A method based on Snake model which is optimized by superpixel is proprosed for segmenting pectoral muscle region.Firs tly,the superpixel segmentation based on fusion of grayscale and texture features is appli ed to efficiently pre-segment original image into homomogenous regions with precise boundary,and these regions are described as superpixels. Secondly,the superpixels are classified based on anatomical information and gray scale of pectoral muscle. Finally,define the initial contour of the Snake model from superpixel′ s edge and optimize it by superpixel classification results.The accurate boundary of pectoral muscle is achieved by evoluting the initial contour of the Snake model.We verify the proposed method on mammograms f rom the mammographic image analysis society (mini-MIAS) database.The experimental results show that the proposed method can get pectoral muscle′s real contour for mammograms with different sizes,shapes and intensities at pectoral muscle,and it shows good robustness and noise immunity.We also compare this method with ot hers,and it shows higher accuracy and better stability.The proposed method is suitable for pectoral muscl e segmentation.