Abstract

Subject-specific finite element models of knee joint contact mechanics are used in assessment of interventions and disease states. Cartilage thickness distribution is one factor influencing the distribution of pressure. Precision of cartilage geometry capture varies between imaging protocols. This work evaluated the cartilage thickness distribution precision needed for contact mechanics prediction in models of the tibiofemoral joint by comparing model outputs to experimental measurements for three cadaveric specimens. Models with location-specific cartilage thickness were compared to those with a uniform thickness, for a fixed relative orientation of the femur and tibia and with tibial freedom of movement. Under constrained conditions, the advantage of including location-specific cartilage thickness was clear. Models with location-specific thickness predicted the proportion of force through each condyle with an average error of 5% (compared to 27% with uniform thickness) and predicted the experimental contact area with an error of 21 mm 2 (compared to 98 mm 2 with uniform thickness). With tibial freedom, the advantage of location-specific cartilage thickness not clear. The attempt to allow three degrees of relative freedom at the tibiofemoral joint resulted in a high degree of experimental and computational uncertainty. It is therefore recommended that researchers avoid this level of freedom. This work provides some evidence that highly constrained conditions make tibiofemoral contact mechanics predictions more sensitive to cartilage thickness and should perhaps be avoided in studies where the means to generate subject-specific cartilage thickness are not available.