Digital Image Analysis System for Skin Disease Detection and Classification Using the KNN Method

Abstract

Skin diseases are a public health issue requiring prompt attention; however, a shortage of medical personnel in remote areas often hinders early diagnosis. This study developed a digital image analysis system for the automated detection and classification of skin diseases using the K-Nearest Neighbor (KNN) method. The research utilized a dataset of 7,913 images covering five disease types: Acne, Measles, Melanoma (skin cancer), Tinea (ringworm), and Varicella (chickenpox). The data was split into 80% for training and 20% for testing. System stages included image preprocessing and color feature extraction using RGB, HSV, and YCrCb color spaces. Feature vectors were constructed based on the mean values ​​of each color channel. Classification was performed using the KNN algorithm with Euclidean distance calculation. Performance was evaluated using a confusion matrix and metrics including accuracy, precision, recall, and F1-score. Test results demonstrated that the system effectively classifies skin diseases. Based on the confusion matrix, the system successfully classified images across the categories of Acne, Measles, Melanoma, Tinea, and Varicella. The Melanoma class yielded the best performance due to its highly distinctive visual characteristics. The system also features an interactive user interface (GUI) to facilitate the analysis process. Consequently, this system has the potential to serve as an effective, rapid, and user-friendly tool for the early diagnosis of skin diseases for both the general public and medical professionals.