Time-Varying Connectivity Models for Brain Imaging Data

• Activity Number: 590
• Type: Topic Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Imaging
• Author(s): Ivor Cribben*+ and Yi Yu
• Companies: Alberta School of Business and University of Cambridge
• Keywords: graphical models ; sparsity ; networks ; dynamic networks ; fMRI

Abstract

Recently, the use of graphical models for estimating functional connectivity or brain networks has become increasingly popular in the field of neuroscience. In most functional magnetic resonance imaging (fMRI) studies, the networks between brain regions are assumed to be stationary over time. However, there is now more evidence that the network or functional connectivity is changing over time even when the subjects are at rest. In the first part of this talk, we introduce a novel method that dynamically clusters brain regions or voxels by their functional connectivity. This new method allows for situations where the number of brain regions is greater than the number of time points in the experimental time course (n< p). This method promises to offer deeper insight into the inner workings of the brain. In the second part of this talk, we propose a new nonparametric method for estimating time varying graphical structures, or networks between brain regions, that allows for smoothly changing graphs over time. We apply both new methods to simulated data and to a resting-state fMRI data set.