Related Experiment Video
Updated: Jan 16, 2026
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
Performance comparison of three artificial intelligence models in predicting gingival-colored porcelain compositions
Boxuan Xu1, Yiqing Wang2, Lei Zhang3
1Predoctoral student, Peking University School and Hospital of Stomatology & National Center of Stomatology & National Clinical Research Center for Oral Diseases &National Engineering Research Center of Oral Biomaterials and Digital Medical Devices & Beijing Key Laboratory of Digital Stomatology & Research Center of Engineering and Technology for Computerized Dentistry Ministry of Health & NMPA Key Laboratory for Dental Materials, Beijing, PR China.
Artificial intelligence (AI) systems accurately predict gingival porcelain color. The Residual Neural Network (ResNet) model demonstrated superior performance in matching colors for dental restorations, enhancing esthetic outcomes.
Area of Science:
- Dental Materials Science
- Artificial Intelligence in Dentistry
- Restorative Dentistry
Background:
- Lack of studies on esthetic outcomes of soft tissue restoration coloration.
- Limited investigation into the relationship between ceramic powder proportions and gingival color.
Purpose of the Study:
- Develop and compare three AI systems: Residual Neural Network (ResNet), Multilayer Perceptron (MLP), and Genetic Algorithm-optimized Backpropagation (GA+BP).
- Predict gingival-colored porcelain compositions to improve color matching accuracy in restorative dentistry.
Main Methods:
- Fabricated 359 gingival-colored porcelain specimens (286 standard, 73 extreme).
- Measured CIELab coordinates and correlated them with powder compositions.
- Developed and evaluated ResNet, MLP, and GA+BP models using 5-fold cross-validation and Mean Squared Error (MSE).
Main Results:
- ResNet achieved the lowest MSE (0.0199) and Mean Absolute Error (MAE) (0.1069), indicating superior prediction accuracy.
- ResNet demonstrated the highest explained variance (0.718), outperforming MLP and GA+BP.
- External validation showed ResNet's average ΔE₀₀ of 1.55, falling within perceptibility thresholds.
Conclusions:
- ResNet exhibited the best accuracy in predicting gingival-colored porcelain compositions.
- AI-driven systems, particularly ResNet, can enhance accuracy and reproducibility in dental color matching.
- Findings support the clinical application of AI for improved esthetic restorative dentistry.
