Related Experiment Video
Updated: Jan 9, 2026
10:42
A Postoperative Evaluation Guideline for Computer-Assisted Reconstruction of the Mandible
Published on: January 28, 2020
6.9K
Predicting postrestorative facial appearance in edentulous patients using deep learning: A prospective cohort study
Hongyang Ma1, Minjie Wu2, Yu Han2
1Professor, Department of Oral Implantology, Second Affiliated Hospital of Harbin Medical University, Harbin, China; Researcher, Second Clinical Division, Peking University School and Hospital of Stomatology, Beijing, PR China.
The Journal of Prosthetic Dentistry
|December 2, 2025
Summary
FacePointNet, an AI model, accurately predicts facial changes in edentulous patients after dental restoration. This tool enhances treatment planning by visualizing esthetic outcomes, improving patient counseling and clinical success.
Area of Science:
- Biomedical Engineering
- Dental Prosthodontics
- Artificial Intelligence in Medicine
Background:
- Predicting postrestorative facial appearance in edentulous patients is challenging due to complex soft tissue dynamics.
- Current methods lack precision in simulating esthetic outcomes, impacting patient well-being and treatment success.
Purpose of the Study:
- Develop FacePointNet, a bidirectional deep learning model, to predict facial changes in edentulous patients post-dental restoration.
- Enhance prerestorative treatment planning using quantitative metrics and visual similarity scores (VSSs).
Main Methods:
- Utilized 3D facial scans from 16 edentulous patients (pre- and post-restoration).
- Employed FacePointNet, a point-set neural network, for bidirectional geometric transformation prediction.
- Evaluated performance using cross-validation, chamfer distance (CD), Euclidean distance (ED), and surgeon-rated VSSs.
Main Results:
- AI-predicted 3D morphology closely matched actual restorative outcomes (mean landmark errors: 3.80–5.98 mm).
- No significant differences observed between predicted and actual results (P > .05).
- Expert reviewers reported high visual concordance (VSS: 4.2/5), especially for nasolabial folds.
Conclusions:
- FacePointNet shows clinical utility for visualizing localized facial changes before dental restoration.
- The AI model advances precision prosthodontics by offering reliable prerestorative predictions.
- Future research should focus on larger datasets and biomechanical integration for enhanced global prediction.
