Detection of Breast Cancer Using Modified Markov Model

Abstract

A new segmentation method for medical image with intensity clustering is enclosed. In the intended approach, an improved K-means and EM algorithm are combined to develop a hybrid strategy for better clumping. The intended advent aims to exploit the ability of providing well distributed clump of K-means and the closeness of clumps provided by EM. The introductory clumps are provided by the improved K-means algorithm. This introductory clumping process outcomes in canters which are distributed in the given data. These canters form the introductory variable for EM, which afterwards uses them and repeats to find the local maxima. Experiments for synthetic and real images make evident the feasibility and superiority of the projected model.