Analysis of Residual Coronary Artery Motion for Breath Hold and Navigator Approaches Using Real-Time Coronary MRI

Transcription

1 Magnetic Resonance in Medicine 55: (2006) Analysis of Residual Coronary Artery Motion for Breath Hold and Navigator Approaches Using Real-Time Coronary MRI R. W. Fischer, 1,2 R. M. Botnar, 1,3 K. Nehrke, 4 P. Boesiger, 2 W. J. Manning, 1 and D. C. Peters 1 * Coronary artery MRI methods utilize breath holds, or diaphragmatic navigators, to compensate for respiratory motion. To increase image quality and navigator (NAV) gating efficiency, slice tracking is used, with more sophisticated affine motion models recently introduced. This study assesses the extent of remaining coronary artery motion in free breathing NAV and single and multi breath hold coronary artery MRI. Additionally, the effect of the NAV gating window size was examined. To visualize and measure the respiratory induced motion, an image containing a coronary artery cross section was acquired at each heartbeat. The amount of residual coronary artery displacement was used as a direct measure for the performance of the respiratory motion correction method. Free breathing studies with motion compensation (slice tracking with 5 mm gating window) had a similar amount of residual motion ( mm) as a single breath hold ( mm) and were superior to multiple breath holds ( mm). Affine NAV methods allowed for larger gating windows ( 10 mm windows) with similar residual motion ( mm). In this healthy adult cohort (N 10), free-breathing NAV methods offered respiratory motion suppression similar to a single breath hold. Magn Reson Med 55: , Wiley-Liss, Inc. Key words: cardiac MR; navigators; respiratory motion; coronary MRI; real-time imaging Suppression of respiratory motion artifacts is one of the major limiting factors in coronary MRI. Two primary approaches are used to suppress respiratory motion: breath hold imaging (1) and free breathing real-time navigator (NAV) gated imaging (2). Repeated breath holds have the disadvantages of limited scan time, need for patient compliance (3), diaphragmatic drift (4), and slice mis-registration for 3D or 2D contiguous slices acquired in consecutive breath holds. Navigator sequences use either intersecting planes excited by a 90 and 180 RF pulse (5) or a 2D selective pencil-beam navigator RF pulse (6). Navigators are commonly placed on the dome of the right hemidiaphragm to monitor respiratory motion. The position of the diaphragm is used to decide in real-time if the interface is within a pre-specified window, and thus whether data are accepted or rejected (7). The rejection of unwanted positions leads to decreased scan efficiency (% of accepted positions) and prolonged scan times. If a simple gating strategy is applied, only small gating windows can be used, frequently decreasing the scan efficiency to 30% or less. To allow for larger gating windows and thus higher NAV scan efficiencies, different models for correction of the respiratory motion have been developed. Since the dominant respiratory induced cardiac motion is in the superior-inferior (SI) direction (8), slicetracking implementations, i.e., 1-dimensional correction methods, were introduced to facilitate larger gating windows (9). Recently, more sophisticated models have been proposed, which use patient specific 3D translations or affine transformations to compensate for respiratory motion (10). The effectiveness of these methods for coronary MRI has been assessed by acquiring 3D coronary MRI volumes with different respiratory motion compensation methods and comparing subjective and objective image quality, e.g., for slice tracking (11), navigator location (12), navigator timing (13), NAV window size (9), and affine motion models (10). However, the residual respiratory motion has not been directly reported as a metric for comparing motion compensation methods. Furthermore, since only comparisons between methods have been reported, the baseline coronary artery motion of each method remains unknown. In this study we sought to directly visualize and to quantify residual coronary artery motion due to respiration after use of prospective motion compensation techniques. In each heartbeat the coronary arteries were imaged in a 150 ms acquisition window, during the subject-specific cardiac rest period. The displacement of the coronary arteries between images due to residual respiratory motion after correction was used to measure the quality of the motion correction. 1 Cardiovascular Division, Beth Israel Deaconess Medical Center, Boston, MA, USA. 2 ETH and University Zurich, Zurich, Switzerland. 3 Technical University Munich, Munich, Germany. MATERIALS AND METHODS 4 Philips Research Laboratories, Hamburg, Germany Grant Sponsor: American Heart Association; Grant Number: AHA Studies were performed on ten healthy adult subjects (6 SDG N. female, 4 male, age years) on a whole-body 1.5-T * Correspondence to: Dana C. Peters, Ph.D., Cardiac MR Center, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, MA, 02215, USA. Gyroscan ACS NT MR Scanner (Philips Medical Systems, dcpeters@bidmc.harvard.edu Best NL) equipped with a Power Trak 6000 gradient Received 24 May 2005; revised 10 November 2005; accepted 11 November (23mT/m, 220 us rise time) and a commercial five-element cardiac phased array coil. Written informed consent was DOI /mrm Published online 1 February 2006 in Wiley InterScience ( obtained from all participants in accordance with our Institutional Review Board. wiley.com) Wiley-Liss, Inc. 612

