Sparse Brain Network using Penalized Linear Regression

Sparse partial correlation is a useful connectivity measure for brain networks, especially, when it is hard to compute the exact partial correlation due to the small-n large-p situation. In this paper, we consider a sparse linear regression model with a l1-norm penalty for estimating sparse brain connectivity based on the partial correlation. For the numerical experiments, we construct the sparse brain networks of 97 regions of interest (ROIs) extracted from FDG-PET data for the autism spectrum disorder (ASD) children and the pediatric control (PedCon) subjects. To validate the results, we check their reproducibilities by leave-one-out cross validation and compare the clustered structures derived from the brain networks of ASD and PedCon.