Skeleton extraction of cerebral vascular image based on level set model

This paper presents a new framework to calculate the continuous brain blood vessel skeleton curve of binary images and gray-scale images. The main idea of the method is: the minimum cost path between any two skeleton points is a skeleton line. Algorithm using two different intermediate functions, one is Euclidean distance field and another is deformed gradient vector flow, gained two different energy function respectively proportional to them. The topology nodes from first energy function control the shape of the object, and the second one control skeleton extraction using topology nodes as the source points. Experiment and analysis can verify the validity and robustness of this method, and not sensitive to the boundary noise.