Development of computational model for total knee arthroplasty design

Abstract

Total Knee Arthroplasty is a common option to treat knee pain and restore knee function for osteoarthritis patients, though failure is inevitable. Implant manufacturers often focus on mechanical performance of implant, such as contact area, stress and fatigue, and consider individual response of knee geometry less, whilst evaluation of design on subject performance could be useful, but require years of continuous assessment and sometimes unfeasible. Computational methods provide an efficient and objective approach to investigate the parametric effects of loading, surgical deviation and design variations in a well-controlled environment, with both consideration of implant and knee components. Understanding these effects and their mechanisms can aid engineers and physicians to design better implants and establish optimal surgical protocols. The 3D geometry of knee was constructed and segmented from Magnetic Resonance (MR) images, whereas the geometry of implant was acquired from scan of digitizer. The model were assembled and simulated with surgical procedures and processed with finite element (FE) analysis. Parametric study was then carried out to evaluate different effects, such as load and bone strength on different flexion angle, and gait conditions. Two kinds of parametric studies have been done in this study. The first study aimed to evaluate the contact stress of the tibial insert at different load, flexion angle and gait conditions. The results could be verified and compared with existing experiments and finite element results. The second study focused on the investigation of effect of variation of the bone stiffness on the stress of trabecular femur, and aimed to study the mechanism of periprosthetic fracture upon reduction of bone stiffness. In the study of contact stress of tibial insert, the FE prediction showed that the contact stress ranged from about 23MPa to about 25MPa. This result was comparable to that of existing literatures, ranged from about 17MPa to about 27MPa. Although the predicted stress exceeded the allowable stress 10MPa of UHMPWE for the tibial insert, it did not reach the yield strength of UHMWPE. In the gait study of tibial insert, the contact stress formed two peaks at about 16% and 45% gait cycle, with contact stress about 53MPa and about 29MPa, respectively. The latter stress was comparable to existing literatures that peak stress occurred at 45-55% of gait. The former stress peak may be due to over-constrain on specific implant kinematics by input boundary conditions of gait. In the study of bone strength on trabecular femur, the predicted stress was concentrated on the apex of screw-hole, posterior supra-condylar region and anterior flange, consistent with clinical observation of peri-prosthetic fracture. The maximum predicted stress increased by about 3.87% and 2.94% at 5-degree and 60-degree flexion, respectively when the bone strength was reduced by half. Although the increase of the predicted stress was relatively small, the reduction in bone strength resulted in reduction in bone sustainability. By comparing the predicted stress and yielding stress upon bone strength reduction, it was shown that the risk of stress yielding or micro-fracture at the trabecular layer was likely even at lower flexion angle. Since the maximum predicted stress shifted to the anterior flange region at higher flexion angle, the design of implant anterior flange contributed to improve the stress distribution anteriorly especially at higher flexion angles. This computational model could provide an efficient platform to investigate the change of stress distribution and magnitude on bone-implant upon sensitivity of different parameters. This model could aid physicians and designers to design better implants and surgical protocols.