HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Reeder, S. B. Articles by Bluemke, D. A. Search for Related Content PubMed PubMed Citation Articles by Reeder, S. B. Articles by Bluemke, D. A. Related Collections Cardiac Radiology Magnetic Resonance Imaging Physics and Basic Science

Advanced Cardiac MR Imaging of Ischemic Heart Disease¹

¹ From the Department of Radiology, Rm H1306, Stanford University, 300 Pasteur Dr, Stanford, CA 94304 (S.B.R.); GE Medical Systems, Milwaukee, Wis (Y.P.D.); and the Departments of Medicine (J.A.C.L.) and Radiology (D.A.B.), Johns Hopkins University, Baltimore, Md. Presented as a scientific exhibit at the 1999 RSNA scientific assembly. Received March 23, 2000; revision requested June 7; final revision received March 2, 2001; accepted March 19. Address correspondence to S.B.R. (e-mail: sreeder@stanford.edu).

	Abstract

Top Abstract Introduction Technical Issues Measurement and Detection of... Contrast-enhanced Perfusion... Coronary MR Angiography Future Directions Conclusions References

 
Important advances in rapid magnetic resonance (MR) imaging technology and its application to cardiovascular imaging have been made during the past decade. High-field-strength clinical magnets, high-performance gradient hardware, and ultrafast pulse sequence technology are rapidly making the vision of a comprehensive "one-stop shop" cardiac MR imaging examination a reality. This examination is poised to have a significant effect on the management of coronary artery disease by means of assessment of wall motion with tagging and pharmacologic stress testing, evaluation of the coronary microvasculature with perfusion imaging, and direct visualization of the coronary arteries with MR coronary angiography. This article reviews current state-of-the-art pulse sequence technology and its application to the evaluation of ischemic heart disease by means of MR tagging with dobutamine stress testing, MR perfusion imaging, and MR coronary angiography. Cutting edge areas of research in coil design and exciting new areas of metabolic and oxygen level–dependent imaging are also explored.

Index Terms: Coronary vessels, stenosis or obstruction, 54.76 • Heart, ischemia, 51.771 • Heart, MR, 51.1214 • Magnetic resonance (MR), perfusion study, 51.12143 • Magnetic resonance (MR), pulse sequences, 51.1214 • Magnetic resonance (MR), vascular studies, 51.12142 • Myocardium, infarction, 511.771 • Myocardium, ischemia, 511.771

	Introduction

Top Abstract Introduction Technical Issues Measurement and Detection of... Contrast-enhanced Perfusion... Coronary MR Angiography Future Directions Conclusions References

 
Cardiovascular disease is the leading cause of death in the United States (1). Each year, over 1.5 million U.S. residents have myocardial infarctions, roughly one-third of whom die. Although there has been a dramatic reduction in the death rate due to cardiovascular disease over the past 30 years, it still accounts for 22% of all deaths.

Diagnostic imaging has played an important role in the proper assessment and management of coronary artery disease. Advances in rapid magnetic resonance (MR) imaging technology and its application to cardiac imaging have shown that MR imaging has tremendous potential for evaluation of coronary artery disease and cardiac disease in general.

The concept of a comprehensive cardiac MR imaging evaluation has been an evolving theme during the past decade, with the vision of a complete cardiac examination that could be performed in a relatively short time (eg, <1 hour). A single examination that provides information on regional myocardial contractility and contractile reserve, regional perfusion reserve, and the detailed anatomy of the coronary arteries can be envisioned on the basis of MR imaging with tagging and dobutamine stress testing, gadolinium-enhanced measurement of regional perfusion, and advances in MR coronary angiography, respectively.

This article reviews the clinical state of the art in cardiac imaging for evaluation of coronary artery disease at 1.5 T. An overview of pulse sequence technology is followed by a description of cardiac MR imaging applications in evaluation of myocardial infarction, evaluation of myocardial ischemia, and coronary angiography. Finally, some examples of exciting advances that demonstrate promising new roles for cardiac MR imaging in management of coronary artery disease are explored.

	Technical Issues

Top Abstract Introduction Technical Issues Measurement and Detection of... Contrast-enhanced Perfusion... Coronary MR Angiography Future Directions Conclusions References

 
The greatest hurdle for cardiac MR imaging was the development of gradient hardware and pulse sequence technology that allows imaging sequences to "freeze" the motion of the heart during the cardiac cycle. In addition, it was necessary that these techniques be insensitive to the high velocities of the myocardium and ventricular blood, which manifest as ghost artifacts with traditional pulse sequences. The development of rapid pulse sequences facilitated breath-hold imaging, which was necessary to accommodate the wide range of motion the heart experiences during respiration. The greatest challenges in the acquisition of MR imaging movies, or cine imaging, have been twofold: (a) cardiac motion caused by diaphragmatic motion during breathing and (b) flow artifacts from myocardial and blood pool velocities.

Improvements in gradient hardware technology, as well as the ability of MR imaging systems to manipulate and process raw data more rapidly, have brought significant advances to cardiac MR imaging. Although rapid imaging techniques are routinely used to capture motion of the heart, areas of active investigation in cardiac MR imaging necessitate reductions in total imaging time, such as bolus tracking of contrast agents through the myocardium (2). Accurate characterization of perfusion defects requires sampling of three to six imaging planes with single-heartbeat temporal resolution (3). In addition, MR dobutamine stress testing requires acquisition and reconstruction of complete cardiac cine loops within one to five heartbeats. Finally, real-time MR "fluoroscopy" is now capable of cine refresh rates on the order of 20–40 frames per second (4,5).

