An automated computer aided diagnosis of skin lesions detection and classification for dermoscopy images

Abstract

Skin cancer is a deadly disease nowadays. So, early detection and prevention are essential. To classify the skin lesions in accurate manner an automatic Computer-Aided Diagnosis (CAD) for dermoscopy images were needed. The lesion segmentation is vital in the classification process. For segmenting the skin lesions many researchers have been developed different methods on melanocytic skin lesions (MSLs) and few methods for non-melanocytic skin lesions (NoMSLs), while the accurate segmentation for the variety of lesions are somewhat risky. In this K-means clustering is used for segmentation. After lesion is segmented extract the features such as color, text and shape. Many methods are used for classification but they focus only on melanocytic skin lesion i.e detecting melanoma only. Other lesion should also be classified for that a novel approach is used in this paper. The support vector machine (SVM) classifier was used for classification of skin lesions such as Melanoma, Basal cell carcinoma (BCC), Seborrhoeic keratosis (SK) and Nevus. The dataset collected from Dermweb. We used 100 NoMSLs and 220 MSLs set of images. Our classification method has achieved better accuracy as compared to others.