(G-243) Micromechanically Accurate Skeletal Surrogates: Imaging, Inverse Design, & Validation

Introduction:: In general, modern techniques are limited in that efficient methods for replicating spatially varying generalized 3D linear elastic anisotropy (SV-G3LEA) do not exist [1], [2]. Examples of these structures include naturally occurring stiff mineral-based composites such as dense earth and bone, where anisotropic spatial microstructural variability is of great importance. The following work is focused on a micromechanically accurate linear elastic reconstruction of bones from computed tomography (CT) data, addressing issues in clinical scanning and 2-level additive manufacturing (AM) based modeling/fabrication.

Materials and Methods:: Modern methods for mechanical mimicry of physical structures rely on: (Stage 1) imaging such as high-resolution CT (HR-CT) scanning, (Stage 2) spatially variate modeling of the design domain via discretization as parametric unit cells, and (Stage 3) fabrication via additive manufacturing (AM) methods [3]–[5]. This work follows the described stages in:

1. Development of a biomechanically optimized super resolution neural network (SR-NN).


2. Development of a highly efficient 3D inverse homogenization (IH) structural optimization algorithm for generalized anisotropic elasticity tensor mimicry independent of printing material.


3. Assembly and mechanical testing of surrogate bones.

Towards Stage 1, structural improvements on SR-NNs [6] are being made. Such an algorithm is hypothesized to promote accurate recreation of micromechanical anisotropy.

Towards Stage 2, efforts in Graphical Processing Unit (GPU) accelerated structural analysis [7] are under way. Towards a novel generalized solution to 3D IH, these developments in structural analysis along with derivations upon the Parametric Level Set Method (PLSM) [1] approach to topological shape optimization are used.

Towards Stage 3, a material of similar elasticity to bone is used, synthesizing outcomes of Stages 1 and 2.

Results, Conclusions, and Discussions:: Figure 1 shows a skeletal surrogate under lateral/sideways impact. Testing indicates that similar failure patterns can emerge if structural and material information are appropriately accounted for.

This work is leading to highly efficient algorithms in level set optimization and 3D structural analysis. As shown above, intuitive fractures result from load. While the results above are qualitative and the displayed fabrication only accounts for spatially varying isotropic elasticity, it is anticipated that once spatially variate generalized anisotropy is achieved, a novel and important 2-level structural optimization approach will result.

Acknowledgements (Optional): : We would like to thank the Wake Forest Center for Biomedical Informatics (WFBMI) group for funding part of this work under the Fall 2022-Fall 2023 AI Pilot Award. Additionally, we would like to thank the North Carolina Space Grant Consortium for supporting the presenting author as a 2023-2024 Graduate Research Fellow.

References (Optional): :

1. Y. Wang, Z. Luo, N. Zhang, and Z. Kang, ‘Topological shape optimization of microstructural metamaterials using a level set method’, Comput Mater Sci, vol. 87, pp. 178–186, May 2014, doi: 10.1016/J.COMMATSCI.2014.02.006.


2. O. Sigmund, ‘Materials with prescribed constitutive parameters: An inverse homogenization problem’, Int J Solids Struct, vol. 31, no. 17, pp. 2313–2329, Sep. 1994, doi: 10.1016/0020-7683(94)90154-6.


3. J. Park, T. Zobaer, and A. Sutradhar, ‘A Two-Scale Multi-Resolution Topologically Optimized Multi-Material Design of 3D Printed Craniofacial Bone Implants’, Micromachines (Basel), vol. 12, no. 2, pp. 1–19, Feb. 2021, doi: 10.3390/MI12020101.


4. H. Jia, A. Misra, P. Poorsolhjouy, and C. Liu, ‘Optimal structural topology of materials with micro-scale tension-compression asymmetry simulated using granular micromechanics’, Mater Des, vol. 115, pp. 422–432, Feb. 2017, doi: 10.1016/J.MATDES.2016.11.059.


5. J. A. Morelli, R. Canfield, and V. Tech, ‘Evaluation of Topology Optimization Filtering with Numeric Examples’, 2019, Accessed: Mar. 02, 2023. [Online]. Available: https://vtechworks.lib.vt.edu/handle/10919/102220


6. C. You et al., ‘CT Super-resolution GAN Constrained by the Identical, Residual, and Cycle Learning Ensemble(GAN-CIRCLE)’, IEEE Trans Med Imaging, vol. 39, no. 1, pp. 188–203, Aug. 2018, doi: 10.1109/TMI.2019.2922960.


7. P. C. F. Lopes, A. M. B. Pereira, E. W. G. Clua, and R. Leiderman, ‘A GPU implementation of the PCG method for large-scale image-based finite element analysis in heterogeneous periodic media’, Comput Methods Appl Mech Eng, vol. 399, p. 115276, Sep. 2022, doi: 10.1016/J.CMA.2022.115276.