Digital Caries Detection in Children using Intraoral Scanners: A Diagnostic Agreement Study (Preprint)

BACKGROUND Dental caries is a common chronic disease in children, yet large-scale surveillance is limited by the resource demands of clinical examinations. Digital tools like intraoral scanners may offer a scalable alternative for caries detection, enabling more efficient data collection for research and population-level oral health monitoring. OBJECTIVE To assess the diagnostic agreement between visual examination and on-screen assessment of 3D models generated by an intraoral scanner (IOS) in colour and supplemented with fluorescence for caries detection in primary teeth. METHODS Children participating in a clinical trial (n=216, mean age 5.6 years) underwent visual examination using the International Caries Detection and Assessment System (ICDAS) and intraoral scanning using the TRIOS 4 IOS (3Shape TRIOS A/S Denmark). Four trained registered dental practitioners independently assessed each participant’s 3D models in colour, then supplemented with fluorescence, using a previously validated ICDAS index modified for on-screen assessments of 3D models. All 3D models were assessed again four weeks later. The time taken for intraoral scanning and on-screen assessment was recorded. Multilevel logistic regression was used to estimate and compare the likelihood of detecting caries between methods, and Bland–Altman plots were used to visualise agreement. Analyses were performed at the initial (ICDAS>01), moderate (ICDAS>03), and extensive (ICDAS>05) dental caries thresholds. Intraclass Correlation Coefficient (ICC) estimated method agreement and examiner reliability. RESULTS 8,209 visible primary tooth coronal surfaces were analysed. The likelihood of detecting caries using colour assessment of 3D models was similar to visual examination at all disease thresholds: initial (odds ratio 1.1, 95% confidence interval (CI) 1.0–1.3), moderate (0.9, 0.7-1.1) and extensive (1.0, 0.7–1.3). When colour assessments were supplemented with fluorescence, the likelihood of detecting caries was 30% higher at the initial threshold relative to visual examination (1.3, 1.1-1.5) and similar at the moderate (0.9, 0.7-1.1) and extensive thresholds (0.9, 0.7-1.3). Bland-Altman plots showed a high level of agreement at both moderate and extensive thresholds. Examiner reliability using intraoral scans ranged from good to excellent. CONCLUSIONS On-screen assessment of 3D models in colour shows greatest agreement with visual examination for caries detection at all disease thresholds. CLINICALTRIAL Australian New Zealand Clinical Trials Registry ACTRN12622001237774 INTERNATIONAL REGISTERED REPORT RR2-10.2196/51578