A bootstrap method to improve brain subcortical network segregation in resting-state fMRI data
Résumé
Brain functional networks are sets of distant cortical, subcortical or cerebellar regions characterized by coherent dynamics. While spatial independent component analysis (sICA) reproducibly detects the cortical components of these networks from resting-state functional magnetic resonance imaging (fMRI) data, little is known about their subcortical (basal ganglia) components. We propose a method to detect cortico-subcortical networks accross subjects. Cortical components are first detected using sICA. Subcortical components are then identified using a general linear model combined with bootstrap to ensure statistical robustness, and then compared with an atlas of the basal ganglia for validation.