Graeme C. McKinnon, Dr. Bryan J. Mock, Dr. Douglas C. Noll, Aziz H. Poonawalla, H. Glenn Reynolds, Dr. Gary X. Shen, Dr. R. Jason Stafford, Dr. S. Lalith Talagala, Dr. Qing-San Xiang, and Dr. Yihong Yang. Finally, my appreciation goes to Sherry Xia Yao who spent countless evenings and weekends taking care of our infant son during the period when the book was written.

The development of MRI pulse sequences over the last 30 years has been truly exciting. The process of writing this book has only enhanced our appreciation of the dedicated scientists, engineers, and clinicians who have advanced MRI into the essential tool that it has become. We hope you will enjoy this book, and find it to be a useful reference for continuing the advancement of this great field.

Matt A. Bernstein

Rochester, Minnesota

**mbernstein@mayo.edu **

Kevin E King

Waukesha, Wisconsin

kevin, f. **king@med.ge.com **

Xiaohong Joe Zhou

Chicago, Illinois

**xjzhou@uic.edu **

May 2004

**PART I **

**BACKGROUND **

Part I of this book contains selected background information that is used in many sections throughout the book. **Chapter 1 (the only chapter in Part I) describes two mathematical tools that are universally used to analyze magnetic resonance imaging (MRI): the Fourier transform (Section 1.1), and the rotating reference frame (Section 1.2). Although neither of these tools is a pulse sequence (or pulse sequence element) itself, they are both described in considerable detail to establish the notation and conventions that are used throughout the book. A more extensive discussion of Fourier transform can be found in Bracewell (1978). A more thorough introduction to the rotating reference frame and other magnetic resonance (MR) physics concepts is contained in many references, for example, Slichter (1989). **

Other mathematical tools (e.g., rotation matrices) used in specific applications are described in their own individual sections (e.g., **Section 2.3), as needed. For additional general mathematical tools used in this book, the reader is encouraged to consult a mathematics handbook such as Arfken and Weber (2001). **

Arfken G.B., Weber H.J. Mathematical methods for physicists. 5th ed. 2001.

Bracewell R.N. *The Fourier transform and its applications*. New York: McGraw-Hill; 1978.

Slichter C.P. *The principles of magnetic resonance*, 3rd ed. Berlin: Springer-Verlag; 1989.

The *Fourier transform *(FT) is a mathematical operation that yields the spectral content of a signal (**Bracewell 1978). It is named after the French mathematician Jean Baptiste Joseph Fourier (1768–1830). If a signal consists of oscillation at a single frequency (e.g., 163 Hz), then its FT will contain a peak at that frequency (Figure 1.1a). If the signal contains a superposition of tones at multiple frequencies, the FT operation essentially provides a histogram of that spectral content (Figure 1.1b). For example, consider the following physical analogy. Suppose several keys on a piano are struck simultaneously and the resultant sounds are sampled and digitized. The FT of that signal will provide information about which keys were struck and with what force. **

**FIGURE 1.1 **Schematic representations of the Fourier transform. (a) If a time domain signal contains a tone at a single frequency, its Fourier transform will contain a peak at that frequency, which in this case is 163 Hz. (b) If the signal contains a superposition of two tones, the Fourier transform displays a second peak. In the case shown, a 15-Hz tone with approximately one-quarter the amplitude is modulating the original tone.

Fourier transforms are ubiquitous in the practical reconstruction of MR data and also in the theoretical analysis of MR processes. This is because the physical evolution of the transverse magnetization is described very naturally by the FT. In Magnetic Resonance Imaging (MRI), we usually use *complex *Fourier transforms, which employ the complex exponential, rather than separate sine or cosine Fourier transforms. This choice is made because a complex exponential conveniently represents the precession of the magnetization vector. **Table 1.1 reviews some basic properties of the complex exponential. Often a magnitude operation (i.e., |Z|) is used on a pixel-by-pixel basis to convert the complex output of the FT to positive real numbers that can be more conveniently displayed as pixel intensities. **

