Validation of three optical marker models in recordings of dynamic 3D angular knee movements using radiostereometric analysis as a reference

Movement analysis with marker-based optical tracking systems (OTS) are commonly used to quantify human movements (1). In OTS, video cameras capture the position of reflective skin-markers placed on the body. The markers define body segments and are the prerequisite for further calculations of kinematics and kinetics (2). A known problem with markers attached to the skin is that they may displace relative to the underlying bone, frequently expressed as soft-tissue artefacts (1, 3). Radiostereometric analysis (RSA) is invasive, but a superior method for measuring skeletal movements with high precision and accuracy (4-6). RSA takes images in two directions simultaneously, enabling a three-dimensional analysis of the movement of the skeletal bones, both statically and dynamically. Only one of three commonly used marker models has previously been validated with RSA for the knee joint. This study analyses angular movements in the knee joint during loaded extension-flexion using the CAST (7), the IOR (8) and the SUhsm model (9) with simultaneously measured data from the RSA-system. To what degree do recordings of angular movements in the knee joint measured with marker-based motion capture system, agree with RSA measurements during loaded knee extension-flexion movement? Twelve participants who had undergone knee replacement surgery were included. At the examination, the participant performed a knee flexion, from an extended to a flexed knee position, while standing on a platform. Angular movements in the knee joint were observed in three anatomic planes (flexion-extension, abduction-adduction, internal-external rotation) and the results were compared within the three marker models and the RSA system. For all three directions of movements, the three marker models were delayed at the start in comparison with RSA. All three marker models also showed an underestimation of movement compared to RSA. For the flexion-extension movement, only small differences were seen between the three marker models. For the abduction-adduction movement, the SUhsm model followed the RSA system’s increasing values in adduction, while the CAST and IOR model instead presented increasing values of abduction. For the external-internal rotation all three marker models presented external rotation instead of the internal rotation of the RSA system. Knee motions captured with an OTS are most reliable in extension-flexion even though it underestimates the values. Soft tissue artifacts affect the outcome in abduction-adduction and external-internal rotation. Data from these directions should be interpreted with caution. The movement analysed in this study is a slow and controlled movement of the knee joint. It is possible that a more high-intensity movement, e.g. a sport-specific movement, has additional difficulties in adequately measuring the movement of the skeletal bones. Selection of marker model and knowledge of different marker models are important when using marker-based OTS.