Multiplane Magnetic Resonance Imaging of the Heart and Major Vessels:

661 Charles B. Higgins1 David Stark Michael McNamara Peter Lanzer Lawrence E. Crooks Leon Kaufman Received October 25, 1983; accepted after revision January 5, 1984. This work was supported in part by Diasonics, Inc.; California Heart Association grant CA 82- Ni 21; and U.S. Public Health Service contract HV 02928 from the National Heart, Lung and Blood Institute. Dr. Lanzer is supported by grant 464/i 2-1 from the Deutsche Forschungsgemeinschaft. All authors: Department of Radiology, University of California School of Medicine, San Francisco, CA 94143. Address reprint requests to C. B. Higgins. AJR 142:661-667, April 1984 0361-803X/84/1 424-0661 American Roentgen Ray Society Multiplane Magnetic Resonance Imaging of the Heart and Major Vessels: Studies in Normal Volunteers The feasibility of magnetic resonance imaging for defining anatomy of internal cardiac structures and major blood vessels was assessed in 14 normal subjects. Both electrocardiogram-gated and standard spin-echo images were obtained. The R-R interval determined the pulse repetition times in gated sequences. Gated images provided better visualization of internal cardiac morphology and of upper mediastinal vessels than did nongated images. Trabecular detail and components of the mitral valve could be resolved. All segments of the left ventricular wall could be evaluated by combining axial, coronal, and sagittal images. Gated acquisition of magnetic resonance images did not increase imaging time; five transverse slices of the left ventricle were obtained in 6.0-8.5 mm. The good image quality, ease of gated acquisition, large field of view, capability of direct imaging in multiple planes, and noninvasiveness of the technique suggest that it will be an important imaging method in cardiovascular disease. Magnetic resonance imaging (MRI) has inherent advantages for evaluating the heart and blood vessels. These include its noninvasive nature, excellent contrast between flowing blood and cardiovascular structures [1-3], and the ability to produce direct images in any plane. The time required to obtain sufficient data to generate MR images encompasses several minutes and hundreds of heartbeats. The continuous motion of the heart results in inadequate signal intensity and blurring of the edges of moving structures such that diagnostic images of the heart generally are not obtained. But with synchronization of each radiofrequency pulse sequence to a specific segment of the cardiac cycle, more informative images are obtained [4-6]. Our initial experience with gated MRI of the heart and great vessels indicates that this noninvasive technique can provide very clear visualization of internal cardiac morphology. By demonstrating this capability, we hope to stimulate its use in evaluating a variety of cardiovascular diseases. Subjects and Methods Gated MR images were obtained in 14 normal volunteers after informed consent was given. Subjects were 26-46 years of age; 12 were men. The study was approved by the institution s committee on human research. Since the volunteers were not imaged by other methods, no attempt was made to compare MRI with other techniques. The imager used (Diasonics) had a superconducting magnet operating at 3.5 kg (0.35 T), corresponding to a hydrogen resonance frequency of 15 MHz. This imager has been described in detail [7, 8]. The images were obtained using spin-echo (SE) pulse sequences at time delays (TE5) of 28 and 56 msec between the application of the initial radiofrequency (RF) pulse and the corresponding refocusing SE signal. At each anatomic level, two images were generated; one was formed from the first spin echo (TE = 28 msec) and the other from the second echo (TE = 56 msec). A gating signal initiated by the A wave of the electrocardiogram was obtained from a device specially designed and constructed for safe use in high electromagnetic fields and

