Preliminary work on dermatoscopic lesion segmentation

Dermoscopy has become the primary tool used for pigmented skin lesion diagnosis providing better quality and accurate images. Computer-Assisted Image Interpretation is a new direction that involves the automatical lesion detection, feature extraction and classification (benign or malignant). This paper refers to several prior pre-processing enhancement techniques and an automated segmentation method. We have tested our methods on 60 dermoscopic images and compared the automated segmentation results with dermatologist-determined segmentation using an area percentage error.