Near-infrared hyperspectral imaging of teeth for dental caries detection

Abstract

Near-infrared (NIR) is preferred for caries detection compared to visible light imaging because it exhibits low absorption by stain and deeper penetration into teeth. Hyperspectral images from 1000 to 2500 nm have been obtained for a total of 12 extracted teeth (premolars and molars) with different degrees of natural lesion. Analysis of the reflectance spectra suggests that light scattering by porous enamel and absorption by water in dentin can be used to quantify the lesion severity and generate a NIR caries score. Teeth were ground for histological examination after the measurements. The NIR caries score obtained correlates significantly (Spearman's correlation of 0.89, p<0.01) with the corresponding histological score. Results yield a sensitivity of >99% and a specificity of 87.5% for enamel lesions and a sensitivity of 80% and a specificity >99% for dentine lesions. The nature of the technique offers significant advantages, including the ability to map the lesion distribution rather than obtaining single-point measurements, it is also noninvasive, noncontact, and stain insensitive. These results suggest that NIR spectral imaging is a potential clinical technique for quantitative caries diagnosis and can determine the presence of occlusal enamel and dentin lesions.