Colloquium: "A Bayesian Modeling Approach of Multiple-Subject fMRI Data"

Abstract:  We present a Bayesian nonparametric regression model for the analysis of multiple-subject functional magnetic resonance imaging (fMRI) data. Our goal is to provide a joint analytical framework that allows the detection of regions of the brain that activate in response to a stimulus, while simultaneously taking into account the association, or clustering, of spatially remote voxels within and across subjects. The proposed modeling strategy incorporates information on both the spatial and temporal correlation structures of the data. It also employs discrete wavelet transforms as a way to decorrelate the data. The high dimensionality of the data and the large amount of parameters to be estimated pose computational  challenges. We employ variational Bayes algorithms as an approximate computational technique and compare efficiency and estimation results with respect to a full Monte Carlo Markov chain algorithm. We explore performances of the proposed model on simulated data, with both block- and event-related designs, and on real fMRI data.

Host: Todd Kuffner

Tea @ 3:45pm in Cupples I, room 200