The diaphragm can move 2–3 cm during quiet breathing, causing substantial displacement of the heart, and strategies are needed to compensate for this phenomenon. Breath-hold imaging has been the most widely used approach and has been demonstrated to be a reliable method of obtaining sets or subsets of images that are well registered and unaffected by respiration. Breath-hold imaging requires very rapid imaging techniques that can acquire a full set of images in about 20 seconds. This approach implicitly assumes that cardiac motion is highly reproducible. In practice, this assumption is generally true, although there are some notable exceptions, including arrhythmias and pericardial effusions, when cardiac motion may vary from beat to beat. The latter phenomenon, known as "swinging heart," occurs as a result of excessive motion of the heart within the pericardial sac in the presence of a large effusion (6). State-of-the-art sequences performed on imaging units with high-performance gradients can now acquire cine data sets in as few as one to three heartbeats, and real-time acquisitions eliminate the need for breath holding altogether, facilitating the capture of beat-to-beat variations in cardiac physiology.

The velocities of ventricular blood during systole and early diastole are approximately 200 cm/sec and 150 m/sec, respectively (7). Traditional imaging sequences are highly susceptible to first moment phase shifts caused by these high velocities. The ghost artifacts that result have been successfully addressed by using sequences with ultrashort acquisition windows with very small first moments, as well as first moment nulling techniques. Blood signal suppression techniques that null the signal from the blood pool are also highly effective for artifact reduction. Fortuitously, imaging sequences with small first moments often have the best speed performance and are best suited for rapid cardiac imaging.

Although a variety of rapid imaging sequences have been applied to capture cardiac motion, two pulse sequence strategies are particularly effective for the demands of cardiac imaging: single- and multiecho gradient-echo (GRE) imaging and spiral imaging.

Single- and Multiecho GRE Segmented k-Space Imaging
The first reliable breath-holding techniques applied to cine cardiac imaging were ultrafast spoiled GRE sequences (8,9). Spoiled GRE imaging is well suited to imaging the heart because it is a robust imaging technique with excellent reproducibility and adequate signal-to-noise ratio (SNR). Short echo times impart insensitivity to motion and minimize the off-resonance effects of susceptibility and chemical shift. The effects of short T2* commonly observed in tissue from the thorax and mediastinum are also minimized. Acquisition begins when the imaging unit is triggered by the electrocardiogram, synchronizing image acquisition with the heart cycle. Multiple repetition times (TRs) are played through the heart cycle, and sequential lines of k-space data from each TR are segmented into individual raw data sets that will form the final frames of the cine acquisition. For single-echo imaging, the number of lines (views) of k space per segment (NVS) determines both the time resolution of the cine acquisition (TR x NVS) and the total number of heartbeats required for the acquisition, which is the total number of k-space lines acquired (N_y) divided by NVS. Images with improved time resolution can be acquired by means of reductions in NVS at the cost of longer breath holds. An acquisition becomes "real time" when all lines of k space are acquired during one heartbeat (ie, NVS = N_y).

An important variation of GRE imaging, generically known as steady-state free precession (SSFP) and commonly known as fast imaging with steady-state precession (FISP) or "true FISP," has recently seen a resurgence in interest. Introduced in the late 1980s (10), this technique preserves the coherence of the transverse magnetization by refocusing the gradients in all three axes. Both the transverse and longitudinal magnetizations reach a steady-state equilibrium that results in image SNR greater than with spoiled GRE sequences (11). Unlike the signal intensity of spoiled GRE sequences, which is primarily T1 dependent, the image intensity with SSFP also depends on T2 as a result of the coherence of the transverse magnetization. The usefulness of this pulse sequence was not fully realized until recently, as a result of severe artifacts caused by field inhomogeneities. With the advent of high-performance gradient hardware and improved field homogeneity, SSFP has had a recent revival with applications in MR fluoroscopy and cardiac imaging (12–14). For cardiac imaging, the increased T2 weighting improves the contrast between myocardium and ventricular blood, potentially eliminating the need for black-blood techniques.

In comparison with standard spin warp imaging, single-echo segmented k-space imaging or SSFP requires little additional postprocessing. Oblique imaging plane optimization schemes for maximizing sequence efficiency by means of reduced sequence dead times have allowed substantial improvements in sequence efficiency and time performance (15).

More recently, the use of multiple echoes after each radio-frequency (RF) excitation has been used to improve both time and SNR performance. The concept is based on the original echo-planar imaging technique, in which an entire image data set is acquired after a single RF excitation (16). The efficiency of echo-planar imaging is very high, unlike that of GRE sequences, which require one RF excitation for every k_y line sampled.

Although echo-planar imaging is capable of high SNR and very low imaging times (40–120 msec), it is inherently sensitive to field inhomogeneities, chemical shift, and short relaxation times (T2*), all of which are problematic when imaging the heart. To circumvent these problems, McKinnon (17) applied interleaved echo-planar imaging to cardiac imaging with an image data set segmented over several "shots," usually one-eighth to one-fourth of the raw data per heartbeat. Interleaving reduces image artifacts caused by field inhomogeneities and short relaxation times. When echo-planar imaging is applied to the heart for cine acquisitions, sequential excitation or sampling of imaging planes is performed in a similar manner as with segmented k-space acquisitions. This fact implies that the magnetization behavior of cardiac echo-planar imaging and spoiled GRE imaging is very similar and can be analyzed on a continuum of echo train lengths.

