Accelerated 4D Flow MRI Using a Shared Subspace Constraint

Abstract

Cardiovascular diseases are the leading cause of death in the
world, more than the next three leading causes of death combined.
Cardiovascular imaging techniques have allowed for the
study and understanding of the function and structure of the
heart as well as the detection, diagnosis, and monitoring of cardiovascular
diseases in patients. One powerful technique for cardiac
imaging is 4D phase contrast magnetic resonance imaging
(PC-MRI) which allows measurement of blood flow velocity in the
heart and vessels. However, 4D PC-MRI is difficult to perform
due to low imaging speed and is therefore often carried out using
accelerated imaging techniques which reconstruct images from
reduced data. One approach for accelerating PC-MRI is explicitsubspace
low-rank imaging; this project focuses on further accelerating
explicit-subspace low-rank PC-MRI through the use of a
shared temporal subspace between PC-MR images with velocity
encoded in different directions. We will: a) investigate the subspace
structure of the differently encoded images to verify that
they indeed live in a shared subspace; b) evaluate the feasibility
of estimating this shared subspace from reduced auxiliary data
(which has direct implications on the frame rate of the resulting
images); and c) demonstrate the utility of exploiting this subspace
structure when performing image reconstruction from reduced
data.