2 Analysis of Residual Respiratory Coronary Artery Motion 613 FIG. 1. The planning of the scan plane is shown. (a) illustrates how the plane was chosen in order to be orthogonal to the proximal and distal right coronary artery (Ao Aorta, LV left ventricle). (b) shows the plane in a coronal view and (c) is a representative real-time image acquired in a single heartbeat. The coronary artery is cut twice, once proximal superior within the heart, and once distal inferior within the heart. The small windows show magnifications of cross sections of the coronary arteries, which were used for motion analysis. MR Sequence The acquisition was vector ECG-triggered (14) with a patient specific trigger delay (15). The sequence consisted of a 50ms T2prep pulse (16) to enhance contrast, the navigator sequence, a spectrally selective fat saturation pulse, and finally the imaging sequence, which was a 2-dimensional undersampled radial balanced steady-state free precession (SSFP) sequence. The following sequence parameters were used: radial SSFP: 90 ; number of projections 48; 192 readout points; BW 1040 Hz/pixel; TR 3.2ms; FOV mm; measured pixel size mm; reconstructed pixel size mm; slice thickness 8 mm; cardiac acquisition window T AQ 152 ms. SSFP catalyzation was performed with 5 dummy RF pulses using a linear ramp (17). Imaging Plane The 2-dimensional images permit assessment of displacement in 2 dimensions. The orientation of the imaging plane determines which components of the displacement can be measured. As the dominant respiratory induced cardiac motion is in the SI direction, 1 dimension of the plane was aligned in this direction. Motion in the anteriorposterior (AP) and the right-left (RL) dimension are of reduced but similar magnitude (18). The second direction of the plane was thus not chosen in a predefined manner, but was aligned to match subject anatomy (a single oblique scan plane). The plane was chosen, as closely as possible, to be orthogonal to the main axis of the coronary artery so as to minimize the effect of through plane motion. The displacement of the coronary arteries due to respiration is dependent on the location of the coronary artery; the base of the heart moves differently than the apex (8). To obtain information about the influence of the location of the coronary artery, the imaging slice was planned to intersect the coronary arterial tree in 2 positions, proximally (further from the diaphragm) and distally (close to the diaphragm). The right coronary artery (RCA) was always targeted. For one subject the left anterior descending artery (LAD) was targeted, because their anatomy prevented targeting of the RCA. Figure 1 shows the planning of the 2D real-time coronary artery slice. To assure that the scan plane was angulated so that 1 axis was aligned with the SI direction, and that a cross-section of the coronaries was imaged, the scan prescription was performed using a slice from a 3D coronary MRI data-set (19) (to visualize the coronaries) and simultaneously a coronal localizer view (to assure scan plane alignment with SI); the angulation was further confirmed by viewing the rotation matrix. A representative real-time image displaying 2 coronary artery intersections is shown in Fig. 1c). Note in Fig. 1a that the imaging plane is not precisely orthogonal to the coronary artery; this is due to the constraint that 1 encoding direction be purely SI. Additional real-time images along the main axis of the RCA (with doubly oblique scan planes) were also acquired to provide simultaneous visual assessment of proximal, mid, and distal coronary artery motion (Fig. 2). Respiratory Compensation Motion Models The displacement of the coronary artery was analyzed within multiple breath holds and also in a single breath hold. Twelve successive images were acquired within a single breath hold, and 5 consecutive breath holds were performed. This allowed comparison of the performance of a multi breath hold approach to provide increased scan time similar to free breathing approaches. For free-breathing approaches, images were obtained over 150 consecutive heartbeats. Coronary artery displace- FIG. 2. Series of consecutive real-time coronary MRI images, each obtained within a single heartbeat. Respiratory motion leads to displacement of the heart. The structures of the heart are displaced relative to the white horizontal lines. RV right ventricle, RCA right coronary artery, LAD left anterior descending artery.

