Development and Comparative Analysis of an AI Based Convolutional Neural Network Algorithm versus Dentists of Varying Experience in Detecting Dental Caries using Bitewing X-ray Images: An In vitro Study.

BACKGROUND

Dental caries, being an infectious, microbiologic disease, can cause dissolution and destruction of calcified tissues. Early detection of dental caries is therefore crucial for maintaining overall tooth health.

MATERIALS AND

METHODS

A randomized, single-blinded retrospective study was conducted. Three groups participated: dentists with <10 years of experience, dentists with more than 10 years of experience, and the AI model. An external dataset of 20 anonymized bitewing radiographs served as the test set, reviewed and annotated by experienced dentists as the gold standard. Performance metrics included Intersection over Union, Confusion Matrix, Sensitivity, Precision, and F1 score.

RESULTS

Among all, the AI model showed a high precision of 0.87, a recall of 0.77, and an F1 score of 0.82. This shows the AI model's superior precision in identifying caries and avoiding misinterpretations, like cervical burnout and Mach band effect.

CONCLUSION

The study achieved high diagnostic accuracy in identifying early carious lesions and offered advantages in efficiency and avoiding misinterpretations. Integrating AI with dental expertise could enhance diagnostic accuracy, optimize workflow, and support preventive interventions, ultimately improving patient care. Future research should focus on expanding the model's capabilities and exploring its seamless integration into clinical practice.