Treffer: Integration of YOLOv8 and FastAPI for Early Detection of Nail Diseases

Nails are important indicators of various health conditions, including fungal infections (onychomycosis), autoimmune disorders (psoriasis), and subungual melanoma (black line). However, early detection of these diseases remains limited due to low accessibility and public awareness. This study aims to develop an end-to-end, web-based early detection system for nail diseases by integrating the YOLOv8 object detection algorithm with the FastAPI framework. A total of 600 annotated nail images obtained from Kaggle were categorized into four classes: healthy nail, psoriasis, black line, and onychomycosis. The model was trained using PyTorch on Google Colab with GPU acceleration and evaluated using precision, recall, and mean Average Precision (mAP@0.5). The model achieved a precision of 93%, recall of 88%, and mAP@0.5 of 89%. Manual testing on 100 images via the deployed web application showed an overall accuracy of 97%. Class-wise accuracy reached 100% for healthy nail and psoriasis, 92% for black line, and 96% for onychomycosis. These results demonstrate that the system performs reliably across various conditions. The main contribution of this study is the implementation of a real-time, web-integrated nail disease detection system that is accessible to both medical professionals and the general public. Future research may focus on expanding the dataset, optimizing model robustness under varied lighting and background conditions, and conducting clinical validation.