AI for caries detection: how close are we to clinical use?

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Dental caries detection is crucial for timely intervention.
• Traditional methods can be subjective and time-consuming.
• AI offers potential for objective and efficient caries diagnosis.

Purpose of the Study:

• To systematically review and meta-analyze the performance of AI platforms in detecting dental caries.
• To assess AI accuracy using both clinical and radiographic imaging modalities.

Main Methods:

• Systematic literature search across eight electronic databases (Jan 2000-Mar 2024).
• Inclusion of 45 studies (33 radiographic, 12 clinical images) after screening 2538 articles.
• Quality assessment using QUADAS-2 and CLAIM; narrative synthesis and meta-analysis (7 studies).

Main Results:

• Mean accuracy for clinical images: 78.2%; for X-ray images: 81.5%.
• Meta-analysis results: Overall sensitivity 76%, specificity 91%.
• Area Under the Curve (AUC) from HSROC curve: 92%.

Conclusions:

• AI models demonstrate significant potential for accurate dental caries detection.
• High sensitivity and specificity indicate AI's reliability in identifying caries.
• AI platforms can aid clinicians in improving diagnostic accuracy and patient care.