Motion robust joint k-q reconstruction for accelerated multi-band diffusion MRI

Motivation: Joint k-q reconstruction accelerates dMRI by exploiting shared information between diffusion directions/b-shells. However, subject motion causes artifacts in reconstruction due to inter-volume geometric inconsistency. Goal(s): To develop a motion-robust and computationally efficient joint k-q reconstruction to accelerate dMRI acquisition. Approach: We proposed a motion-robust joint k-q reconstruction method for accelerated multi-band dMRI with a multi-shell k-q model. Rigid motion parameters, motion induced phase errors, dMRI images were jointly estimated in iterative reconstruction. Variable-splitting approach was utilized to mitigate computational burden. Results: Simulation and in vivo results demonstrate that our proposed method enables robust joint k-q reconstruction in the presence of inter-volume subject motion. Impact: The developed method addresses the primary challenge in joint k-q reconstruction - subject motion, enabling robust acceleration of dMRI. This advancement facilitates translation of advanced dMRI methods to less cooperative subjects, expanding accessibility and utility of advanced dMRI in clinical setting.