**TABLE 1.1 **Properties of Complex Numbers

When we are provided with a function of a continuous variable, its FT is calculated by a process that includes integration. This continuous FT is widely used for theoretical work in MRI. The actual MRI signal that is measured, however, is sampled at a finite number of discrete time points, so instead a *discrete Fourier transform *(DFT) is used for practical image reconstruction. With the DFT, the integration operation of the FT is replaced by a finite summation. An important special case of the DFT is called *the fast Fourier transform *(FFT) (**Cooley and Tukey 1965; Brigham 1988). The FFT is an algorithm that calculates the DFT of signals whose lengths are particular values (most typically equal to a power of 2, e.g., 256 = 2⁸). As its name implies, the FFT is computationally faster than the standard DFT. **

Let *g(x) *be a function of the real variable *x. *The output of the function *g(x) *can have complex values. The complex Fourier transform of *g(x*) is another function, which we call *G(k): *

(1.1)

The two real variables *x *and *k *are known as Fourier *conjugates *and represent a pair of FT *domains. *Examples of domain pairs commonly used in MR are (time, frequency) and (distance, k-space). If the physical units of the pair of variables that represent the two domains are multiplied together, the result is always dimensionless. For example, with the time-frequency pair, the product:

(1.2)

The two functions *g(x*) and *G(k) *in **Eq. (1.1) are called Fourier transform pairs. Knowledge about one of the pair is sufficient to reconstruct the other. If G(k) is known, then g(x) can be recovered by performing an inverse Fourier transform (IFT): **

(1.3)

The IFT undoes the effect of the FT, that is:

(1.4)

and vice versa:

(1.5)

Note that the right sides of **Eqs. (1.4) and (1.5) are simply g(x) and G(k), respectively, and are not multiplied by any scaling factors. This is because the IFT definition in Eq. (1.3) is properly normalized. A further discussion of the normalization is given in subsection 1.1.10. **

Note the factor of *2π *that appears in the argument of the exponentials in **Eqs. (1.1) and (1.3). If instead domain variables such as time and angular frequency (ω, measured in radians/second) are used, then the form of the FT appears somewhat differently. The FT and its inverse become: **

(1.6)

Note the absence of the 2π factor in the exponential in **Eq. (1.6) and the extra multiplicative normalization factor in front of the FT. Equation (1.6) could be recast into a more symmetric form by splitting the 2π factors in the denominators of both the FT and IFT definitions. Alternatively, we can recast Eq. (1.6) by making the familiar substitution from angular frequency ω to standard frequency f (measured in cycles/second or hertz): **

(1.7)

Substituting **Eq. (1.7) into Eq. (1.6) yields the symmetric FT pairs **

(1.8)

and

(1.9)

In this book, we mainly use the form of the FT and IFT with the factor of *2π *in the exponential, such as **Eqs. (1.1) and (1.8). **

Multidimensional FTs often arise in MRI. For example, the *two-dimensional FT *(2D-FT) of a function of two variables can be defined as:

(1.10)

where

are vectors. The inverse 2D-FT is given by:

(1.11)

**Eqs. (1.10) and (1.11) are readily generalized to three or more dimensions. **

If the function *g *is *separable *in *x *and *y*:

(1.12)

then the FT is also separable:

(1.13)

An example of a separable two-dimensional function is the Gaussian:

(1.14)

In contrast,

(1.15)

is not separable.

An important property of the FT is the *shift theorem. *A shift or offset of the coordinate in one domain results in a multiplication of the signal by a *linear phase ramp *in the other domain, and vice versa:

(1.16)

A second useful property of the FT is that *convolution *in one domain is equivalent to simple *multiplication *in the other. If *f(x) *and *g(x) *are two functions, then convolution is defined as:

(1.17)

and

(1.18)

Parseval’s theorem (named after Marc-Antoine Parseval des Chêsnes, 1755–1836, a French mathematician) is a third commonly used property of the FT. It states that if *f *and *g *are two functions with Fourier transforms *F *and *G*, respectively, then

(1.19)

where * denotes complex conjugation. Letting *g *= *f *in **Eq. (1.19) results in a useful special case, which shows that the FT operation conserves normalization: **

(1.20)

