Sensitivity of the stress field of the proximal femur predicted by CT‐based FE analysis to modeling uncertainties. Issue 5 (13th July 2021)

• Record Type: Journal Article
• Authors: Youssefian, Sina; Bressner, Jarred A.; Osanov, Mikhail; Guest, James K.; Zbijewski, Wojciech B.; Levin, Adam S.
• Is Part Of: Journal of orthopaedic research. Volume 40:Issue 5(2022)
• Page Start: 1163
• Page End: 1173
• Publication Date: 2021-07-13
• Subjects: density–elasticity relationship -- finite element model -- proximal femur -- stress field; Orthopedics -- Periodicals; Musculoskeletal system -- Periodicals; 616.7
• DOI: 10.1002/jor.25138
• Languages: English
• ISSNs: 0736-0266

Abstract

Abstract: Proximal femur anatomy and bone mineral density vary widely among individuals, precluding the use of one predefined finite element (FE) model to determine the stress field for all proximal femurs. This variability poses a challenge in current prosthetic hip design approach. Given the numerous options for generating computed tomography (CT)‐based FE models, selecting the best methods for defining the mechanical behavior of the proximal femur is difficult. In this study, a combination of computational and experimental approaches was used to explore the susceptibility of the predicted stress field of the proximal femur to different combinations of density–elasticity relationships, element type, element size, and calibration error. Our results suggest that FE models with first‐order voxelized elements generated by the Keyak and Falkinstein density–elasticity relationship or quadratic tetrahedral elements generated by the Morgan density–elasticity relationship lead to accurate estimations of the mechanical behavior of human femurs. Other combinations of element size, element type, and mathematical relationships produce less accurate results, especially in the cortical bone of the femoral neck and calcar region. The voxelized model was more susceptible to variation of element size and density–elasticity relationships than FE models with quadratic tetrahedral elements. Regardless of element type, the stress fields predicted by the Keyak and Falkinstein and the Morgan relationships were the most robust to calibration error when deriving material density from CT‐generated Hounsfield data. These results provide insight into the implementation of a robust platform for designing patient‐specific implants capable of maintaining or modifying the stress in bones.