Abstract
INTRODUCTION: This in vitro study evaluated the validity and reliability of spectral-domain optical coherence tomography (SD-OCT) to detect approximal carious lesions compared to bitewing radiography. METHODS: Forty extracted human premolars and molars with fifty-four unrestored approximal surfaces (ICDAS ll score 0-3) were mounted in a patient-equivalent simulator and imaged by digital radiography (DR) and the prototype of an intraoral OCT probe. The lesion status of the tooth surfaces was validated histologically and by microtomography (µCT). Five calibrated raters with low to high level of clinical expertise analyzed the DR and OCT images at intervals of 2-4 weeks each. Intra- and interrater agreement (weighted Cohen's kappa κ, Fleiss' kappa), sensitivity, and specificity were calculated. RESULTS: The sensitivity of OCT (0.66-0.91) was higher than that of DR (0.46-0.82), as was the specificity (OCT: 0.67-0.92; DR: 0.33-0.58). Intrarater agreement with OCT was moderate to substantial (κ: 0.53-0.77; p < 0.001) and fair to substantial for DR (κ: 0.36-0.78; p < 0.05). The DR interpersonal agreement ranged from slight to substantial (κ: 0.1-0.74, Fleiss' κ: 0.23/0.24; p < 0.001) toward fair to substantial with OCT (κ: 0.27-0.62; Fleiss' κ: 0.23/0.18; p < 0.001). Agreement between histology and µCT was almost perfect (κ: 0.82; p < 0.001). CONCLUSIONS: In a clinically oriented simulation, OCT was more sensitive in detecting early approximal carious lesions than DR without impairing specificity or reproducibility. Clinical studies will have to show whether OCT can confirm these promising results.