**Table 1.2 provides several 1D-FT pairs that are commonly used in MRI. These relationships can be applied to multidimensional FTs if the variables are separable. **

**TABLE 1.2 **Fourier Transform Pairs Commonly Used in Magnetic Resonance Imaging

In MRI, the sampling process provides a finite number (e.g., 256) of complex data points, rather than a function of a continuous variable. Consequently, the MR image is normally reconstructed with a DFT. Given a string of *N *complex data points:

(1.21)

the *J *th element of DFT is defined as:

(1.22)

Note that the index *J *= 0 represents the DC, or zero-frequency element, of the DFT (DC is adopted from the abbreviation for direct current used in electrical engineering). The exponential factor in **Eq. (1.22) is sometimes called a twiddle factor. The Kth element of the inverse DFT (IDFT) is defined as **

(1.23)

The factor of *1/N *in **Eq. (1.23) is required for normalization, so that **

(1.24)

In analogy to the manipulation of the *2π *− normalization factor of the complex FT described for **Eqs. (1.6)-(1.9), the normalization factor of 1/ N factors on both the DFT and IDFT. It is best to check the documentation of the particular numerical routines that you use. The 1/N normalization factor is required (somewhere) for Eq. (1.24) to hold, because: **

(1.25)

Although the DFT is typically evaluated numerically, it does have some useful analytical properties, which are summarized in **Table 1.3. **

**TABLE 1.3 **Discrete Fourier Transform Pairs

In the complex Fourier transform of **Eq. (1.1), it is easy to identify physical units with x (e.g., cm) and k (e.g., cm−1). With the DFT of Eq. (1.22), however, J and K are simply dimensionless integer indices. When we perform a DFT on a signal that has physical meaning, such as MRI data, how do we associate physical units with the string of numbers that is the output of the computer? Consider a one-dimensional MR signal S(k), which can be reconstructed with an IFT: **

(1.26)

In order to see how **Eq. (1.26) relates to the DFT, first approximate the continuous variables x and k with their discrete representations: **

(1.27)

Approximating the integral by a discrete summation and substituting **Eq. (1.27) into Eq. (1.26) yields: **

(1.28)

where the constant *C *is determined by the normalization conditions; that is, it is equivalent to a factor of Δ*k *that converts the integral to a sum. The important point is, comparing the exponentials in **Eqs. (1.23) and (1.28), we conclude that: **

(1.29)

**Equation (1.29) provides the link between the step sizes of the input and output of the DFT operation and the number of complex points in the data string. It is a very useful relationship for MRI. For example, it tells us that the product of the pixel size and the step size in k-space is equal to the inverse of the number of sampled points. Equation (1.29) can be rearranged as: **

(1.30)

which says that the total extent in k-space is equal to the inverse of the spatial pixel size. Similarly, the field of view is the inverse of the step size in k-space.

Suppose 256 complex points are sampled for a total duration of 8.192 ms. An image is reconstructed with a 256-point DFT. Find the bandwidth, and express it in two common forms: bandwidth per pixel Δ*vpp *and the half-bandwidth ±Δν.

Applying **Eq. (1.30) with the time-frequency domain pairs, the total bandwidth is: **

(1.31)

Thus Δν = ±15.625 kHz. The bandwidth in units of hertz per pixel is:

Like the continuous FT, the DFT obeys a shift theorem as well. Similar to **Eq. (1.16), multiplying the data by a linearly increasing phase ramp results in the shift: **

(1.32)

A common case is called the *half field of view *or *Nyquist shift *(named after Harry Nyquist, 1889–1976, a Swedish-born American engineer). Setting *a *= *N*/2, **Eq. (1.32) becomes: **

(1.33)

The Nyquist shift is used when we want the DC component of the DFT to be centered, instead of occurring at the zeroth point. Because this is typically the case in MR images, the two-dimensional raw data are multiplied by a checkerboard [1, −1] pattern before reconstruction (see **Section 13.1). Note that if N is an even integer and the (–1)K multiplier is used to center the DFT response, then the DC component will not be exactly centered but, rather, offset by one-half pixel. Figure 1.2 illustrates using an example with N = 8. The shift constant a in Eq. (1.32) need not necessarily be an integer, so shifts by fractions of a pixel can be accomplished with the appropriate linear phase ramp. This is sometimes used to correct the one-half pixel offset. **

**FIGURE 1.2 **Demonstration of the one-half pixel shift that arises with the discrete FT. Here the DFT length is *N *= 8. The zero frequency, or DC, component of the DFT occupies the zeroth pixel, whose center is indicated with a dot. After a Nyquist shift, the DC signal occupies the fourth pixel. The center of the fourth pixel is shifted by one-half pixel compared to the image center.

