The aim of this study was to evaluate the diagnostic accuracy of cone-beam computerized tomography (CBCT) for the detection of noncavitated proximal carious lesions and to compare the accuracy with conventional film-based radiographs and the digital phosphor-plate system.
Methods:
135 extracted, noncavited, unrestored human teeth (67 premolars and 68 molars) were placed with approximal contacts and imaged with two intraoral modalities: Digora Optima (Soredex) and Kodak Insight film, and the NewTom 3G CBCT system (Quantitative Radiology) with 9 inch FOV. The volumetric data were reconstructed and sectioned (0.25mm) in the mesiodistal plane. Evaluation of all intraoral and cone-beam CT images was performed twice by two observers using a 5-point rating scale. Scores from each observer and each image modality were compared with an histological examination, as the reference/gold standard, using receiver operating characterictic (ROC) analysis to evaluate diagnostic ability. Parameters expressing other aspects of accuracy (sensitivity, specificity and predictive values) were also computed, separately, for enamel and dentinal lesions. Weighted kappa coefficients were calculated to assess intra- and interobserver agreement for each radiographic modality.
Results:
The mean Az values were 0.629 ± 0.015 for NewTom 3G, 0.665 ± 0.017 for phosphor plates and 0.667 ± 0.017 for film, respectively. Analysis revealed statistically significant differences between the CBCT system and both intraoral modalities (p<0.05). For enamel lesions, the average sensitivity score using NewTom 3G (0.068) was approximately one third of that for phosphor plates (0.189) and the differences were significant. For lesions extending into dentin, the average sensitivity score using the NewTom 3G system was a little better than that of intraoral radiographs.
Conclusions:
For detecting incipient, noncavitated proximal carious lesions, cone-beam computed tompography was less accurate than intraoral modalities.