In this light, an alternative paradigm approaches the situation from the bottom up by adding echoes to fast GRE sequences (18). In this paradigm, multiple TRs, each with multiple echoes, are acquired for each phase of the cine animation. SNR gains with multiecho GRE imaging and echo-planar imaging are achieved mainly by means of improvements in sequence efficiency and the increased recovery of longitudinal magnetization (T1) permitted by a longer TR. State-of-the-art multiecho sequences can now produce good-quality real-time acquisitions. For example, a newly developed multiecho sequence (19) run on a high-performance imaging unit can acquire good-quality real-time images at a rate of 23 frames per second. This sequence has a TR of 10.7 msec, an echo train length of eight, a 125-kHz bandwidth, and a 36 x 18-cm field of view. The matrix size is 160 x 48 with partial k_y acquisition, and an entire image is acquired every 43 msec. When new postprocessing algorithms (20) are used, this acquisition can produce effective frame rates of around 40 frames per second.

Figure 1 is a schematic of k-space segmentation for a multiecho sequence, which is similar to k-space segmentation for single-echo sequences (8,9). The number of views per segment (NVS) is the product of the number of echoes per RF excitation, or echo train length, and the number of TRs in each segment (NVS = echo train length x number of TRs per segment). In this figure, NVS is nine, the echo train length is three, and the number of TRs is three. Depending on the imaging parameters, a single line of k space is acquired, on average, every 0.5–1.5 msec in imaging units with high-performance gradients. The time resolution of a multiecho image (in milliseconds) is equal to TR x NVS/echo train length.

View larger version (80K): [in this window] [in a new window] [Download PPT slide]   Figure 1.   Schematic of a cine acquisition with corresponding k-space trajectory for a three-echo spoiled GRE sequence, with nine lines of k space per movie frame (NVS) and Ny = 27. Only 27 lines are shown for clarity; typical acquisitions use an Ny of 60-120. Consecutive echo trains are offset horizontally in the diagram for clarity. Temporal resolution (in milliseconds) is equal to TR x NVS/echo train length. RF pulses are represented by solid boxes, and each echo is read out during open boxes. Typical short-axis cine images are shown. 2D = two-dimensional.

An advantage of multiecho sequences when used in conjunction with myocardial tagging (which is discussed later in this article) is improved tag contrast. The closely spaced RF pulses of single-echo imaging drive magnetization toward steady state, causing rapid tag fading. Without special strategies, diastolic motion becomes difficult to sample as a result of tag fading. Single-echo echo-planar imaging has the best tag contrast (22), and multiecho spoiled GRE imaging has improved tag contrast characteristics over single-echo imaging.

Single- and multiecho spoiled GRE sequences are well suited to gadolinium-enhanced perfusion imaging for two reasons. First, they are sufficiently rapid to acquire two to six imaging planes every one to two heartbeats. Second, they are inherently T1 sensitive, making them ideal for perfusion imaging with contrast agents because these agents act by modulating T1 in a concentration-dependent fashion, as described later in this article.

Spiral Cardiac Imaging
Spiral imaging is an alternative imaging modality (23,24) that samples k space along trajectories that start at the center of k space and spiral outward. This scheme is a more natural sampling pattern that has high sampling densities near the center of k space, where the power spectrum of the image is generally highest. Spiral imaging, like echo-planar imaging, is highly efficient, but it is less sensitive to T2* losses than echo-planar imaging owing to the fact that sampling begins at the center of k space immediately after RF excitation. With echo-planar imaging, a significant fraction of signal is lost due to T2* decay by the time the center of k space has been sampled. In addition, spiral imaging is inherently insensitive to motion artifacts, unlike most spin warp techniques. This insensitivity is a result of the fact that all moments of spiral gradient waveforms naturally refocus. For a detailed explanation, see the article by Meyer et al (24).

Despite the intuitive appeal and theoretical advantages of spiral imaging, its implementation has been technically challenging. The use of spatial-spectral pulses has been necessary for eliminating chemical shift from fat to prevent blurring artifacts (25). This approach has the added advantage of eliminating the signal from pericardial fat in which coronary arteries lie, providing excellent vessel-wall contrast. In addition, new developments in the use of rapid field map correction algorithms (26) and rapid non-Fourier grid reconstruction algorithms (24,27) have helped spiral imaging mature into a viable and robust technique for cardiac imaging, especially in the area of coronary angiography (24,28,29). For motion studies, a spiral sequence can produce a high-quality cine loop in approximately 10–20 heartbeats with 15–20-msec temporal resolution and 1 x 1-mm in-plane spatial resolution. Spiral imaging is well suited to real-time fluoroscopic cardiac imaging (4,30,31), and the recent application of spiral imaging with high-performance gradients (G_max = 4 G/cm, dG/dT = 150 T/m per second) has produced real-time images at frame rates of 20–30 frames per second (31). Figure 2 is a schematic of cine spiral k-space trajectories, with examples of cine spiral images.

View larger version (61K): [in this window] [in a new window] [Download PPT slide]   Figure 2.   Spiral k-space cardiac cine acquisition. Three spiral arms are segmented into each frame with every heartbeat. In three heartbeats, all nine spiral arms have been acquired. Spatial-spectral pulses are represented by solid boxes, and each spiral arm is read out during open boxes. (Images courtesy of Craig H. Meyer, PhD, Stanford University, Stanford, Calif.)