3 614 Fischer et al. ment was measured and compared for 6 different methods of respiratory motion compensation: Model 1. Within 5 breath holds lasting 12 heartbeats each, without any motion compensation. Model 2. Within 5 breath holds lasting 12 heartbeats each, with navigator and slice tracking (0.6 SI, see Model 4 for description). Model 3. Free breathing, without any respiratory motion compensation. Model 4. Free breathing, with navigator slice tracking. For slice tracking, the entire slice prescription was shifted along the SI dimension by a distance of 0.6 times the SI displacement of the diaphragm in realtime, as estimated by the NAV (12). Model 5. Free breathing, with patient specific 2D translation tracking (see below). Model 6. Free breathing with a 2D affine motion model with patient specific parameters (see below). Patient Specific Models Our scanner is equipped with commercial software providing NAV and slice tracking with NAVs (motion Models 2, 3, and 4). To employ motion Models 5 and 6 for prospective motion correction, customized research software was used on our scanner (10). A calibration scan preceding the actual scan determined the dependency between the measured navigator positions and the heart motion. Model 5 used a simple translation to model the heart motion; Model 6 used an affine transformation. The calibration for both Models 5 and 6 was performed using the same imaging plane as for the real-time coronary artery imaging. A second navigator on the diaphragm preceding the first by 300 ms was used when calibrating the parameters, allowing for additional information about the respiratory state. Twenty dynamics were used for the calibration. The most end-expiratory image was used as a reference image. For every image Model 5 determined the translation, which leads to an optimal match with the reference image. In a second step the linear dependency between the optimal translations and the NAV-positions was determined. Model 6 is the same except that an affine transformation is used to match the images instead of a translation. The resulting calibrated motion model was used in the actual scan for prospective correction of translatory or affine motion, respectively. Displacement Measurement of the Coronary Arteries To identify the coronary artery, a region of interest (ROI) was placed around the cross-section of the artery. Subsequent images were shifted in both dimensions by multiples of a fourth of a pixel; this shift was applied by adding a linear phase in the Fourier domain. After shifting the image, the image correlation with the reference ROI was calculated: R a i *b i 2 [1] a 2 2 i * b i where a i pixel i of the reference ROI, b i pixel i of the shifted image ROI (i runs through all the pixels of the ROI), and R resulting image correlation. The shift that resulted in the highest correlation was assumed to be the displacement of the coronary artery. All processing was performed using MATLAB 6.5 (Mathworks, Inc. Natick, MA, USA). To measure the amount of motion remaining within an experiment, all measured positions of the coronary arteries were averaged and the mean magnitude of the deviation from this average was calculated: r i r mean i x [2] N Gating Window To minimize respiratory motion, navigator methods only accept data if the diaphragm position is within a certain range (gating window). The free breathing part of the study (Models 3 6) examined the influence of this gating window size on the amount of remaining motion. During the acquisition, images were acquired for all navigator positions. All navigator positions were logged and matched to each image. The highest of all measured diaphragm positions was used as the reference position. To analyze the amount of motion within a specific gating window, only the images for which the navigator was within the specified gating window were used. Statistical Methods All measurements are presented as mean SD. Measurements of mean deviation with each respiratory motion compensation technique were compared with a (1-way between groups) parametric analysis of variance (ANOVA) using the Tukey test for pair-wise comparisons to correct for multiple