Support Vector Machine: Melanoma Skin Cancer Diagnosis based on Dermoscopy Image

Melanoma is one of the most deadly types of skin cancer. Until now, the diagnosis of melanoma skin cancer is still using the biopsy method, which is the procedure of taking a small portion of tissue from the patient's body to be examined using a microscope. This method is considered less effective because finding the type of melanoma skin cancer requires a long preparation and a slightly longer wound healing time. Suppose this melanoma can be diagnosed from the start for patients suffering from malignant melanoma tumors. In that case, it can be treated immediately so that the percentage of patient recovery will be higher. Therefore, in this study, a rapid dermoscopy image computerized analysis system was designed so that it can be used to classify the type of melanoma image or not (nevus) with Gray Level Co-occurrence Matric (GLCM). For the classification method used, the support vector machine (SVM) is one of the supervised learning that is usually used for classifications such as (Support Vector Classification) and regression (Support Vector Regression). This method is believed to be quite good and accurate in performing dermoscopy image recognition which is used as input and for image processing. The system for diagnosing melanoma will be implemented in a graphic user interface that will facilitate the user interface experience. The results of this study are information about skin cancer from the image, whether it is melanoma or not (nevus) which can help dermatologists diagnose melanoma quickly. In this study, 2589 data training was conducted, and testing was carried out on 192 testing data. The test results obtained an accuracy rate of 83% and an error percentage of 17%.