Classification of PH2 Images for Early Detection of Skin Diseases
Ebrahim Mohammed Senan, Mukti E. Jadhav, Avinash A. Kadam
Abstract
Melanoma is one of the most deadly types of skin diseases, but there is an opportunity to survive in case of an early diagnosis. Our paper offers a contribution to assist dermatologists and experts to save time and effort to diagnose and treat melanoma in its early stages. Segmentation is necessary to focus the system on the skin lesion only, so we apply the Active contours algorithm to isolate the lesion area from the rest of the image. The PH2 dataset contains many features. In our proposed system, we focus on extracting color features through the Fuzzy color histogram (FCH) technique. The SVM, K-NN, ANN and FFNN classifiers present promising results, with 100% accuracy, 100 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> sensitivity and 100% specificity.