Can proton density-weighted MRI-based finite element models predict bone strength?

Abstract

Hip fractures are a major clinical concern, and patient-specific finite element (FE) modeling of the proximal femur is increasingly used to assess bone strength and fracture risk at the hip. Although most patient-specific FE models rely on computed tomography (CT) imaging, CT requires relatively high doses of radiation, limiting its application. Magnetic resonance imaging (MRI)-based FE models present a radiation-free alternative that could expand clinical applicability. This study evaluated the predictive capability of proton density (PD)-weighted MRI-based FE models of the proximal femur under sideways fall. Cadaveric femora (n = 14) were imaged using MRI (n = 10) and CT (n = 14) to construct patient-specific FE models, and then mechanically tested to obtain experimental measures of stiffness and failure load. Existing relationships between image voxel intensity and tissue modulus were used; in the case of the MRI models, the relationship was based on a BV/TV-modulus equation previously applied in a T2-weighted MRI-FE study. Digital image correlation (DIC) was used to measure full-field strain distributions during mechanical testing to compare with FE predictions. Values of the whole-bone stiffness and failure load predicted by the MRI-based models showed poor agreement with experimentally measured values and paired CT-based FE predictions. However, when homogeneous material properties were assigned to the FE models, predicted stiffness and failure load no longer differed between MRI and CT. Notably, both MRI- and CT-based FE models qualitatively approximated the spatial distribution of principal strain patterns measured experimentally using DIC. These results indicate that the femoral geometry is captured well by PD-weighted MRI, but the appropriate material mapping relationship to use with this imaging sequence is unknown. Thus, future work should explore the relationship between PD-weighted MRI signal values and measures of modulus or density to advance an MRI-based FE framework for fracture risk assessment.