Penerapan Convolutional Neural Network (CNN) Untuk Deteksi dan Klasifikasi Penyakit Kulit Pada Citra Dermatoskop

Abstract

Early detection of skin diseases is crucial to improving treatment success and reducing the risk of complications. With advances in image processing technology and deep learning, automated diagnosis can help reduce the burden on medical personnel and improve screening efficiency. This study developed a skin disease classification system using a Convolutional Neural Network (CNN) on the BCN20000 dataset (ISIC collection 249), which includes 18,946 dermatoscopic images of various types of skin lesions. This research was conducted through several main stages, namely image pre-processing, which included resizing, normalization, and augmentation; dividing the dataset into training, validation, and testing sets; training the CNN model by applying regularization techniques and hyperparameter optimization; and concluding with model evaluation using classification metrics. The test results showed that the proposed model achieved an accuracy of 98.6%, with high precision, recall, and F1-score values for each skin lesion class. This model shows potential as a tool for initial screening of skin diseases and can be further integrated into medical decision support systems.