A brain segmentation algorithm based on Markov model fused with fuzzy similarity dynamic weights

According to the fuzziness of medical image itself, this paper fused the dynamic connectedness in the Markov models. The method used the dynamic connectedness method to estimate fuzzy similarity between the pixels, and used this information to control the potential energy parameter in Markov model. The spatial correlation parameters can be changed with the image intensity and shape information. At last, we analyzed the result of experiments using the simulated images and actual clinical images of human brain MR images. The experiment result indicated that the method we proposed was better than the traditional Markov image segmentation method. It had some improvement of having higher segmentation accuracy and achieved a relatively satisfactory result.