Semiautomated finite element mesh generation methods for a long bone

The objective of this work was to develop and test a semi-automated finite element mesh generation method using computed tomography (CT) image data of a canine radius. The present study employs a direct conversion from CT Hounsfield units to elastic moduli. Our method attempts to minimize user interaction and eliminate the need for mesh smoothing to produce a model suitable for finite element analysis. Validation of the computational model was conducted by loading the CT-imaged canine radius in four-point bending and using strain gages to record resultant strains that were then compared to strains calculated with the computational model. Geometry-based and uniform modulus voxel-based models were also constructed from the same imaging data set and compared. The nonuniform voxel-based model most accurately predicted the axial strain response of the sample bone (R(2)=0.9764).