Identification of cortical landmarks based on consistent connectivity to subcortical structures

Quantitative assessment of structural connectivities between cortical and subcortical regions has been of increasing interest in recent years. This paper proposes an algorithmic pipeline for identification of reliable cortical landmarks based on the consistent structural connectivity between cortical and subcortical regions. First, twelve subcortical regions are segmented from MRI data, and cortical surface and white matter fibers are reconstructed and tracked from magnetic resonance diffusion tensor imaging (DTI) data. Second, given that subcortical structures are relatively consistent across individual subjects, the structural connectivity from cortical to subcortical regions is extracted as the connectional attribute for each cortical region. Third, the cortex is segmented into different regions based on their cortico-subcortical connection attributes, and regions with the most consistent connectivity patterns across different subjects are selected as cortical landmarks. Experimental results from eight healthy subjects show that our approaches can identify 22 reliable cortical landmarks, a portion of which are validated via task-based fMRI data.