662 HIGGINS ET AL. AJR:142, April1984 Fig. 1-Comparison of nongated (upper) and electrocardiogram-gated (lower) images at two transverse levels through ventricles. Internal structure of ventricles is not Jiscernible on nongated images. Gated images show parts of mitral valve (straight arrow), tricuspid valve or anulus (curved arrow), and trabecular pattern of both ventricles. rapidly changing RF pulses. The instrumentation is described in a separate report [4j. After skin preparation, three electrodes were attached to the left arm and both legs or, altematively, in the right and left subclavicular and right abdominal regions. Inscription of a deflection at the time of each AF pulse on the baseline of the electrocardiographic tracing permitted exact localization of the specific time of gating. The RF and gradient pulse were prevented from activating the gating module by a suppression circuit, which kept the module from accepting any signal for 500 msec after the inscription of the A wave. The gating signal was transmitted to a triggering module, which is sensitive to the rising or falling edge of an amplitude change in the input signal and produces a square wave-pulse output to start the MAI sequence. The triggering module further permits the operator to choose a time delay between the gating signal and the start of the imaging sequence. The repetition time (TA), determined by the subject s heart rate, varied from about 0.6 to 1.0 sec (60-100 beats/mm). The TA was equal to or double the A-A interval of the electrocardiogram. In order to form images, AF and gradient magnetic field pulses were applied in 512 cardiac cycles. Total imaging time, a product of repetition time, the number of lines along the y axis (vertical dimension of the matrix), and number of times the signal was averaged, was about 5-10 mm. Using the multisection imaging technique, five adjacent 7-mm-thick slices were imaged by fitting five sections in the repetition interval, withal 00-msec delay for each subsequently irradiated section. Thus, the imaging period for all five sections extended over about 500 msec. (With this technique, each slice is out of phase by 100 msec with the adjacent slice.) The MA images were displayed in a matrix of 128 vertical x 256 horizontal picture elements in black and white, using 256 shades of gray; the brightest areas represented tissues with the highest MA signal. Each imaging plane had a section thickness (z axis) of 7 mm with a 2.5-mm space between adjacent sectors. The horizontal spatial resolution was 1.7 mm in the imaging plane. Results Images of the Heart Nongated images of the thorax generally show such low signal from the beating heart that its anatomic features cannot be discerned (figs. 1 and 2). Without gating, each sequence is initiated randomly during the cardiac cycle; consequently, points on the cardiac wall occupy different positions in the image matrix from one sequence to the next. This results in the appearance of signal within the cardiac chambers and indistinct luminal margins. On the other hand, the electrocardiogram-gated MR images discretely demonstrate internal cardiac morphology, including the trabecular patterns of the right and left ventricles (figs. 1 and 2). This is illustrated by the visualization of the moderator band of the right ventricle (fig. 1). Similarly, chordal structures, parts of the leaflets of the mitral valve, and papillary muscle structures are demonstrated within the left ventricle (figs. 1-3). In normal patients the pencardium is represented by a thin rim of low signal intensity over the ventral aspect of the heart (fig. 3). Images with sharper margins and better delineation of the cardiac walls are obtained from the first spin echo (TE = 28 msec) than from the second spin echo (TE = 56 msec). This is probably a consequence of three factors: (1) The second SE sequence has a longer gating window; the time from the initiating pulse to the refocusing pulse is 56 msec, which permits more time for a point in the cardiac wall to have moved before the completion of the imaging sequence; (2) the signal-to-noise ratio is less for images formed from the second SE signal [8]; and (3) signal from flowing blood is

AJR:142, April 1984 MRI OF HEART AND MAJOR VESSELS 663 Fig. 2.-Comparison of nongated (upper) and electrocardiogram-gated (lower) images at two transverse levels through caudal aspect of ventricles. Juncture of coronary sinus (curved arrow) and inferior vena cava (straight arrow) with right atrium. Fig. 3.-Optimizing TR and TE values for imaging heart. Gated images obtained at same axial level through ventricles and representing two TR and two TE values. Upper images were obtained by gating to every heartbeat (TR = 0.5), lower images by gating to every second beat (TR = 1.0). First SE images (TE = 28 msec, left) as compared with second SE images (TE = 56 msec, right) contain less signal within chamber and better delineation of endocardial interface. Signal produced by myocardium is greatest with longer TA value (every second beat). more likely to generate signal on the second SE image, which would tend to diminish the sharp interface between the myocardial wall and blood within the cardiac chamber. The second SE image (TE = 56 msec) in figure 3 shows partial filling of the ventricular chamber, with high-intensity signals resulting in poorer distinction of the endocardial interface of the posterior and lateral segments of the left ventricle. Images have been obtained gated to every heartbeart (TR = A-A interval) and to every second heartbeat (TA = A-A interval x 2). Because the time in which protons can regain magnetization between AF sequences is doubled with gating to every second beat, signal intensity of the myocardium is greater in the latter than in images gated to every heartbeat (fig. 3). To calculate the Ti (longitudinal) relaxation time of the myocardium it is necessary to perform data acquisition at the same anatomic level using two different TA values. Usually

664 HIGGINS ET AL. AJR:142, April 1984 Fig. 4.-Optimizing imaging technique for blood vessels. Nongated (upper) and gated (lower) images of upper mediastinum. First SE images (TE = 28 msec) (left); second SE images (FE = 56 msec) (right). Great vessels are more clearly defined on gated images. Signal-to-noise ratio for vascular structures is greatest for gated first SE image (lower left). we do this by gating initially to every heartbeat, then to every second beat (fig. 3). Images of the Great Vessels As was observed for the heart, there is also clearer discrimination of mediastinal vessels in gated MR images (fig. 4). In Fig. 5.-Pulmonary artery anatomy. Gated images at two adjacent transverse levels of pulmonary bifurcation (above) and two levels containing pulmonary veins (below). nongated images the margins of vessel walls are less sharply defined and there is extraneous signal within the vascular lumina and adjacent to the great vessels (fig. 4). Presumably, the edge-blurring and extraneous signal result from the pulsation and consequent inconstant position of the vascular walls during the cardiac cycle, and perhaps from the excitation of stationary blood when the imaging sequence falls in diastole during the several-minute imaging period.

