Abstract
Background
Usually, the pulmonary venous and left atrial (PV–LA) anatomy is assessed with contrast-enhanced computed tomographic imaging for catheter ablation of atrial fibrillation (AF). A non-contrast-enhanced magnetic resonance (MR) imaging method has not been established. Three-dimensional balanced steady-state free precession (3D b-SSFP) sequences cannot visualize the PV–LA anatomy simultaneously because of the signal intensity defect of pulmonary veins. We compared two-dimensional (2D) b-SSFP sequences with 3D b-SSFP sequences in depicting the PV–LA anatomy with non-contrast-enhanced MR imaging for AF ablation.
Methods
Eleven healthy volunteers underwent non-contrast-enhanced MR imaging with 3D b-SSFP and 2D b-SSFP sequences. The MR images were reconstructed on the 3D PV–LA surface image. Two experienced radiological technicians independently scored the multiplanar reformatted (MPR) images on a scale of 1–4 (from 1, not visualized, to 4, excellent definition). The overall score was a sum of 5 segments (LA and 4 PVs).
Results
In the 2D b-SSFP method, MR imaging was successfully performed, and the 3D PV–LA surface image was precisely reconstructed in all healthy volunteers. The image score was significantly higher in the 2D b-SSFP method compared to the 3D b-SSFP method (19 [19; 20] vs. 12 [11; 15], p=0.004, for both observers). No PV signal intensity defects occurred in the 2D b-SSFP method.
Conclusions
The 2D b-SSFP sequence was more useful than the 3D b-SSFP sequence in adequately depicting the PV–LA anatomy.
Abbreviations: AF, atrial fibrillation; 3D, three-dimensional; MR, magnetic resonance; CT, computed tomography; PV–LA, pulmonary venous and left atrial; b-SSFP, balanced steady-state free precession; MPR, multiplanar reformatted; LAA, left atrial appendage; LSPV, left superior pulmonary vein; TE, echo time; TR, repetition time; FA, flip angle; FOV, field of view; SENSE, sensitivity encoding
Keywords: Atrial fibrillation, Magnetic resonance imaging, Pulmonary vein, Left atrium, Balanced SSFP
1. Introduction
Catheter ablation of atrial fibrillation (AF) has been rapidly popularized by the appearance of three-dimensional (3D) mapping systems. Wide encircling isolation of the entire pulmonary vein (PV) antrum from the left atrial side provides the ability to eliminate AF trigger as well as treat the AF substrate. Complete isolation of the entire circle is considered the optimal electrophysiological end point. However, given the complex and highly variable individual 3D PV antrum anatomy, achievement of continuity and transmurality along the entire encircling ablation line is challenging. Hence, high-resolution images from magnetic resonance (MR) or computed tomographic (CT) imaging integrated with 3D mapping systems represent a highly desirable technique to maximize the efficacy and minimize the risks of AF ablation procedures [1–3].
Pulmonary venous and left atrial (PV–LA) anatomy is usually assessed with contrast-enhanced CT, because the non-contrast-enhanced MR imaging approach has not been well established [4]. Recently, examination using a 3D balanced steady-state free precession (b-SSFP) sequence has been reported (3D-B method) [5,6]. b-SSFP imaging techniques have progressed for evaluating the thoracic vasculature, including the coronary arteries and thoracic aorta, because of their inherent high signal-to-noise and contrast-to-noise ratios [7–9]. However, a PV signal intensity defect frequently occurs with practical use of the 3D-B method [10]. Actually, in cases of patients with renal failure, we must attempt the AF ablation and complete the procedure safely with compromising the image quality. The precise anatomical information of the LA antrum region and anterior ridge between the left atrial appendage (LAA) and left superior PV (LSPV) is necessary to achieve continuity of the ablation line. We assessed the feasibility of an MR imaging acquisition and processing protocol, a b-SSFP sequence without a contrast agent (2D-B method), to depict the accurate PV–LA anatomy for AF ablation.
