Functional Connectivity-Based Parcellation of Human Medial Frontal Cortex via Maximum Margin Clustering

Recently, the utilization of resting-state functional connectivity MRI (rs-fcMRI) for defining specialized functional regions has been explored. In this study, we employed a novel clustering algorithm, maximum margin clustering (MMC), for group parcellation of human medial frontal cortex using rs-fcMRI data. Experimental results showed that the medial frontal cortex can be divided into anterior and posterior clusters. The border between the two clusters located close to the vertical commissure anterior line, suggesting the homologues of the pre-supplementary motor area (pre-SMA) and supplementary motor area (SMA). The pre-SMA cluster exhibited stronger connectivity with cognitive-processing regions, whereas the SMA cluster exhibited stronger connectivity with the motor cortex. Our findings demonstrate that rs-fcMRI represents a useful and effective tool for parcellation of human cortex into functional regions.