Graphical-model-based Multivariate Analysis of Functional Magnetic-Resonance Data

Rong Chen,Edward H. Herskovits
DOI: https://doi.org/10.1016/j.neuroimage.2006.11.040
IF: 5.7
2007-01-01
NeuroImage
Abstract:We propose a method for the analysis of functional magnetic-resonance (fMR) data, based on a Bayesian-network representation. Our method identifies multivariate linear/nonlinear voxel-activation pattern differences across groups, which may provide information complementary to that resulting from a general linear model (GLM)-based analysis. In addition, we describe a model-stabilization method based on data resampling, which may be helpful in the presence of small numbers of subjects, or when data are noisy.
What problem does this paper attempt to address?