Meeting Banner
Abstract #4247

Parallel Reconstruction Observing Self Consistency and Temporal Smoothness (PROST)

Mitchell A. Cooper1, 2, Thanh D. Nguyen2, Pascal Spincemaille2, Keigo Kawaji1, 2, Jonathan W. Weinsaft3, Martin R. Prince2, Yi Wang1, 2

1Biomedical Engineering, Cornell University, Ithaca, NY, United States; 2Radiology, Weill Cornell Medical College, New York, United States; 3Cardiology, Weill Cornell Medical College, New York, United States

PROST (Parallel Reconstruction Observing Self consistency and Temporal smoothness) is a SPIRiT based paralell imaging technique implemented by extending the SPIRiT self-consistency kernel to the time domain. PROST is based on the assumption that each temporal phase has small changes with respect to its neighbors (temporal smoothness). PCA is incorporated into PROST to further express the information redundancy in the temporal dimension. PROST improves error performance compared to regular SPIRiT. When used for CINE imaging, PROST gives similar ejection fraction values compared to fully sampled data.

Abstract PDF E-Poster Video

Keywords

acquisition action adding allow allowing always arbitrary assumption basis biomedical breath cardiology carried cine coil college components compress compressible conjugate consistency constraining cooper course derived develop dimension domain easily ejection enforcing enter error experienced explored express extending extends facto fraction fully gave gives golden gradient healthy heart holds imposes improved include incorporate increasingly inherent iterations keeping kernel kernels martin matrix meters minim minimization neighbors observing onto optimal optimization oral original pace parallel patterns previously principal prior problem profile propose pros pulse quired radial radiology reconstructed reduced reduction redundancy regular resolved retrospectively sample sampled scanned segment self semi slices slower smoothness space spar sparsity spatial spiral spirit squares statistically strict temp tempo temporal theory thereby thick train training true truly umber unknown unknowns updates utilizing varied varying view volunteer volunteers whole