AJR:142, April 1984 MRI OF HEART AND MAJOR VESSELS 665 Fig. 6.-Mediastinal vessels. Gated images at Fig. 7.-Coronary arteries. Axial images. A, Right coronary artery (arrowhead) in right atrioventricular level of bifurcation of pulmonary artery, obtained in groove in region of left ventricular outflow tract. B, Proximal part of left anterior descending coronary artery diastole. Substantial signal in all upper mediastinal and a diagonal branch (arrowheads) at base of aorta. vessels; signal intensities are particularly high in superior vena cava (curved arrow) and descending aorta (straight arrow). Fig. 8.-Sagittal images of left ventricle. Gated images extend from lateral (left upper) to medial (right lower) aspect of left ventricle and provide visualization of its posterior and diaphragmatic walls. Signal intensity within the lumina of mediastinal vessels vanes greatly depending on the phase of the cardiac cycle used for gating. Images corresponding to systole generally have no intraluminal signal (figs. 4 and 5), whereas images acquired in diastole show intense intraluminal signal (fig. 6). In gated images the central pulmonary arteries and veins connected to the left atrium are clearly demonstrated (fig. 5). The proximal parts of the coronary arteries can be discemed in most patients, especially when the image volume is centered at the base of the heart. The right coronary artery can be visualized in the right atrioventricular groove in most subjects (fig. 7). In this position the right coronary artery is particularly conspicuous because of the high contrast differential within the lumen of the vessel, where flowing blood produces little or no signal, and the surrounding bright signal produced by fat in the groove. Less often, the origin of the coronary arteries from the aortic sinuses and the proximal parts of the left anterior descending coronary arteries are visualized (fig. 7). Sagittal and Coronal Images Sagittal and coronal images of the heart display cardiac structures in a manner different from most other imaging methods. Sagittal images of the left ventricle demonstrate the full length of the inferior and posterior segments and most of the anteroseptal segment (fig. 8). Coronal images display the size of the cardiac chambers, the base of the aorta, and permit visualization of the full length of the diaphragmatic segment of the left ventricle (fig. 9). In contrast to computed tomography (CT), in which sagittal and coronal images are

666 HIGGINS ET AL. AJR:142, April1984 Fig. 9.-Coronal image of heart and ascending aorta. Visualization of entire length of diaphragmatic. septal, and lateral segments of left ventricle. Sinuses of Valsalva and ascending aorta. reconstructed from a series of axial scans, MA images are direct images; thus, there is no loss of image quality or spatial resolution. Sagittal images also depict the entire length of the thoracic aorta; the smooth anterior and posterior walls of the normal aorta are clearly demonstrated (fig. 10). Discussion MRI distinguishes clearly between flowing blood and the vascular wall because blood flowing at normal velocities produces a low-intensity or no MA signal [1-3]. The natural contrast between flowing blood and the walls of the cardiovascular system have proven advantageous for defining cardiovascular lesions such as mural thrombus and wall thinning at the site of myocardial infarctions [1 ], thoracic aortic aneurysms and dissections [1 ], and atherosclerotic plaques [1]. Cardiac anatomy can be imaged in sharp detail using MAI with electrocardiographic gating. The extraneous signals within the lumina and edge-blurring of the heart and great vessels produced by nongated images are avoided on gated images. It is likely that these distracting features on nongated images of the thorax result from motion of the walls of the heart and great vessels during the cadiac cycle and the reception of signal from stationary blood when the imaging sequence falls during diastole in the case of the great vessels and during isovolumic systole or isovolumic diastole in the case of the ventricles. Transverse images generally have been most useful for defining internal cardiac morphology. However, coronal and sagittal images better demonstrate the parts of the heart that are oriented parallel to the transverse plane, such as the diaphragmatic (inferior) wall of the left ventricle. This aspect of the ventricle generally has been demonstrated inadequately by CT [9]. Our images also show the potential of multiplane imaging for the thoracic aorta and its branches. This format is similar to a lateral aortogram and displays the contour of the anterior Fig. 10.-Sagittal image. Anterior and posterior walls of aortic arch and descending aorta; main pulmonary artery (arrow). and posterior walls of the aorta. Sagittal images of the thoracic aorta should be useful for demonstrating coarctations and interruption of the aorta, aneurysms, and dissections. Even at the current spatial resolution, the coronary arteries are clearly visualized on MA images. When the image volume is centered at the base of the heart, the proximal parts of the coronary arteries can be discerned. However, diagnostically useful visualization of the coronary arteries will require threedimensional imaging and improvement of spatial resolution to the 0.5 mm range. The advantages of MRI over CT for cardiac imaging are quite apparent: Contrast medium is not required; direct rather than reformatted images can be obtained in the transverse, sagittal, and coronal planes; and gating can be applied more easily. For MRI, the electrocardiogram is used to define the interpulse delay (TA). At normal resting heart rates, the A-A interval is less than 1 sec, and the TA used for nongated imaging of the body is usually 1-2 sec. Therefore, the imaging time required for gated MR images, unlike that associated with gated CT scans, is no greater than that needed for nongated images. Using a multislice technique, most of the heart can be imaged in one or two sequences. Consequently, imaging of the entire heart is achieved in about 6-1 5 mm. Uke echocardiography, gated MRI is noninvasive and does not involve ionizing radiation. Echocardiography also provides images in multiple planes, but the area encompassed on MRI is greater than that obtained with sector-scan echocardiography. Therefore, orientation and interpretation of cardiac images relative to surrounding structures may be better with gated MRI. On the other hand, echocardiographic equipment is less costly and more portable than MRI equipment, and the rapid sampling frequency of echocardiography is ideal for evaluating cardiac valves. An anticipated advantage of MRI is the ability to provide better tissue characterization than has been attained with sonography and other imaging techniques. Early reports show that MAI can differentiate ischemically damaged from normal myocardium by magnetic relaxation

