Related Experiment Video
Updated: Aug 5, 2026
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Artificial intelligence-based photographic detection of pink esthetic score attributes using a hybrid deep learning
Mihad Ibrahim1, Ghada Gamal Adayil2, Nouran Hany3
1Department of Periodontology, Faculty of Dentistry, Cairo University, 11 Saray Street, Almanial, Cairo, Egypt. mihad.ibrahim@dentistry.cu.edu.eg.
Scientific Reports
|August 1, 2026
Summary
This study introduces an AI system using anatomy-driven measurements for automated Pink Esthetic Score (PES) evaluation from photos. The novel approach shows high accuracy in assessing key esthetic attributes, offering a transparent method for research.
Area of Science:
- Biomedical Engineering
- Artificial Intelligence in Dentistry
- Digital Dentistry
Background:
- Accurate assessment of the Pink Esthetic Score (PES) is crucial for evaluating esthetic outcomes in dental procedures.
- Existing automated methods often lack transparency and anatomical grounding.
- Intraoral photographs are a common source for esthetic evaluation, but manual assessment can be subjective.
Purpose of the Study:
- To develop and validate an anatomy-driven artificial intelligence (AI) system for automated postoperative Pink Esthetic Score (PES) evaluation using intraoral photographs.
- To create a transparent and reproducible AI pipeline for PES assessment by deriving attributes from anatomically grounded measurements.
- To compare the AI system's performance against expert assessments of PES.
Main Methods:
- A hybrid analytical pipeline integrating instance segmentation (Mask R-CNN for teeth, YOLOv11 for gingiva) and rule-based measurements was developed.
- Anatomical features were extracted from segmented images and converted into ordinal PES attributes via data-driven threshold calibration.
- The system was validated against 82 postoperative intraoral photographs scored by independent experts, using diagnostic accuracy, confusion matrix, and regression metrics.
Main Results:
- The AI system achieved high accuracies for individual PES attributes: 91.5% (mesial papilla), 85.4% (distal papilla), 86.6% (gingival margin), and 82.9% (gingival color).
- Exact agreement with expert total PES scores was 57.3%, increasing to 79.3% with a ±1 PES point tolerance.
- Regression analysis showed a mean absolute error of 0.76, root mean squared error of 1.35, and R² of 0.49.
Conclusions:
- An anatomy-driven, segmentation-based AI framework offers a transparent and reproducible method for automated photographic PES attribute assessment.
- This AI system demonstrates promising performance for objective esthetic evaluation in research settings.
- The hybrid approach integrating segmentation and rule-based measurements provides a robust foundation for future dental AI applications.