Ultrafast time-resolved contrast-enhanced 3D pulmonary venous cardiovascular magnetic resonance angiography using SENSE combined with CENTRA-keyhole

Purpose: To evaluate the diagnostic benefit of time-resolved CENTRA-keyhole contrast-enhanced cardiovascular magnetic resonance angiography (CE-CMRA) for improving arterial-venous separation of pulmonary vessels. Methods: Twenty-three patients (18 males; age = 58 +/- 11y) after radiofrequency pulmonary vein isolation to treat atrial fibrillation were examined using CENTRA-keyhole based multi-phase 3D CE-CMRA yielding 6 near-isotropic 3D datasets every 1.6 s (50-60 coronal partitions, 1.4 x 1.4 x 1.3 mm, SENSE-factor 3). Results were compared with conventional non-keyhole CE-CMRA (identical parameters, SENSE-factor 2). Results: Data acquisition was accelerated by a speedup factor of similar to 9 compared with the reference CE-CMRA (SENSE 1.5*, keyhole 6*). No pulmonary venous stenoses were detected by either method, overall pulmonary venous diameters were 17.1 +/- 3.6 mm. Applying Bland-Altman analysis, vessel diameters differed by a mean of 0.1 mm + 2.1 mm/-2.0 mm (mean +/- 2 SD), indicating close agreement between both techniques. Interobserver variability was higher for CENTRA-keyhole (mean = 0.1 mm; mean +/- 2 SD: +2.5 mm/-2.3 mm) compared to conventional technique (0.0 mm; +1.6 mm/-1.5 mm), corresponding to a percentual deviation (mean +/- 2 SD) of the mean diameter of approximately +/- 15% (keyhole CE-CMRA) and +/- 10% (conventional CE-CMRA), respectively. Using keyhole-based time-resolved CE-CMRA, the contrast between pulmonary veins versus aorta/pulmonary artery was significantly increased (p < 0.05), which improved vessel depiction. In 12 cases, the contrast bolus arrival was delayed in one of the pulmonary veins by 1 dynamic frame (= 1.6 seconds); in 7 cases by 2 frames (= 3.2 seconds) and in 1 subject by 3 frames (= 4.8 seconds). The bolus usually appeared first in the upper right pulmonary vein whereas a delay occurred most often in the lower left pulmonary vein. Conclusions: Conventional CE-CMRA may be advantageous for accurate vessel size measures as evidenced by superior interobserver reproducibility in this study. Multi-dynamic CE-CMRA using CENTRA-keyhole with SENSE, however, allows for improved arterio-venous separation of pulmonary vessels and additional dynamical information on pulmonary venous perfusion, while maintaining high spatial resolution. Exact bolus timing is no longer needed.