Nagulan Ratnarajah1, Andy Simmons2, Ali Hojjat3
1Department of BioSciences, University of Kent, Canterbury, Kent, UK; 2Centre for Neuroimaging Sciences, Kings College London, UK; 3Department of BioSciences, University of Kent, UK
To resolve orientational uncertainties in diffusion tensor imaging, probabilistic fiber tracking has received considerable interest. Current approaches model the uncertainty at each voxel with a probability density function (PDF) for the fiber orientations, then propagate a streamline in randomly sampled directions, and the most probable path is estimated using the connectivity map. We consider the random path as a set of curves, and find the most probable path estimated directly from the generated curves using average curves. A simple stochastic model is presented here and two implementations evaluated using phantom and in vivo diffusion MRI data