A Randomized Controlled Trial Comparing AI-Assisted and Conventional Radiographic Interpretation in Early Caries Detection

Irosha Gajjada ¹, Ambreen Rehman ², Pavithra Tadakamalla ³, Faheem Mir ⁴, Ankita Jain ⁵, Arindam Banik⁶

¹Senior Lecturer, Department of Oral and Maxillofacial Pathology, MNR Dental College and Hospital Sangareddy
²Assistant Professor, Department of Oral Biology, Fujairah University
³BDS, Kamineni Institute of Dental Sciences, DR.NTR University of Health Sciences
⁴Preventive Dental Sciences Department, College of Dentistry, Dar AlUloom University, Riyadh, SA.
⁵Professor, Department of Public Health Dentistry, Teerthanker Mahaveer University, Dental College
⁶Senior Lecturer, Department of Conservative Dentistry and Endodontics, Guru Nanak Institute of Dental Sciences and Research, Kolkata, West Bengal, India

ABSTRACT

Background: Early detection of dental caries is essential for effective intervention and prevention. Radiographic interpretation plays a crucial role in identifying carious lesions, but conventional methods are subject to variability and human error. Artificial Intelligence (AI) has the potential to improve diagnostic accuracy by automating radiographic analysis. This study aimed to compare the diagnostic accuracy, time efficiency, and clinician confidence between AI-assisted and conventional radiographic interpretation for early caries detection. Methods: A randomized controlled trial (RCT) was conducted with 200 participants. Participants were randomly assigned to one of two groups: AI-assisted interpretation (n=100) and conventional interpretation (n=100). All participants underwent standardized bitewing radiographs, which were analyzed by AI and human clinicians. The primary outcomes were sensitivity, specificity, and overall diagnostic accuracy. Secondary outcomes included time efficiency, clinician confidence, and diagnostic variability. Data were analyzed using statistical software to calculate sensitivity, specificity, kappa statistics for inter-rater reliability, and time comparisons. Results: AI-assisted interpretation demonstrated a higher sensitivity (91%) compared to conventional interpretation (85%). However, conventional interpretation showed higher specificity (88%) compared to AI (84%). AI-assisted interpretation was significantly faster (2 minutes per radiograph) than conventional interpretation (5 minutes). Clinician confidence was slightly higher in the AI group (4.5 vs. 4.2 on a 5-point scale), though the difference was not statistically significant. Diagnostic consistency was higher in the AI group (kappa = 0.93) compared to conventional interpretation (kappa = 0.72). Conclusion: AI-assisted interpretation provided higher sensitivity and faster diagnosis compared to conventional radiographic interpretation for early caries detection. However, conventional methods maintained higher specificity. The findings suggest that AI could improve early detection and clinical efficiency, though challenges with specificity remain. Further research is needed to refine AI algorithms and improve diagnostic precision....

Keywords: Artificial Intelligence, Dental Caries, Radiographic Interpretation, Sensitivity, Time Efficiency

How to cite this article: Gajjada I, Rehman A, Tadakamalla P, Mir F, Jain A, Banik A; A Randomized Controlled Trial Comparing AI-Assisted and Conventional Radiographic Interpretation in Early Caries Detection.Int J Drug Deliv Technol. 2026;16(1s): 567-571; DOI: 10.25258/ijddt.16. 567-571