Toward a combined tool to assist dermatologists in melanoma detection from dermoscopic images of pigmented skin lesions

► Classification of melanocytic lesions as malignant or benign from dermoscopic images. ► Border, texture, color and structure features inspired by the clinical practice. ► Method designed and tested on a database of 655 images of melanocytic lesions. ► Automatic lesion segmentation. Classification based on AdaBoost with decision trees. ► Sensitivity/specificity: 90%/77% (90%/85%) for automatic (manual) segmentation.