Related Experiment Video
Updated: Oct 7, 2025

05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
1.0K
Automatic Dental Plaque Segmentation Based on Local-to-Global Features Fused Self-Attention Network
IEEE Journal of Biomedical and Health Informatics
|January 11, 2022
Summary
Early dental plaque detection is crucial for preventing gum disease and cavities. This study introduces an AI-powered network for accurate plaque segmentation from oral images without dyes, improving early diagnosis.
Area of Science:
- Biomedical Engineering
- Artificial Intelligence in Dentistry
- Medical Imaging Analysis
Background:
- Accurate early dental plaque detection is vital for preventing periodontal diseases and dental caries.
- Current methods struggle with low contrast between plaque and healthy teeth, often requiring dyeing reagents.
- Intelligent segmentation of dental plaque from oral endoscope images is challenging due to visual similarities.
Purpose of the Study:
- To propose a novel deep learning network with a self-attention module for intelligent dental plaque segmentation.
- To enable accurate, pixel-level plaque detection directly from oral endoscope images without dyeing reagents.
- To improve early detection of dental plaque, aiding in the prevention of periodontal diseases and dental caries.
Main Methods:
- Developed a novel network architecture incorporating a self-attention module for super-pixel level analysis.
- Fused multiple-scale complementary information: plaque color distribution, Heat Kernel Signature (HKS) for structure, and Circle-Local Binary Patterns (LBP) for texture.
- Utilized a CNN-based attention module to refine fused features and focus on regions of interest.
Main Results:
- The proposed method achieved superior performance compared to state-of-the-art methods, even with a small training dataset.
- Experimental evaluations confirmed the network's ability to accurately segment dental plaque.
- User studies indicated that the method is more accurate than conventional dental examinations by experienced dentists.
Conclusions:
- The novel deep learning network effectively segments dental plaque from oral endoscope images without requiring dyeing reagents.
- The self-attention mechanism and multi-scale feature fusion enhance segmentation accuracy, particularly in challenging cases.
- This AI-driven approach offers a promising tool for early and accurate dental plaque detection, supporting preventative dental care.

