Artificial Intelligence-Aided Tooth Detection and Segmentation on Pediatric Panoramic Radiographs in Mixed Dentition...

Area of Science:

• Dentistry
• Artificial Intelligence
• Medical Imaging

Background:

• Accurate tooth identification in mixed dentition panoramic radiographs (PRs) is crucial for detecting eruption disturbances.
• Current methods rely heavily on clinician experience due to developmental variability.

Purpose of the Study:

• To develop a deep learning model for automated tooth detection and segmentation in pediatric PRs during mixed dentition.

Main Methods:

• A customized YOLOv11 model was trained on 250 pediatric PRs using transfer learning from adult PRs and manual refinement.
• Performance was evaluated using mean average precision (mAP) and F1-score.

Main Results:

• The model achieved high detection performance (mAP = 0.963, F1 = 0.953) and segmentation accuracy (mAP = 0.890).
• Excellent accuracy was observed for permanent teeth (F1 = 0.977) and clinically acceptable accuracy for deciduous teeth (F1 = 0.884).

Conclusions:

• The automated system demonstrated near-expert accuracy in tooth detection and segmentation.
• This AI framework supports early detection of anomalies and reduces clinician variability in mixed dentition analysis.