A Bayesian Hierarchical Framework for Modeling Brain Connectivity of Neuroimaging Data

• Activity Number: 120
• Type: Topic Contributed
• Date/Time: Monday, July 30, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Shuo Chen*+ and F. DuBois Bowman
• Companies: Emory University and Emory University
• Keywords: brain imaging ; functional connectivity ; Bayesian hierarchical model ; resting-state fMRI ; MCMC

Abstract

We propose a novel Bayesian hierarchical model that unifies voxel level and region level brain connectivity analysis, and yield population level inferences based on brain imaging data sets. The first level of our model summarizes functional connectivity cross-region voxel pairs using a two component mixture model consisting of connected and non-connected voxel mixture model. Next, we use the proportion of connected voxel pairs to quantify the between region connectivity extent. Furthermore, we evaluate the impacts of clinical covariates on region pair connectivity at a population level. We perform the massive parameter estimation using Markov Chain Monte Carlo (MCMC) techniques, which can be executed quickly despite the numerous number of voxel pairs. We apply our method to a resting-state fMRI data set of 32 subjects for a depression study and simulated data sets to demonstrate the properties of our method. Our proposed method can reveal more brain connectivity information than conventional methods.