Early stage detection and classification of melanoma

Abstract

This paper proposes and describes a system, to classify skin lesion as malignant or benign. There are over 200 different known cancer in which malignant melanoma, is the most dangerous form of skin cancer, and have a good prognosis if treated in the curable early stages. The purpose of the research is to develop, implement and test an application, on clinical images taken using high speed cameras, to detect melanoma in early stage. The proposed system is based on TDS (Total Dermoscopic Score) calculation with preprocessing and OTSU's segmentation method for classification of skin lesion as non-cancerous (Benign), suspicious or cancerous (melanoma). The system has been tested on a dataset of 30 lesions (16 images of malignant lesions and 14 benign lesions). The result of experiment shows that preprocessing steps and OTSU's segmentation with ABCD algorithm provide more accurate result.