Related Experiment Video
Updated: Jan 16, 2026
05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
1.3K
Artificial intelligence for detection and classification of furcation defects using radiographic imaging: A
Georgios S Chatzopoulos1,2, Vasiliki P Koidou3, Lazaros Tsalikis1
1Department of Preventive Dentistry, Periodontology and Implant Biology, School of Dentistry, Aristotle University of Thessaloniki, Thessaloniki, Greece.
Imaging Science in Dentistry
|January 15, 2026
Summary
Artificial intelligence (AI) algorithms show promise for detecting furcation defects on dental radiographs, especially deep learning models on 2D images. However, variations in AI tools and imaging methods necessitate standardized research for clinical use.
Area of Science:
- Dental radiology
- Artificial intelligence in medicine
- Diagnostic accuracy studies
Background:
- Traditional methods for detecting furcation defects have limitations.
- Accurate diagnosis of furcation involvement is crucial for periodontal health.
- Artificial intelligence (AI) offers potential for improving radiographic analysis.
Purpose of the Study:
- To systematically review and synthesize evidence on the diagnostic accuracy of AI algorithms for detecting and classifying furcation defects.
- To address limitations associated with traditional methods of furcation defect identification.
- To evaluate AI performance across various algorithms and radiographic modalities.
Main Methods:
- Comprehensive literature search conducted through April 2025.
- Inclusion of diagnostic accuracy studies comparing AI algorithms to reference standards for furcation defects on dental radiographs.
- Risk of bias assessed using QUADAS-2; meta-analysis not performed due to study heterogeneity.
Main Results:
- Eight retrospective studies were included, using diverse AI algorithms (ResNet, UNet, YOLO-v4, Vision Transformers) and radiographic types (periapical, panoramic, CBCT).
- Advanced deep learning models on 2D radiographs generally demonstrated high diagnostic accuracy for furcation involvement detection, with notable sensitivity, specificity, and AUC values.
- Performance varied significantly based on the AI model and imaging modality; proprietary AI tools showed inconsistent results, and classification of furcation severity was less consistently reported.
Conclusions:
- AI algorithms, particularly deep learning on 2D radiographs, show potential for accurate furcation defect detection.
- Methodological and reporting heterogeneity exists within current research.
- Future research should focus on standardized protocols, clinical comparisons, and developing translatable AI tools for improved diagnosis.
