5-D free-breathing and ungated cardiac MRI using diffeomorphicmotion-compensated (DMoCo) algorithm

Motivation: Free-breathing and ungated 3D radial acquisitions are emerging as powerful alternatives to gold-standard breath-held 2D cine cardiac-MRI. Current motion-resolved algorithms for 5D MRI suffer from low sampling of bins, when scan time is limited. Goal(s): To develop a data-efficient diffeomorphic motion-resolved reconstruction algorithm for free-breathing and ungated cardiac MRI data. Approach: Each volume is represented as the deformed version of the static template, where the deformations are constrained by phase-dependent diffeomorphisms. The static template and the deformations are estimated from the measured k-t space data. Results: The proposed algorithm enables the recovery of 3D cardiac cine MRI data from six minutes of scan-time. Impact: The proposed diffeomorphic motion compensated recovery scheme is more data and computationally efficient thancurrent motion resolved reconstruction methods used in 5D cardiac MRI, and can offer good recovery from short 3D radialungated scans.