Melanoma skin cancer segmentation with image region growing based on fuzzy clustering mean

Melanoma is the leading cause of skin-cancer-related death worldwide. On the other hand, if found in an early stage, there is a higher likelihood of cure. For that reason, various types of imaging techniques have been investigated. Dermoscopy is the examination of skin lesions with a dermatoscope. Use of dermoscopy provides a valuable aid in diagnosing and distinguishing skin lesions, especially in the diagnosis of melanoma. The accuracy of diagnosis using dermoscopy is very important and depends on the experience of dermatologists. Visual examination is a waste of time, so there is currently wide attention paid to the development of computer-aided diagnostic systems to aid the clinical evaluation of dermatologists. Image Segmentation is very important in digital-image processing and self-discovery, with an important role to play in solving many difficult problems, particularly those related to chronic diseases, such as skin cancer. Analysis of automatic dermoscopy images usually has three stages: a) feature selection and extraction, b) image segmentation, and c) feature classification. In this thesis, using the MatLab simulation program, we developed a new algorithm to determine more accurate location of cancer area and to determine the correctness of treatment by different image methods. This thesis we combined the fuzzy clustering method with image region growing method. The performance of these methods are tested based on the accuracy, specificity and sensitivity for greater than 200 images. As results the proposed method is strong to finding the boundary of the melanoma skin cancers. We tested our method on Pedro hospital Portugal. We tested many clusters and finally 5 cluster are chosen to results. Also image region growing method based on the fuzzy had high performance than the other methods which we compared in this study. Also the best performance for accuracy, sensitivity and specificity was respectively 0.9685, 0.9542 and 0.9829. Key Words: Melanoma skin cancer, image region growing, fuzzy C-mean.