Short-Wave Infrared Reflectance at 1050–1550 nm for Proximal Caries Detection: An In Vitro Diagnostic Accuracy Study Validated by Micro-CT

<b>Background/Objectives:</b> Short-wave infrared reflectance (SWIRR) imaging is a non-ionising approach for proximal caries detection; however, the diagnostic impact of wavelength selection in reflectance imaging has not been systematically evaluated. <b>Methods:</b> This in vitro diagnostic accuracy study assessed SWIRR at 1050, 1200, 1300 and 1550 nm for proximal caries detection, using micro-computed tomography as the reference standard and digital bitewing radiography (BWR) as the clinical comparator. A total of 250 extracted permanent posterior teeth with sound or carious proximal surfaces were examined. SWIRR and BWR images were independently evaluated twice by two calibrated examiners using method-specific criteria. Diagnostic performance was quantified by sensitivity, specificity and accuracy; examiner reliability was analysed using kappa statistics, and pairwise comparisons were performed using McNemar tests with Holm-Bonferroni correction. <b>Results:</b> Diagnostic performance of SWIRR was wavelength dependent, showing increasing sensitivity and decreasing specificity with longer wavelengths. The highest overall accuracy was observed at 1050 nm (80.0%), exceeding that of BWR (73.8%) while maintaining comparable specificity and higher sensitivity. At 1550 nm, sensitivity was highest but accompanied by an increase in false-positive findings. <b>Conclusions:</b> SWIRR demonstrates high diagnostic potential for proximal caries detection, with 1050 nm providing the most favourable balance between diagnostic accuracy and specificity.