Automated Melanoma Types and Stages Classification for dermoscopy images

Skin is important organ of our body which covers muscles, bones and also other parts of body[4]. Skin is more exposed to outer environment. So, skin is affected to UV rays, dust and pollution. Chances of skin dieses is more than other organ of our body. Melanoma is a kind of skin cancer that begins in melanocytes cell[18]. It can influence on the skin only, or it may expand to the bones and organs. It is less common, but more serious and aggressive than other types of skin cancer. Majority of deaths related to skin cancer occur due to Melanoma over the world. For effective treatment it is very important to melanoma identified earlier as possible. As well as detection of the stages of melanoma to recognize depth of spreading of melanocyte cell in other organ of body. Process of Detection of Skin cancer is difficult, expensive and time consuming process [17]. Computerized image analysis is diagnose skin diseases earlier based on past information. Purpose of of this research is to extract more features of skin lesion images for more accurate recognition the types of Melanoma and decrease ratio of false diagnosis. Using YCbCr color space for segmentation and detection of skin color segmentation. Then k-means clustering techniques are used to reduce the computational cost by avoiding feature calculation for every pixel in the image. Texture Haralic 13 feature are extracted .In, this system gives accuracy of classification of melanoma types using SVM 87.05%, with ANN 75%, R-F classifier 90%. So, R-F classifier gives better classification result.