An Algorithm for Step-wise Skeletonization of Blood Vessel Network

Abstract

Skeletonization is an iterative process frequently used in image processing to reduce the pictorial content of an object to a graph representing only the general shape of the object. We present an algorithm that extracts the skeleton from a set of segmented blood vessels. To accommodate the complexities of the blood vessels’ shape, our solution combines some properties of the algorithms based on the distance-transform, iterative thinning, and Voronoi-diagram. In addition, the algorithm employs procedures for smoothing and centering of the estimated skeleton’s points. These ensure the skeleton points are kept inside the region of interest. Numerical experiments with simulated 2D and real data proved the efficiency of the algorithm in extracting skeletons of blood vessels. Next we plan to extend the algorithm so that it is usable in 3D space as well. Ill. 7, bibl. 7 (in English; summaries in English, Russian and Lithuanian).