Melanoma skin cancer detection using color and new texture features

Melanoma is the most prevalent skin cancer and sometimes it is very difficult to diagnose. Noninvasive dermatoscopy is used to diagnose type of cancer. Since proposed method is based on eye-deduction, diagnosis of melanoma in early stage is difficult for dermatologist. A new algorithm is presented to classify dermoscopic images into malignant and benign. Initially the images were segmented using active counter model and two features such as texture and colorful components were extracted. Texture-based features were first in this area used to diagnose disease and its results indicated high-efficacy. In the international skin imaging collaboration dataset we achieve accuracy of 97% by support vector machine classifier.