A Novel Cerebrovascular Segmentation Algorithm Based on Gauss-Markov Random Field Model

In order to solve the thorny cerebrovascular segmentation problems about cerebral vessels of many branches, small shape, special position and complex patterns, this paper presents a novel statistical method to achieve effectively the accurate segmentation of cerebral vessels. Firstly, the Markov random field information is added to the statistical model which makes the full use of the spatial neighborhood information of each pixel and a new Markov statistical model is proposed; then Stochastic versions of the Expectation Maximization (SEM) algorithm is used to estimate parameters of the Markov model and the optimal solution is found, which finishes the three-dimensional cerebrovascular segmentation. Experimental results show that the proposed method can not only segment the large vessel branches, but also have a good effect on small vessels segmentation because of considering neighborhood information of each pixel. Therefore, the proposed method also has the far-reaching significance to the clinical prevention and diagnosis of cerebrovascular diseases.