Error introduced by common reorientation algorithms in the assessment of rodent trabecular morphometry using micro-computed tomography.

Abstract

Quantitative analyses of bone using micro-computed tomography (μCT) are routinely employed in preclinical research, and virtual image reorientation to a consistent reference frame is a common processing step. The purpose of this study was to quantify error introduced by common reorientation algorithms in μCT-based characterization of bone. Mouse and rat tibial metaphyses underwent μCT scanning at a range of resolutions (6-30 μm). A trabecular volume-of-interest (VOI) was manually selected. Image stacks were analyzed without rotation, following 45° In-Plane axial rotation, and following 45° Triplanar rotation. Interpolation was performed using Nearest-Neighbor, Linear, and Cubic interpolations. Densitometric (bone volume fraction, tissue mineral density, bone mineral density) and morphometric variables (trabecular thickness, trabecular spacing, trabecular number, structural model index) were computed for each combination of voxel size, rotation, and interpolation. Significant reorientation error was measured in all parameters, and was exacerbated at higher voxel sizes, with relatively low error at 6 and 12 μm (max. reorientation error in BV/TV was 2.9% at 6 μm, 7.7% at 12 μm and 36.5% at 30 μm). Considering densitometric parameters, Linear and Cubic interpolations introduced significant error while Nearest-Neighbor interpolation caused minimal error, and In-Plane rotation caused greater error than Triplanar. Morphometric error was strongly and intricately dependent on the combination of rotation and interpolation employed. Reorientation error can be eliminated by avoiding reorientation altogether or by "de-rotating" VOIs from reoriented images back to the original reference frame prior to analysis. When these are infeasible, reorientation error can be minimized through sufficiently high resolution scanning, careful selection of interpolation type, and consistent processing of all images. © 2018 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 36:2762-2770, 2018.