2. Material and methods
2.1. Study subjects
The study population consisted of 11 healthy volunteers (11 men, 29.1±6.1 years). This study was approved by the ethics committees of Himeji Cardiovascular Center (approval date, November 19, 2013; approval number, 20). A written informed consent to participate was obtained from all subjects. Every researchers involved in this study acted in conformity with the Declaration of Helsinki (adopted by the 18th WMA General Assembly, Helsinki, Finland).
2.2. MR imaging
All healthy volunteers underwent non-contrast-enhanced MR imaging with the 3D-B and 2D-B methods on a 1.5 T MR system (Intera Achieva; Phillips Medical Systems). The detailed scan parameters are summarized in Table 1.
Table 1.
MRI scan parameters.
| 3D-B method | 2D-B method | |
|---|---|---|
| Contrast medium | No | No |
| Scan mode | 3D | M2D |
| Scan technique | FFE | FFE |
| Contrast enhancement | Balanced | Balanced |
| Fast imaging mode | TFE | TFE |
| TFE shot mode | Multi-shot | Single-shot |
| Acquisition time | 146.7 ms | 202.1 ms |
| Profile order | Low_high (Radial) | Linear |
| Fat suppression | SPIR | No |
| Band width | 723.4 Hz | 964.5 Hz |
| T2-prep (echo time/refocusing pulses) | Yes (50/4) | No |
| TR/TE/FA | 4.6 ms/2.3 ms/80° | 3.1 ms/1.53 ms/80° |
| FOV | 280 mm | 300 mm |
| Scan matrix (voxel or pixel size) | 256×256 (1.09×1.09×3.00 mm) | 192×192 (1.56×1.56 mm) |
| Reconstruction matrix (voxel or pixel size) | 512×512 (0.55×0.55×1.5 mm) | 512×512 (0.59×0.59 mm) |
| Slices | 70 | 40–60 |
| Slice thickness | 3.0 mm | 6.0 mm |
| Slice gap | – | −3.0 mm |
| Slice orientation | Transverse | Sagittal |
| Phase encoding direction | RL | AP |
| Slice scan order | – | Ascend |
| SENSE reduction factor | No | 3 |
| NEX | 1 | 3 |
| Cardiac synchronization (trigger delay/cardiac phase) | ECG-Trigger (user defined/mid-diastole) | ECG-Trigger (user defined/mid-diastole) |
| Navigator respiratory compensation | Gate and track | Gate and track |
FFE=fast field echo, TFE=turbo field echo, SPIR=spectral pre-saturation with inversion recovery spectrally selective inversion recovery, TR=repetition time, TE=echo time, FA=flip angle, FOV=field of view, NEX=number of excitations, SENSE=sensitivity encoding.
2.2.1. 3D-B method
The PV–LA anatomical images were acquired using a respiration-navigated, electrocardiogram-gated, 3D b-SSFP sequence in the transverse plane with the following parameters: 2.3-ms echo time (TE); 4.6-ms repetition time (TR); 80° flip angle (FA); 280×280-mm field of view (FOV); 256×256 scan matrix; 70 slices; 1.09×1.09×3-mm effective spatial resolution; and 0.55×0.55×1.5-mm reconstructed spatial resolution. To enhance the contrast between blood and the surrounding tissue, T2-prep and fat saturation pre-pulse were applied. The navigator was placed over the dome of the right hemidiaphragm (gating window, 3.0 mm; tracking factor, 0.6). An acquisition window of approximately 150 ms using the multi-shot technique was placed at the mid-diastole. These techniques have been used to evaluate the coronary arteries. Scan parameters of the 3D-method are documented in a previous study [8,9]. The typical scan time for the 3D-B method was 7–15 min depending on the healthy volunteer heart rate and respiration pattern.