	Measurement and Detection of Myocardial Wall Motion Abnormalities

Top Abstract Introduction Technical Issues Measurement and Detection of... Contrast-enhanced Perfusion... Coronary MR Angiography Future Directions Conclusions References

View larger version (171K): [in this window] [in a new window] [Download PPT slide]   Figure 3.   Breath-hold tagged cine MR images obtained in a healthy human volunteer. Two short-axis views of the same imaging plane (top two rows) and one long-axis view (bottom row) are shown; each displays tag lines from a different set of mutually orthogonal planes. Only three (of 25) phases during the cardiac cycle are shown (early, middle, and late systole [left, middle, and right columns, respectively]). LA = left atrium, LV = left ventricle, RA = right atrium, RV = right ventricle. (Courtesy of Cengizhan Ozturk, MD, PhD, Johns Hopkins University, Baltimore, Md, and Elliot R. McVeigh, PhD, National Institutes of Health, Bethesda, Md, and Johns Hopkins University; reprinted, with permission, from reference 33.)

Cardiac MR Dobutamine Stress Testing
Dobutamine stress echocardiography and dobutamine stress nuclear cardiography have been routinely employed over the past decade for evaluation of coronary artery disease by means of detection of wall motion abnormalities under pharmacologic stress (37,38). Dobutamine increases both contractility and heart rate, eliciting significant increases in myocardial oxygen consumption. Typically, the patient’s heart is monitored in real time with an ultrasound (US) imaging probe (37,38) while the rate of intravenous administration of dobutamine is stepped from 0 µg/kg per minute to 5, 10, 20, and 40 µg/kg per minute, with 3 minutes at each level. If a wall motion abnormality develops before the maximal rate is reached, dobutamine administration is stopped and the test is considered positive for hemodynamically significant coronary artery disease. Dobutamine administration is discontinued for new onset of symptoms such as chest pain or dyspnea or if the target heart rate of the patient has been reached ([220 - age] x 0.85). Blood pressure is checked frequently throughout the examination.

There are two general responses that are detected with dobutamine stress testing. First, wall motion abnormalities can be induced in coronary beds fed by arteries with subcritical but hemodynamically significant stenoses. This situation occurs when the coronary reserve is insufficient to increase blood flow to match increased oxygen demands during pharmacologic stress. Detection of induced wall motion abnormalities with high-dose dobutamine echocardiography (0–40 µg/kg per minute) has been shown to have a sensitivity of 83% and specificity of 85% for detection of ischemic heart disease (39).

Thallium-201 and technetium-99m scintigraphy have similar sensitivity and specificity for detection of coronary artery disease, albeit at higher cost and with radiation exposure, which may limit repeat use of this technique (40). In addition, gated wall motion imaging during resting states is routinely performed in nuclear cardiology. The presence of resting wall motion abnormalities in combination with reversible perfusion defects may enhance the specificity and sensitivity of nuclear medicine in detection of coronary artery disease.

The second response seen with dobutamine is directly related to assessment of myocardial viability. Tissue that contracts normally during rest is viable. Myocardium that is dysfunctional during rest in the setting of recent or ongoing ischemia is of great interest, since the viability of this tissue is uncertain. Contractile dysfunction of viable tissue comes in distinct entities. Hibernating myocardium is viable myocardium with depressed contractile dysfunction resulting from chronic ischemia (months to years) insufficient to cause tissue necrosis. Contractility recovers quickly once coronary flow is restored. Stunned myocardium exhibits sustained (hours to days) reduced regional contractility after coronary reperfusion from an acute ischemic insult. However, it will respond to inotropic stimulation, and this behavior of the myocardium can be exploited to differentiate between stunned and infarcted myocardium (37). Low-dose dobutamine stress echocardiography (0–20 µg/kg per minute) has been used to test for viability in the form of transient recovery of function (41). Assessment for reversible dysfunction is important for prognosis and to identify high-risk patients, who may benefit most from aggressive revascularization therapy.

Unfortunately, transthoracic echocardiography is inherently limited by poor image quality and provides only qualitative estimates of myocardial motion. (In dobutamine echocardiography, a qualitative "wall motion index" is used to assess contractility.) Use of this technique leads to subjective measures of wall motion abnormalities rather than quantifiable assessment of the significance of detected abnormalities. The presence of multivessel disease may also affect the sensitivity of this test. Approximately 10%–15% of dobutamine echocardiographic studies are considered "technically poor," and studies are often unsuccessful due to lack of adequate acoustic windows (37,42). Cardiac MR imaging has consistently superior image quality over that of echocardiography and has the added advantage of tissue tagging, providing accurate quantitative motion estimates that allow detection of subtle wall motion abnormalities. Preliminary results of MR dobutamine stress testing in 208 patients have demonstrated sensitivity and specificity superior to those of dobutamine stress echocardiography in detection of coronary artery disease (43).

Accurate measurement of regional strain requires that movie loops of seven to nine imaging planes (four or five short-axis planes and three or four long-axis planes) be acquired during the plateau phase of each 3-minute dobutamine titration, which is necessary for the heart to reach a stable hemodynamic state. During this interval, approximately three 15-second breath holds can be performed, requiring that a cine loop of each imaging plane be acquired in five or fewer heartbeats. Multiecho spoiled GRE imaging with high-performance gradients (18) can easily provide the speed performance necessary for MR dobutamine stress testing.

