of Computer Science and Centre for Medical Image Computing, University
College London, London, United Kingdom
We propose a new tractography algorithm leveraging parametric models of dispersion fit to diffusion weighted magnetic resonance imaging to guide streamline propagation probabilistically. Many current tractography techniques rely on a few discrete directions per voxel which can misrepresent the underlying anatomy, opening a risk of false negative connections. We test the algorithm on synthetic data and in vivo data of a human subject. The algorithm shows advantages in tracking through the corona radiata, a region of white matter known to exhibit a significant degree of fiber dispersion. We also demonstrate that the algorithm succeeds in tracking the major white matter pathways for which standard techniques work well.