The two-dimensional DFT can be defined as:

(1.34)

Three-dimensional (and higher) DFTs are defined analogously to **Eq. (1.34). Note that the data input for the 2D-DFT is a rectilinear N × M matrix of complex numbers, dKL. Because this matrix can be rearranged either as series of rows or columns, 2D-DFTs can be evaluated by performing either the row or column 1D-DFT first. Similarly, 3D (or higher dimensionality) DFTs can be evaluated by performing the 1D-DFTs in any order. This is a very useful property for partial Fourier reconstructions (Section 13.4), where it is necessary to perform the reconstruction along the partial Fourier direction last. **

If the signal in one domain consists of discretely sampled points, then the FT (or IFT) in the other domain is *periodic. *In MRI, the signal is discretely sampled, so the image is periodic or consists of *replicates *that can lead to aliasing artifacts. (This is discussed in more detail in **Section 11.1.) **

To understand how this property of periodicity arises, consider the continuous signal *S*(*t*), -−∞ < *t *< ∞. In practice, the signal will be sampled over a finite time interval, − *T < t *< *T. *To represent the finite sampling interval, first we multiply the signal *S(t) *by a rectangle (RECT) function, which is defined in **Table 1.2 and which is zero for | t| > T. Next, the discreteness of the sampling process can be mathematically represented by multiplication by a sampling comb, which consists of a series of Dirac deltas (further described in Section 1.1.10). If the samples are separated in time by Δt, then the sampled signal becomes: **

(1.35)

According to the convolution theorem, the FT of product in **Eq. (1.35) is the convolution of the FTs of the three factors. From Table 1.2, the FT of the RECT function is a sine x over x, or SINC function. Convolution with this factor gives rise to the shape of the point-spread function. The FT of the sampling comb is another series of deltas, this time spaced by 1/Δt. Convolution with this comb gives rise to the periodic nature of the image. In order to avoid aliasing, the replicates must not overlap. This is accomplished by satisfying the Nyquist criterion—that is, the sampling rate 1/Δt must be at least twice the highest frequency contained in the signal. (These properties are further discussed in Section 11.1 on bandwidth and sampling.) **

A modern MRI scanner must perform a large number of DFTs to reconstruct an image. Computational speed is an important issue. Consider the DFT from **Eq. (1.22) **

(1.36)

where we have abbreviated the twiddle factor

(1.37)

If the twiddle factors are precalculated and stored in a table of complex numbers, then it takes approximately *N*² complex multiplications to evaluate **Eq. (1.36) for all values of J. The FFT algorithm reduces the number of operations and thereby increases the computational efficiency. **

If *N *is even, then the signal can be split into two subsignals: its even-indexed elements and its odd-indexed elements. It can be shown that the DFT can be expressed as a linear combination of two half-length DFTs of the even-and odd-indexed signals. The total number of complex multiplications is then approximately *2(N/2)² = N²/*2. If *N *is a power of 2 (1, 2, 4, 8, 16, 32,…), then this process can be continued recursively all the way down to single-point DFTs. When the sub-FFTs are finally reassembled, the number of complex multiplications is on the order of *N *log2 *N*. For *N *= 512 = 2⁹, this increases the speed by a factor of approximately 512/9 ≈ 57, which is very substantial. Clinical high-resolution MRI would not be feasible without the invention of the FFT.