testing, with significance chosen as P RESULTS Cross-sectional images (Fig. 1c) of the proximal and distal RCA allowed for tracking of residual coronary artery motion. Figure 2 shows consecutive real-time right coronary artery MRI images with a large segment of the RCA, demonstrating proximal, mid, and distal coronary artery motion during 1 respiratory cycle. These images, thus, were useful in visualizing the respiratory induced motion extent after each compensation technique. They also allowed visualization of the effect of other sources of motion, e.g., arrhythmias (20). Two-dimensional coronary artery displacement is shown for 1 subject for (a) free-breathing with no NAV gating, (b) NAVs with affine correction model and a 10 mm gating window, and (c) a 12 heartbeat breath hold in Fig. 3. Each point represents a measured displacement. A marker indicates the average position of the coronary artery, and the mean deviation from the average displacement is shown as the radius of the circle. Breath holds Figure 4 shows the diaphragmatic displacement measured by the NAV for an excellent, average, and poor breath-

4 Analysis of Residual Respiratory Coronary Artery Motion 615 FIG. 3. The plots show all measured positions of a coronary artery in 3 experiments from 1 subject. The black dot is the mean position of all measurements; the mean deviation corresponds to the radius of the circle. The results of 3 different motion correction methods for the same subject are illustrated. (a) shows the displacement of the coronary artery during a free breathing experiment; no gating window and no motion correction method was applied; the mean deviation was 2.22 mm. (b) shows the displacement if the affine motion correction method was used with a 10 mm gating window; the mean deviation decreased to 0.53 mm. (c) shows the 12 positions measured during a single breath hold. The line connects the positions measured in consecutive heartbeats. Fewer positions can be seen because some overlap. The mean deviation for this breath hold was 0.91 mm. holder. Pooled breath hold data for coronary artery displacement are presented in Table 1. The mean deviation was much smaller within a single breath hold ( mm) than within all 5 breath holds ( mm). Navigator slice tracking led to a significant decrease of the deviation if all breath holds were examined ( mm). There was, however, no improvement using slice tracking for motion for a single breath hold ( mm). FIG. 4. The position of the diaphragm for 5 successive breath holds is shown for 3 subjects. The vertical lines mark the end of each breath hold. There are differences between subjects. Subject 1 has minimal diaphragm motion during each breath hold and small displacements between breath holds. Subject 2 has minor diaphragmatic motion during a breath hold, but large differences between breath holds. Subject 3 has large diaphragmatic displacement both within and between breath holds. Free Breathing The basal and apical motion of the heart due to respiration is not uniform. The data were, therefore, divided into 2 groups, one containing all proximal coronary artery data and another for distal coronary artery data (see Fig. 1). Figure 5a shows results for proximal coronary artery data, plotting mean deviation for all subjects against navigator acceptance window size. Average measured navigator efficiency is also shown. Figure 5b shows that for free breathing the mean deviation is larger for the distal coronary artery than for the proximal coronary artery. This effect is partially compensated for by the correction methods. Overall, the NAV methods are effective; they can reduce the amount of motion remaining. Inspection of Fig. 5b shows that a simple gating strategy with a 3 mm window decreases the residual motion compared to imaging over the entire respiratory cycle from mm to mm. For 3 mm gating windows, all methods performed equally well. As the size of the gating window increased, the amount of residual motion increased with all methods. The correction methods were useful, resulting in slower increase of the mean deviation. Combining data of both coronary artery locations for a typical gating window