Figure 4 is a panel of short-axis images acquired in a healthy human volunteer at rest and during dobutamine-induced stress (20 µg/kg per minute). Five (of eight) movie frames acquired through systole are shown in this figure, each with a time resolution of 41 msec. In this particular examination, movies from eight imaging planes were acquired at four dobutamine levels. Increased tag detection and cavity obliteration are seen with dobutamine administration, indicating an increase in contractility. Less than 2 minutes was needed at each stress level to perform cine acquisitions of all eight imaging planes.

View larger version (78K): [in this window] [in a new window] [Download PPT slide]   Figure 4.   Cine short-axis images acquired in a healthy volunteer at rest and during 20 µg/kg per minute of dobutamine stress. The first five of eight time frames are shown, each separated by 41 msec. A noticeable increase in contractility, evident as increased tag deflection and cavity obliteration, is seen in the bottom row, where end-systole has been reached by 173 msec after the QRS complex. LV = left ventricle, RV = right ventricle. (Courtesy of Elliot R. McVeigh, PhD.)

View larger version (31K): [in this window] [in a new window] [Download PPT slide]   Figure 5.   Maximal strain calculated at five locations around the left ventricle in the midwall at different levels of dobutamine. Ant = anterior, Inf = inferior, Lat = lateral, Pos = posterior, Sep = septal. (Courtesy of Michael A. Guttman, MSE, National Institutes of Health, Bethesda, Md.)

View larger version (103K): [in this window] [in a new window] [Download PPT slide]   Figure 6.   Cine short-axis images acquired in a patient with coronary artery disease at rest (top row) and during 40 µg/kg per minute of dobutamine stress (bottom row). During dobutamine administration, abnormal tag deflection, as evidenced by reduced tag curvature, is seen in the anteroseptal region (black arrow) but not in the lateral wall (white arrow) on the end-systolic images. This wall motion abnormality is a result of dobutamine-induced ischemia.

View larger version (53K): [in this window] [in a new window] [Download PPT slide]   Figure 7.   Acute myocardial infarction with myocardial tagging. The end-systolic image shows abnormal poor wall motion in the anteroseptal distribution (black arrow). Normal contraction of the myocardium is seen in the lateral wall (white arrow).

View larger version (71K): [in this window] [in a new window] [Download PPT slide]   Figure 8.   Plot of circumferential strain for a patient (red line) during the cardiac cycle shows paradoxical stretching in the septal region from apex to base. In other areas, circumferential strain is reduced from the strain evolution of the normal population (green lines span two standard deviations), indicating global hypokinesis. (Courtesy of Elliot R. McVeigh, PhD.)

	Contrast–enhanced Perfusion Imaging

Top Abstract Introduction Technical Issues Measurement and Detection of... Contrast-enhanced Perfusion... Coronary MR Angiography Future Directions Conclusions References

Accurate characterization of regional myocardial perfusion necessitates repeated acquisition of images over the entire heart with high temporal resolution to capture the first pass of contrast agent through the myocardium (3). Advances in hardware and pulse sequence technology now allow acquisition of multiple imaging planes every heartbeat during the tracking of a contrast agent bolus. Figure 9 diagrams the timing of a typical single- or multiecho spoiled GRE perfusion imaging sequence. In this example, a complete k-space data set composed of N_y echoes is acquired, followed by N_y echoes for a second imaging plane and finally a third plane. Depending on the heart rate, image resolution, and imaging unit performance, current technology allows acquisition of two to six complete images every heartbeat, permitting accurate tracking of the contrast agent bolus through the myocardium. An added advantage of single-heartbeat acquisition is that breath holding is no longer required during administration of the bolus and subsequent image acquisition. Trade-offs for higher image resolution can be made by reducing the temporal sampling rate. Segmentation to more than one heartbeat per image requires breath holding, however.

View larger version (44K): [in this window] [in a new window] [Download PPT slide]   Figure 9.   Timing schematic for perfusion imaging with saturation recovery. Here, two imaging planes are acquired every heartbeat with a single-echo GRE sequence (Ny = 96, Nx [matrix size in the readout direction] = 256, TR = 3.1 msec, echo time = 1.2 msec, field of view = 36 cm). Two additional imaging planes are acquired in the next heartbeat, so that four planes are acquired with two-heartbeat resolution. Each acquisition is preceded by a 90° nonselective presaturation pulse to zero longitudinal magnetization and to enhance T1 contrast. Breath holding was not required for this acquisition. (Images courtesy of Anthony Z. Faranesh, MSE, Johns Hopkins University, Baltimore, Md.)

The acquisition of each imaging plane is preceded by a 90° saturation pulse or a 180° inversion pulse to enhance the sensitivity of the pulse sequence to changes in T1. Sequences that use the 90° saturation pulse are insensitive to variations in heart rate, whereas inversion preparation increases sensitivity to T1 but is more susceptible to image artifacts and variations in heart rate. Increasing the imaging tip angle also increases the T1 sensitivity, as does centric phase encoding, which delays acquisition of the central lines of k space to the last few TRs, where T1 contrast is the highest.

Figure 10 is an example of a first-pass perfusion acquisition in a patient with an old infarction of the left circumflex arterial distribution. The bolus of gadopentetate dimeglumine can be followed as it passes through the right ventricle, pulmonary vasculature, and left ventricle and finally into the left and right ventricular myocardium. A well-defined, crescent-shaped defect correlates with the left circumflex artery occlusion, as confirmed with conventional angiography.

