Reliability and accuracy of cone beam computed tomography versus conventional multidetector computed tomography for...

To assess the reliability and accuracy of linear measurements on three-dimensional (3D) cross-sectional images, both acquired with cone beam computed tomography (CBCT) and multi-detector row CT (MDCT). Bone thickness was evaluated with regard to image-guided planning of craniofacial implant surgery.

Five dry human skulls were used. Cuts were made with a circular bone saw at the ideal implant positions in the nasal, orbital, and temporal regions prior to acquisition of CBCT and MDCT scans. After imaging examination, bone width was assessed by three independent observers using a caliper and defined as a reference. In the next step, cross-sectional images of the regions with the aforementioned cuts were reconstructed from 3D virtual models generated from the digital DICOM datasets with the use of 3D image-based planning software. Subsequently, linear measurements were performed. The systematic difference and interobserver and intraobserver variation of MDCT and CBCT linear measurements were compared with the physical measurements at different locations in the nasal, orbital, and temporal region, respectively. Also, the potential influence of different gray-level values was investigated. The quantitative accuracy of distance measurements was performed using a two-way analysis of variance (ANOVA) and variance component analyses. Only differences with P values < .05 were considered significant.

All radiologic measurements showed a significant overestimation of the bony dimensions, reaching more than the used voxel sizes of 0.3 mm for CBCT and 0.5 mm for MDCT. For CBCT, an average measurement bias of 0.39 to 0.53 mm and for MDCT of 0.57 to 0.59 mm was found. MDCT images showed less interobserver variation in linear measurements on cross-sectional images from 3D virtual models compared with CBCT images. Contrast settings statistically significantly influenced linear measurements of bone width for CBCT images (P < .0015) and interobserver variation on MDCT imaging (P < .029).

Both CBCT images (KaVo 3D eXam Imaging System) and MDCT images (Aquilion ONE, Toshiba) showed a highly consistent submillimeter overestimation of the anatomical truth in assessing bone thickness of nasal, orbital, and temporal regions of ex vivo specimens. When using CBCT and MDCT images for presurgical assessment, one should be aware of the overestimation of the cortical bone thickness.