Functional Mixed-Effects Model for Analyzing fMRI Data

• Activity Number: 466
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Ragnheidur Haraldsdottir*+ and Martin Lindquist
• Companies: Columbia University and Columbia University
• Keywords: fMRI ; Brain ; FDA ; Mixed-effects model

Abstract

In this work we introduce a functional mixed-effects approach for analyzing fMRI data. For each individual the goal is to model the fMRI response curve using a series of basis functions corresponding to main effect, subject effect and condition effect. We derive an EM-algorithm to efficiently estimate model parameters. We apply this method on data from a thermal pain study where the individual subjects are inflicted with pain at each of 4 different temperatures. This method allows us to decompose the fMRI response into subject specific and temperature specific components in a flexible manner.