A Bayesian Approach for Multi-¬Subject Effective Connectivity Inference Using Multi-¬Modal Neuroimaging Data

• Activity Number: 202 - New Exploratory Methods and Inference Approaches for Massive Multi-Modal Data with Applications to Brain Imaging
• Type: Invited
• Date/Time: Monday, July 31, 2017 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Michele Guindani*
• Companies: University of California, Irvine

Abstract

We discuss the use of multi-¬subject vector autoregressive (VAR) models for inference on effective connectivity based on resting-¬state functional MRI data. In particular, we discuss the use of a Bayesian variable selection approach to allow for simultaneous inference on effective connectivity at both the subject-¬ and group-¬level. Furthermore, our proposal accounts for multi--modal data by integrating structural imaging information into the prior model, encouraging effective connectivity between structurally connected regions. The effectiveness of the approach is explored through simulation studies showing improved inference on effective connectivity at both the subject-¬ and group-¬level, compared to currently used methods. Our motivating application is from temporal lobe epilepsy data, where we use resting-¬state functional MRI and structural MRI. The group-¬level effective connectivity we infer include both known relationships between resting-¬state networks, as well as relationships of potential interest for future investigation.