Connectivity-based parcellation of functional SubROIs in putamen using a sparse spatially regularized regression model

Abstract

In this paper, we present a novel framework for parcellation of a brain region into functional subROIs (Sub-Region-of-Interest) based on their connectivity patterns with other brain regions. By utilising previously established neuroanatomy information, the proposed method aims at finding spatially continuous, functionally consistent subROIs in a given brain region. The proposed framework relies on (1) a sparse spatially-regularized fused lasso regression model for encouraging spatially and functionally adjacent voxels to share similar regression coefficients; (2) an iterative merging and adaptive parameter tuning process; (3) a Graph-Cut optimization algorithm for assigning overlapped voxels into separate subROIs. Our simulation results demonstrate that the proposed method could reliably yield spatially continuous and functionally consistent subROIs. We applied the method to resting-state fMRI data obtained from normal subjects and explored connectivity to the putamen. Two distinct functional subROIs could be parcellated out in the putamen region in all subjects. This approach provides a way to extract functional subROIs that can then be investigated for alterations in connectivity in diseases of the basal ganglia, for example in Parkinson's disease.