2.2.2. 2D-B method
The PV–LA anatomical images were acquired using a respiration-navigated, electrocardiogram-gated, 2D b-SSFP sequence in the sagittal plane with the following parameters: 1.53 ms TE; 3.1 ms TR; 80° FA; 300×300 mm FOV; 192×192 scan matrix; 6.0 mm slice thickness; −3.0 mm gap; 40–60 slices; 1.56×1.56 mm effective spatial resolution; and 0.59×0.59 mm reconstructed spatial resolution. A parallel imaging technique, sensitivity encoding (SENSE), with a reduction factor of 3.0 was used to shorten the acquisition window. For the same reason, TE/TR, FOV, and scan matrix were determined. An acquisition window of approximately 200 ms using the single-shot technique was placed at the mid-diastole. The navigator was placed over the dome of the right hemidiaphragm (gating window, 3.0 mm; tracking factor, 0.6). The saturation bands were placed in the phase-encoding (anterior-posterior) line to minimize back-folding artifacts from the body walls. The typical scan time for the 2D-B method was 4–9 min depending on the healthy volunteer heart rate and respiration pattern. Source images were transferred to a workstation (Advance viewing; Phillips Medical Systems) and reformatted to axial images with a thickness of 1.0 mm for later image segmentation.
2.3. Image segmentation
The MR images were transferred to a 3D mapping system (Carto 3 version 2.3; Biosense Webster Inc.), which was equipped with image integration software (CartoMerge™) [11]. Proprietary software tools on the 3D mapping system allowed for cardiac image segmentation to separate the LA and PVs from the surrounding cardiac structures for the catheter ablation of AF. Fig. 1 shows the reconstructed 3D PV–LA surface image obtained from the contrast-enhanced CT image. The quality of the 3D PV–LA surface image for AF ablation required the following: (1) adequate visualization of all PVs and (2) precise separation of the LAA from the LSPV.
Fig. 1.
A segmented 3D PV–LA surface image using a 3D mapping system. (a) The 3D PV–LA surface image from the PA view. (b) The 3D PV–LA surface image from the inferior view. The data were reconstructed from the contrast-enhanced computed tomographic image. Of note, all 4 PVs and an adequate anatomy of the anterior ridge between the LSPV and left atrial appendage were clearly depicted. The red arrow indicates the anterior ridge. The 3D pink and red points indicate the ablation points. LI=left inferior pulmonary vein; LS=left superior pulmonary vein; RI=right inferior pulmonary vein; RS=right superior pulmonary vein; LAA=left atrial appendage.
2.4. Image analysis of the 3D-B and 2D-B methods
The 3D-B and 2D-B methods were analyzed in random order. Two experienced radiological technicians independently scored the multiplanar reformatted (MPR) images using the 3D-B and 2D-B methods on a scale of 1–4: 1=poor image quality, the geometry and contour not discernible; 2=moderate image quality, the geometry and contour only partly discernible; 3=good image quality, clear delineation of most parts of the geometry and contour; and 4=excellent image quality, clear delineation of the entire geometry and contour. The overall score was a sum of 5 segments (LA and 4 PVs). An adequate image was defined as an image without a score of 1 or 2 in any segments.
2.5. Intervariability in image quality assessment
To assess interobserver variability, image scores of the LA and 4 PVs were independently assessed by 2 observers.
2.6. Statistical analysis
The data were tested with the Kolmogorov–Smirnov test and are presented as mean±SD for normally distributed variables. Median and quartiles are given for non-normally distributed variables. Categorical variables are expressed as number and percentage of patients. Continuous data were analyzed with the Wilcoxon rank sum test to test for significant differences. Categorical variables were analyzed with the Fisher exact test. Using a linear regression model, we evaluated interobserver variability in assessing MR image quality. A p value <0.05 was considered statistically significant. All statistical analyses were performed with SPSS, release 11.0.