Although 2 is by far the most common base, or *radix*, for the length of the FFT, some computational speed can be gained whenever *N *is not a prime number. Because of the wide availability and maximal computational efficiency of the radix-2 FFT, however, signals of arbitrary length are often extended to the next power of 2 by using zero filling (see **Section 13.1.2) prior to reconstruction. **

This section describes the Dirac delta function (named after Paul Adrien Maurice Dirac, 1902–1984, an English physicist and mathematician) and the normalization of the continuous FT. Those who are not interested in these mathematical properties can skip to the next section.

The Dirac delta function (**Dirac 1957, 58–61) (or perhaps more properly the Dirac delta distribution) is zero everywhere except at the origin, yet has unit area: **

(1.38)

Thus, the δ(*x*) must not be finite at the origin. The Dirac delta has the property of picking out a single value of a function under multiplication and integration:

(1.39)

The normalization of continuous FTs, for example, **Eqs. (1.1), (1.3), and (1.4), requires that: **

(1.40)

By rearranging **Eq. (1.40): **

(1.41)

and comparing this with **Eq. (1.39), we conclude that the contents of the square brackets in Eq. (1.41) must be a representation of the Dirac delta: **

(1.42)

or, substituting *(x − x’) = a, *and 2π*k *= *u: *

(1.43)

To show that **Eq. (1.43) is true, it is useful to first state the result: **

(1.44)

which can be demonstrated either with contour or numerical integration.

Then consider:

(1.45)

Letting *u = ag, ***Eq. (1.45) becomes: **

(1.46)

Letting *g *→ ∞ and using the result from **Eq. (1.44): **

(1.47)

Combining Eqs.(**1.45) and (1.47) : **

(1.48)

Thus, the integral of the complex exponential is a Dirac delta function. Specifically, **Eq. (1.43) is verified, and the continuous FT and IFT pair of Eqs. (1.1) and (1.3) is properly normalized. **

Bracewell R.N. *The Fourier transform and its applications*. New York: McGraw-Hill; 1978.

Brigham E.O. *The fast Fourier transform and its applications*. Englewood Cliffs, NJ: Prentice Hall; 1988.

Cooley J.W., Tukey J.W. An algorithm for the machine calculation of complex Fourier series. *Mathematics of Computation*. 1965;19:297-301.

Dirac P.A.M. *The principles of quantum mechanics*, 4th ed. Oxford: Oxford, University Press; 1957.

**Section 11.1 Bandwidth and Sampling **

**Section 13.1 Fourier Reconstruction **

**Section 13.4 Partial Fourier Reconstruction **

The description of many physical quantities and processes requires a coordinate system, or *reference frame. *Depending on the choice of reference frame, the description of a physical process can be drastically different. For example, suppose a bicyclist is traveling east as viewed from a building’s window. The same bicyclist will appear to be going west when viewed from a car traveling east at a higher speed. In this example, the building and the car are two different reference frames. They give entirely different descriptions of the same physical process.

Two reference frames, the *laboratory reference frame *and the *rotating reference frame, *are often employed to describe MRI phenomena. The laboratory reference frame is defined with respect to the scanner or the magnet. By convention, when discussing the rotating frame, the *B*o-field direction is always chosen to be the *z *axis (also known as the *longitudinal axis). *(This is somewhat different from the definition of logical gradient axes in **Section 7.3, in which z axis may or may not correspond to the B0-field direction.) The x and y axes in the laboratory reference frame are selected as a pair of orthogonal vectors in a plane normal to the B0-field (denoted by x’ and y′ in Figure 1.3a). In a horizontal magnet with a cylindrical bore, the y axis is usually chosen to be from floor to ceiling (or from down to up). When the z axis is pointing toward the viewer, the x axis is selected to be from left to right. These three axes conform to the right-hand rule, which states that if the fingers of the right hand are curled from the positive x axis to the positive y axis, the thumb points along the positive z axis. Such a coordinate system is called a right-handed Cartesian coordinate system, and the plane defined by the x and y axes is known as the transverse plane. **

**FIGURE 1.3 **(a) The laboratory reference frame and (b) a rotating reference frame. These frames are related by a rotation about the *z *axis with an angular frequency ω0.

