MSR-Hardi: Accelerated Reconstruction of Hardi Data Using Multiple Sparsity Regularizers

Brain neural connectivity patterns are increasingly analyzed with diffusion magnetic resonance imaging (dMRI) via the estimation of local fiber-tract orientations. High angular resolution diffusion imaging (HARDI), a variant of dMRI, is known to produce better representation of fiber orientations than the traditionally used diffusion tensor imaging (DTI). However, it requires a large number of samples leading to longer scan times. In this paper, we propose a new method, namely, MSR-HARDI, for the accelerated reconstruction of HARDI data using multiple sparsity regularizers in the k - q space. Combination of regularizers is observed to provide improved reconstructions as compared to individual regularizers. The proposed method is also observed to provide better reconstruction than the existing state-of-the-art methods in terms of the normalized mean squared error.