Abstract:
|
The Bayesian approach is not uncommon in finding the direction of fiber in the area of in vivo neuroimaging using DT-MRI data, however, the prior is often chosen as the most proximal distribution known beforehand. Though such a method is widely used in the Statistical world, it does not use spatial information from neighboring voxel in the form of Bayesian prior. To incorporate the fiber orientation from the neighboring voxels while finding the fiber orientation within a voxel of interest, we used the Watson distribution, the parameter of which is calculated by using the neighboring voxels fiber orientation information. In our work, we present a comparison of our model with its counterpart of the single fiber model where no spatial information has been used. The framework developed here is quite general in nature, and we are hopeful about its application over a broad range of areas where connectivity is an essential part of the data.
|