Melanoma classification using texture and wavelet analysis

Melanoma is the deadliest skin cancer that develops in melanocytes cell that produces melanin. This cancer is caused by ultraviolet light (UV) radiation. We use wavelet method (Discrete Wavelet Packet Transform (DWPT)) and texture feature extraction (Gray-Level Co-occurrence Matrix (GLCM), and Local Binary Pattern (LBP)) as feature extraction in this study to classify melanoma cancer. We divided this research into 4 phases. First, we used hair removal data and hair removal data with augmentation for 3 classes. Second, we used combination data with and without hair removal for three classes. Third, we used hair removal data and hair removal data with augmentation for 2 classes, and the last we used combination data with and without hair removal for two classes. The classification results of 2 classes (Melanoma and Non-Melanoma) were better than the results of the classification of 3 classes (Atypical Nevus, Common Nevus, and Melanoma). The system cannot distinguish between Atypical Nevus and Common Nevus, that's why the classification results of 2 classes were better than three classes. These two classes have characteristics that are almost similar to color and texture. When these two classes are combined, the classification results are better because the classes Nevus and Melanoma, have quite clear differences in terms of color and texture.