Elsevier

Gait & Posture

Volume 32, Issue 2, June 2010, Pages 231-236
Gait & Posture

Repeatability and reproducibility of OSSCA, a functional approach for assessing the kinematics of the lower limb

https://doi.org/10.1016/j.gaitpost.2010.05.005Get rights and content

Abstract

Marker-based gait analysis of the lower limb that uses assumptions of generic anatomical morphology can be susceptible to errors, particularly in subjects with high levels of soft tissue coverage. We hypothesize that a functional approach for assessing skeletal kinematics, based on the application of techniques to reduce soft tissue artefact and functionally identify joint centres and axes, can more reliably (repeatably and reproducibly) assess the skeletal kinematics than a standard generic regression approach.

Six healthy adults each performed 100 repetitions of a standardized motion, measured on four different days and by five different observers. Using OSSCA, a combination of functional approaches to reduce soft tissue artefact and identify joint centres and axes, the lengths of the femora and tibiae were determined to assess the inter-day and inter-observer reliability, and compared against a standard generic regression approach. The results indicate that the OSSCA was repeatable and reproducible (ICC lowest bound 0.87), but also provided an improvement over the regression approach (ICC lowest bound 0.69). Furthermore, the analysis of variance revealed a statistically significant variance for the factor “observers” (p < 0.01; low-reproducibility) when using the regression approach for determining the femoral lengths.

Here, this non-invasive, rapid and robust approach has been demonstrated to allow the repeatable and reproducible identification of skeletal landmarks, which is insensitive to marker placement and measurement session. The reliability of the OSSCA thus allows its application in clinical studies for reducing the uncertainty of approach-induced systematic errors.

Introduction

The reliable assessment of skeletal motion is important for a wide range of clinical investigations in studies where the correct interpretation of results relies on the ability to detect small differences between cohorts, but also in longitudinal studies when multiple measurements are required, possibly performed by different observers. By providing access to subject specific motion patterns, movement analysis allows quantitative assessment of a range of functional and kinematic parameters and can therefore provide a direct comparison for the efficacy of interventional approaches [1], [2], or complement imaging and physical examination findings to support clinical decisions. Furthermore, the additional registration of synchronous external kinetic data can provide essential information for the determination of internal loading conditions using musculoskeletal analyses [3], [4], [5].

The non-invasive capture of human motion is generally performed by attaching reflective markers to the subject's skin. To assess the skeletal motion from these marker positions, a number of techniques have been developed to identify specific bone structures or landmarks. Geometric regression relationships, that are still used in the majority of motion capture systems, describe the location of the hip joint centre relative to specific skin markers placed over the left or right superior inferior iliac spine and the posterior superior iliac spine [6], [7]. Such regression methods, however, are subject to errors [8] due to their generic nature, since bony deformities or gender-based skeletal differences [9], [10] are not considered, but also the variation of marker placement. Furthermore, these approaches are subject to soft tissue artefact (STA) [11], [12], [13], where the attached markers move relative to the underlying skeletal structures, resulting in errors in the determination of the skeletal motion. The efficacy of these techniques is therefore limited, particularly in subjects with high soft tissue coverage.

Recently, methods to minimize the errors associated with skin marker artefact have been proposed using the point cluster technique [14] or the optimal common shape technique (OCST) [15]. In the latter, the motion of markers attached to the skin relative to one another is minimized using a so-called Procrustes analysis [16] to generate a rigid configuration of the segment marker set, formed from all time points of the trial. By replacing the original segment marker set with this OCST configuration, the relative motion of markers, and therefore the effects of skin elasticity can be minimized.

Furthermore, rapid [17], accurate, and robust methods to functionally determine spherical joint centres, using the symmetrical centre of rotation estimation (SCoRE) [18] and joint axes, using the symmetrical axis of rotation approach (SARA) [19], have been developed and verified in silico. The combination of these techniques allows the location of joints to be estimated from kinematic data alone, without the assumptions associated with generic anatomical relationships. However, it remains unknown whether the combination of these functional methods to identify the joint locations can provide an improvement in reliability, as well as a reduction of user-induced variability over regression methods. We therefore hypothesized that a combination of the OCST, SARA and SCoRE is more reliable (repeatable and reproducible) than a common regression approach [20] for the identification of skeletal structures of the lower limb. The aim of this study was thus to develop the OCST, SARA and SCoRE combined approach (OSSCA) to motion analysis and assess its repeatability and reproducibility.

Section snippets

Study protocol and participants

Optical markers were attached to the skin of the right lower limb of six healthy male subjects (age: 30.0 ± 4.5; BMI: 22.8 ± 0.4) on multiple days, followed by motion analysis (12 FX20 cameras, Vicon, Oxford, UK) of standardized movements, as well as a single recording of a static standing posture. All subjects provided written informed consent prior to their participation in the study, and the study was approved by the local ethics committee.

Each entire marker set was attached by five independent

Results

The SCoRE residual measured over all 600 measurements was 7.1 ± 1.2 mm (Fig. 2), equivalent to an average error in the hip joint estimation of approximately 3.5 mm.

The femoral and tibial segment lengths were determined using the OSCCA and the regression approach. These segment lengths were compared to each other during the repeatability test for the variation in measurements between different days (Fig. 3a) and measurements between different observers (Fig. 3b). The ICC for all tests showed that

Discussion

The reliable assessment of skeletal motion is important in studies where the ability to detect small differences between groups is essential for the correct interpretation of results. Furthermore longitudinal studies, especially when multiple measurements performed by different observers are required, necessitate methods that can ensure high repeatability and reproducibility. Indeed, the conclusions of a recent study have indicated that poor reliability in clinical gait analysis is one of the

Conclusions

In this study, the OSSCA (OCST, SARA and SCoRE combined approach), a functional approach for assessing skeletal kinematics from skin marker based motion capture data, has been introduced. Key components in this approach are the minimization of soft-tissue artefact together with accurate, fast and robust methods for the functional identification of joint centres and axes. Our hypothesis that the OSSCA is more repeatable and reproducible than a commonly used regression approach has been confirmed

Acknowledgements

The authors are grateful to the German Research Foundation (SFB 760) and the EU (FP6 DeSSOS IST-027252) for funding this project. They would also like to thank Verena Schwachmeyer, Philippe Moewis, Florian Kugler, and Sandra Zdzieblik for their valuable contribution to measurement of the subjects.Conflict of interest: The authors declare that they have no conflict of interest.

References (28)

Cited by (66)

View all citing articles on Scopus
View full text