Inelastic mechanical descriptors for osteoporotic hip fracture discrimination with 3D-DXA-based nonlinear finite element models.

Introduction: Dual-energy X-ray absorptiometry (DXA) is the gold standard for diagnosing osteoporosis. Advances in 2D-3D modelling to generate patient-specific 3D-DXA models out of DXA images enable accurate volumetric representations of the femur, with potential for fracture risk prediction when combined with finite element (FE) analyses. This study evaluates the ability of 3D-DXA-based FE models to discriminate hip fractures under side-fall loading. Methods: We used a retrospective case-control study including 128 women, 64 of whom suffered a hip fracture. Mechanical descriptors, including strength, nonlinear deformation, residual displacement, and energy absorption under elastic-plastic assumptions, were derived from force-displacement curves. Results: The area under the receiver operating characteristic curve (AUROC) showed that strength and trabecular volumetric bone mineral density (vBMD) equally discriminated between fracture and control subjects. Residual displacement due to plastic strain accumulation at failure emerged as a key descriptor which, when combined with strength, significantly improved fracture discrimination (ΔAUROC = 0.11 vs. areal bone mineral density (aBMD); ΔAUROC = 0.08 vs. trabecular vBMD). Discussion: These findings highlight the potential of 3D-DXA and FE modelling to improve fracture assessment within current DXA-based clinical workflows.