Influence of Voxel Size and Filter Application in Detecting Second Mesiobuccal Canals in Cone-beam Computed...

This study assessed the influence of voxel size and filter application in detecting second mesiobuccal (MB2) canals in cone-beam computed tomographic (CBCT) images.

Using the OP300 CBCT system (Instrumentarium, Tuusula, Finland) and 3 voxel size protocols (80 μm, 125 μm, and 200 μm), we scanned 40 first molars: 20 with an MB2 canal and 20 without. All molars received silver palladium pins on the palatal root, whereas the non-MB2 molars were also filled with gutta-percha. Five oral radiologists assessed the presence of an MB2 canal under 3 filter application conditions: without filter, with sharpen 1 × filter, and with sharpen 2 × filter. Intra- and interobserver reproducibility was evaluated using the weighted kappa index. We compared the area under the receiver operating characteristic curves with SPSS Statistics v.20.0 (IBM Corp, Armonk, NY) using 2-way analysis of variance and the Tukey post hoc test with 5% significance level.

Our analysis found median intra- and interobserver agreement values of 0.70 and 0.56, respectively. The 80-μm voxel with sharpen 1 × filter image group had the highest sensitivity, accuracy, and negative predictive values. As for specificity and positive predictive, the 80-μm voxel group without filter application presented the highest values. The areas under the receiver operating characteristic curve were higher in the 80-μm groups than in the 125-μm and 200-μm voxel size groups (P < .05). We found no differences among the filters used (P = .22) or for the filter-voxel size interactions (P = .88).

A smaller voxel size increased the accuracy in detecting MB2 canals, whereas the enhancement filters did not.