of 5 mm results in residual motion of mm with gating alone, mm with slice tracking, mm with patient specific translation, and mm with patient specific affine correction. The difference between the methods becomes more evident as the gating window is increased. The mean deviation shows the smallest increase for the affine correction method. Table 2 displays at what window size the improvement becomes statistically significant. Applying the affine motion correction method allows increasing the gating window to 10 mm with similar residual motion ( mm) as the use of a5mmwindow with slice tracking ( mm). The acquired data during free breathing allowed calculating the tracking factor of the coronary artery. Figure 6 Table 1 Residual Motion During Breath Holds Single breath hold 5 breath holds No tracking mm mm Tracking mm mm

5 616 Fischer et al. FIG. 5. The mean deviation of the coronary artery segment (proximal, distal) measured during free breathing experiments is shown (pooled data N 10). (a) shows the group of proximal coronary arteries, located superior within the heart; (b) shows the group of distal coronary arteries, located inferior within the heart. In all cases an increase of the gating window size led to an increase of the mean deviation; however, the increase was smallest for the affine method. FB free breathing. shows the SI displacement of the coronary artery as a function of the navigator position. The hysteresis of the curve is caused by differences between inspiration and expiration (21). The slope of the linear regression is the ideal tracking factor for this subject and location. Figure 7 displays how this tracking factor varies between subjects and coronary artery locations. The tracking factor at the proximal coronary artery is in all cases smaller ( ) than the tracking factor of the distal coronary artery ( ). Differences Between Subjects The effectiveness of certain motion models is also strongly subject dependent. Figure 8 displays the mean displacement of the proximal coronary artery segment under 4 of the motion models for individual subjects; the method with the least motion varies among patients. DISCUSSION In this study, we sought to quantify residual proximal and distal coronary artery motion using 6 different motion suppression methods. We found that quantifiable residual respiratory motion persists with all methods. Free breathing NAV correction methods with small gating windows reach a similar degree of accuracy in motion suppression ( 0.6 mm residual motion) as single breath holds ( 0.5 mm residual motion). This value is high compared to the minimal respiratory motion encountered, which could be as low as 0.2 mm with the best breath hold. This also shows that the accuracy of the cross-correlation method is reasonable, and close to its nominal accuracy ( 1/4 pixel or 0.35 mm). Since these low values can be obtained it is reasonable to assume that higher values of residual motion are due to motion and not to inaccuracy of the measurement method. Measurement and inclusion of motion in the third dimension would increase this estimate however, only slightly, since SI motion is predominant. Furthermore, the maximal deviation is much greater than the mean deviation. For example, if the displacement of the coronary artery is equally distributed within 2.4 mm, the mean deviation results to be 0.6 mm. Compared to the size of a Table 2 Results of ANOVA Test for Free-Breathing Methods of Fig. 5, Providing the Gating Window Size for Which Differences Are Statistically Significant Proximal No correction Tracking Translation Tracking 9mm Translation 9mm NS Affine 9mm 11 mm NS Distal No correction Tracking Translation Tracking 7mm Translation 7mm NS Affine 5mm 9mm NS NS not statistically significant. FIG. 6. The position of the distal coronary artery displacement versus the displacement of the diaphragm is shown for 1 subject. The slope of the linear regression (0.65) is the ideal tracking factor. The preceding and following navigator positions were used to decide if the subject was inspiring or expiring. The hysteresis of the curve shows that there is a difference between inspiration and expiration.