When the transverse plane of the laboratory reference frame rotates about the *z *axis with a nonzero angular frequency ω0 (**Figure 1.3b), a rotating reference frame results. Although ω0 is usually a constant, it does not have to be. **

The rotating reference frame greatly simplifies the description of many MRI phenomena. For example, in the laboratory reference frame, spins will precess about the magnetic field with their Larmor frequency ω (**Figure 1.4a, named after Joseph Larmor, 1857–1942, an Irish physicist). In the reference frame rotating with the Larmor frequency, the precession of the spins appears stopped (Figure 1.4b). Using the rotating frame is analogous to stepping onto a moving merry-go-round. From that reference frame, the rotation of the carousel also appears stopped. A mathematical description of this simplification is given in subsection 1.2.2. The concept of the rotating frame is widely used in analyzing and describing the precession and nutation of magnetization, the radiofrequency (RF) magnetic field (i.e., B1 field), and, in particular, the interaction between spin systems and magnetic fields. Many examples can be found throughout this book (e.g., in Sections 3.1, 3.3, and 6.3). **

**FIGURE 1.4 **Spin precession as viewed from (a) the laboratory reference frame and (b) a rotating reference frame. The rotating reference frame slows down the precession. When the angular frequency of the rotating reference frame is equal to the spin precession frequency, the rotational motion of the spin is frozen.

*(Note that the angular frequency vector **follows the right-hand rule in the convention that we use.) *

Let *x*′, *y’, *and *z’ *denote the three Cartesian axes in the laboratory reference frame, and *x, y*, and *z *whose three components in the laboratory reference frame are *px’(t), py’(t), *and *pz’(t), *respectively:

(1.49)

(Recall that the symbol ∧ denotes a unit vector.) The same vector can also be expressed in the rotating reference frame:

(1.50)

Because the reference frame rotates about the *z *axis, it can be readily seen that:

(1.51)

The two sets of transverse axes are related to one another by a rotation:

(1.52)

(1.53)

where ω0 is the angular frequency of the rotating frame.

In light of **in the rotating frame are related to their laboratory-frame counterparts by: **

(1.54)

where ℜ is a rotation matrix. If the components in the rotating reference frame are known, the components in the laboratory frame can be determined from:

(1.55)

where the inverse of the rotation matrix ℜ−1 equals its transpose ℜ*T*.

A useful relation between the rates of change of a vector in the two reference frames can be derived (**Slichter 1989) by taking the time derivative of Eq. (1.49): **

(1.56)

(1.57)

is a rotational angular velocity vector. For a clockwise-rotating reference frame (as viewed from the positive *z *points to the negative *z *axis as indicated in **.) The relationships given by Eqs. (1.54) and (1.56) are often used to simplify the descriptions of MRI phenomena, some of which are detailed in the following subsection. **

**Description of Spin Precession **

Consider a group of spins precessing with the same phase in a static magnetic field *B0. *, will precess about the magnetic field direction (i.e., the *z *axis) with a Larmor frequency ω = γ*B*0 (where γ is the gyromagnetic ratio). In the laboratory reference frame, this rotational motion can be mathematically described by:

(1.58)

According to **in the rotating reference frame is: **

(1.59)

to derive **Eq. (1.59).) If we choose the angular frequency of the rotating frame to be equal to the Larmor frequency (i.e., ω0 ≈ ω), then the magnetization becomes a stationary . This example illustrates how the rotating reference frame simplifies the description of spin precession. **

**Radiofrequency Magnetic Field **

is initially applied along the *x *axis and has a frequency of ωrf (also known as the carrier frequency). In the laboratory frame, a *circularly polarized*, or *quadrature, B1 *field can be expressed as:

(1.60)

Using **Eq. (1.54), we can readily transform this B1 field to the rotating frame: **

(1.61)

If we set the rotating frame angular frequency ω0 to be equal to the carrier frequency ωrf, **Eq. (1.61) then becomes: **

(1.62)

Clearly, the rotating frame has demodulated the RF oscillation and transformed the rapidly oscillating RF field into a much simpler form—the time-dependent envelope *B1(t). *This simplification is widely used in analyzing RF pulses throughout Part II of this book (see examples in **Section 3.1, 6.1, and 6.3). **

