Elsevier

Journal of Biomechanics

Volume 44, Issue 5, 15 March 2011, Pages 935-942
Journal of Biomechanics

Three-dimensional micro-level computational study of Wolff's law via trabecular bone remodeling in the human proximal femur using design space topology optimization

https://doi.org/10.1016/j.jbiomech.2010.11.029Get rights and content

Abstract

The law of bone remodeling, commonly referred to as Wolff's Law, asserts that the internal trabecular bone adapts to external loadings, reorienting with the principal stress trajectories to maximize mechanical efficiency creating a naturally optimum structure. The goal of the current study was to utilize an advanced structural optimization algorithm, called design space optimization (DSO), to perform a micro-level three-dimensional finite element bone remodeling simulation on the human proximal femur and analyse the results to determine the validity of Wolff's hypothesis. DSO optimizes the layout of material by iteratively distributing it into the areas of highest loading, while simultaneously changing the design domain to increase computational efficiency. The result is a “fully stressed” structure with minimized compliance and increased stiffness. The large-scale computational simulation utilized a 175 μm mesh resolution and the routine daily loading activities of walking and stair climbing. The resulting anisotropic trabecular architecture was compared to both Wolff's trajectory hypothesis and natural femur samples from literature using a variety of visualization techniques, including radiography and computed tomography (CT). The results qualitatively revealed several anisotropic trabecular regions, that were comparable to the natural human femurs. Quantitatively, the various regional bone volume fractions from the computational results were consistent with quantitative CT analyses. The global strain energy proceeded to become more uniform during optimization; implying increased mechanical efficiency was achieved. The realistic simulated trabecular geometry suggests that the DSO method can accurately predict bone adaptation due to mechanical loading and that the proximal femur is an optimum structure as the Wolff hypothesized.

Introduction

Internal bone architecture, with its unique complex material matrix, has long been studied to determine the underlying principles of its adaptation and “remodeling”. Wolff, 1892, Wolff, 1986 proposed that trabecular bone in the proximal femur functionally adapts to external mechanical loading stimuli, orientating to align with the principal stress trajectories. Wolff had observed the “self-optimizing” property of bone and theorized that bone achieves maximum mechanical efficiency with minimal mass: a naturally optimum structure. These hypotheses known as Wolff's Law has been analyzed, critiqued, and refined using clinical, experimental, and analytical means. In the past few decades, computational techniques have been developed that utilize the finite element (FE) method to simulate this “bone remodeling” process.

Carter and Hayes (1977), Carter (1984) proposed some of the earliest theoretical frameworks for adaptive bone remodeling, including developing the apparent bone density elasticity formulation utilizing an exponential penalization factor, similar to the penalized material model used in topology optimization (Bendsøe, 1989). They completed several of the first single load continuum-level analyses using peak effective stress for remodeling (Carter, 1987, Carter et al., 1987, Carter et al., 1989). Huiskes et al. (1987) proposed a time-dependent strain energy density (SED) approach for remodeling, which has become one of the dominant methods for computational bone remodeling. Additionally, several stress-based methods have also been developed (Beaupré et al., 1990, Adachi et al., 1997). With increasing computing power, recent studies have been able to use micro-FE models that accurately represent trabecular bone architecture; however, these are still predominantly two-dimensional models that inadequately describe actual femur loading conditions and feature simplified geometries (Tsubota et al., 2002, Jang and Kim, 2008, Jang and Kim, 2009). Previous three-dimensional studies have typically used a macro-level or hierarchical multi-scale approach to reduce computational cost (Bagge, 2000, Coelho et al., 2009). These methods provide adequate insight into overall mechanical properties and density distribution but cannot accurately represent the individual trabecular architecture. Three-dimensional micro-FE models of the proximal femur have been created using computed tomography (CT) scans to complete a single stress analysis; such simulations can only be achieved by utilizing computationally efficient algorithms and parallel computing (van Rietbergen et al., 2003, Verhulp et al., 2006). Computational bone remodeling requires numerous FE analyses in an iterative fashion and additional optimization calculations; as a result, the computational cost is much greater. Only one study has achieved three-dimensional results using a micro-level resolution using local-stress criteria (Tsubota et al., 2009). The common feature of these various numerical approaches is that they are empirically derived specifically to emulate the bone remodeling process. These phenomenological approaches can predict the progress of bone remodeling; however, they do not prove that bone is an optimum structure, which is an essential assumption in Wolff's Law.

This study utilized design space optimization (DSO), a rigorous mathematical structural optimization technique developed by Kim and Kwak (2002), in order to determine the global optimum structure of the cancellous bone in the proximal femur. DSO is a specialized topology optimization algorithm that attempts to distribute a finite amount of material into the areas of highest loading to achieve an optimal strength to weight ratio by minimizing the global strain energy (SE). This method has the unique ability to describe the intermediate structural adaptation progress in the time domain and incorporate multi-disciplinary and multi-objective models (Jang and Kim, 2009).

