Acquisition and reconstruction of undersampled radial data for myocardial perfusion magnetic resonance imaging.

PURPOSE To improve myocardial perfusion magnetic resonance imaging (MRI) by reconstructing undersampled radial data with a spatiotemporal constrained reconstruction method (STCR). MATERIALS AND METHODS The STCR method jointly reconstructs all of the time-frames for each slice. In 7 subjects at rest, on a 3-T scanner, the method was compared with a conventional (GRAPPA) Cartesian approach. RESULTS Increased slice coverage was obtained, as compared with Cartesian acquisitions. On average, 10 slices were obtained per heartbeat for radial acquisitions (8 of which are suitable for visual analysis with the remaining 2 slices, in theory, usable for quantitative purposes), whereas 4 slices were obtained for the conventional Cartesian acquisitions. The new method was robust to interframe motion, unlike using Cartesian undersampling and STCR. STCR produced images with an image quality rating (1 for best and 5 for worst) of 1.7 +/- 0.5; the Cartesian images were rated 2.6 +/- 0.4 (P = 0.0006). A mean improvement of 44 (+/-17) in signal-to-noise (SNR) ratio and 46 (+22) in contrast-to-noise ratio (CNR) was observed for STCR. CONCLUSION The new radial data acquisition and reconstruction scheme for dynamic myocardial perfusion imaging is a promising approach for obtaining significantly higher coverage and improved SNR ratios. Further testing of this approach is warranted during vasodilation in patients with coronary artery disease.