AI-Based Intraoral Videography for Automated Dental Inspection and Charting in Children With Mixed Dentition

Examination and documentation (charting) of teeth are indispensable yet labor-intensive processes, especially in pediatric patients, where mixed dentition, anatomical variability, and limited cooperation during imaging present unique challenges. Despite their distinguished potential, existing AI models lack applicability to pediatric dentistry because they are mostly designed for adult dentition and use radiographic images. To fill this gap, we developed an AI-powered model for automated pediatric dental charting using real-time intraoral videos, utilizing a YOLO-based object detection framework. The model was trained to classify 44 tooth types, including primary and permanent dentitions. The dataset is composed of 112,538 frames extracted from 89 intraoral footages of children aged 6-12 years at diverse dental development stages. Of this dataset, 80% was allocated for training and 20% for testing. The model achieved a mAP@0.5 of 0.405, with a precision of 0.495 and a recall of 0.405 across all tooth classes. Notably, the model performed substantially better in detecting primary teeth, achieving a mAP@0.5 of 0.616, compared to 0.255 for permanent teeth, due to the latter's ongoing eruption and inconsistent appearance. Despite these limitations, this study is a major advancement toward automating pediatric dental charting and will pave the way for future developments in AI applications for pediatric dentistry, facilitating early caries detection for children in schools and in large-scale public health screening programs.