Motion analysis: accuracy of joint centre estimation

Central to all areas of (human) motion analysis is the objective to reconstruct the bone motion in the laboratory reference system, i.e., the complexity of the captured data is reduced and mapped to a simplified model allowing for the computation of joint kinetics and kinematics. In clinical gait analysis, the calculation of hip joint moments may form the basis of the therapy or for the determination of the lower limb alignment axes during surgical intervention. Therefore the exact location of the hip joint centre is of primary importance. Data acquisition suffers from several errors arising from palpation, movement between markers and underlying bone, misplaced alignment devices, marker location, anthropometric regression equations, and optimization algorithms. Clinical movement analysis requires the most accurate knowledge of joint parameters. Mathematically the problem that is dealt with can be formulated as improving the accuracy and precision in the transformation between technical (defined by surface mounted markers) and anatomical (defined by the underlying anatomy) coordinate systems. In order to assign a local coordinate system to a segment in three dimensions a cluster consisting of at least three markers is required. The aim of the project is to investigate existing and to develop more accurate methods for estimating joint centres from marker points. A substantial part of the studies is devoted to find recommendations or a standardized protocol for the placement of the marker sets in relation to specific movement types, e.g., in gait analysis, in cycling or in running. In order to account for errors in the centre estimation induced by different sources, the problem will be assessed in two steps. Errors which are not caused by biological properties (such as marker shifting with respect to the bone of the very segment due to soft-tissue movement) are the topic of the first examination. A representative joint to be considered will be the hip joint. A mechanical analogue will be constructed. Investigations regarding the precision and the accuracy of the (optical) measurement system will be performed. These measurements will be based on different protocols. The second step will focus on errors originating from biological sources. Instead of using the mechanical analogue, the markers will now be attached directly to the skin of human subjects. Again the measurements will be based on different protocols. Special emphasis will be put on marker placement. Redundant markers will also be applied.

The present study is devoted to the analysis of the complex movements of the human knee joint playing a key role within the human musculo-sceletal system. While observing ones knee joint, the almost horizontal flexion axis is quite obvious. Looking closer, it can be noticed that a flexed knee allows for rotation of the shank about its long axis such that the foot tip can turn inwards and outwards. In the stretched position the side bands of the knee prohibit this rotation. Here inward and outward rotation of the foot tip is enabled by rotation in the hip joint. Arthritic modifications, injuries or wear compromise the functionality of this joint. Today the scientific analysis of the musculo-sceletal system cannot longer be provided from an exclusively medical point of view but necessitates a multidisciplinary research approach. In this context the present study is integrated.A crucial point in human joint analysis is the determination of a reference system where joint displacements are related to. Rotation angles and translations depend upon the selected joint axes. For this reason, in the past much effort was put in prescriptions how those axes should be determined. At first, landmarks in the human anatomy were used to define axes, however, human anatomy exhibits large inter-individual variations. Therefore, mathematical methods have been designed to find functional knee joint axes instead. This axis is something like the optimum rotation axes. However, the knee joint is not only no uniaxial joint, moreover, the rotation axes may differ for different movements (level walking, squats, cycling, running, stair climbing). In the present project numerical routines have been developed allowing for the calculation of joint axes. At first the program was tested on simulated data in order to test the sensitivity and numerical stability of the code. Then data of squats, rowing and cycling were analysed. Here retro-reflective markers have been fixed to the skin of the subjects (in-vivo) and the movements have been recorded by infra-red cameras (Vienna University). Furthermore, in-vitro data of a squat assessed at the University of Tübingen have been analysed. Especially the latter results are promising. The due to the anatomical reference method strongly divergent behaviour of the individual in-vitro knees was homogenized by application of the in this project developed mathematical axis determination method. This is a success for in-vitro endo-prosthesis testing since here accuracy is the most important prerequisite when attempting to observe small nuances in design. Only then individual prosthesis design can be in the scope of research.