Data-consistent Retrospective Motion Correction and Co-Registration

Motivation: Motion correction algorithms based on Image-based co-registration of retrospectively ordered motion states have limited effectiveness for highly accelerated scans. Goal(s): To develop a robust motion correction algorithm for highly accelerated dynamic MRI scans. Approach: We developed an approach that jointly, and data-consistently, estimates motion-corrected images and motion fields. Results: Simulated and experimental results demonstrate that the proposed approach yields improved motion-corrected images at high acceleration factors during dynamic MRI scans. Impact: The proposed approach could remove previously reported limitations on the use of retrospectively re-ordered dynamic MRI scans.