Analysis and Diagnosis of Dental Panoramic Radiographs Based on Convolutional Neural Networks

Many people often underestimate the importance of oral health, but in fact, oral diseases are closely related to overall health. If not detected and treated in a timely manner, they may have long-term serious effects on overall health. This paper proposes an automated model for diagnosing oral diseases and analyzing oral images based on convolutional neural networks. The model is designed for analyzing dental panoramic radiographs, covering four key modules: tooth enumeration, disease detection and lesion image optimization, caries segmentation, label matching and statistical analysis. It also provides a visualization interface to display the results. Compared to traditional target detection methods, YOLOv8 offers higher detection speed and better performance. Dental panoramic radiographs contain a large amount of detail, and this paper utilizes YOLOv8 and U-Net for detection and segmentation.The tooth enumeration precision of this study reaches 0.986, disease detection achieves 0.787, and the caries segmentation mPA reaches 88.05%. With this model, patients can fully understand their own oral health issues, while doctors can obtain more accurate auxiliary diagnostic and treatment references. This innovation brings significant progress to the field of oral healthcare, making oral health management more intelligent and precise.