View larger version (72K): [in this window] [in a new window] [Download PPT slide]   Figure 10.   Eight time frames of a midventricular imaging plane of a human heart, obtained by using the saturation recovery technique after bolus intravenous injection of gadopentetate dimeglumine. The time interval between images is four heartbeats. A large, crescent-shaped defect is visible in the lateral wall of the left ventricle (arrow), which corresponds to a documented old occlusion of the left circumflex artery. LV = left ventricle, RV = right ventricle. (Courtesy of Anthony Z. Faranesh, MSE.)

View larger version (82K): [in this window] [in a new window] [Download PPT slide]   Figure 11a.   (a) First-pass (top row) and delayed (bottom row) images through four imaging planes of the ventricle, obtained in a patient 5 days after a documented inferior wall myocardial infarction. A region of hypoenhancement is seen on the first-pass images (arrows), and a larger area of hyperenhancement is evident on the delayed images (arrows). (b) Images from a cine tagging study of the same patient in a midventricular imaging plane. A wall motion abnormality (arrows) is seen in the same region as the perfusion abnormality in a. The tag pattern in this area shows poor contractile function. (Reprinted, with permission, from reference 33.)

View larger version (57K): [in this window] [in a new window] [Download PPT slide]   Figure 11b.   (a) First-pass (top row) and delayed (bottom row) images through four imaging planes of the ventricle, obtained in a patient 5 days after a documented inferior wall myocardial infarction. A region of hypoenhancement is seen on the first-pass images (arrows), and a larger area of hyperenhancement is evident on the delayed images (arrows). (b) Images from a cine tagging study of the same patient in a midventricular imaging plane. A wall motion abnormality (arrows) is seen in the same region as the perfusion abnormality in a. The tag pattern in this area shows poor contractile function. (Reprinted, with permission, from reference 33.)

View larger version (151K): [in this window] [in a new window] [Download PPT slide]   Figure 12.   Long-axis delayed perfusion image of a canine heart, obtained 48 hours after 1-hour occlusion of the left anterior descending coronary artery followed by reperfusion and 15 minutes after bolus injection of gadopentetate dimeglumine. A zone of hyperenhancement surrounds a dark no-reflow core (arrows). LV = left ventricle, RV = right ventricle.

Progression. It is well documented that the extent of residual small-vessel obstruction plays an important role in left ventricular remodeling and long-term clinical outcomes (50). Initially, microvascular obstruction was believed to finish shortly after the onset of arterial reflow, but later experiments suggested that this process is dynamic and develops up to 3.5 hours after reperfusion (51). The ability to assess accurately regions of no reflow with contrast-enhanced MR imaging has been demonstrated (47,52). In addition, it has been shown that infarct size and the extent of microvascular obstruction increase significantly over the first 48 hours after 90 minutes of coronary artery occlusion and reperfusion (53). These results demonstrate that progressive microvascular and myocardial injury occurs well beyond the period of coronary artery occlusion and reperfusion.

Outcomes. The risk of future adverse events increases with infarct size, and the presence of hypoenhanced areas is associated with fibrous scar formation and left ventricular remodeling. Wu et al (54) recently determined that the presence and size of a hypoenhanced core seen during the first pass correlate with outcome. This study showed that after infarction, the degree of microvascular obstruction—determined by the size of hypoenhanced zones (if present)—allows prediction of future cardiovascular complications. Patients with hypoenhanced regions were found to have an increase in future cardiovascular events relative to those with no hypoenhanced regions (45% vs 9%). Infarct size, defined by the size of late hyperenhanced areas, was also shown to correlate directly with long-term prognosis in patients with acute myocardial infarction. Interestingly, the size of the microvascular obstruction was the strongest prognostic marker, independent of infarct size.

Stress Testing for Perfusion Reserve
Perfusion imaging can be used in assessment of perfusion reserve by means of potent vasodilators, usually dipyridamole or adenosine. Preliminary studies in a swine model are being performed by using dipyridamole stress testing (55). Coronary artery stenosis is created by placing a custom-made nylon flow-reducer fitting in the left anterior descending artery with standard coronary angiographic techniques. Dipyridamole is administered at 0.57 µg/kg per minute for 4 minutes. At 6–8 minutes after initiation of dipyridamole administration, Tc-99m sestamibi is injected. Immediately after injection of the radiotracer, an intravascular gadolinium T1 contrast agent (AngioMARK; EPIX Medical, Cambridge, Mass) is injected at 5 mL/sec with a power injector and MR perfusion imaging is started. After 30 minutes, resting perfusion images are obtained in the same manner. Figure 13 is an example of early-phase gadolinium-enhanced images in four (of five) imaging planes spanning the left ventricle and corresponding Tc-99m sestamibi single photon emission computed tomographic (SPECT) images, obtained during dipyridamole administration. A well-defined perfusion defect is seen in the anterolateral wall on the MR images, which correlates closely with the large defect seen on the SPECT images. No perfusion deficit was seen on resting images. Preliminary human myocardial stress perfusion studies with dipyridamole and adenosine have shown high sensitivity and specificity in detection of coronary artery disease (40).

View larger version (80K): [in this window] [in a new window] [Download PPT slide]   Figure 13.   Gadolinium-enhanced perfusion images and Tc-99m sestamibi SPECT images acquired during dipyridamole stress. The MR images were acquired with a multiecho spoiled GRE sequence (echo train length of four, TR = 6.2 msec, echo time = 1.2 msec) during the early phase after bolus injection. A well-defined perfusion defect (arrows) in the anteroseptal region is clearly seen on the MR images, which correlates closely with defects on the SPECT images. LV = left ventricle, RV = right ventricle. (Courtesy of Dara Kraitchman, VMD, PhD, Johns Hopkins University, Baltimore, Md.)

