A Radial Based Multi-spatial, Multi-temporal Resolution Aligned Reconstruction Scheme for Accelerated Motion Correction in Brain MRI

Motivation: Joint motion/image optimization in aligned reconstruction can be computationally inefficient; scout-based method uses additionally acquired data to accelerate calculation, which may also be motion-corrupted. Goal(s): To develop a 3D motion correction method which takes the advantages of both self-navigation and fast calculation. Approach: A 3D radial acquisition and a multi-temporal, multi-spatial resolution scheme was used to formulate convex motion optimization subproblem. The temporal continuity of motion was introduced to constrain the resulted motion patterns. A motion-informed CS reconstruction was performed for accelerated image reconstruction. Results: The proposed method achieved joint optimization in 2 joint iterations at time resolution of 0.7s with 7.5x undersampling. Impact: A flexible and time-efficient method based on aligned reconstruction framework was developed for rigid-body motion correction in accelerated brain MRI, which may be beneficial to the exams of clinical uncooperative patients as well as brain MRI research community.