Related Experiment Video
Updated: Nov 9, 2025
09:34
Semi-Automated Planimetric Quantification of Dental Plaque Using an Intraoral Fluorescence Camera
Published on: January 27, 2023
2.2K
Automated caries detection with smartphone color photography using machine learning
Duc Long Duong, Malitha Humayun Kabir1, Rong Fu Kuo1,2
1Department of Biomedical Engineering, National Cheng Kung University.
Health Informatics Journal
|April 17, 2021
Summary
A new computational algorithm uses smartphone images to detect dental caries with high accuracy. This cost-effective method aids in diagnosing carious lesions, supporting clinical decisions and research.
Area of Science:
- Dentistry
- Computational Imaging
- Biomedical Engineering
Background:
- Dental caries affects billions globally, necessitating accurate detection methods.
- Current caries detection requires specialized equipment and expertise.
- Accurate diagnosis is crucial for clinical decision-making, epidemiology, and research.
Purpose of the Study:
- To develop and evaluate a computational algorithm for automated caries detection.
- To classify carious lesions on tooth occlusal surfaces using smartphone images.
- To align automated detection with the International Caries Detection and Assessment System (ICDAS) criteria.
Main Methods:
- Extracted teeth (620 unrestored molars/premolars) were photographed using a smartphone.
- Images were labeled into three classes: No Surface Change (NSC), Visually Non-Cavitated (VNC), and Cavitated (C) based on ICDAS II codes.
- A two-step Support Vector Machine (SVM) classification scheme was employed: C versus (VNC + NSC), and VNC versus NSC.
Main Results:
- The SVM model achieved 92.37% accuracy, 88.1% sensitivity, and 96.6% specificity for detecting cavitated lesions (C versus VNC + NSC).
- For non-cavitated lesions (VNC versus NSC), the model achieved 83.33% accuracy, 82.2% sensitivity, and 66.7% specificity.
- The algorithm demonstrated auspicious potential for clinical diagnostics using smartphone-imaged data.
Conclusions:
- A smartphone-based computational algorithm shows promise for accurate and cost-effective dental caries detection.
- The SVM system, while requiring further validation, offers a viable tool for aiding clinical diagnostics and epidemiological studies.
- This approach could significantly improve accessibility to caries detection, particularly in resource-limited settings.
