Comparison of the Artificial Intelligence Versus Traditional Radiographic Interpretation in Detecting Periapical Periodontitis: A Diagnostic Accuracy Study

Abstract

Objective: To compare the performance of an AI model with that of experienced radiologists in detecting periapical periodontitis using radiographic images. Methods: Thirty radiographic images (CBCT, panoramic, and periapical) were analyzed by an AI model and two experienced radiologists. Diagnostic accuracy, sensitivity, specificity, and confidence levels were evaluated. Statistical analyses included Chi-square tests, independent samples t-tests, and Pearson correlation analysis. Results: The AI model achieved 89.6% accuracy, 86.5% sensitivity, and 88.1% specificity. Radiologist 1 showed the highest performance (accuracy 98.5%, sensitivity 93.8%, specificity 96.7%), while Radiologist 2 performed slightly lower than the AI model (accuracy 81.7%, sensitivity 83.3%, specificity 80%). The AI model demonstrated the highest mean confidence level (86.5% ± 9.18). Moderate positive correlations were observed between the AI’s confidence and that of Radiologist 1 (0.383) and Radiologist 2 (0.347). Conclusions: The AI model demonstrated comparable performance to experienced radiologists in detecting periapical periodontitis. These findings suggest that AI could serve as a valuable tool in dental diagnostics, potentially improving efficiency and consistency. However, further research is needed to refine AI models and evaluate their performance across diverse clinical scenarios.

Keywords

• ai
• artificial intelligence
• diagnosis
• diagnostic test accuracy
• periapical periodontitis