Micro-CT Image-Based Mesh Generation and Finite Element Analysis of Bone

This study investigated the reduced platen compression (RPC) test of the distal femur metaphysis (DFM) of adult male rats. The objectives were to develop a finite element (FE) mesh generation algorithm that would take existing µCT images and create a model for use in commercial finite element analysis (FEA) software, and to investigate the RPC test using different boundary conditions (BC) for comparison with modern bone FEA studies. The MATLAB Image Processing Toolbox was used to manipulate the µCT images from their exported format into a usable volume in the form of a 3D binary image. The voxels in this image were directly converted to hexahedral elements in a FE mesh using custom developed MATLAB scripts. The meshes were imported to ABAQUS for FE processing. Post-processing was done in both ABAQUS and MATLAB. ABAQUS was used to visualize the results in contour plots. The stress and strain values for each element were exported for distribution analysis in MATLAB. To validate the FEA techniques developed in this study, two animals were chosen based on their RPC mechanical testing results; one with significantly above average and one with significantly below average maximum load. Both whole specimen compression (WSC) and RPC tests were simulated using rough and smooth BCs. The reaction force results of the simulations mirrored past mechanical testing, but the outlier effect was muted due to greater consistency in platen sizing in the simulations than in RPC testing. The contour plots and strain distributions indicated that the RPC test has little influence from the cortical tissue, supporting the assumption that the RPC test effectively measures cancellous tissue properties. The computational requirements of the image processing and mesh generation techniques, along with the FE processing in ABAQUS are analyzed for future study feasibility. Computation time is not a pressing issue, but memory becomes the limiting factor for both simulation processing and result storage. Overall, the mesh generation technique developed in this study is powerful due to its generality; any binary image can be converted in a FE mesh. Future possibilities include extending the analysis to nonlinear FEA and incorporating tissue failure models