6 Analysis of Residual Respiratory Coronary Artery Motion 617 FIG. 7. The ideal tracking factors for all 10 subjects for proximal and distal coronary artery locations is shown. The tracking factor for the coronary arteries located distal is larger than the tracking factor of the coronary arteries located proximal. The tracking factor varies depending on the subject. The mean tracking factor 1 SD is also shown. coronary artery (3 5 mm) and the typical spatial resolution of coronary MRI ( mm) (15), the displacements remaining after use of any of these respiratory motion compensation techniques is likely to be important. This is especially true for the distal coronary artery segments, where the respiratory motion tends to be larger due to the proximity to the diaphragm. For most subjects a single short breath hold remained the best method of respiratory motion suppression (Fig. 8); but for scans that require longer time than a feasible single breath hold, free-breathing NAV methods provide better FIG. 8. The results of 4 different motion compensation techniques for the proximal coronary artery segment are shown. The methods are: (a) a single breath hold without motion correction, (b) 5 different breath holds with slice tracking, (c) slice tracking with a tracking factor of 0.6 using a gating window of 5 mm, and (d) the affine motion correction model using a 10 mm gating window. The differences between subjects are considerable. For most subjects, breath holding is the best method; but for poor breath holders, the free breathing techniques can lead to improvements. The mean deviation never falls below 0.5 mm for the free breathing methods; a good breath hold can, however, be nearly free of motion. For most of the volunteers, the affine motion correction method allowed increasing the gating window from 5 mm to 10 mm with similar residual motion. motion compensation than multi-breath hold approaches. Multiple breath holds perform poorly, because the position of the heart varies between breath holds (22). Slice tracking with a constant factor of 0.6 reduces this effect, but does not eliminate it. The size of the gating window has a strong influence on residual motion for all free breathing methods. An increase of the gating window, which leads to an improved NAV efficiency, always led to an increase of residual motion (Fig. 5). The usage of correction methods led to a smaller increase, but could never completely compensate for a larger gating window. The patient specific models led to the smallest increase of residual motion for larger gating windows. However, these methods did not decrease the baseline motion measured for small gating windows. Instead of providing greater NAV efficiency, it might be helpful to design a patient specific method that reduces the baseline motion. Calibrating the model within a smaller gating window might decrease the baseline motion, because then the model only has to represent the motion within this gating window instead of the motion over the entire respiratory cycle. One important cause of the baseline motion for all free breathing methods is that there is respiratory motion during the acquisition window (13). Monitoring of the diaphragm shows that it can move up to 2 mm during the 150 ms of the acquisition window used here. Assuming the tracking factor of 0.6 to be accurate, this would result in more than 1 mm of coronary artery displacement during the acquisition window. Therefore, these measurements of residual motion might be smaller if more typical acquisition windows for coronary MRI (70 to 120 ms) were used. Furthermore, methods to compensate for motion during the acquisition window should be investigated. Since the displacement of the diaphragm during a breath hold is slow, motion during image acquisition is less of a problem during breath hold imaging. One reason for the improved performance of the patient specific, and particularly the affine, method is that the tracking factor may vary considerably between subjects and locations (Fig. 7) (8). In this study the average tracking factor of the proximal coronary artery was found to be This value lies between the value of 0.25 reported by Keegan et al. (23) and the values reported by Danias et al. (24) and by Wang et al., which both found a tracking factor close to 0.6. As noted by Keegan et al., a possible explanation is that the methods to measure the diaphragm displacement varied. In the studies of Danias et al. and of Wang et al., the diaphragmatic displacement was measured from the acquired images; in the study of Keegan et al. and in this study, a navigator on the dome of the right hemi-diaphragm was used. As expected, the tracking factors found were smaller for proximal coronary arteries versus distal coronary arteries, the latter being closer to the diaphragm. The heart is not only displaced during breathing but also stretched. Only the affine model could potentially compensate for this motion, which may be one reason why this model led to the best results for the free breathing part of the study.

