Optimal binary thresholding segmentation for medical images in rough fuzzy set framework

Medical image segmentation is a key procedure for medical image comprehending and explaining. Its main aim is to segment the interesting objects from the surroundings by describing the information of the object. Binary thresholding segmentation technique, due to its advantages of finding close edges of objects as well as computational complexity, has been extensively investigated. How to select an appropriate thresholding to partition a medical image is addressed. This paper interprets an image to be a fuzzy set and introduces definitions of object rough fuzzy set and background rough fuzzy set of the image under a certain granularity. Meanwhile, a kind of rough entropy that makes a compromise between object roughness and background roughness is presented to determine an optimal thresholding of image segmentation. Experimental results demonstrate effectiveness and feasibility of the proposed algorithm of medical image segmentation.