PURPOSE--Current clinical practice uses crude radiographic measures to assess the integrity of bones with metastatic lesions. To address the inadequacy of these methods, this study investigated the use of finite element (FE) models for the prediction of strength of femoral shafts both with and without cortical defects. These models utilized CT scan data in the characterization of bone geometry and heterogeneity to achieve greater precision in the prediction of the flexural and torsional behavior of femoral shafts.

METHODOLOGY--This research involved three distinct phases: 1) validating the use of CT scan data in linear FE models of femoral shafts in flexion, 2) evaluating the effects of nonlinear material properties on model behavior, and 3) extending the linear CT scan-based FE models to a torsional loading configuration.

  The first phase involved six matched pairs of femoral shafts, one of each pair containing a hemispherical defect. CT scans were taken of all the bones prior to mechanical testing and were used for the generation of FE models. A cylindrical model was generated by using the CT scan data solely to ascertain periosteal and endosteal diameters. A second model utilized CT scan data to generate geometries more representative of the true bone geometry. A third model used the CT scan data both to characterize the true geometry and allow variation in material properties. The strengths predicted by all three models were then correlated to failure loads measured in four-point bending.

  For the second phase of the research, nonlinear material properties were used in the modeling of bones tested in flexion. Bilinear stress-strain relationships were assigned to each element, and loads were applied in 1 kN increments. Yield and failure loads determined from the models were then correlated to measured yield and failure loads.

  The final phase of this research involved five pairs of bones tested in torsion. Preparation of specimens and formulation of models was similar to the flexural experiment of the first phase. Likewise, correlations between predicted and measured failure torques were derived.

PROGRESS--All work on this study has been completed.

RESULTS--This investigation demonstrated the benefit of using CT scan data to desribe bone geometry and heterogeneity in the FE modeling of femoral shafts, both with and without defects. The linear models for the flexural loading configuration showed that model predictions were most precise when CT data was utilized for both geometry and heterogeneity (r=0.97). Nonlinear models were also precise in the prediction of ultimate failure load (r=0.99) and were more descriptive of structural behavior. Linear predictions of ultimate torque were precise (r=0.99) but were twice the magnitude of measured torques, perhaps as consequence of the anisotropic nature of cortical bone.

  Because of its ability to characterize irregular and changing cross-sections and material heterogeneity, this promising FE technique can potentially be extended to bones whose defects manifest odd geographies, moth-eaten borders, or permeative qualities.

FUTURE PLANS--The precision of the models in these two loading configurations suggests that clinical application of this technology may improve patient care. Future work may focus on improving model accuracy through revised material properties assumptions (i.e., plasticity, asymmetry, and anisotropy).

RECENT PUBLICATIONS FROM THIS RESEARCH

• Improved method for predicting the strength of femoral shafts with cortical defects. Rossi SA, Jones KA, Keyak JH, Skinner HB. Trans Orthop Res Soc 1996:21:602.

Go to TOP.

Previous	Contents	Next