Isotropic Heterogeneous FEM Model versus Isotropic and Anisotropic Homogeneous Models

Objectives: Previous 3D finite element analyses (FEAs) have shown that isotropic heterogeneous (IsoHet) models with implant allowed more detailed trabecular stress analyses than isotropic homogeneous (IsoHom) models. However, larger heterogeneous models are difficult to construct, while IsoHom models approximate bone only roughly. Thus, this study investigated whether an anisotropic homogeneous (AnisoHom) model can be a better approximation for an IsoHet model.

Methods: Three-dimensional images were reconstructed from micro-CT images of a solid-screw implant and abutment placed into the molar region of a partially edentulous human mandible (dried skull) by bone structure analysis software (TRI/3D-Bon, RATOC System Engineering). The IsoHet model was created by binarizing implant with abutment, cortical bone, and trabecular structure. In the AnisoHom and IsoHom models, instead of the trabecular structure, anisotropic (transversely isotropic) and isotropic homogenous materials were modeled, respectively. A FEA-software (TRI/3D-FEM, RATOC) was used to axially apply 200 N on abutment top, and calculate principal strains in each model.

Results: In IsoHom and AnisoHom, high maximum principal (tensile) strains tended to concentrate in the cancellous bone around the implant threads, mainly lingually in AnisoHom, but lingually, buccally and apically in IsoHom. In IsoHet these strains were found mainly lingually from the implant, but over very limited areas. In all the models, high minimum principal (compressive) strains were mainly concentrated in the cortical bone distally from implant neck and around the implant apex. However, in the trabeculae of IsoHet, their distribution was more limited.

Conclusions: For cortical bone, similar strain distributions may be predicted by the 3 models, but for cancellous bone, the homogeneous models are a rough approximation. However, if maximum principal strains are of interest, an anisotropic homogeneous model may better approach the predictions of a heterogeneous model.

Supported by a Grant-in-Aid (No. 18592117), from Japan Society for the Promotion of Science.