A Multiscale Analysis of Functional MRI with a Three Components Hemodynamic Response Function

• Activity Number: 20
• Type: Topic Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Marco A.R. Ferreira*+ and Yuan Chen and Jeff Johnson and Jeff Rouder
• Companies: University of Missouri and University of Missouri and University of Missouri and University of Missouri
• Keywords: Bayesian inference ; massive datasets ; Wavelets

Abstract

We develop a novel multiscale methodology for functional magnetic resonance imaging (fMRI) analysis based on a three components hemodynamic response function. Specifically, we propose a novel hemodynamic response function that is a mixture of three gamma densities. In addition, we link the voxel-wise fMRI parameters with a multiscale spatial prior. Further, to estimate the model parameters we develop a Markov chain Monte Carlo algorithm. Our hemodynamic response function and multiscale spatial prior are flexible enough to accommodate distinct physiological responses in different parts of the brain. We illustrate our methodology with the analysis of a single-subject fMRI visual task experiment.