POSTPROCESSING OF BRAIN WHITE MATTER FIBER ORIENTATION DISTRIBUTION FUNCTIONS

By acquiring high angular resolution diffusion weighted magnetic resonance images (HARDI), Q-ball analysis can disclose the 3D organization of fibrous tissue such as the brain white matter. The resulting orientation distribution functions (ODFs) typically contain a lot of noise. Therefore we here present an ODF noise filtering technique as well as a regularization algorithm. In order to facilitate efficient fiber tractography (the 3D reconstruction of axonal fasciculi) a data reduction scheme was developed that extracts the most important fiber directions together with an estimate of their uncertainty as well as 2 new scale invariant white matter characteristics. We also present a new framework, with proof of concept, for fully ODF-based local clustering of major fiber tracts, based on matching neighbouring voxels using symmetric criteria.