3. Results
3.1. MR imaging and image segmentation data
All healthy volunteers were in sinus rhythm during the MR imaging. The heart rate ranged from 48 to 83 beats/min. In the 2D-B method, the MR imaging was successfully performed, and the LA with PVs was segmented completely from the reconstructed 3D surface images with the 3D mapping system in all healthy volunteers. In contrast, in the 3D-B method, an adequate image was achieved in only 2 of 11 volunteers (18%) (Table 2). The clinical characteristics did not differ between the 2 cases and the remaining 9 cases. The overall image quality score was significantly higher on the MPR images using the 2D-B method compared to that using the 3D-B method (3D-B method vs. 2D-B method: 12 [11;15] vs. 19 [19;20], p=0.004, for both observer 1 and observer 2). Representative cases of the 2D-B and 3D-B methods are shown in Fig. 2. The 2D-B method could completely depict the PV–LA anatomy. In the 3D-B method, adequate segmented images were not acquired in most cases because of PV signal intensity defects, especially the left superior and right superior PVs (Tables 2 and 3). In the 2D-B method, motion artifacts due to respirations and heart beat fluctuations were observed in some cases (Figs. 3a, b). Fortunately, the motion artifact was acceptable in the segmented 3D PV–LA surface image (Fig. 3c). The mean image segmentation processing time was 3.5±0.8 min.
Table 2.
Adequate image quality by observer and technique.
|
Observer 1 |
p-Value |
Observer 2 |
p-Value | |||
| 3D-B method | 2D-B method | 3D-B method | 2D-B method | |||
| (n=11) | (n=11) | (n=11) | (n=11) | |||
| Overall image quality | ||||||
| Adequate | 2 (18%) | 11 (100%) | <0.001 | 2 (18%) | 11 (100%) | <0.001 |
| Inadequate | 9 (82%) | 0 | 9 (82%) | 0 | ||
Fig. 2.
Comparison of the image quality between the 3D-B method and 2D-B method in the same healthy volunteer. (a) Axial image using the 3D-B method. (b) Axial image using the 2D-B method. (c) Coronal image using the 3D-B method. (d) Coronal image using the 2D-B method. (e) 3D PV–LA surface image using the 3D-B method. (f) 3D PV–LA surface image using the 2D-B method. The red arrows indicate a signal intensity defect of the superior pulmonary veins using the 3D-B method. The dotted red arrows indicate an adequate signal intensity of the superior pulmonary veins using the 2D-B method. The white arrows indicate the pulmonary artery. LI=left inferior pulmonary vein; LS=left superior pulmonary vein; RI=right inferior pulmonary vein; RS=right superior pulmonary vein; LAA=left atrial appendage.
Table 3.
Image quality score by observer and technique.
|
Observer 1 |
p-Value |
Observer 2 |
p-Value | |||
| 3D-B method | 2D-B method | 3D-B method | 2D-B method | |||
| (n=11) | (n=11) | (n=11) | (n=11) | |||
| Image quality of PV and LA | ||||||
| LA | 3 (3; 4) | 4 (4; 4) | 0.083 | 3 (3; 4) | 4 (4; 4) | 0.083 |
| LSPV | 2 (1; 3) | 4 (4; 4) | 0.007 | 2 (1; 3) | 4 (4; 4) | 0.007 |
| LIPV | 3 (3; 4) | 4 (4; 4) | 0.035 | 3 (3; 4) | 4 (4; 4) | 0.035 |
| RSPV | 1 (1; 2) | 4 (4; 4) | 0.004 | 1 (1; 2) | 4 (4; 4) | 0.004 |
| RIPV | 3 (3; 3) | 4 (4; 4) | 0.004 | 3 (3; 3) | 4 (4; 4) | 0.004 |
| Overall image quality score | 12 (11; 15) | 19 (19; 20) | 0.004 | 12 (11; 15) | 19 (19; 20) | 0.004 |
PV=pulmonary vein, LA=left atrium, LSPV=left superior pulmonary vein, LIPV=left inferior pulmonary vein, RSPV=right superior pulmonary vein, RIVP=right inferior pulmonary vein.
Fig. 3.
