Rapid velocity-encoded cine imaging with turbo-BRISK

Abstract

Velocity-encoded cine (VEC) imaging is potentially an important clinical diagnostic technique for cardiovascular diseases. Advances in gradient technology combined with segmentation approaches have made possible breathhold VEC imaging, allowing data to be obtained free of respiratory artifacts. However, when using conventional segmentation approaches, spatial and temporal resolutions are typically compromised to accommodate short breathhold times. Here we apply a sparse sampling technique, turbo-BRISK (i.e., segmented block regional interpolation scheme for k-space) to VEC imaging, allowing increased spatial and temporal resolution to be obtained in a short breathhold period. BRISK is a sparse sampling technique with interpolation used to generate unsampled data. BRISK was implemented to reduce the scan time by 70% compared with a conventional scan. Further, turbo-BRISK scans, using segmentation factors up to 5, reduce the scan time by up to 94%. Phantom and in vivo results are presented that demonstrate the accuracy of turbo-BRISK VEC imaging. In vitro validation is performed using conventional magnetic resonance VEC. Pulsatile centerline flow velocity measurements obtained with turbo-BRISK acquisitions were correlated with conventional magnetic resonance imaging measurements and achieved r values of 0.99 +/- 0.004 (mean +/- SD) with stroke volumes agreeing to within 4%. A potential limitation of BRISK is reduced accuracy for rapidly varying velocity profiles. We present low- and high-resolution data sets to illustrate the resolution dependence of this phenomenon and demonstrate that at conventional resolutions, turbo-BRISK can accurately represent rapid velocity changes. In vivo results indicate that centerline velocity waveforms in the descending aorta correlate well with conventional measurements with an average r value of 0.98 +/- 0.01.