Abstract
Melanoma is the most dangerous of skin cancer types and causes the most deaths. This paper aimed to provide a usable Computer Aided Diagnosis (CAD) system that helps dermatologist in the diagnosis of skin cancer. The proposed CAD system called Skin Cancer Computer Aided Diagnosis support system (SCCAD) consisted of six components, namely; image acquisition, image pre-processing, segmentation, features extraction, image classification and viewing result. Image pre-processing is achieved by various pre-processing approaches. Image segmentation is based on Otsu's threshold method. The extracted features were texture, color, and shape. These features became the input to the Support Vector Machine (SVM) classifier to classify the lesions as melanoma or non-melanoma. We obtained the dermoscopic images from the PH2 and the digital image archive of the Department of Dermatology of the University Medical Center Groningen (UMCG) databases. We evaluated the performance of the classification model by using 10-Fold cross-validation and the confusion matrix. In addition, we compared between the SVM and ensemble classifier. The accuracy values of SVM and ensemble are 92.6%, 91.1% respectively. In addition, we evaluated the usability of the CAD system by informal study with Human Computer Interface (HCI) experts.