Motion artifact using the 2D-B method. (a) Coronal image. (b) Axial image. (c) 3D PV–LA surface image. The red arrow indicates a motion artifact during respirations and heart beat fluctuations. The motion artifact was not critical in the segmented 3D PV–LA surface image. LI=left inferior pulmonary vein; LS=left superior pulmonary vein; RI=right inferior pulmonary vein; RS=right superior pulmonary vein.
3.2. Interobserver variability in image quality assessment
When estimating the image quality in the 3D-B method, there was excellent correlation between observer 1 and observer 2 (LA and 4 PVs: r=0.972, p<0.001; LA: r=0.796, p<0.001; LSPV: r=0.796, p=0.003; RSPV: r=0.898, p<0.001; LIPV: r=0.825, p=0.002; RIPV: r=0.787, p=0.004). Intervariability in the 2D-B method was not assessed, because the score between observer 1 and observer 2 was completely identical. This suggested that interobserver variability was within an acceptable range.
4. Discussion
4.1. Main findings
Our main findings were that an MR imaging acquisition and processing protocol, the 2D b-SSFP sequence without a contrast agent, was useful in acquiring adequate PV–LA anatomical information for AF ablation.
4.2. Image analysis of 3D-B and 2D-B methods
In the 3D-B method, PV signal intensity defects were remarkable. Hu et al. reported that PV off-resonance causes signal void artifacts [10]. In conventional b-SSFP acquisitions, without shifts in the b-SSFP signal profile, the PV signal was severely suppressed. In the b-SSFP acquisitions with a shifted signal profile of 25–50 Hz, the PV signal voids were reduced. With a shift of 75–125 Hz, the PV signal intensity was greatly enhanced, while the on-resonance structures such as the LA were suppressed. Furthermore, a unique bloodstream pattern from the PV to LA was considered to influence the PV signal intensity defects. We recognized that the pulmonary venous flow is biphasic: a rapid filling wave is observed during systole when the mitral valve is closed; a second wave is observed in diastole during the rapid ventricular filling phase of the mitral flow but is significantly delayed. Keren et al. reported that the pulmonary venous flow is influenced by dynamic changes in the left atrial pressure created by the contraction and relaxation of the atrium and ventricle [12]. We speculated that the existence of a valve could create a one-directional blood flow. A valve is not located between the PV and LA. Thus, we considered the PV signal intensity defect to be caused by this PV blood flow turbulence. Actually, the 3D-B method could depict the pulmonary artery more precisely compared to the 2D-B method (Fig. 2a–d). The pulmonary artery flow is monophasic and the valve could prevent blood regurgitation. Storey et al. reported that the inflow effect was important in the b-SSFP sequence [13]. Unlike in 3D acquisition, an excitation pulse in 2D acquisition is applied to only the target slice. Thus, the inflow effect of 2D acquisition was considered to be higher than that of 3D acquisition. As a result, the 2D-B method was likely to avoid the PV intensity defects. For AF ablation, the simultaneous visualization of PVs and the LA is necessary. Thus, 3D PV–LA surface image segmentation using the 3D-B method was not acceptable in most cases. We considered the 3D-B method to have several limitations for the simultaneous visualization of the LA and PV in the b-SSFP sequence. As for improving the image quality, we set the imaging section in the sagittal section. The reasons were as follows: (1) the imaging section was perpendicular to the bloodstream to maximize inflow effect, and (2) it was easy to clearly separate the LAA and PV. The LAA was located anteriorly and very close to the LSPV in some cases. In addition, each slice was overlapped with 3.0-mm widths. This contrivance could compensate the data between the slices with 6.0-mm thickness and improve the effective spatial resolution in the reconstructed axial images. In general, the image quality using the 3D-B method is superior to that using the 2D-B method owing to the high signal-to-noise ratio, the thin slice thickness, etc. The advantage to using the 2D-B method is that “pulmonary vein” off-resonance does not occur. This indicates that simultaneous visualization of the PV and LA can be achieved in the 2D-B method as compared to the 3D-B method (Table 4).
