B0351
Title: Fused graphical lasso for brain networks with symmetries
Authors: Saverio Ranciati - Universita di Bologna (Italy) [presenting]
Alberto Roverato - University of Bologna (Italy)
Alessandra Luati - Imperial College London (United Kingdom)
Abstract: Neuroimaging is the growing area of neuroscience devoted to producing data with the goal of capturing the processes and dynamics of the human brain. We consider the problem of inferring the brain connectivity network from time-dependent functional magnetic resonance imaging (fMRI) scans. To this aim, we propose the symmetric graphical lasso, a penalized likelihood method with a fused type penalty function that takes into explicit account the natural symmetrical structure of the brain. A symmetric graphical lasso allows one to simultaneously learn the network structure and a set of symmetries across the two hemispheres simultaneously. We implement an alternating directions method of multipliers algorithm to solve the corresponding convex optimization problem. Furthermore, we apply our methods to estimate the brain networks of two subjects, one healthy and one affected by mental disorder, and compare them with their symmetric structure. The method applies once the temporal dependence characterizing fMRI data has been accounted for. We compare the impact of the analysis of different detrending techniques on the estimated brain networks. Although we focus on brain networks, the symmetric graphical lasso is a tool that can be more generally applied to learn multiple networks in the context of dependent samples.