	Coronary MR Angiography

Top Abstract Introduction Technical Issues Measurement and Detection of... Contrast-enhanced Perfusion... Coronary MR Angiography Future Directions Conclusions References

Cardiac catheterization is the current standard of reference for assessment of the coronary arterial circulation; over 1 million diagnostic catheterizations are performed every year in the United States (1), at a cost of about $4,000–$5,000 per procedure. Of all patients undergoing coronary artery catheterization, 20% have normal coronary artery anatomy, whereas 30% are diagnosed with coronary artery disease but no intervention is pursued (56). In the remaining 50%, lesions are present and attempts at revascularization are made, either with percutaneous transluminal coronary angioplasty or coronary artery bypass grafting. This situation has important implications for potential revascularization therapy, which is very expensive and carries a high risk of morbidity and mortality (56). A screening examination that allows accurate assessment of coronary artery disease and myocardial viability would reduce the number of patients undergoing unnecessary catheterization and those undergoing unnecessary revascularization, greatly reducing morbidity and mortality. Such an examination could allow assessment of the outcomes of interventions and presents an opportunity for tremendous reductions in medical expenditures. (The cost of a typical MR imaging examination is $1,000.) Reduction of unnecessary procedures and accurate assessment of revascularization prognosis is dependent on knowledge of the status of the myocardium: Is the tissue in question salvageable, and, if so, will it recover function on revascularization?

Tremendous advances in MR coronary angiography during the past decade have demonstrated promising potential for noninvasive diagnosis of coronary artery disease. The in-plane image resolution of current coronary artery MR imaging techniques is about 1 mm, sufficient for detection of stenoses in large coronary arteries but inadequate for accurate detection of disease in smaller branches of the coronary vasculature. Accordingly, the future role of MR coronary angiography will likely be as an important adjunct to the comprehensive cardiac MR imaging examination, which combines angiography with stress tagging and perfusion imaging. Direct correlation of MR angiographic abnormalities with perfusion and wall motion abnormalities may prove to be a powerful combination in evaluation of coronary artery disease.

Figure 14 shows several maximum-intensity projection images from a high-resolution 3D spoiled GRE study of an ex vivo human heart, performed after injection of gadopentetate dimeglumine into the coronary vessels. Exquisite detail of the coronary anatomy is visualized, especially when it is rotated in three dimensions. These images offer insight into the best-possible coronary angiogram that MR imaging might produce in the future.

View larger version (36K): [in this window] [in a new window] [Download PPT slide]   Figure 14.   Maximum-intensity projection images of an ex vivo human heart after injection of gadopentetate dimeglumine into the coronary vasculature, viewed from above (left) and the side (middle and right). The heart was imaged with a high-resolution 3D spoiled GRE sequence at 1.5 T. (Courtesy of Guy Shechter, BS, Johns Hopkins University, Baltimore, Md.)

Two-dimensional Spoiled GRE Coronary Angiography
Standard 2D spoiled GRE imaging has been successfully used to image the coronary arteries by relying on the inflow of fresh spins into the imaging plane. The "2D time of flight" inflow phenomenon is used in many areas of MR angiography and relies on spins outside the imaging plane, not previously saturated from RF pulses. As these spins flow into the imaging plane, they generate high signal, which produces excellent vessel-myocardial contrast. By using multiple breath holds, several planes that span the heart are acquired during diastole, when coronary arterial flow is maximal (57–59). Increased tip angles, chemical presaturation of epicardial fat, and first-order flow compensation are often used to improve vessel-myocardial contrast.

Venous injection of gadopentetate dimeglumine prior to imaging will enhance the coronary vasculature by shortening the T1 of the blood pool, further increasing the vessel-myocardial contrast (60,61). Recently developed intravascular gadolinium contrast agents, such as MS-325, have reduced extravasation with high relaxivities. These agents are highly effective for shortening blood T1 and significantly improving vessel-myocardial contrast for both 2D and 3D MR coronary angiography (62). Figure 15 shows some examples of anomalous coronary arteries imaged with a 2D spoiled GRE sequence with high tip angle, first-order moment nulling, and intravenous gadolinium contrast material.

View larger version (143K): [in this window] [in a new window] [Download PPT slide]   Figure 15a.   Anomalous left main coronary artery (arrows) arising from the right coronary artery origin. The images were acquired with a 2D spoiled GRE sequence after intravenous injection of gadolinium contrast material. Ao = aorta, LV = left ventricle, PA = pulmonary artery.

View larger version (134K): [in this window] [in a new window] [Download PPT slide]   Figure 15b.   Anomalous left main coronary artery (arrows) arising from the right coronary artery origin. The images were acquired with a 2D spoiled GRE sequence after intravenous injection of gadolinium contrast material. Ao = aorta, LV = left ventricle, PA = pulmonary artery.

Spiral Coronary Artery Imaging
As described earlier, spiral imaging is well suited to coronary artery imaging because of its insensitivity to flow and motion artifacts and high data acquisition efficiency. Initial efforts toward coronary angiography with spiral imaging involved 2D cine acquisitions, in a manner similar to that described earlier (24). More recently, 3D spiral stack imaging has been applied to 3D spiral coronary imaging (29). Acquisition times exceed that allowed by breath holding, and respiratory compensation methods are necessary (29). The application of high-performance gradient systems with 3D spiral imaging may provide complete coronary angiograms within a breath hold. Preliminary efforts with "fluoroscopic" spiral coronary angiography have shown promise (30).

