Detection of Dental Caries Based on GV Black Classification Utilizing Yolov8 from X-ray Images

Tooth decay, a prevalent ailment, often eludes detection through conventional visual and tactile examination, especially in its early stages. This study presents a new approach utilizing Yolov8-based multiclass classification on X-ray images to identify dental caries according to the GV Black classification. The classification has seven categories: classes from I-VI, and the normal dental structure. The study implemented the Y olov8 model on the datasets and assessed its accuracy, precision, and recall for detecting dental caries. The performance metrics evaluation shows good result for the You Only Look Once (Yolov8), a real-time object detection model. The model displays a recall of 84.9%, mAP50 of 91.5%, mAP50-95 of 73.6%, precision of 96.3%, and mAP50 of 91.5% without data augmentation such as flip, rotation, and shear. The scores improved to 96.6% precision, 92.1% recall, 95.1% mAP50, and 75.9% mAP50-95 after data augmentation. These results highlight the efficacy of the Yolov8 model for detecting dental caries based on GV Black Classification, while at the same time indicating further scopes of refinement. The performance according to the values presented can be compared to existing models as it also surpassed the average benchmarks. It is one of the most promising applications of artificial intelligence and deep learning in the medical field, especially in dental diseases detection.