Accuracy of linear measurements in stitched versus non-stitched cone beam computed tomography images

Abstract

Cone beam computed tomography (CBCT) images are useful in clinical dentistry. Linear measurements are necessary for accurate treatment planning. Current program called stitching program in Kodak 9000C 3D systems automatically combines up to three localized volumes to construct larger images with small voxel size. The purpose of this study was to assess the accuracy of linear measurements from stitched and non-stitched CBCT images in comparison to direct measurements. This study was performed in 10 human dry mandibles. Gutta-percha rods were marked at reference points to obtain 10 vertical and horizontal distances. Direct measurements by digital caliper were served as gold standard. All distances on CBCT images obtained by using and not using stitching program were measured, and compared with direct measurements. The intraclass correlation coefficients (ICCs) were calculated. The ICCs of direct measurements were 0.998 to 1.000. The ICCs of intra-observer of both non-stitched CBCT images and stitched CBCT images were 1.000 indicated strong agreement made by a single observer. The inter-method ICCs between direct measurements vs non-stitched CBCT images and direct measurements vs stitched CBCT images ranged from 0.972 to 1.000 and 0.967 to 0.998, respectively. No statistically significant differences between direct measurements and stitched CBCT images or non-stitched CBCT images (P > 0.05).