Extraction of 3d vascular tree skeletons based on the analysis of connected components evolution

摘要: The article is dealing with the automated extraction of branching structures in 3D medical images. A generic object-oriented programming framework proposed, which most existing iterative algorithms for centerline tubular objects can be efficiently implemented, and bifurcations handled. New thus easily derived. We describe a simple algorithm fast structure vascular tree, has been implemented within this framework. recursively tracks branches detects by analyzing binary connected components on surface sphere that moves along vessels. It assumes vessels locally separated from background an appropriate adaptive threshold. originality resides analysis evolution during growth allows it to cope local abrupt changes vessel diameter shape. was successfully tested 16 magnetic resonance angiography Its accuracy assessed comparing resulting axes those extracted reference algorithm. distance between them less than one voxel except bifurcations, where maximum 3.8 voxels.