**Two-Dimensional Vectors and Their Complex Representation **

Often two-dimensional vectors given by **Eq. (1.63) are used to describe quantities in the transverse plane, such as the B1 field or the magnetization. A two-dimensional vector precessing with an angular frequency ω can be expressed as: **

(1.63)

If the following identifications are made:

(1.64)

in **Eq. (1.63) is equivalent to a complex variable pc(t) given by: **

(1.65)

**Equation (1.65) is called the complex representation . This representation is particularly convenient to work with because of properties of the complex exponential. For example, a straightforward multiplication can translate a vector from the laboratory frame to the rotating frame: **

(1.66)

We can use this relationship to rederive the results in the earlier part of **subsection 1.2.2 with a greater simplicity. **

The complex representation is a natural way to describe the precessing transverse magnetization. The MRI signals induced by the precessing transverse magnetization are also often expressed in the complex form and simplified using **Eq. (1.66). An example is provided in the following subsection. **

**The Bloch Equations **

The Bloch **equations (1.67), named after Felix Bloch 1905–1983, a Swiss born American physicist, relate the time evolution of magnetization to the external magnetic fields, the relaxation times ( T1 and T2), the molecular self-diffusion coefficient (D), and other parameters (Slichter 1989; Torrey 1956). **

(1.67)

If we apply a static *B*0 field and the RF *B1 *field given by **Eq. (1.60) to a spin system, then the overall magnetic field becomes: **

(1.68)

Let us ignore the relaxation and diffusion processes for now. Then, **Eq. (1.67) reduces to: **

(1.69)

Next, we convert **Eq. (1.69) from the laboratory reference frame into a rotating reference frame. Applying Eq. (1.56) to Eq. (1.69), we obtain: **

(1.70)

This conversion, however, is not yet complete because the *B1 *field is still in the laboratory frame. The components of the *B1 *field in the rotating frame can be obtained from **Eq. (1.61): **

(1.71)

Incorporating the *B1 *field of **using Eq. (1.57), the rotating reference frame Bloch equation is derived: **

(1.72)

The magnetic field in the square brackets is known as the *effective field*:

(1.73)

and the magnetization precesses about the applied *B1 *field, as if there were no static magnetic field *B*0 present. The concept of effective field is frequently used to describe the interactions between RF pulses and the spin systems (see, for example, **Sections 6.1 and 6.2). **

**Equation (1.72) can be equivalently expressed as three scalar equations after explicitly carrying out the vector cross product: **

(1.74)

(1.75)

(1.76)

Let us define a complex quantity to represent the vector component of the magnetization in the transverse plane:

(1.77)

Using this complex quantity, we can derive **Eq. (1.78) by multiplying Eq. (1.75) by i and adding the result to Eq. (1.74): **

(1.78)

**Equation (1.78) is particularly useful in solving for the transverse magnetization after an RF excitation pulse (see Section 3.1). **

Finally, let us consider two special cases. In the first case, we set the rotating frame frequency to be identical to the carrier frequency of the RF field (i.e., ω0 = ωrf). Thus, **Eq. (1.72) becomes: **

(1.79)

This reference frame is sometimes referred to as the *RF reference frame *or *B*1 reference frame because the *B1 *field is stationary in this frame. Note that the *B*1 field lies along the *x *axis because we have chosen a specific initial condition for *t = *0, as implied by **, where α is the angle between the initial B1 vector and the x axis. Equation (1.79) shows that in a rotating reference frame with ω0 = ωrf, the B1 field is demodulated and the main magnetic field is effectively reduced by ωrf/γ. At resonance (i.e., ωrf = γB0), the z component of the effective magnetic field disappears and the magnetization will precess about the B1 field. **

In the second case, we can set the rotating frame frequency equal to the Larmor frequency (i.e., ω0 = ω). Thus, **Eq. (1.72) becomes: **

(1.80)

This reference frame is known as the *Larmor reference frame *or *B*0 reference frame because the *B*0 field disappears in this reference frame. If we keep the RF field at a fixed frequency and sweep the *B*0 field (i.e., changing ω), resonance can also occur when ω = ω0 = ωrf.

Slichter C.P. *Principles of magnetic resonance*. Berlin: Springer-Verlag; 1989.

