Fast Spatially Coherent Fiber Orientation Estimation in Diffusion MRI from kq-Space Sampling

Diffusion Magnetic Resonance Imaging is a state-of-the-art technique that can provide accurate identification of complex neuronal fiber configurations in the human brain. Typical acquisition times are however too long for the clinical application. We propose a method to recover the fiber orientation distribution (FOD) at high spatio-angular resolution via practical kq-space under-sampling patterns that enable both acceleration and super-resolution. The inverse problem for FOD reconstruction is regularized by a structured sparsity prior promoting simultaneously voxelwise sparsity and spatial smoothness of fiber orientations. A convex minimization problem is formulated and solved via a forward-backward algorithm. Real data analysis suggest that high spatio-angular resolution FOD mapping can be achieved from severe kq-space acceleration.