Online Program Home
  My Program

Abstract Details

Activity Number: 63 - Statistical Methods for Brain Connectivity and Network Analysis
Type: Topic Contributed
Date/Time: Sunday, July 30, 2017 : 4:00 PM to 5:50 PM
Sponsor: Section on Bayesian Statistical Science
Abstract #324666 View Presentation
Title: Clustering DTI Data to Identify White Matter Fiber Bundles Using a Mixture of Langevin Distribution
Author(s): Subhadip Pal*
Companies: University of Louisville
Keywords: DTI ; NEUROIMAGING ; STIEFEL MANIFOLD ; MATRIX LONGEVIN DISTRIBUTION ; CLUSTERING
Abstract:

Estimating major fiber bundles inside the white matter region of the brain based on the Diffusion Tensor has been a challenge to the Neuroimaging community. A dominant approach in literature that has been developed to tackle the problem is to employ probabilistic and deterministic tractography algorithms. However, the tractography methods are typically slow and mainly use only one principle direction of the Diffusion Tensors. In this project, we adopt a different approach where we cluster the Diffusion Tensors based on the first two principal directions, which are elements in the space of Stiefel manifold. We use a mixture of Matrix Langevin distribution, as it is one of the popular distribution defined on the Stiefel manifold. A major contribution of this project is to come up with a feasible Bayesian computation related to the Matrix Langevin distribution, which involves intractable normalizing constant. The method we have developed has been successfully applied to various simulated and real data sets.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association