FMRI Connectivity Relation Analysis with Advanced Structural Equation Modeling

Functional magnetic resonance imaging (fMRI) has become an important investigative tool in neuroscience. A neuro-scientific interest shared by many fMRI researchers is the elucidation of causal interactions between cortical areas. Structural equation model (SEM) is a powerful technique that allows us to examine causal relations between variables, and thus is applicable to uncovering connectivity relation between brain regions.In most research in applying SEM to fMRI problems, individual SEM models were developed at fixed time points. In this research, we propose to utilize the two-level longitudinal SEM to model fMRI data. This allows direct incorporation of time effect into the SEM model, and thus provides more flexibility for investigating temporal and causal relations on region connectivity. We will also investigate the Bayesian approach for obtaining inferences of the two-level SEM model, and apply the newly developed SEM methodologies to two categories of fMRI data sets, to answer open neuro-scientific questions.