Hill Climbing Optimization and Fuzzy C-Means Clustering for Melanoma Skin Cancer Identification and Segmentation

Abstract

Even though melanoma is not common as other skin cancers, it is one of the most dangerous skin cancer that grow any part of the skin such as chest, back, face, neck and legs. According to World Health Organization, the extreme sun UV radiation causes about 60,000 early demises every year worldwide. More than 80% of these deaths accounted for melanoma cancer. The proposed work explains the detection and segmentation of Melanoma skin cancer using hill climbing and FCM combined process. The process initiated with the transformation of image from RGB to CIELab space. The 3D histogram of CIELab image provides seeds for FCM. They act as initial clusters for the segmentation process. The experimental results of the proposed approach clearly illustrates that outcome of the process mainly depends on the number initial seeds.