Deep learning for osteoporosis screening in dental practice: a systematic review

Area of Science:

• Artificial Intelligence in Dentistry
• Medical Imaging Analysis
• Osteoporosis Diagnosis

Background:

• Osteoporosis screening is crucial for early intervention.
• Dental imaging offers a potential, non-invasive screening modality.
• Deep learning (DL) models are increasingly explored for medical image analysis.

Purpose of the Study:

• To systematically review DL tool performance in osteoporosis screening via dental imaging.
• To assess the clinical implementation of these DL models in dental practice.

Main Methods:

• A systematic search of seven databases and gray literature was conducted.
• Studies included DL algorithms applied to dental radiographs or CT scans for osteoporosis screening.
• Reference standards included dual-energy X-ray absorptiometry (DXA) or expert assessment.

Main Results:

• Thirteen studies met inclusion criteria; DXA was the common reference standard.
• Panoramic radiography was the primary imaging modality; accuracy was the main metric.
• ResNet and EfficientNet architectures showed superior performance, but no external validation or clinical implementation was reported.

Conclusions:

• DL models show potential for supportive osteoporosis screening in dentistry.
• Lack of external validation and clinical integration limits real-world applicability.
• Future research should prioritize standardization and development of validated, accessible systems for dental practice.