Estimation of hemodynamic response functions for un-delineated overlapping rois in fMRI data based on sparse dictionary learning

Extracting region-specific hemodynamic response function (HRF) from overlapping ROIs of activated areas in the brain in noisy functional Magnetic Resonance Imaging (fMRI) data is essential when analyzing the temporal dynamics of a brain region response and its neuronal coupling for functional and effective connectivity. Based on the assumption of spatially sparse brain hemodynamics, HRFs from jointly activated overlapping regions are separated based on sparse dictionary learning. The proposed HRF estimation procedure is tested on both simulated and real fMRI data. The results reveal the efficiency of the proposed method in separating temporally-dependent inter-region HRFs from fMRI data.