Parcellation of Human Inferior Parietal Lobule Based on Diffusion MRI

摘要

The function of a brain region can be constrained by its anatomical connections. The Inferior Parietal Lobule (IPL) is a cortical region with marked functional heterogene- ity, involved in visuospatial attention, memory, language and mathematical cognition. In this work three different variants of the normalized graph-cut clustering algorithm were applied to obtain a parcellation of the IPL of living subjects into com- ponent regions based on the estimate of anatomical connectivity obtained from diffusion tensor tractography. Results over the three different algorithms were compared and a new metric proposed to measure the quality of individual parcellations by comparing to standard atlas regions. In this study of 19 subjects, an average of 64% overlap with the Juelich brain atlas was observed.