Influence of voxel size on the accuracy of linear measurements of the condyle in images of cone beam computed tomography: A pilot study

Background: To analyze the influence of voxel size and exposure time on the accuracy of linear measurements of the condyle. Material and Methods: Four macerated hemi-mandibles of pigs were scanned in nine different voxel size protocols. Three-dimensional models of the condyle were generated in order to establish a comparison between linear measurements obtained with each voxel protocol and those obtained with a caliper (gold standard). The comparison between the protocols was performed considering the average of the two measurements of the condyle in the latero-medial (LM) and antero-posterior (AP) axes and also through repeated measurement ANOVA with rank transformation. The level of significance was 5%. Results: A significant difference was found between the protocols regarding the LM and AP variables (p-values = 0.0027 and 0.0263, respectively). In the LM axis, the protocol P6 (voxel size of 0.3 mm with scan time of 4.8 seconds) did not show statistical difference compared to the gold standard. The protocols P4 and P5 (voxel size of 0.25 mm with scan times of 14.7 and 26.9 seconds, respectively) were both statistically similar compared to caliper, although they have presented a longer scan time. In the AP axis, the protocol P8 (voxel size of 0.4 mm with time scan of 4.8 seconds) was statistically similar to the gold standard. Conclusions: A smaller voxel size does not necessarily mean more accuracy regarding the linear measurements of the condyle. It is possible to obtain an acceptable level of accuracy with a larger voxel size and a shorter exposure time to radiation.