Skeletonization for reinitialization in level-set-based vascular tree segmentation

摘要: An extension to level set based segmentation is proposed for vascular tree delineation. The method starts with topology extraction, by a shape constrained evolution steered strictly positive, image base speed function ensure some oversegmentation. Next, the skeleton of resulting oversegmentation determined, which then used initialise another both negative and positive forces on features, obtain most accurate segmentation. novelty our approach lies in constraint that imposed implicitly first evolution. We apply repeatedly re-initializations this preserving current zero set. compare plain same full range function. Both are initialised placing single seed point at root vessel tree. Pilot experiments twelve multislice CT data sets Circle Willis show capable segmenting smaller branches distal part structures has potential segment vessels severe stenosis or occlusion.