Effect of ultra-low dose CBCT versus conventional CBCT imaging on the accuracy and reliability of semi-automatically segmented mandibular condyle

Aim: This study evaluated the accuracy and reliability of semi-automated mandibular condyle segmentation using conventional and ultra-low-dose (ULD) cone-beam computed tomography (CBCT) protocols, with dry skull measurements as the reference standard. Methods: Twenty-nine dry human mandibles with intact condyles and glenoid fossae were scanned with two CBCT protocols: conventional (0.2 mm voxel size, 90 kV, 8 mA) and ULD (0.4 mm voxel size, 90 kV, 4 mA, ~50% dose reduction). Semi-automated segmentation was performed in 3D Slicer software. Linear condylar measurements (width, length, height) were obtained by two blinded evaluators, and their averages were analyzed against the reference standard. Descriptive statistics (mean, SD, CI), agreement analysis (ICC), and intergroup comparisons (ANOVA with Bonferroni correction) were performed. Results: Real measurements showed a mean condylar size of 12.08±0.99 mm. Conventional CBCT produced 11.88±0.91 mm, and ULD CBCT yielded 11.67±0.89 mm, with differences 0.8, p