<![CDATA[According to the WHO, about 2 to 3 million non-melanoma.non-melanoma skin cancers and 132,000 melanoma skin cancers occur globally every year, making up one out of every three cancers.globally each year, and account for one in every three cancers diagnosed.diagnosed. In Indonesia, skin cancer is listed as the cancer with the third highestincidence after uterine cervical and ovarian cancer, and breast cancer.Skin cancer can be detected with dermoscopy. Dermoscopy is a non-invasive diagnostic technique using optical magnification that allows visualization of morphologicHowever, this cannot be done optimally because it still relies on manual analysis so it cannot classify skin cancer types on larger datasets with potential errors and low accuracy. To accurately determine the type of skin cancer,a better classification method is needed. The purpose of this research is to determine the accuracy of skin cancer calcification using Convolutional Neural Network (CNN), support vector machine (SVM), K-nearest neighbor (KNN) models. The datasheet used amounted to 2,239 containing skin cancer images with class division 114 actinic keratosis, 376 basal cell carcinoma, 95 dermatofibroma, 438 melanoma, 357 nevus, 462 pigmented benign, 77 seborrheic keratosis, 181 squamos cell, 139 vascular lesion. The results showed that the convolutional neural network (CNN) algorithm model obtained a sensitivity of 92.59%, specificity of 99%, precision of 93%, F1-Score of 93.01%, and accuracy of 98.35%. For the KNN algorithm model, 57.77% sensitivity, 94.53% specificity, 64.25% precision, 55.99% F1-Score, and 90.45% accuracy were obtained. And for the SVM algorithm model, 61% sensitivity, 94.81% specificity, 70.23% precision, 61.26% F1-Score, and 91.17% accuracy were obtained.]]>
Copyrights © 2025
