Automatic extraction of sulcal lines on the cortical surface using shortest path probability maps

This paper describes an automatic procedure for extracting sulcal lines from cortical surface meshes of the human brain, which will serve as a tool for landmark extraction as well as for investigating the morphometry of sulci. The procedure consists in a sequence of steps, including sulcal basin segmentation based on local curvature information, estimating a bundle of depth-constrained geodesic paths and determining a robust probability map of sulcal lines crossing. In this experiment, we present quantitative validations on two main sulci to observe the agreement of our method with manually traced curves.