(k,q)-Compressed Sensing for dMRI with Joint Spatial-Angular Sparsity Prior

摘要

Advanced diffusion magnetic resonance imaging (dMRI) techniques, likediffusion spectrum imaging (DSI) and high angular resolution diffusion imaging(HARDI), remain underutilized compared to diffusion tensor imaging because thescan times needed to produce accurate estimations of fiber orientation aresignificantly longer. To accelerate DSI and HARDI, recent methods fromcompressed sensing (CS) exploit a sparse underlying representation of the datain the spatial and angular domains to undersample in the respective k- andq-spaces. State-of-the-art frameworks, however, impose sparsity in the spatialand angular domains separately and involve the sum of the corresponding sparseregularizers. In contrast, we propose a unified (k,q)-CS formulation whichimposes sparsity jointly in the spatial-angular domain to further increasesparsity of dMRI signals and reduce the required subsampling rate. Toefficiently solve this large-scale global reconstruction problem, we introducea novel adaptation of the FISTA algorithm that exploits dictionaryseparability. We show on phantom and real HARDI data that our approach achievessignificantly more accurate signal reconstructions than the state of the artwhile sampling only 2-4% of the (k,q)-space, allowing for the potential of newlevels of dMRI acceleration.