Near-infrared transillumination with high dynamic range imaging for occlusal caries detection in vitro.

The aim of this study was to assess near-infrared transillumination with high dynamic range imaging (NIRT-HDRI) for occlusal caries detection in vitro and to compare it with visual inspection using the International Caries Detection and Assessment System II (ICDAS) and digital bitewing radiography (BWR). Sixty-one extracted permanent molars with sound or occlusal carious surfaces without severe cavitation were visually assembled. Two examiners assessed twice these surfaces independently using ICDAS, BWR and NIRT-HDRI. The latter was performed with a prototype consisting of two laser sources (780 nm), a CCD sensor and subsequent processing with image analysis software. Thresholds for carious surfaces, enamel and dentin lesions were defined for all methods. Micro-computed tomography served as the reference standard. Linear weighted Kappa analysis of the methods versus the reference at the threshold carious surface, enamel and dentin lesion revealed 0.59/0.08/0.12 for ICDAS, 0.37/- 0.06/0.58 for BWR and 0.33/- 0.01/0.51 for NIRT-HDRI. Sensitivity values at the three thresholds were 0.85/0.78/0.13 for ICDAS, 0.59/0.00/0.69 for BWR and 0.98/0.33/0.78 for NIRT. Specificity values at the three thresholds were 0.70/0.40/1.00 for ICDAS, 0.9./0.96/0.90 for BWR and 0.30/0.65/0.72 for NIRT-HDRI. Reliability analysis revealed substantial agreement for BWR and NIRT and almost perfect agreement for ICDAS. NIRT exhibited a strong ability to identify occlusal dental decay in general; however, it revealed a tendency towards overestimation. Its strength was the detection of dentin caries lesions compared with ICDAS and BWR. NIRT-HDRI seems to be a suitable method to detect hidden dentin caries as a supplement to visual examination.