Yogesh Rathi1, James Malcolm2, Sylvain Bouix1, Gordon Kindlmann3, Carl-Fredrik Westin3, Marek Kubicki1, Martha E. Shenton1
1Psychiatry, Harvard Medical School, Boston, MA, USA; 2Psychiatry, Brigham and Women's Hospital; 3Radiology, Harvard Medical School
In this work, we propose to use a mixture model of Watson directional functions for representing the signal and computing the diffusion ODF and fiber ODF. The parameters of the model can be directly used for multi-directional streamline tractography, which allows to track fibers not possible with single-tensor tractography methods. In addition, the model parameters live in a vector space and hence interpolation between ODF's is straightforward. The number of parameters required to represent each fiber is 3 (2 for direction in spherical coordinates and 1 scaling). Thus, the representation is also concise.