Differentiation of Occlusal Discolorations and Carious Lesions with Hyperspectral Imaging In Vitro

Stains and stained incipient lesions can be challenging to differentiate with established clinical tools. New diagnostic techniques are required for improved distinction enable early noninvasive treatment. This in vitro study evaluates the performance of artificial intelligence (AI)-based classification hyperspectral imaging data occlusal lesion detection differentiation from stains. Sixty-five extracted permanent human maxillary mandibular bicuspids molars (International Caries Detection Assessment System [ICDAS] II 0–4) were imaged a camera (Diaspective Vision TIVITA® Tissue, Diaspective Vision, Pepelow, Germany) at distance 350 mm, acquiring spatial spectral information wavelength range 505–1000 nm; 650 fissural spectra used train algorithms (models) automated between but sound enamel lesions. Stratified 10-fold cross-validation was used. The model highest performance, fine k-nearest neighbor algorithm, classify five additional tooth areas. Polarization microscopy ground sections served as reference. Compared lesions, intact showed higher reflectance 525–710 nm lower 710–1000 nm. A algorithm achieved Matthews correlation coefficient (MCC) 0.75, sensitivity 0.95 specificity 0.80 when distinguishing spectra. superposition color-coded results on further projections enabled qualitative assessment entire fissure’s health. AI-based evaluation images is highly promising complementary method visual radiographic examination detection.