Improved Modified Suppressed Fuzzy C-Means

This paper presents a study on the fuzzy classification techniques that have been applied to the MR images. The goal is to improve the fuzzy techniques in inventing a new classification method, called the Improved Modified Suppressed Fuzzy C-Means (IMS-FCM) which modifies another algorithm called Modified Suppressed Fuzzy C-Means (MS-FCM). The latter one works with a common parameter α based on the exponential separation strength between clusters in each iteration in order to modify the memberships degrees of the pixels and to accelerate in consequence the convergence of the standard algorithm FCM to the optimum. It´s not logical because the context differs from one pixel to another. To overcome this disadvantage we propose a new version of MS-FCM taking account of noise aspect. The former aspect is treated by a new parameter called the degree of cleanness of the pixel relatively to a class instead of α. We test the new algorithm and the FCM, S-FCM and MS-FCM algorithms in many magnetic resonance images. Overall, the new algorithm gives better results than the others.