Generation of TACT image slices using different reconstruction algorithms: Effects on natural caries detection

Objective. The purpose of this study was to determine whether the method of reconstruction of tuned-aperture computed tomography (TACT) slices has an influence on observer performance in caries detection. Study design. Eight images of each of 40 extracted posterior teeth were acquired by using a solid state intraoral detector. Stacks of tuned-aperture computed tomography slices were generated by using the minimum and average reconstruction methods. Slices of the 2 experimental conditions were presented to 8 observers in a balanced order. Images were viewed on a high-resolution 21-in color monitor. Observers scored the presence/absence of caries by using a 5-point confidence scale. Observers' assessments were compared with histologic examinations of tooth sections. Receiver operating characteristic curves measured observers' diagnostic performance. Analysis of variance was used to test for possible significant differences between observers and between experimental conditions. Sensitivity, specificity, accuracy, and interobserver reliability were also calculated. Results. No statistically significant differences between the 2 methods of reconstruction were found for the detection of either occlusal (P = .07) or proximal (P = .52) caries. Interobserver reliability was similar for both experimental conditions. Conclusions. The minimum and average methods of TACT reconstruction provide comparable performances for caries detection tasks.