Three-dimensional semi-automatic segmentation of intracranial aneurysms in CTA

This paper proposes a method for the semi-automatic segmentation of cerebral aneurysms from CTA datasets. The method consists of two phases: a region growing-based approach followed by a level set method, firstly to extract the cerebral artery and then to segment the aneurysm. The first phase automatically locates a seed point for initialization of the region growing in a seed slice, detection of which is based on a measure of maximum entropy, combined with prior anatomical knowledge. Two masks are defined to confine the region growing. A halting criterion is defined as the intensity difference between each voxel and its neighbour.