Linear measurements of CBCT superimposition

Abstract

Objectives: To assess accuracy and reliability of linear measurements for analysis of CBCT superimposition and determine if linear measurements are enough without requiring advanced analytical tools. Methods: Pre and post-operative CBCT scans of 20 orthodontic cases were used for CBCT superimposition. The gold standard was linear CBCT measurements on the axial section of each separate scan (pre and post) before superimposition. Voxel based and point based superimposition were done using Dolphin 3D (Dolphin Imaging and Management, Chatsworth, California, USA), In Vivo (Anatomage, San Jose, California, USA), and Ondemand 3D (Ondemand 3D; Cybermed Co., Seoul, Korea). Analysis of the superimposition was obtained on axial section of the superimposed models. Results were compared to the gold standard using error assessment according to Dahlberg error, Bland &Altman method and intra-class correlation coefficient. Results: Dolphin side by side method showed 25.4% relative Dahlberg error (RDE), 4.28 limit of agreement (LOA) and 0.768 intra-class correlation coefficient (ICC). In-Vivo point method showed 26.75% RDE, 4.48 LOA and 0.87 ICC. Whereas In-Vivo volume showed 24.66% RDE, 4.15 LOA and 0.822 ICC. On Demand showed 28.44%, 4.78 LOA and 0.7 ICC. Intra observer reliability of all methods regardless of the observers are poor. Conclusion: There is weak accuracy of the linear measurements on axial plane of CBCT superimposed models and it is not recommended alone for analysis.