Active contour based appearance priors applied to tumours segmentation

Segmentation of tumor from Magnetic Resonance (MR) images is difficult task. In fact this due to the large diversity in shape and appearance of tumors regions with intensities overlapping the normal brain tissues. An expanding tumor can also prevent and deform nearby tissue. In this paper, we proposed Active Contour segmentation based method that incorporates histograms of clustered features and appearance priors to separate the tumor from brain tissue. Experimental results on difficult cases have drawn a very good performance of proposed segmentation method.