Finite Element Analysis for the Load-Bearing Femur and Tibia in the Human Knee Using an in vivo HR-pQCT Protocol

Abstract

Abstract. High-resolution peripheral quantitative computed tomography (HR-pQCT) combined with micro-finite element (μFE) analysis enables noninvasive assessment of bone mechanics. While widely used at the distal radius and tibia, knee modeling is challenging due to complex geometry, large model sizes, and the need for physiologically relevant boundary conditions. Standardization and reproducibility are critical for interpreting mechanical predictions. HR-pQCT images were collected from one participant with recent anterior cruciate ligament (ACL) injury for boundary condition testing and 28 healthy volunteers for reproducibility. Patient-specific μFE models incorporated anatomically shaped or rectangular simulated-polymethyl methacrylate (PMMA) support layers with variable stiffness (1500–3000 MPa) and extrusion lengths (1–7 mm). Sensitivity analyses quantified the influence of layer geometry, stiffness, and length on strain energy density (SED). Reproducibility of repeated tibial scans was assessed using root-mean-square percentage coefficient of variation (RMS %CV) and intraclass correlation coefficients (ICCs). μFE knee models contained hundreds of millions of degrees-of-freedom, with solution times of 3.6–16.4 h for anatomically shaped layers and 4.1–20.1 h for rectangular layers (32 cores). Support layer stiffness had the greatest effect on tibial SED, whereas geometry and length had minimal influence. Reproducibility across tibial scans ranged from 5.5% to 18.4% RMS %CV, with ICCs of 0.72–0.94; peri-articular trabecular regions 5–7.5 mm below the articular surface were most stable. Standardized HR-pQCT-based μFE knee modeling is essential for reproducibility. An anatomically shaped, 3‐mm extruded PMMA-like support layer with ∼2500 MPa stiffness provides stable, computationally efficient load distribution and will facilitate comparability in future longitudinal knee studies.