Performance evaluation of AI-based caries detection technology and its educational training module: a dual-phase...

Area of Science:

• Dental diagnostics
• Artificial intelligence in healthcare
• Radiographic interpretation

Background:

• Accurate radiographic diagnosis of dental caries is crucial for effective treatment planning.
• Dental education requires robust training methods for interpreting complex radiographic images.
• Existing methods for caries detection may have limitations in accuracy and efficiency.

Purpose of the Study:

• To evaluate the diagnostic accuracy of the Overjet Caries Assist (OCA) AI system.
• To assess the efficacy of the OCA AI-based training module for dental students.
• To enhance the radiographic diagnosis of dental caries among dental learners.

Main Methods:

• 1604 proximal tooth surfaces on bitewing radiographs were independently assessed by two calibrated observers and compared with OCA.
• Sensitivity, specificity, and diagnostic accuracy metrics were calculated for the AI system.
• 82 dental students completed caries diagnosis assessments before and after training with the OCA module.

Main Results:

• The OCA system demonstrated high accuracy, with average sensitivities of 0.69 for enamel and 0.91 for dentinal lesions.
• Average specificities were 0.99 for enamel and 0.98 for dentinal lesions.
• Dental students showed a 43%-46% improvement in detecting radiographic caries after using the OCA training module.

Conclusions:

• The Overjet Caries Assist (OCA) module serves as an effective AI tool for supporting caries diagnosis in dental practice.
• The OCA training module is a valuable resource for improving dental students' radiographic interpretation skills.
• AI-assisted tools show promise in both clinical diagnostics and dental education.