Verification and Validation of Microcomputed Tomography-Based Analyses of Bone Morphology, Apparent Elastic Modulus, and Von Mises Stress.

Abstract

Patient-specific image-based finite element analysis (FEA) offers a noninvasive method to estimate bone stiffness and strength, but its clinical translation requires thorough verification and validation. While previous studies provide resolution-dependent errors in trabecular bone morphometric indices, there remains a need to extend these analyses to include mechanical property predictions. This study examined microcomputed tomography (μCT) voxel size effects on trabecular bone morphometric analyses and predictions of apparent elastic modulus (Eapp) and trabecular stress. Voxel (v)- and geometry (g)-based FEA of trabecular bone cores under compression were compared. Convergence analyses were performed on element size and boundary conditions; and, bulk density, tissue density, and FEA were validated against experimental measurements. Trabecular bovine bone cores (n = 22, Ø10 mm × 10 mm) were tested quasi-statically under uni-axial compression below the yield limit, and μCT scanned with isotropic voxel sizes of 20 μm, 50 μm, and 100 μm. Increasing voxel size resulted in significant reductions in connectivity density, anisotropy, and trabecular number, while trabecular thickness increased. Predicted Eapp from both v-FEA and g-FEA were highly correlated with measurements (R2 = 0.8, p < 0.001). A maximum mean error of +6.35% was found in Eapp with 100 μm voxel size. Despite its high computational cost (∼120 min), 20 μm v-FEA was the most accurate for predicting Eapp and trabecular stress. Alternatively, both 50 μm v-FEA and g-FEA at low mesh density (h = 0.0011, where h is 1/√number of elements) predicted both Eapp and trabecular stress behavior with less than ±5% error in 10-20 min. Findings support efficient, accurate FEA strategies for predicting trabecular bone mechanics.