AJR:142, April 1984 MRI OF HEART AND MAJOR VESSELS 667 times [1 0, 1 1 ]. An additional advantage of MAI is its ability to image atherosclerotic plaque directly [12]. Gated MRI cannot be performed in every patient. Adequate gating cannot be achieved in the presence of marked arrhythmias, especially atrial fibrillation and frequent premature yentricular contractions. However, a rapid heart rate does not preclude satisfactory gated acquisition of MA images, since the option exists to gate to every second or third heartbeat. Gating to every second or third A wave results in a longer TA interval, which permits a longer time for recovery of magnetization between pulses, thereby creating greater signal intensity from the heart than would otherwise be possible in the presence of tachycardia. AEFERENCES 1. Herfkens RJ, Higgins GB, Hricak H, et al. Nuclear magnetic resonance imaging of the cardiovascular system. Normal and pathologic findings. Radiology 1983;l 47:747-759 2. Kaufman L, Crooks LE, Sheldon PE. The potential impact of nuclear magnetic resonance imaging on cardiovascular diagnosis. Circulation 1983;67:251-259 3. Kaufman L, Crooks LE, Sheldon PE, Aowan W, Miller T. Evaluation of NMA imaging for detection and quantification of obstruction in vessels, Invest Radiol 1982;l7:554-560 4. Lanzer P, Botvinick E, Kaufman L, Davis P, Lipton MJ, Higgins GB. Cardiac imaging using gated NMR. Radiology 1984 (in press) 5. Alfidi RJ, Haaga JR, El Yousef SF, et al. Preliminary experimental results in humans and animals with superconducting, whole body, nuclear magnetic resonance scanner. Radiology 1982; 143:175-1 81 6. Pohost GM, Goldman MA, Pykett IL, et al. Gated NMA imaging in canine myocardial infarction. Circulation 1982 [Suppl];66:539 7. Crooks LE, Arakawa M, Hoenninger J, et al. Nuclear magnetic resonance whole body imager operating at 3.5 kgauss. Radiology 1982;143:169-174 8. Crooks LE, Mills CM, Davis PL, et al. Visualization of cerebral and vascular abnormalities by NMA imaging. The effect of imaging parameters on contrast. Radiology 1982;144:843-852 9. Lipton MJ, Brundage B. Evaluation of ischemic heart disease in man by CT. In: Higgins GB, ed. CT of the heart and great vessels. New York: Futura, 1983:255 10. Higgins GB, Herfkens A, Lipton MJ, et al. Nuclear magnetic resonance imaging of acute myocardial infarction in dogs: alterations in magnetic relaxation times. Am J Cardiol 1983;52: 184-188 11. Williams ES, Kaplan JI, Thatcher F, Zimmerman G, Knobel SB. Prolongation of spin-lattice relaxation times in regionally ischemic tissue from dog hearts. J Nucl Med 1980;21 :449-453 12. Herfkens AK, Higgins GB, Hricak H, et al. Nuclear magnetic resonance imaging of atherosclerotic disease. Radiology 1983;148:161-166