The objective of this study is to conduct the first micro-level three-dimensional FE bone remodeling simulation of the proximal femur using DSO topology optimization to address Wolff's hypothesis of self optimization using proven mathematical theory. All loading and geometric simplifications required for previous studies were addressed by utilizing accurate three-dimensional loads and an accurate femur model based on clinical and CT scan data. We analyzed the results by comparing them to natural bone via cross-sectional imagery and by using new novel methods, such as simulated radiographs and quantitative comparison to morphologic parameters, to fully address the hypothesis of natural optimality using structural optimization, while identifying key limitations and future areas of improvement.

Section snippets

Modeling and simulation conditions

The femur model was created using CT scans from a 90.8 kg, 183 cm male (Heiner and Brown, 2001). A micro-FE model was constructed using 23.3 million finite elements, representing over 84 million degrees of freedom, with an individual cubic voxel size of 175 μm. The initial cancellous region was populated with randomly positioned tori, consistent with previous studies, to create a primarily isotropic cancellous structure, as shown in Fig. 1 (Jang and Kim, 2008, Tsubota et al., 2002, Tsubota et al.,

Results

Ward's classification (Ward, 1838, Whitehouse and Dyson, 1974) of the trabecular architecture contains four main loading groups: the principal compressive group (PC), the principal tensile group (PT), the secondary compressive group (SC), and the secondary tensile group (ST) (Fig. 3). An additional area of interest known as Ward's Triangle, in the central neck region, is also noted; this area is known to be a location of minimum bone density in the proximal femur (Tobin, 1955).

The initial

Discussion

A three-dimensional FE bone remodeling simulation in the proximal femur was completed using the DSO topology optimization method. The computational simulation utilized a large-scale micro-finite element model with two daily loading regimes, walking and stair climbing. The resulting adapted femur architecture reveals similar trabecular patterns in the four anisotropic loading groups as discussed above. Measurement of the simulation arch angles in the neck region reveals an average range from 84°

Conflict of interest statement

There are no conflicts of interest.

Acknowledgements

This work is funded through the Nature Sciences and Research Council of Canada (NSERC). The authors thank Dr. K. Svanberg at KTH (Stockholm, Sweden) for providing the MMA code for academic research. The authors also thank Dr. J. Skedros, Dr. C. Lovejoy, and Dr. F. Baruffaldi for providing the various images of human proximal femur.

References (47)

  • M. Heller et al.

    Determination of muscle loading at the hip joint for use in pre-clinical testing

    Journal of Biomechanics

    (2005)
  • R. Huiskes et al.

    Adaptive bone remodeling theory applied to prosthetic design analysis

    Journal of Biomechanics

    (1987)
  • I.G. Jang et al.

    Computational study of Wolff's law with trabecular architecture in human proximal femur using topology optimization

    Journal of Biomechanics

    (2008)
  • I.G. Jang et al.

    Application of design space optimization to bone remodeling simulation of trabecular architecture in human proximal femur for higher computational efficiency

    Finite Elements in Analysis and Design

    (2010)
  • I.G. Jang et al.

    Computational simulation of simultaneous cortical and trabecular bone change in human proximal femur during bone remodeling

    Journal of Biomechanics

    (2010)
  • R.M.V. Pidaparti et al.

    Cancellous bone architecture: advantages of nonorthogonal trabecular alignment under multidirectional joint loading

    Journal of Biomechanics

    (1997)
  • J.G. Skedros et al.

    Mathematical analysis of trabecular ‘trajectories’ in apparent trajectorial structures: the unfortunate historical emphasis on the human proximal femur

    Journal of Theoretical Biology

    (2007)
  • K. Tsubota et al.

    Functional adaptation of cancellous bone in human proximal femur predicted by trabecular surface remodeling simulation toward uniform stress state

    Journal of Biomechanics

    (2002)
  • K. Tsubota et al.

    Computer simulation of trabecular remodeling in human proximal femur using large-scale voxel FE models: approach to understanding Wolff's Law

    Journal of Biomechanics

    (2009)
  • C.H. Turner et al.

    The elastic properties of trabecular and cortical bone tissues are similar: results from two microscopic measurement techniques

    Journal of Biomechanics

    (1999)
  • E. Verhulp et al.

    Comparison of micro-level and continuum-level voxel models of the proximal femur

    Journal of Biomechanics

    (2006)
  • T. Adachi et al.

    Simulation of trabecular surface remodeling based on local stress nonuniformity

    JSME International Journal

    (1997)
  • F. Baruffaldi et al.

    An innovative CCD-based high-resolution CT system for analysis of trabecular bone tissue

    IEEE Transactions on Nuclear Science

    (2006)
  • Cited by (111)

    • Simulation on bone remodeling with stochastic nature of adult and elderly using topology optimization algorithm

      2022, Journal of Biomechanics
      Citation Excerpt :

      Obtaining the similar solutions with surface remodeling algorithms, topology optimization (TO) algorithm is more efficient to predict bone remodeling, due to better iterative convergence (Bagge, 2000; Coelho et al., 2009). With the prerequisite condition of perimeter constraint, the micro-level trabeculae structure of femoral was reconstructed by means of TO algorithm (Boyle and Kim, 2011b; Jang and Kim, 2008; Park et al., 2018). Wu et al. introduced a local density constraint into TO algorithm to reproduce bone-like cancellous structure and claimed the superior robustness and manufacturability of the artificial trabeculae structure (Wu et al., 2018).

    View all citing articles on Scopus
    View full text