Accuracy and precision of linear measurements in cone beam computed tomography Accuitomo® tomograms obtained with different reconstruction techniques

Abstract

Objectives: To assess the accuracy and precision of linear measurements, the influence of different object positions in cone beam CT (CBCT) images and the effect of different systems for reformatting and viewing the tomograms on measurement values. Methods: An object consisting of Plexiglas® plates with metal spheres was radiographically examined in three positions using the CBCT unit Accuitomo®. Tomograms were reformatted using the i-Dixel software and Sectra® picture archiving and communication system (PACS) multiplanar reformatting (MPR). Two observers measured 20 linear distances in axial, frontal and sagittal planes. Physical measurements of actual distances by means of a digital caliper served as the gold standard. Results: For measurements performed on the Accuitomo and Sectra PACS workstations, the mean differences between repeated measurements were below 0.13 mm (standard deviation (SD) 0.13 mm) and 0.21 mm (SD 0.18 mm), respectively. Mean differences between the two observers were below 0.11 mm (SD 0.07 mm) and 0.15 mm (SD 0.09 mm). The mean difference between the gold standard and the basic position when using the Accuitomo workstation, was -0.08 mm (SD 0.21 mm) and for a deviated/rotated position -0.08 mm (SD 0.23 mm) and -0.09 mm (SD 0.20 mm). None was statistically significant (P > 0.05). For measurement performed on MPR tomograms using the Sectra PACS workstation, mean differences were -0.09 mm (SD 0.17 mm) for basic position, and -0.08 mm (SD 0.19 mm) and -0.13 mm (SD 0.15 mm) for deviated and rotated positions, respectively. The mean difference between gold standard and basic/rotated positions was statistically significant, with P = 0.032 and P = 0.002. Conclusions: Our study confirms previous findings indicating that measurements in 3DX Accuitomo FPD (flat panel detector) tomograms show a high level of agreement with actual distances. Our findings suggest the possibility of minor distortion in tomograms reformatted by means of Sectra PACS MPR. © 2009 The British Institute of Radiology.