RESUMO
OBJECTIVES: To determine whether the number of iterative restorations performed on TACT images affects observers' ability to detect dental caries. METHODS: Eight TACT basis images of 40 extracted human posterior teeth were acquired using a CCD sensor. TACT slices of each tooth were generated and subsequently submitted to 1, 2 or 3 iterative restorations. Stacks of images from the three experimental conditions were presented to six observers in a balanced order, on a high-resolution 21' color monitor. Observers scored the presence or absence of approximal caries using a 5-point confidence scale. Observers' assessments were compared with the results of histological examination. Receiver operating characteristic (ROC) curves were generated and possible significant differences between observers and between modalities tested by ANOVA. The level of significance was set at alpha=0.05. Interobserver reliability was calculated as intraclass correlation. RESULTS: Mean areas under the ROC curves (A(z)) for the three experimental conditions were 0.791 (one iterative restoration), 0.81 (two iterative restorations), and 0.778 (three iterative restorations). ANOVA did not demonstrate any significant difference between modalities (P=0.25) but did so between observers (P=0.031). Interobserver reliability was similar for all experimental conditions tested. CONCLUSIONS: Varying the number of iterative restorations from one to three does not affect observers' ability to detect approximal caries with TACT slices.