Connectivity-based parcellation of functional sub-regions from brain fMRI signals

In this paper, we propose a connectivity-based framework to extract functional sub-regions (subROIs) in the putamen from fMRI signals. The proposed framework aims to generate a network that represents the connectivity patterns of the putamen voxels among themselves, and with the other brain regions. A spatial constraint is introduced into the network generation framework to ensure the spatial continuity of the final subROIs. A eigenvalue-based community detection approach is then incorporated to sub-divide the network into two functionally connected and spatially continuous sub-regions. The framework is applied to synthetic datasets to evaluate its performance wih respect to other literature-based approaches. The proposed framework is finally applied to resting state fMRI data from five healthy subjects to parcellate the putamen region into two functional subROIs. Although the framework is developed for putamen, it can generally be applied to other brain region subROI parcellation tasks.