Table 4.
The advantage/disadvantage for image quality using the 2D-B method compared to the 3D-B method.
| 3D-B method | 2D-B method | |
|---|---|---|
| Scan time (minutes) | 7–15 | 4–9 |
| Signal/noise ratio | Better | |
| Slice thickness | Thinner | |
| Resolution | Better | |
| Inflow effect | Better | |
| Contrast | Better | |
| PV off-resonance | Frequently | None |
PV=pulmonary vein.
Finally, Fig. 4 shows the case with persistent AF and chronic renal dysfunction. The LA was enlarged and the rhythm was AF during MRI acquisition. The acquisition window setting during AF was different from that during sinus rhythm. The acquisition window was placed automatically as the shortest interval after the R-wave sensing for minimalizing the variability of cardiac motion due to the unstable diastolic phase during AF. Of note, signal void improved by using the 2D-B method, even in the patient with persistent AF. Thus, PVs and the LA could be simultaneously visualized.
Fig. 4.
Comparison of the image quality between the 3D-B method and 2D-B method in the patients with persistent AF and renal dysfunction. The image using the 3D-B method (left panel a, c, and e) and the 2D-B method (right panel b, d, and f). The motion artifacts caused by heart beat fluctuation were observed in both 3D and 2D methods. Still, signal void was not observed around the pulmonary vein in the image using the 2D-B method even in the patient with persistent AF. The red arrow and dotted red arrow indicate the left superior pulmonary vein with and without signal void, respectively. AF=atrial fibrillation; LI=left inferior pulmonary vein; LS=left superior pulmonary vein; RI=right inferior pulmonary vein; RS=right superior pulmonary vein.
4.3. Clinical implication
Generally, in spatial resolution and signal-to-noise ratio, non-contrast MR was reported to be inferior to contrast CT. Still, we considered the quality of the PV and LA anatomy using the 2D-B method to be acceptable for the integration with 3D mapping systems. If paroxysmal AF patients, especially those with chronic renal failure, are scanned at the time of sinus rhythm, this method contributes to the prevention of preprocedural radiation exposure and contrast medium-related side effects.
4.4. Study limitations
Our study had 4 major limitations. Firstly, the sample size was rather small. However, the acquired PV–LA images were compared in the same volunteer. Thus, improvements in image quality using the 2D-B method were obvious. Secondly, volunteers suffering from AF were not included in this study. However, 60% of patients with AF ablation have paroxysmal AF. In such paroxysmal AF patients with renal failure and/or heart failure, we believe that this MR imaging method is necessary to improve safety before AF ablation. The feasibility of this method in persistent AF cases should be investigated with further study. Thirdly, the voxel size is too large for measuring PV diameter. We focused on the simultaneous visualization of PVs and the LA for AF ablation; the impact of this voxel size was unclear in this study. Fourthly, this study showed that the image quality of the 2D-B method was superior to that of the 3D-B method. However, if new techniques or technologies for improving B0 inhomogeneity are developed, those may yield significantly different results. Thus, an optimal sequence needs to be further studied.
5. Conclusion
By using the 2D-B method, the PV–LA anatomical image without any signal intensity defect and the 3D PV–LA surface image segmentation could be acquired in all healthy volunteers. Our MR imaging method was a feasible technique for reducing the risks of radiation exposure and contrast agent use. We hope that this noninvasive imaging method can improve the safety and efficacy of AF ablation without compromising the quality of the PV–LA anatomical image.
Conflict of interest
None.
Acknowledgments
We would like to thank Mr. John Martin for his linguistic assistance.
