Comparison of quantitative light-induced fluorescence and near-infrared intraoral scanner imaging for proximal caries across predefined ICCMS radiographic thresholds: a diagnostic accuracy study

BACKGROUND: Proximal caries detection remains challenging because early lesions are often inaccessible to visual inspection and bitewing radiography exposes patients to ionizing radiation. Quantitative light-induced fluorescence (QLF) and near-infrared imaging integrated into intraoral scanners are proposed as non-ionizing adjuncts. This study compared the diagnostic performance of a handheld QLF device (QrayPen C, Aiobio) and an intraoral scanner with near-infrared imaging (NIRI) (iTero Element 5D, Align Technology) for proximal caries across clinically relevant ICCMS radiographic thresholds. METHODS: In this single-center comparative diagnostic accuracy study, 66 adults contributed 297 posterior proximal sites. Each site was assessed using QrayPen (QS score, 0-3) and the iTero Element 5D near-infrared imaging system (NIRI score, 0-3). Bitewing radiographs were scored according to the International Caries Classification and Management System (ICCMS) and served as the reference standard. Diagnostic accuracy was evaluated at three pre-specified thresholds: any caries (R0 vs. R1-R6), dentin involvement (R0-R2 vs. R3-R6), and deeper dentin involvement (R0-R3 vs. R4-R6). The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and predictive values were calculated. Statistical significance was set at p < 0.05. RESULTS: For any caries, AUCs were similar for QS and NIRI (0.729 vs. 0.727; p = 0.902). At the pre-specified cut-off (score ≥ 1), QS showed higher specificity than NIRI (88.1% vs. 78.2%), while sensitivities were comparable (54.7% vs. 61.1%). For dentin involvement (cut-off ≥ 2), AUC was 0.909 for QS and 0.896 for NIRI (p = 0.605); both methods showed sensitivity of 81.8%, with specificity of 95.5% for QS and 93.9% for NIRI. For radiographically deeper dentin involvement (cut-off = 3), NIRI demonstrated a slightly higher AUC than QS (0.978 vs. 0.961; p = 0.036); sensitivity/specificity were 72.2%/96.4% for QS and 83.3%/97.1% for NIRI. QS and NIRI scores were strongly correlated (Spearman ρ = 0.735; p < 0.001) with substantial ordinal agreement (linear-weighted κ = 0.707). CONCLUSIONS: Both QS (QrayPen C) and NIRI (iTero Element 5D) demonstrated moderate accuracy for identifying any radiographic proximal caries and higher accuracy for radiographically deeper lesions. NIRI showed a small but statistically significant advantage for the radiographically deeper dentin involvement threshold. These non-ionizing methods may be useful adjuncts for proximal caries assessment, but do not eliminate the need for bitewing radiography when radiographic staging information is clinically required. Bitewing radiography under-detects enamel lesions, and site-level analyses did not model within-participant clustering; therefore, estimates should be interpreted as exploratory.