Related Experiment Video
Updated: Jan 8, 2026
05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
1.3K
Caries Detection in Primary Molars with Bitewing Radiographs through Deep Learning Based-Object Detectors.
Zhi Qin Tan1,2, Ilana Felix Pinho3, Ryan Banks2
1Faculty of Dentistry, Oral and Craniofacial Sciences, King's College London, London, UK.
Caries Research
|December 18, 2025
Summary
Artificial intelligence, specifically DINO and YOLOv7 algorithms, shows promise for detecting and staging dental caries in children
Area of Science:
- Dentistry
- Artificial Intelligence
- Medical Imaging
Background:
- Automated caries detection aids in prioritizing treatment for children.
- AI algorithms can improve diagnostic accuracy for caries lesions.
Purpose of the Study:
- Develop AI algorithms using object detection for caries detection and staging in primary molars.
- Utilize bitewing radiographs for AI model training and evaluation.
Main Methods:
- Trained five deep learning object detection algorithms on 1,023 primary molar bitewing radiographs.
- Radiographs were annotated for four caries severity stages.
- Evaluated models using sensitivity, specificity, accuracy, and weighted kappa scores across different lesion thresholds.
Main Results:
- The DINO model achieved the highest concordance (kappa=0.513) for staging caries.
- DINO demonstrated high sensitivity for detecting all caries (0.509) and dentine caries requiring treatment (0.659).
- YOLOv7 showed excellent performance with specificity >0.98 and accuracy >0.91.
Conclusions:
- DINO and YOLOv7 algorithms are effective for detecting caries in primary molars on bitewing radiographs.
- These AI tools have potential clinical applications to assist dentists in daily practice.
