A six degrees-of-freedom marker set for gait analysis: Repeatability and comparison with a modified Helen Hayes set

Abstract

Kinematic gait analysis is limited by simplified marker sets and related models. The majority of sets in clinical use were developed with low resolution imaging systems so required various assumptions about body behaviour. Further major limitations include soft tissue artefact and ambiguity in landmark identification. An alternative is the use of sets based on six degrees-of-freedom (DOF) principles, primarily using marker clusters for tracking.

This study evaluates performance of a 6DOF set, based largely on CAST/ISB recommendations, through comparison with a conventional set and assessment of repeatability. Ten healthy subjects were assessed in treadmill walking, with both sets applied simultaneously on two occasions. Data were analysed using repeatability coefficients, correlation of key features, and comparison of joint angle curves and difference curves with confidence bands.

Apart from pelvic tilt all segment and joint angles from both sets showed high within and between session repeatability (CMC > 0.80). Hip rotations showed clear differences between the two sets with indications in support of the 6DOF set. Knee coronal angles showed evidence of cross-talk in the conventional set, highlighting difficulties with anatomical identification despite control measures such as a foot alignment template. Knee transverse angles showed inconsistent patterns for both sets. At the ankle the conventional set only allowed true measurement in two planes so with high repeatability the 6DOF set is preferable.

The 6DOF set showed comparable performance to the conventional set and overcomes a number of theoretical limitations, however further development is needed prior to clinical implementation.

Introduction

Kinematic gait analysis is often based on markers located according to a standard arrangement (marker set). A variety of sets have been proposed but the majority in clinical use are based on some variation of the Helen Hayes (HH) set [1]. This was developed for low resolution imaging systems so had to have few markers, as far apart as possible [2]. As a result, joint motion is constrained with only three rotational degrees-of-freedom (DOF); thigh segment definition relies on hip joint centre estimated from pelvis markers, shank definition relies on a knee marker shared with the thigh, and foot definition relies on an ankle joint centre estimated from shank markers. These constraints introduce errors to joint angle calculations [3], cause errors to propagate to more distal joints [4] and require the entire model to be used even if only a subset is of interest.

Substantial work has been carried out to develop sets that track each segment independently, allowing 6DOF (rotational and translational) at each joint. In clinical gait analysis it is generally only the three rotational DOF that are assessed, however allowing linear DOF avoids introducing unnecessary errors [3] and minimises error propagation. Distinct from this, major limitations in kinematic gait analysis result from soft tissue artefact (STA) and landmark identification [5], [6], [7], [8], [9]. For STA; movement over bone violates a rigid body assumption and for some cases the errors can be as large as the range of movement (ROM) of the joint [6]. More faithful tracking may be achieved with clusters of markers on rigid plates, placed away from bony landmarks (technical set) [9], [10], [11], as long as a relationship is established with anatomical landmarks to allow segment reconstruction. For landmark identification; it is difficult to place markers accurately because anatomical landmarks tend to be large curved areas [5], [12] and this variability contributes to errors in joint angles; for example cross-talk [13]. Techniques are being considered to improve landmark identification [14] but have not yet become widely applicable in clinical gait laboratories. Each centre has familiarity with their set and a wealth of past data so any new set must offer enough advantages to overcome this historical background.

To evaluate current status for practical application in a clinical gait laboratory a 6DOF set was produced, using proposals in literature, particularly the ‘calibrated anatomical systems technique’ (CAST) [6], [11] and International Society of Biomechanics (ISB) recommendations [15]. This study compares the 6DOF set with a clinically used variation of HH. A recent study compared five marker sets/protocols including the CAST and Plug-in Gait (similar to HH) with three subjects [16]. Results presented here add further detail for the two sets over a sample of 10 subjects with repeatability assessed between sessions. The authors are not aware of other studies reporting inter-session coefficient of multiple correlation (CMC) repeatability values for a 6DOF set, nor directly comparing HH and 6DOF with relatively large sample size.