A Review Different Methods to Detect Skin Cancer using Image Processing

Malignant melanoma is that the foremost hazardous kind of human carcinoma and its incidence has been rapidly increasing. Early detection of melanoma in dermos copy images is extremely important and important, since its detection within the early stage are often helpful to cure it. Computer Aided Diagnosis systems are often very helpful to facilitate the first detection of cancers for dermatologists. In this paper, system uses a novel method for the detection of melanoma skin cancer. To detect the hair and a number of other noise from images, pre-processing step is administered by applying a bank of directional letters and thus, Image in painting method is implemented to ll within the unknown regions. System will test on publicly available PH2 dataset in terms of accuracy, sensitivity, specificity and Area under ROC curve (AUC). It is observed that good results are achieved using de-duplication concept on features to avoid same feature or image in training and testing phase, hence proving the validity of the proposed system.