In vitro assessment of cone beam local computed tomography for proximal caries detection

To compare the accuracy of local computed tomography (LCT) and conventional radiography for proximal caries detection and depth assessment.

An in vitro model was used consisting of 20 extracted posterior teeth with 18 caries lesions. Local computed tomography slices were reconstructed in axial and parasagittal planes from 100 basis projections. Conventional radiographs were also acquired. Eight observers determined the presence and depth of caries lesions. Receiver operating characteristic analysis and weighted kappa statistics were used.

Local computed tomography had a mean A(z) score of 0.82 (SD = 0.07) and conventional radiography of 0.79 (SD = 0.08; analysis of variance: P > .05). Interobserver agreement was moderate. The mean kappa for depth assessment was 0.68 (SD = 0.06) for LCT and 0.47 (SD = 0.08) for conventional radiography (analysis of variance: P < .05). Local computed tomography resulted in substantial and conventional radiography in moderate interobserver agreement.

There is no difference between LCT and conventional radiography for proximal caries detection. Local computed tomography was more accurate for assessing caries lesion depth.