A Bayesian Nonparametric Approach to the Analysis of fMRI Data

• Activity Number: 38
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Linlin Zhang*+ and Michele Guindani and Marina Vannucci
• Companies: Rice and The University of Texas MD Anderson Cancer Center and Rice University
• Keywords: Bayesian Inference ; fMRI ; functional Regression Models ; clustering

Abstract

In this paper we propose a novel Bayesian approach to functional regression models, and apply the models to analyze functional Magnetic Resonance Imaging (fMRI) whose primary form is Blood Oxygenation Level Dependent (BOLD) contrast. We model the relationship between stimulus and BOLD response, and incorporate the complex temporal and spatial correlation structure in the model by employing appropriate distributions which embody our knowledge about the structure of the brain. The goals are to detect locations of the brain with strong activation by a task, and capture connectivity via clustering of brain regions.