Related Experiment Video
Updated: Aug 12, 2025
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
1.8K
Quantitative level determination of fixed restorations on panoramic radiographs using deep learning
International Journal of Computerized Dentistry
|January 27, 2023
Summary
Deep learning models, specifically convolutional neural networks (CNNs), show high accuracy in detecting dental restorations on panoramic radiographs. This technology can assist dentists with preliminary diagnostic information.
Area of Science:
- Dentistry
- Artificial Intelligence
- Medical Imaging
Background:
- Deep learning applications are limited in dental imaging.
- Panoramic radiographs are common diagnostic tools.
Purpose of the Study:
- Evaluate convolutional neural network (CNN) effectiveness for dental restoration detection and diagnosis.
- Develop a novel dataset for training and testing CNN models.
Main Methods:
- Trained AlexNet, VGG-16, and ResNet models on 20,973 panoramic radiographs.
- Utilized 10-fold cross-validation and data augmentation.
- Classified dental restorations into five categories.
Main Results:
- ResNet-101 achieved the highest accuracy (92.7%) and macro-average AUC (0.989).
- Other models showed varying accuracies: AlexNet (75.5%), VGG-16 (85.0%), ResNet-18 (92.1%), ResNet-50 (91.7%), InceptionResNet-v2 (92.1%).
Conclusions:
- Achieved 92.7% accuracy, indicating a promising computer-aided diagnostic system.
- The system can provide dentists with supportive preliminary information from panoramic radiographs.
- The novel dataset is versatile and can be repurposed for other dental imaging studies.