Vessel-myocardial contrast is enhanced in spiral imaging by means of spatial-spectral pulses, which eliminate signal from pericardial fat (24,25). In addition, T2 preparation with spiral imaging has been shown to be an effective method of suppressing myocardial and venous structures, improving the contrast between arteries and surrounding structures (64). Although not limited to spiral imaging, these methods are used most frequently in conjunction with this technique.

Figure 16 contains examples of spiral coronary angiograms obtained in healthy volunteers.

View larger version (116K): [in this window] [in a new window] [Download PPT slide]   Figure 16a.   Spiral coronary angiograms of the left anterior descending artery (LAD) and several of its diagonals (a) and the right coronary artery (RCA) (b). (Courtesy of Craig Meyer, PhD.)

View larger version (127K): [in this window] [in a new window] [Download PPT slide]   Figure 16b.   Spiral coronary angiograms of the left anterior descending artery (LAD) and several of its diagonals (a) and the right coronary artery (RCA) (b). (Courtesy of Craig Meyer, PhD.)

Three-dimensional imaging eliminates the problem of imaging plane misregistration and is better suited for imaging tortuous vascular structures. In addition, prescription of image volumes is simplified because an image volume containing the coronary arteries is easy to define. The SNR of 3D acquisitions increases by a factor of N_z (N_z is the number of depth-encoding steps), further enhancing image quality.

Unfortunately, an entire N_x x N_y x N_z 3D data set must be acquired free from the effects of breathing. Recent efforts have been made to acquire a 3D coronary image within a breath hold (61), although the imaging time is still too long for many patients. Novel approaches with directed double-oblique 3D volumes and multiecho spoiled GRE sequences have helped reduce total imaging time (65), but 3D data sets are generally beyond acquisition within a breath hold.

To overcome this problem, "navigator" echo techniques have been developed to compensate for respiratory motion (65–67). The principle behind most navigator techniques is to track the precise position of the diaphragm during respiration by using a pencil beam excitation and single readout. Lines of k space are accepted only when the position of the diaphragm falls within a narrow window. It has been shown that the position of the diaphragm correlates closely with the position of the heart during respiration (67). Thus, 3D data sets can be acquired during free breathing by accepting data only when the diaphragm is in the same position. In a typical examination, a navigated 3D acquisition accepts data about 20%–25% of the time. An alternative approach has been to provide feedback to patients with a light-emitting diode monitor, which allows the patient to adjust the position of the diaphragm (67). Navigator techniques are well tolerated by patients, particularly patients who find it difficult to hold their breath, even for short periods.

Figure 17 contains reformatted images from a 3D navigator acquisition of the right and left coronary arterial structures in a healthy volunteer (65). Figure 18 contains zoomed MR and conventional angiograms of the right coronary artery in a patient with known disease (65). Three stenotic regions are identified in both images.

View larger version (126K): [in this window] [in a new window] [Download PPT slide]   Figure 17a.   Three-dimensional reformatted images of the right and left coronary arteries in a healthy volunteer. The in-plane resolution is 0.7 x 1 mm. (a) Image shows the left circumflex artery (LCx), left main coronary artery (LM), right coronary artery (RCA), acute marginal artery into the right ventricle (RV), and sinus node (SN). (b) Image shows the aorta (Ao), first diagonal (D1), left anterior descending artery (LAD), left circumflex artery (LCx), left main coronary artery (LM), and right ventricular outflow tract (RVOT). (Courtesy of Matthias Stuber, PhD, Beth Israel Deaconess Medical Center, Boston, Mass; adapted and reprinted, with permission, from reference 65.)

View larger version (158K): [in this window] [in a new window] [Download PPT slide]   Figure 17b.   Three-dimensional reformatted images of the right and left coronary arteries in a healthy volunteer. The in-plane resolution is 0.7 x 1 mm. (a) Image shows the left circumflex artery (LCx), left main coronary artery (LM), right coronary artery (RCA), acute marginal artery into the right ventricle (RV), and sinus node (SN). (b) Image shows the aorta (Ao), first diagonal (D1), left anterior descending artery (LAD), left circumflex artery (LCx), left main coronary artery (LM), and right ventricular outflow tract (RVOT). (Courtesy of Matthias Stuber, PhD, Beth Israel Deaconess Medical Center, Boston, Mass; adapted and reprinted, with permission, from reference 65.)

View larger version (84K): [in this window] [in a new window] [Download PPT slide]   Figure 18a.   MR (a) and corresponding conventional (b) angiograms of the right coronary artery in a patient with known right coronary artery disease. Three stenotic areas are identified on both images (arrows). (Courtesy of Matthias Stuber, PhD; adapted and reprinted, with permission, from reference 65.)

View larger version (79K): [in this window] [in a new window] [Download PPT slide]   Figure 18b.   MR (a) and corresponding conventional (b) angiograms of the right coronary artery in a patient with known right coronary artery disease. Three stenotic areas are identified on both images (arrows). (Courtesy of Matthias Stuber, PhD; adapted and reprinted, with permission, from reference 65.)

	Future Directions

Top Abstract Introduction Technical Issues Measurement and Detection of... Contrast-enhanced Perfusion... Coronary MR Angiography