References
- 1.Martinek M., Nesser H.J., Aichinger J. Impact of integration of multislice computed tomography imaging into three-dimensional electroanatomic mapping on clinical outcomes, safety, and efficacy using radiofrequency ablation for atrial fibrillation. Pacing Clin Electrophysiol. 2007;30:1215–1223. doi: 10.1111/j.1540-8159.2007.00843.x. [DOI] [PubMed] [Google Scholar]
- 2.Khaykin Y., Oosthuizen R., Zarnett L. CARTO-guided vs. NavX-guided pulmonary vein antrum isolation and pulmonary vein antrum isolation performed without 3-D mapping: effect of the 3-D mapping system on procedure duration and fluoroscopy time. J Interv Card Electrophysiol. 2011;30:233–240. doi: 10.1007/s10840-010-9538-9. [DOI] [PubMed] [Google Scholar]
- 3.Kistler P.M., Rajappan K., Jahngir M. The impact of CT image integration into an electroanatomic mapping system on clinical outcomes of catheter ablation of atrial fibrillation. J Cardiovasc Electrophysiol. 2006;17:1093–1101. doi: 10.1111/j.1540-8167.2006.00594.x. [DOI] [PubMed] [Google Scholar]
- 4.Kiuchi K., Kircher S., Watanabe N. Quantitative analysis of isolation area and rhythm outcome in patients with paroxysmal atrial fibrillation after circumferential pulmonary vein antrum isolation using the pace-and-ablate technique. Circ Arrhythm Electrophysiol. 2012;5:667–675. doi: 10.1161/CIRCEP.111.969923. [DOI] [PubMed] [Google Scholar]
- 5.Francois C.J., Tuite D., Deshpande V. Pulmonary vein imaging with unenhanced three-dimensional balanced steady-state free precession MR angiography: initial clinical evaluation. Radiology. 2009;250:932–939. doi: 10.1148/radiol.2502072137. [DOI] [PubMed] [Google Scholar]
- 6.Krishnam M.S., Tomasian A., Malik S. Three-dimensional imaging of pulmonary veins by a novel steady-state free-precession magnetic resonance angiography technique without the use of intravenous contrast agent: initial experience. Invest Radiol. 2009;44:447–453. doi: 10.1097/RLI.0b013e3181a7c6cb. [DOI] [PubMed] [Google Scholar]
- 7.Jan B., Steven D., Andrew M.T. Coronary artery disease. In: Jan B., Steven D., Andrew M.T., editors. Clinical cardiac MRI. Springer; Berlin Heidelberg (New York): 2005. pp. 381–437. [Google Scholar]
- 8.Bi X., Deshpande V., Carr J. Coronary artery magnetic resonance angiography (MRA): a comparison between the whole-heart and volume-targeted methods using a T2-prepared SSFP sequence. J Cardiovasc Magn Reson. 2006;8:703–707. doi: 10.1080/10976640600723706. [DOI] [PubMed] [Google Scholar]
- 9.Liu X., Zhao X., Huang J. Comparison of 3D free-breathing coronary MR angiography and 64-MDCT angiography for detection of coronary stenosis in patients with high calcium scores. Am J Roentgenol. 2007;189:1326–1332. doi: 10.2214/AJR.07.2805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hu P., Stoeck C.T., Smink J. Noncontrast SSFP pulmonary vein magnetic resonance angiography: impact of off-resonance and flow. J Magn Reson Imaging. 2010;32:1255–1261. doi: 10.1002/jmri.22356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sra J., Krum D., Hare J. Feasibility and validation of registration of three-dimensional left atrial models derived from computed tomography with a noncontact cardiac mapping system. Heart Rhythm. 2005;2:55–63. doi: 10.1016/j.hrthm.2004.10.035. [DOI] [PubMed] [Google Scholar]
- 12.Keren G., Sherez J., Megidish R. Pulmonary venous flow pattern—its relationship to cardiac dynamics. A pulsed Doppler echocardiographic study. Circulation. 1985;71:1105–1112. doi: 10.1161/01.cir.71.6.1105. [DOI] [PubMed] [Google Scholar]
- 13.Storey P., Li W., Chen Q. Flow artifacts in steady-state free precession cine imaging. Magn Reson Med. 2004;51:115–122. doi: 10.1002/mrm.10665. [DOI] [PubMed] [Google Scholar]