7 618 Fischer et al. CONCLUSIONS This study demonstrated, quantitatively and visually, the residual non-cardiac motion of the coronary arteries remaining after respiratory motion compensation techniques. Free breathing navigator gated techniques can reduce the respiratory motion to a magnitude similar to that encountered in a single brief breath hold, and to a magnitude smaller than that encountered in multiple breathholds. REFERENCES 1. Edelman RR, Manning WJ, Burstein D, Paulin S. Coronary arteries: breath-hold MR angiography. Radiology 1991;181: Sachs TS, Meyer CH, Hu BS, Kohli J, Nishimura DG, Macovski A. Real-time motion detection in spiral MRI using navigators. Magn Reson Med 1994;32: Taylor AM, Keegan J, Jhooti P, Gatehouse PD, Firmin DN, Pennell DJ. Differences between normal subjects and patients with coronary artery disease for three different MR coronary angiography respiratory suppression techniques. J Magn Reson Imaging 1999;9: Holland AE, Goldfarb JW, Edelman RR. Diaphragmatic and cardiac motion during suspended breathing: preliminary experience and implications for breath-hold MR imaging. Radiology 1998;209: Feinberg DA, Hoenninger JC, Crooks LE, Kaufman L, Watts JC, Arakawa M. Inner volume MR imaging: technical concepts and their application. Radiology 1985;156: Nehrke K, Bornert P, Groen J, Smink J, Bock JC. On the performance and accuracy of 2D navigator pulses. Magn Reson Imaging 1999;17: Ehman RL, McNamara MT, Pallack M, Hricak H, Higgins CB. Magnetic resonance imaging with respiratory gating: techniques and advantages. AJR Am J Roentgenol 1984;143: Wang Y, Riederer SJ, Ehman RL. Respiratory motion of the heart: kinematics and the implications for the spatial resolution in coronary imaging. Magn Reson Med 1995;33: Danias PG, McConnell MV, Khasgiwala VC, Chuang ML, Edelman RR, Manning WJ. Prospective navigator correction of image position for coronary MR angiography. Radiology 1997;203: Manke D, Nehrke K, Bornert P. Novel prospective respiratory motion correction approach for free-breathing coronary MR angiography using a patient-adapted affine motion model. Magn Reson Med 2003;50: Stuber M, Botnar RM, Danias PG, Kissinger KV, Manning WJ. Breathhold three-dimensional coronary magnetic resonance angiography using real-time navigator technology. J Cardiovasc Magn Reson 1999;1: Stuber M, Botnar RM, Danias PG, Kissinger KV, Manning WJ. Submillimeter three-dimensional coronary MR angiography with real-time navigator correction: comparison of navigator locations. Radiology 1999;212: Spuentrup E, Manning WJ, Botnar RM, Kissinger KV, Stuber M. Impact of navigator timing on free-breathing submillimeter 3D coronary magnetic resonance angiography. Magn Reson Med 2002;47: Fischer SE, Wickline SA, Lorenz CH. Novel real-time R-wave detection algorithm based on the vectorcardiogram for accurate gated magnetic resonance acquisitions. Magn Reson Med 1999;42: Kim WY, Stuber M, Kissinger KV, Andersen NT, Manning WJ, Botnar RM. Impact of bulk cardiac motion on right coronary MR angiography and vessel wall imaging. J Magn Reson Imaging 2001;14: Brittain JH, Hu BS, Wright GA, Meyer CH, Macovski A, Nishimura DG. Coronary angiography with magnetization-prepared T2 contrast. Magn Reson Med 1995;33: Deshpande VS, Chung YC, Zhang Q, Shea SM, Li D. Reduction of transient signal oscillations in true-fisp using a linear flip angle series magnetization preparation. Magn Reson Med 2003;49: Shechter G, Ozturk C, Resar JR, McVeigh ER. Respiratory motion of the heart from free breathing coronary angiograms. IEEE Trans Med Imaging 2004;23: Spuentrup E, Bornert P, Botnar RM, Groen JP, Manning WJ, Stuber M. Navigator-gated free-breathing three-dimensional balanced fast field echo (TrueFISP) coronary magnetic resonance angiography. Invest Radiology 2002;37: Leiner T, Katsimaglis G, Yeh EN, Kissinger KV, van Yperen G, Eggers H, Manning WJ, Botnar RM. Correction for heart rate variability improves coronary magnetic resonance angiography. J Magn Reson Imaging 2005; 22: Nehrke K, Bornert P, Manke D, Bock JC. Free-breathing cardiac MR imaging: study of implications of respiratory motion initial results. Radiology 2001;220: McConnell MV, Khasgiwala VC, Savord BJ, Chen MH, Chuang ML, Edelman RR, Manning WJ. Prospective adaptive navigator correction for breath-hold MR coronary angiography. Magn Reson Med 1997;37: Keegan J, Gatehouse P, Yang GZ, Firmin D. Coronary artery motion with the respiratory cycle during breath-holding and free-breathing: implications for slice-followed coronary artery imaging. Magn Reson Med 2002;47: Danias PG, Stuber M, Botnar RM, Kissinger KV, Edelman RR, Manning WJ. Relationship between motion of coronary arteries and diaphragm during free breathing: lessons from real-time MR imaging. AJR Am J Roentgenol 1999;172: