Automatic Tooth Enumeration on Panoramic Radiographs Using Deep Learning

Accurate tooth numbering is essential for dental procedures, treatment planning, and patient record management. Automatic tooth enumeration in dental panoramic radiographs is crucial in modern dental image analysis and diagnosis. Tooth enumeration in panoramic radiographs has relied on handcrafted features and traditional image processing techniques. Still, these methods often lack the accuracy and robustness required for complex cases and varied image qualities. Despite the advancements in deep learning-based object detection algorithms, a significant research gap remains in the specific domain of automatic tooth enumeration on panoramic radiographs. This research paper presents an innovative approach for automatic tooth enumeration on panoramic radiographs using the state-of-the-art You Only Look Once (YOLO) object detection framework focused on implementing the YOLOv5 library. The YOLOv5 model is evaluated as an efficient system capable of accurately detecting and enumerating individual teeth from panoramic radiographs. The 612 panoramic radiograph images evaluation shows the bounding-box results show the best tooth detection rate in the YOLOv5x model. The YOLOv5 model is generally very good at predicting tooth enumeration on panoramic radiographs in a relatively small dataset. This advancement is expected to enhance dental diagnosis and treatment planning, benefiting dental professionals and patients.