Related Experiment Video
Updated: Jan 12, 2026
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
Integration of deep learning and Bayesian networks for personalized tooth color prediction in dental esthetics: A
Chenglu Ruan1,2, Jianying Xiong1, Lin Wang3
1Department of Stomatology, Sanming Integrated Medicine Hospital, Sanming, China.
Purpose:
The relationship between tooth color selection and individual satisfaction remains critical in dental esthetics. This study developed a hybrid approach combining deep learning with Bayesian network analysis to investigate how skin tone, age, and gender influence tooth color preferences.
Materials And Methods:
A total of 128 participants (62 males, 66 females; mean age 32.4 ± 7.8 years) evaluated 16 standardized smile images with different VITA classical shades. Participants included 60 dental professionals and 68 non-dental individuals. A hybrid model integrating convolutional neural networks with Bayesian networks was constructed to analyze color preferences. The model was validated through leave-one-out cross-validation and compared with traditional Bayesian analysis.
Results:
The hybrid model achieved superior prediction accuracy (89.2%) compared to traditional Bayesian networks (82.5%, p < 0.01). B1, A1, and B2 shades received the highest overall ratings (8.63 ± 0.92, 8.21 ± 1.04, 7.85 ± 1.12, respectively). Males showed stronger preference for B1 shade (8.92 ± 0.78 vs. 8.37 ± 0.96 for females, p = 0.003), while females demonstrated more diverse preferences across multiple shades. Age negatively correlated with preference for lighter shades (r = -0.42, p < 0.001). Dental professionals exhibited greater discrimination between shades compared to non-dental participants (score range: 2.14-9.03 vs. 3.56-8.42, p < 0.01).
Conclusion:
This novel hybrid framework significantly improved tooth color prediction accuracy compared to traditional methods. The findings provide quantitative guidance for personalized shade selection based on individual characteristics, potentially enhancing clinical outcomes and patient satisfaction in esthetic dentistry.
