Identifying Functional Co-Activation Patterns in Neuroimaging Studies via Poisson Graphical Models

• Activity Number: 362
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Wenqiong Xue and Jian Kang*+ and DuBois Bowman and Tor D. Wager and Jian Guo
• Companies: Emory University and Emory University and Emory University and University of Colorado, Boulder and Harvard University
• Keywords: Functional co-activation patterns ; Functional brain networks ; Penalized multivariate Poisson ; EM algorithm ; Permutation test

Abstract

Meta analysis plays an important role in neuroimaging research. Several approaches have been developed to determine the consistency in activated brain regions for a particular type of task, cognition, emotion or social process. In this paper, we focus on identifying the functional co-activation patterns and building a functional network in the human brain. We adopt a penalized likelihood approach to impose sparsity on the covariance matrix for region-level peak activations based on an extended multivariate Poisson model. The sparse covariance matrix is in turn used to construct a brain network. We obtain the penalized maximum likelihood estimates via the EM algorithm and optimize an associated tuning parameter by maximizing the predictive log-likelihood. We conduct permutation tests on the brain co-activation pattern network. We also discuss the choice of the penalty term and its impact on identifying the network from simulation studies. We apply our proposed method to a meta analysis of 162 functional neuroimaging studies on emotions. Our model identifies a functional network that consists of regions from the basal ganglia, limbic system, and other emotion related brain regions.