Related Experiment Video
Updated: Aug 18, 2025
05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
940
Caries detection with tooth surface segmentation on intraoral photographic images using deep learning
Eun Young Park1, Hyeonrae Cho2,3, Sohee Kang1
1Department of Dentistry, College of Medicine, Yeungnam University, Daegu, South Korea.
BMC Oral Health
|December 8, 2022
Summary
This study demonstrates that a deep learning algorithm, utilizing tooth surface segmentation on intraoral images, significantly improves artificial intelligence-based caries detection. This AI-powered method offers a promising, time-saving, and cost-effective approach for dental diagnostics.
Area of Science:
- Dental Diagnostics
- Artificial Intelligence in Medicine
- Medical Imaging Analysis
Background:
- Intraoral photographic images are crucial for clinical caries diagnosis.
- Artificial intelligence (AI) has been increasingly applied to analyze these dental images.
- Evaluating AI algorithms for enhanced caries detection is an ongoing area of research.
Purpose of the Study:
- To assess the efficacy of a deep learning algorithm for detecting dental caries.
- To evaluate the impact of tooth surface segmentation on AI-based caries detection using intraoral images.
Main Methods:
- A prospective study collected 2348 intraoral images from 445 participants.
- Convolutional Neural Networks (CNNs), including U-Net, ResNet-18, and Faster R-CNN, were employed for image segmentation, classification, and localization.
- Images were divided into training, validation, and test datasets.
Main Results:
- Tooth surface segmentation using CNNs improved the accuracy and area under the receiver operating characteristic curve for caries classification.
- Specifically, accuracy increased from 0.758 to 0.813 and AUC from 0.731 to 0.837.
- The localization algorithm also showed enhanced performance, with improved sensitivity and average precision.
Conclusions:
- Deep learning models incorporating tooth surface segmentation show significant promise for caries detection from intraoral camera images.
- This AI-driven approach can serve as an effective aided diagnostic tool for caries.
- The method offers potential advantages in terms of time and cost savings for dental professionals.
