Accuracy of Different Imaging CBCT Systems for the Detection of Natural External Radicular Resorption Cavities: An Ex Vivo Study

The aim of this study was to compare the sensitivity and specificity of 3 different systems of cone-beam computed tomography (CBCT) in the detection of natural external root resorption (ERR) cavities using microtomography as the gold standard. A sample of 126 ex vivo teeth were submitted to a microtomography examination to verify the presence/absence of ERR cavities. Then, they were divided into the control group: 85 teeth that did not present with an ERR cavity; and experimental group: 41 teeth that presented with 1 or more ERR cavities. The size of the natural ERR cavities varied from 2.46 mm3 to 3.11 mm3, which corresponded to cavities of 1.67 mm and 1.81 mm in diameter, respectively. The teeth were placed on a dry human mandible for scanning in each of the 3 protocols with different voxel sizes: 0.25 mm, 0.20 mm, and 0.166 mm. The accuracy of the 3 protocols evaluated in this study are listed in decreasing order: 60.3% for a voxel size of 0.20 mm, 56.7% for a voxel size of 0.166 mm, and 46.7% for a voxel size of 0.25 mm; these are smaller values than previous studies have obtained using artificial ERR cavities. Statistically significant results were not found among the 3 CBCT protocols that were used (P > .05), and the receiver operating characteristic curve shows the small differences found between the protocols. The results indicate that CBCT presents, for natural ERR, lower sensitivity and specificity values than those detected in previous studies of artificial cavities. The results demonstrate that natural ERR is neither easily observed nor accurately located by CBCT, as previous studies using artificial ERR indicated. •Compare sensitivity and specificity of 3 different CBCTs in the detection of natural ERR cavities.•Architecture of ERRs is different from each other in shape and size.•According to the findings of this study, voxel size 0.25 mm is not valuable or useful for the ERR evaluation.