Evaluation of the Performance of a YOLOv10-Based Deep Learning Model for Tooth Detection and Numbering on Panoramic Radiographs of Patients in the Mixed Dentition Period

<b>Objectives:</b> This study evaluated the performance of a YOLOv10-based deep learning model in detecting and numbering teeth in the panoramic radiographs of pediatric patients in the mixed dentition period. <b>Methods:</b> Panoramic radiographic images from 200 pediatric patients in the mixed dentition period, each with at least 10 primary teeth and underlying permanent tooth germs, were included in the study. A total of 8153 teeth in these panoramic radiographs were manually labeled. The dataset was divided for the development of a YOLOv10-based artificial intelligence model, with 70% used for training, 15% for testing, and 15% for validation. <b>Results:</b> The precision, recall, mAP50, mAP50-95, and F1 score of the model for tooth detection and numbering were found to be 0.90, 0.94, 0.968, 0.696, and 0.919, respectively. <b>Conclusions:</b> YOLOv10-based deep learning models can be used to accurately detect and number primary and permanent teeth in the panoramic radiographs of pediatric patients in the mixed dentition period, which can support clinicians in their daily practice. Future works may focus on model optimization across varied pediatric cases to enhance clinical applicability.