A0779
Title: A fully Bayesian tensor basis model for multi-subject task fMRI data
Authors: Michelle Miranda - University of Victoria (Canada) [presenting]
Jeffrey Morris - University of Pennsylvania (United States)
Abstract: Task-evoked functional magnetic resonance imaging (fMRI) studies are a powerful tool for understanding human sensory, cognitive, and emotional processes. A Bayesian approach is introduced to analyze task fMRI data that simultaneously detects activation signatures and background connectivity. The joint modelling involves a subjective-specific tensor spatial-temporal basis strategy that enables scalable computing yet captures spatial correlation from nearby voxels, distant ROIs, and long-memory temporal correlation. The spatial basis involves a composite hybrid transform with two levels: the first accounts for within-ROI correlation, and the second is a between-ROI distant correlation. The proposed basis space regression modelling strategy increases sensitivity for identifying activation signatures, partly driven by the induced background connectivity that can be summarized to reveal biological insights. This strategy leads to computationally scalable fully Bayesian inference at the voxel or ROI level that adjusts for multiple testing. Moreover, a joint, fully Bayesian multi-subject model is introduced and used to gain insights into the working memory task of the Human Connectome Project.