Torrey H.C. Bloch equations with diffusion terms. *Physical Review*. 104, 1956.

**Section 3.1 Excitation Pulses **

**Section 3.2 Inversion Pulses **

**Section 3.3 Refocusing Pulses **

**Section 4.1 Composite Pulses **

**Section 4.3 Spectrally Selective Pulses **

**Section 6.1 Adiabatic Excitation Pulses **

**Section 6.2 Adiabatic Inversion Pulses **

**Section 6.3 Adiabatic Refocusing Pulses **

**PART II **

**RADIOFREQUENCY PULSES **

Part II of this book describes radiofrequency (RF) pulses. **Chapter 2 focuses on several commonly used RF pulse shapes including hard pulses (Section 2.1), SINC pulses (Section 2.2), and a family of tailored pulses generated with the Shinnar-Le Roux (SLR) algorithm (Section 2.3). The SLR section is more mathematically involved than many others in the book, and that section can be skipped unless the reader has a special interest in this topic. Some of the theoretical results from Section 2.3, however, have wider practical significance and reappear in other sections of Part II. For example, SLR analysis can be used to explain when and how an excitation pulse also can be used for refocusing. A discussion of variable-rate pulses (Section 2.4) completes Chapter 2. Variable rate is not a specific pulse shape but rather a method to modify the shape of any spatially selective RF pulse in order to reduce its RF power deposition. In addition to the basic pulse shapes covered in Chapter 2, other pulse shapes are used in MRI and several less commonly used pulse shapes are discussed in the other sections of the book (for example, a discussion of Gaussian and Fermi pulse shapes is provided in Section 4.2). **

**Chapter 3 discusses three basic RF pulse functions: excitation (Section 3.1), inversion (Section 3.2), and refocusing (Section 3.3) of the magnetization. These sections describe nonadiabatic RF pulses; the adiabatic pulses performing these same three functions are the focus of Chapter 6. The slice profile and nonlinearity of the Bloch equations are explored in Section 3.1; Section 3.3 introduces the formation of RF spin echoes. The relationship among the three RF pulse functions is explored in both Section 3.2 and 3.3. All the sections in Chapter 3 examine the effect of spins being off-resonance on their respective functions. **

Spectrally selective RF pulses (i.e., pulses that are played without a concurrent slice-selection gradient and affect spins only within a specific frequency range) are discussed in **Chapter 4. Composite pulses (Section 4.1) are more widely used in magnetic resonance (MR) spectroscopy, but have found application to imaging as well. Magnetization transfer pulses (Section 4.2) are mainly used in 3D time-of-flight MR angiography. A general discussion of spectrally selective pulses is given in Section 4.3, which includes a discussion of lipid suppression with spectrally selective excitation and saturation, widely used in imaging applications. **

Spatially selective RF pulses are the topic of **Chapter 5. This chapter covers a number of RF pulses that are played along with a gradient pulse to affect the magnetization in a spatially dependent fashion. The pulses included in this chapter are multidimensional pulses (Section 5.1), ramp pulses, also known as tilted optimized non saturating excitation (TONE) pulses (Section 5.2); spatial Saturation pulses (Section 5.3); spatial-spectral pulses (Section 5.4); and spatial tagging pulses (Section 5.5). Multidimensional pulses are selective in more than one spatial direction. Ramp pulses intentionally yield a nonuniform spatial profile and are used for MR angiography. Spatial saturation pulses attenuate signal from specified regions and are widely used in imaging applications to suppress artifacts from aliasing, flow, and motion. Spatial-spectral (SPSP) pulses are simultaneously selective in one spatial and the spectral dimension. (The placement of the spatial-spectral section in Chapter 5 rather than in Chapter 4 is somewhat arbitrary). The chapter concludes with a discussion of spatial tagging pulses, which are used to prepare the longitudinal magnetization with a set of stripes or grids prior to an imaging sequence. Unlike the other pulses described in Chapter 5, tagging pulses are not necessarily played with a concurrent slice-selection gradient but rather can **

You've reached the end of this preview. Sign up to read more!

Page 1 of 1

Close Dialog## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Loading