Comparing reproducibility among observers for low-, standard-, and high-dose protocols of CBCT assessing anatomical landmarks and pathology related to third molars: An ex vivo study

Objective To assess the reliability of low-dose(LD) cone-beam computed tomography (CBCT) protocols for radiographic assessment of third molars, including anatomical structures and related pathology, compared with standard-dose(SD) and high-dose(HD) protocols across three CBCT units. Materials and Methods This ex vivo study included 25 teeth from nine anonymized human skulls, each with one to four impacted third molars. Each tooth was scanned using HD, SD, and LD protocols on three CBCT units. Radiographic parameters were assessed by multiple observers, and inter- and intra-observer agreement was statistically analysed. Results Inter-observer agreement varied substantially across diagnostic parameters, CBCT units, and dose protocols, with κ values generally ranging from moderate to almost perfect agreement. Pronounced variability among observer pairs was observed for certain parameters within the same protocol and unit. No consistent reduction in reproducibility was observed when lowering the dose from HD to SD to LD. Intra-observer reproducibility was generally high. Conclusion Reproducibility in the assessment of third molars was primarily influenced by the diagnostic parameter and the observer rather than by CBCT dose level or unit. Lowering dose did not result in a consistent change in reproducibility. These findings indicate that LD protocols may provide sufficient reproducibility for third molar-related assessments.