Parametric investigation of load-induced structure remodeling in the proximal femur

The process of adaptive bone remodeling can be simulated with a self-optimizing finite element method. The basic remodeling rules attempt to obtain a constant value for the strain energy per unit bone mass, by adapting density. The precise solution is dependent on the loads, initial conditions, and the parameters of the remodeling rule. While there are several investigations on developing algorithms leading to the bone density distribution in the proximal femur, these algorithms often require a large number of iterations. The aim of this study was to develop a more efficient adaptive bone remodeling algorithm, and to identify how the bone density distribution of the proximal femur was affected by parameters that govern the remodeling process. The forces at different phases of the gait cycle were applied as boundary conditions. The bone density distributions from these forces were averaged to estimate the density distribution in the proximal femur. The effect of varying the initial bone density, spatial influence function, non-linear order of the adaptive algorithm, and the influence range on the converged solution were investigated. The proposed procedure was shown to converge in a fewer number of iterations and requiring less computational time, while still generating a realistic bone density distribution. It was also shown that varying the identified parameters within reasonable upper and lower bounds had very little impact on the qualitative form of the converged solution. In contrast, the convergence rate was affected to a greater degree by variation of these parameters. In all cases, the solutions obtained are comparable with the actual density in the proximal femur, as measured by Dual-energy X-ray absorptiometry (DEXA) scans.