Tooth segmentation by low-dose CBCT for orthodontic treatment planning

Three-dimensional imaging has become an increasingly important component of orthodontics. Associated with this, however, is a higher radiation exposure for patients. New cone-beam computed tomography (CBCT) devices have been developed that can provide low-dose CBCT (LD-CBCT). We hypothesized that LD-CBCT is as precise and reproducible as standard high-dose CBCT (HD-CBCT) in segmenting roots and crowns as well as measuring tooth length.

HD-CBCT and LD-CBCT scans were taken of four human cadaveric heads. Thirty single-rooted teeth were segmented twice by one investigator. The length of each tooth was also measured. Lin’s concordance correlation coefficient (CCC) was calculated to assess the agreement of HD-CBCT and LD-CBCT measurements and the intraclass correlation coefficient (ICC) was calculated to assess intrarater reliability. Analyses were supported by Bland–Altman plots.

Volume measurements obtained using HD-CBCT were significantly higher than those obtained using LD-CBCT (p < 0.001). CCC was 0.975 (95% confidence interval [CI] = 0.956–0.986) indicating excellent agreement between the two modalities. Intrarater reliability between the two sets of LD-CBCT and HD-CBCT volume measurements was excellent (ICC = 0.998, 95%CI = 0.995–0.999 [HD-CBCT], ICC = 0.997, 95%CI = 0.992–0.998 [LD-CBCT]). CCC for tooth length measurements was 0.991 (95% CI = 0.983–0.995), indicating excellent agreement between HD-CBCT and LD-CBCT. Intrarater reliabilities between the two sets of tooth length measurements were also excellent for both methods (ICC = 0.998, 95%CI = 0.995–0.999 [HD-CBCT], ICC = 0.997, 95%CI = 0.992–0.998 [LD-CBCT]).

Within the limitations of this experimental setting, LD-CBCT is as valid as HD-CBCT for measuring tooth length. Regarding the volume differences, in vivo studies are required to determine their clinical relevance.