Knee joint kinematics: comparison of two optimization models with respect to data noise

The interpretation of human knee joint kinematics in terms of displacements is an outcome of the underlying model of the joint and the measurement technique. Measurement errors and noise challenge the development of optimization procedures which, based on a reduction in degrees of freedom, aim for the reproducibility of joint displacements by computational techniques. So far, optimization algorithms have been applied which are based on a kinematic model of the healthy human tibio-femoral joint (TFJ) as a compound hinge with two xed orthogonal axes. On the other hand, empirical studies nd non-orthogonal rotational axes. Therefore, it was the aim of the present study to investigate the implications of a rened kinematic model on the accuracy of computed joint rotation angles. For the purpose of quantitative comparison, kinematic data of a TFJ with two axes intersecting at an arbitrary angle were simulated. The joint rotations were optimized for the assumption of (a) two orthogonal and intersecting axes (model A), and (b) two axes intersecting at an arbitrary angle (model B). Model B recovers the original input data closer in case of low noise level as encountered in invasive measurement techniques. Skin mounted markers tracking involves non-normally distributed noise which is typically larger by on order of magnitude. In this case, model A exhibits a more favorable performance. These observations motivate the search for alternative kinematic descriptions of the TFJ.