Diagnostic Accuracy of Deep Learning Models in Detecting Peri-Implant Marginal Bone Loss: A Systematic Review and...

Area of Science:

• Radiology
• Artificial Intelligence
• Dental Implantology

Background:

• Peri-implantitis is a frequent complication of dental implants.
• Early detection of peri-implantitis is crucial to prevent bone loss and implant failure.
• Deep learning (DL) models show potential for improving radiographic diagnostic accuracy.

Purpose of the Study:

• To systematically review the diagnostic performance of DL models in detecting marginal bone loss on radiographic images.
• To assess the clinical utility of DL models for peri-implantitis diagnosis.

Main Methods:

• A systematic literature search was conducted across multiple databases (PubMed, EMBASE, CENTRAL, etc.) for studies published between 2010 and July 2025.
• Two independent reviewers screened studies, extracted data, and assessed quality using QUADAS-2.
• Random-effects meta-analysis synthesized diagnostic metrics (sensitivity, specificity, AUC); heterogeneity and bias were evaluated.

Main Results:

• Five studies involving 12,545 radiographs met the inclusion criteria.
• DL models demonstrated high diagnostic performance with pooled sensitivity of 88%, specificity of 91%, and AUC of 0.95.
• Dataset size influenced accuracy, while imaging type did not; no publication bias was detected.

Conclusions:

• DL models exhibit high accuracy in detecting radiographic marginal bone loss, a key indicator of peri-implantitis.
• These models serve as a valuable adjunct for early diagnosis and timely intervention, complementing clinical assessments.