Comparison of the Diagnostic Accuracy of an AI-Based System for Dental Caries Detection and Clinical Evaluation Conducted by Dentists
Jakub Kwiatek, Marta Leśna, Wiktor Piskórz, Justyna Kaczewiak
Abstract
Background/Objectives: Artificial intelligence (AI)-based software is increasingly used for radiographic analysis in dentistry. This study aimed to evaluate the diagnostic accuracy of an AI-powered radiographic analysis system, using Diagnocat (DGNCT LLC, Miami, FL, USA) as an example, compared with clinical evaluations performed by three experienced dentists. The assessment focused on primary caries detection and the total number of primary and secondary caries based on panoramic radiographs (OPGs). Methods: Three dentists with similar expertise independently classified teeth for treatment using only panoramic radiographs and their clinical knowledge. The study was conducted under single-blind conditions, where clinicians were unaware that their diagnoses would be compared to the AI system’s analysis. Results: The AI system’s agreement with human evaluations varied depending on tooth location, patient age, and gender. The lowest agreement was observed for premolars, likely due to limitations of 2D imaging, while higher accuracy was found for molars and incisors, particularly in younger patients. The system showed limitations in detecting occlusal, labial, and lingual caries. Conclusions: AI-assisted radiographic analysis has the potential to enhance diagnostic efficiency and automation in dentistry. However, its accuracy is influenced by tooth location and imaging modality. Further research is needed to explore the benefits of integrating AI with 3D imaging techniques to improve diagnostic reliability.