The aim of this study was to evaluate the reproducibility of the third molar assessment, comparing five observers and two cone beam CT (CBCT) units. 28 patients, each with two impacted mandibular third molars, were included. Each patient was randomly examined with a Scanora(®) 3D (Soredex, Helsinki, Finland) CBCT unit in one mandibular third molar region and with a Cranex(®) 3D (Soredex) CBCT unit in the other region. Five observers with varying CBCT experience assessed all third molars and recorded the following variables: number and morphology of the roots, relation to the mandibular canal in two directions, shape of the canal and whether there was a direct contact between the roots of the molar and the mandibular canal. The radiographic assessments were compared pairwise among all observers for all variables. Wilcoxon's signed-rank test was used to test the differences in observer accordance percentages among the recorded variables in the images from the two units, and kappa statistics expressed interobserver reproducibility. The mean percentages for observer accordance ranged from 65.4 to 92.9 for Scanora 3D and 60.3 to 94.8 for Cranex 3D. There was no significant difference between the observer accordance in the two CBCT units (p > 0.05), except for assessing root flex in the mesiodistal direction, for which the observer accordance was higher for Scanora 3D (p < 0.05). Kappa values ranged from 0.1 to 1.0 for Scanora 3D and from 0.2 to 0.9 for Cranex 3D. For the variable "direct contact", the interobserver reproducibility was excellent for the two trained radiologists. The two units had almost similar interobserver reproducibility for mandibular third molar assessment. Observer variation existed, and experienced radiologists demonstrated the highest interobserver reproducibility for canal-related variables.
Keywords: cone-beam computed tomography; molar, third; radiography.