Accuracy of Artificial Intelligence in Fixed Dental Prosthesis Design
Conditions
Keywords
Artificial Intelligence, CAD/CAM, Fixed Dental Prosthesis, Dental Crown Design, 3D Superimposition, Deep Learning, Generative Adversarial Network (GAN), Prosthodontics
Brief summary
This in vitro study aims to evaluate the accuracy of an Artificial Intelligence (AI)-based automatic design system for fixed dental prosthesis (FDP) compared with conventional computer-aided design (CAD) software. Digital scans of teeth requiring fixed dental prosthesis will be collected and used to generate prosthetic designs using two approaches: human-designed CAD restorations and AI-generated restorations. The primary outcome is design accuracy assessed using 3D superimposition and Intersection over Union (IOU) percentage. Secondary outcomes include margin detection performance measured using F1 score, precision, and recall. A total sample size of 438 scans will be analyzed. The study will determine whether AI-generated prosthesis designs demonstrate comparable accuracy to conventional digital designs.
Detailed description
This study is designed as an in vitro comparative study to assess the accuracy and performance of an Artificial Intelligence (AI)-based automatic design system for fixed dental prosthesis (FDP) in comparison with conventional computer-aided design (CAD) software. Digital scans of patients requiring fixed dental prosthesis will be collected from the production laboratory of the Faculty of Dentistry. Eligible scans will include adults aged 18-65 years with damaged teeth requiring FDP and adequate occlusal anatomy for analysis. The AI workflow consists of three sequential phases: training (60%), validation (10%), and testing (30%). The AI model will be trained using natural spatial tooth morphology and historical human-designed FDP datasets. The conventional group will consist of FDPs manually designed by experienced dental professionals using CAD software. Primary Outcome: The primary outcome is crown design accuracy measured using 3D superimposition analysis and quantified using Intersection over Union (IOU) percentage. Secondary Outcome: Margin detection accuracy will be assessed using F1 score, precision, and recall metrics. Statistical analysis will be performed using MedCalc software (Version 22). Continuous variables will be presented as mean, root mean square, and standard deviation. Comparisons between groups will be conducted using paired t-test with a significance level set at P ≤ 0.05 (two-tailed). The null hypothesis states that there will be no statistically significant difference between AI-designed and human-designed fixed dental prostheses.
Interventions
OTHERConventional CAD-Based Fixed Dental Prosthesis Design
Fixed dental prostheses will be digitally designed using conventional computer-aided design (CAD) software by experienced dental professionals. Designs will be based on occlusal anatomy and patient-specific intraoral scan data. These manually generated digital designs will serve as the comparator for evaluating accuracy against AI-generated designs using 3D superimposition and quantitative accuracy analysis.
OTHERArtificial Intelligence-Based Fixed Dental Prosthesis Design
An artificial intelligence-based automated design system will generate fixed dental prosthesis designs using deep learning algorithms. The AI model will be trained (60%), validated (10%), and tested (30%) on occlusal scan datasets and historical human-designed prostheses. Generated designs will be evaluated for accuracy and marginal precision using 3D superimposition and Intersection over Union (IoU) analysis.
Sponsors
October University for Modern Sciences and Arts
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
DIAGNOSTIC
Masking
NONE
Eligibility
Sex/Gender
ALL
Age
18 Years to 65 Years
Healthy volunteers
No
Inclusion criteria
* Adults aged 18-65 years Patients with a damaged tooth requiring a fixed dental prosthesis Available digital intraoral scans Adequate occlusal anatomy for analysis of opposing teeth
Exclusion criteria
* Incomplete or poor-quality digital scans Severe occlusal abnormalities affecting analysis Patients outside the specified age range
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Accuracy of AI-Designed Fixed Dental Prosthesis Compared to Human-Designed Prosthesis | Immediately after crown design generation (at time of digital analysis) | Accuracy will be assessed by superimposing AI-generated crown designs and human-designed crowns using 3D imaging software. The Intersection over Union (IOU) percentage will be calculated to evaluate morphological agreement and occlusal fit between the two design approaches. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Margin Detection Performance of AI System | Immediately after digital crown design generation | Margin detection accuracy will be evaluated using F1 score, precision, and recall metrics to assess the AI model's ability to accurately identify preparation margins compared to human-designed reference models. |
Countries
Egypt
Outcome results
None listed