Related Experiment Video
Updated: Oct 12, 2025
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
Deep-learning approach for caries detection and segmentation on dental bitewing radiographs
Ibrahim Sevki Bayrakdar1,2, Kaan Orhan3,4, Serdar Akarsu5
1Department of Oral and Maxillofacial Radiology, Faculty of Dentistry, Eskisehir Osmangazi University, 26240, Eskisehir, Turkey. ibrahimsevkibayrakdar@gmail.com.
This study introduces an Artificial Intelligence (AI) system using Convolutional Neural Network (CNN) algorithms for accurate dental caries detection and segmentation in radiographs. The AI model demonstrated superior performance compared to human observers, offering potential as a clinical decision support tool.
Area of Science:
- Dentistry
- Radiology
- Artificial Intelligence
Background:
- Dental caries detection and segmentation from bitewing radiographs are crucial for diagnosis and treatment planning.
- Accurate and efficient methods are needed to support clinicians in routine dental practice.
Purpose of the Study:
- To develop and evaluate an automatic caries detection and segmentation model using Convolutional Neural Network (CNN) algorithms.
- To assess the clinical performance of the proposed AI model against human observers using VGG-16 and U-Net architectures.
Main Methods:
- Utilized 621 anonymized bitewing radiographs for training an AI system (CranioCatch).
- Implemented VGG-16 and U-Net architectures with PyTorch for caries detection and segmentation, respectively.
- Evaluated model performance by comparing its accuracy with that of experienced human observers.
Main Results:
- The AI model achieved high performance rates for caries detection and segmentation.
- Sensitivity, precision, and F-measure rates were reported as 0.84, 0.81; 0.84, 0.86; and 0.84, 0.84, respectively.
- AI models demonstrated superiority over assistant specialists when tested on an external radiographic dataset.
Conclusions:
- CNN-based AI algorithms show significant potential for accurate and effective dental caries detection and segmentation in bitewing radiographs.
- Deep-learning AI algorithms can serve as valuable tools to assist clinicians in the timely and reliable identification of tooth decay.
- The integration of these AI algorithms into clinical practice can enhance diagnostic capabilities and function as a clinical decision support system in dentistry.
