| Three dimensional imaging in the form of Cone Beam Computed Tomography has become prevalent in the field of orthodontics. Analytical methods of resulting volumetric data sets have not kept pace with the technology capable of producing them. Current 3D analysis techniques are largely adaptations of existing 2D methods, offering no clear diagnostic advantage over traditional imaging techniques in light of increased radiation exposure, and cannot be compared with norms generated from 2D image capture sources. In order to study morphology in 3D, data sets must be generated for longitudinal studies and native 3D analytical methods must also be developed. Existing methods of CBCT volume superimposition are cumbersome, involving complex software pipelines and multiple systems to complete the process. The goal of the current study was to develop a reproducible method of CBCT volume superimposition in the posterior cranial base in a single software package, and construct an easy to follow, step-by-step manual to facilitate future studies in craniofacial morphology.;Existing anonymized sequential CBCT volumes of three subjects meeting inclusion criteria were obtained from the Kornberg School of Dentistry Department of Radiology. Volumes for each subject were imported into AMIRA software, resampled to a standardized 0.5mm voxel size and superimposed with a mutual information algorithm. Posterior cranial base surface data was extracted using a semi-automatic technique. Resulting surface distance data was compiled and visualized through application of color maps. A streamlined image processing protocol was produced and documented in a detailed step-by-step manual.;Surface distance analysis of serial segmentations was performed to verify reliability of the process. Surface distance deviations greater than 0.5mm consistently fell below 0.2 percent of the total surface area.;Sequential scan superimpositions of all three subjects exhibited mean surface distances of less than 0.15mm. Two out of three subjects exhibited deviations of greater than 0.5mm in less than 1 percent of the total surface area, suggesting consistent subvoxel accuracy of the protocol. |
