Sparse Riemannian manifold clustering for HARDI segmentation

We address the problem of segmenting high angular resolution diffusion images of the brain into cerebral regions corresponding to distinct white matter fiber bundles. We cast this problem as a manifold clustering problem in which distinct fiber bundles correspond to different submanifolds of the space of orientation distribution functions (ODFs). Our approach integrates tools from sparse representation theory into a graph theoretic segmentation framework. By exploiting the Riemannian properties of the space of ODFs, we learn a sparse representation for the ODF at each voxel and infer the segmentation by applying spectral clustering to a similarity matrix built from these representations. We evaluate the performance of our method via experiments on synthetic, phantom and real data.