Abstract
This paper describes an ITK implementation of a 3D vascular segmentation filter using a graph-based energy minimization algorithm. The method first computes the shortest path between two user provided points in the vessel and then performs a graph min-cut based segmentation of the vessel based on the intensity information coupled with the computed path as a spatial constraint. The shortest path computation is adapted from the algorithm in [1] to use a vesselness based weighting function. The min-cut algorithm uses the Boost graph library to calculate the minimal cut. Several examples of applications are provided, along with images of the resulting segmentations.