Quality assurance: Repeatability measures across two Vicon® motion capture systems

Abstracts / Gait & Posture 30S (2009) S1–S153

S13

collection between two Vicon® motion capture systems. This study reaffirms the importance of rigorous QA procedures, particularly when upgrading laboratory equipment. Conclusions The findings verify the need for caution when comparing data collected with different motion capture systems. Ideally a laboratory should collect a new normal data cohort following the installation of a new system, and the effects on kinematic and kinetic reporting should be highlighted to all involved in the decision making process, particularly in cases of pre- and post-operative comparison.

Fig. 1. Excursion of knee femoral rotation (ext+, int−) for 11 movements.

ing and walking were similar (RMS = 5.0–5.2◦ ). The internal pivots produced larger axial rotation ROM than walking by 8–9◦ but were similarly repeatable. Such turns are therefore suitable for future studies addressing relevant alterations of knee axial rotation. Ascent, chair rise, squat, and the lunge showed larger flexion ROMs than walking, and among these tasks ascent was found the most repeatable. The tasks other than walking, ascent, or sidestep turns appear to require more carefully designed protocols before any reliable knee kinematics can be claimed. Discussion Researchers and clinicians who aim to compare knee axial rotation among populations should analyze pivot turns which impose considerable in/external rotations on the knee joint. Analyzing a set of tasks can provide more information about knee function than analyzing one task alone. Future studies should include validating these results with more accurate methods, such as fluoroscopy, and analyzing how these tasks may differ among older or pathologic populations. References [1] Kadaba MP, et al. J Orthop Res 1989. [2] Schache AG, et al. Gait & Posture 2006. [3] Zurcher AQ, et al. Gait & Posture 2008.

doi:10.1016/j.gaitpost.2009.08.021 O18 Quality assurance: Repeatability measures across two Vicon® motion capture systems Colin Davenport 1,∗ , Emma Pratt 1 , Wendy Dickens 2 , Jill van der Meulen 1 , Mike Bell 2 1 Sheffield Teaching Hospitals NHS Foundation Trust, Sheffield, United Kingdom 2 Sheffield Children’s NHS Foundation Trust, Sheffield, United Kingdom

Summary Quality assurance (QA) checks at the Sheffield Children’s Hospital Gait Laboratory (SCH) highlighted differences in kinematic data

Introduction SCH upgraded from a Vicon® Workstation® system (6× 370 cameras; 25 mm diameter markers) to a Nexus® system (8× MXF40 cameras; 9 mm diameter markers) in September 2008. As part of annual laboratory QA, data are collected to ascertain the repeatability of marker placement by the laboratory clinician. In response to previous research [1] the aim of this study was to compare the performance of the Nexus® system with that of Workstation® and identify any differences in the kinematic data by visual observation. Materials and methods Data collected with the Workstation® system from the same healthy adult subject was available over a period of 3 years. At each annual QA session, kinematic and kinetic data collection was performed at three separate occasions during the day. Each collection was preceded by a full system calibration and application of markers. At each session data had been collected with the subject walking at a self-selected preferred speed that was recorded with optical light gates. The performance of both the subject and the clinician for each year had been analysed with Coefficient of Multiple Correlation (CMC) analysis to quantify their repeatability both within and between sessions [2]. Following installation of the Nexus® system, data were collected following the same protocol to obtain clinician and subject repeatability measures. In order to compare the Workstation® system with Nexus® the data was processed using the same filtering techniques. Session data collected from both systems respectively were averaged for each year and displayed in Polygon® for visual comparison. Results Subject walking speeds were recorded as 1.62 ± 0.1, 1.64 ± 0.1, 1.60 ± 0.1 and 1.59 ± 0.1 m s−1 for the years 2005–2008, respectively. In summary the following observations were noted: increased ankle dorsi-plantar flexion range; increased knee flexion in loading response and mid-swing; increased hip abduction in terminal swing; increased range of foot rotation and progression; increased range and variability in hip and knee rotation (Fig. 1). Discussion The differences noted in the kinematics can be attributed to technical improvements in data reconstruction software and cameras, together with physical factors such as reduction in marker size; error associated with wand mounted markers; and the alteration in camera layout within the laboratory. Consultation with Vicon® technical support (Oxford, UK) confirmed our findings. The effects of walking speed, subject variability and data filtering can reasonably be annulled.

S14

Abstracts / Gait & Posture 30S (2009) S1–S153

Fig. 1. Yearly average kinematic curves for Workstation® system (dashed) and the Nexus® system (solid, ±1 S.D.).

References [1] Bucknall V, et al. Gait Posture 2008;28(2):S33–4. [2] Pratt E, et al.; in process.

doi:10.1016/j.gaitpost.2009.08.022 O19 Intra- and inter-laboratory repeatability of gait analysis data in normal adults Joseph Bevins 1 , Sarah Churchill 1 , Mark Corbett 1,∗ , Ralph Palmer 2 , David Pratt 2 , Anne Uutela 2 1 2

University of Worcester, Worcester, United Kingdom South Birmingham Primary Care Trust, Birmingham, United Kingdom

Summary Gait analysis data has been compared from two independent laboratories with similar measurement protocols (Vicon Plugin Gait) but significant differences in hardware and software. A hierarchical approach has been used to determine inter-trial, intersession and ultimately inter-laboratory repeatability. Conclusions Inter-trial and inter-session repeatability were both good and compare well to published data. Inter-laboratory repeatability varied from fair to poor and was characterised by fixed angular offsets, indicative of systematic differences in marker application. Introduction Reproducibility and error control are critical to delivering quality clinical gait analysis. Several studies have reported intrarepeatability (within) and inter-repeatability (between) across trials, sessions and observers under various conditions. However, data is scarce when comparing repeatability across different laboratories, where additional compounding factors are present. Patients/materials and methods Ten healthy adults were measured at both laboratories over 10 trials walking at their natural speed. 3D movement analysis (Vicon) and force vector (Kistler/AMTI) data were collected. Markering, data capture and processing were all conducted according to the respective laboratory standard protocols. Five measurements were also repeated on selected subjects at each laboratory over several weeks to establish inter-session repeatability. Absolute values and standard coefficients of variation of thirteen temporo-spatial parameters (e.g. cadence, walking speed) were compared directly. Lower body kinematic waveforms were compared using coefficients of multiple correlation (CMCs) [1]. Results All parameters and measures of repeatability were bilaterally symmetric and similar across subjects. Repeatability between the

laboratories in temporo-spatial parameters was extremely good with an overall difference of less than 0.5%. Inter-trial repeatability (intra-session, subject and laboratory) of kinematics was high (CMC > 0.9) at both laboratories across all sagittal and frontal plane parameters except for pelvic tilt (CMC ∼ 0.5). Intra-trial repeatability in the transverse plane was good (>0.8). Inter-session repeatability was again high (>0.9) in the sagittal and frontal planes, except for pelvic tilt which was poor. Transverse plane repeatability was fair (>0.7). Subtraction of means improved repeatability to high (>0.9) for all segments indicating the presence of fixed angular offsets. Inter-laboratory repeatability was good (>0.8) in the sagittal plane apart from pelvic tilt. Frontal plane was fair (>0.7) apart from knee ab/adduction (∼0.5). The transverse plane was fair for pelvic and foot rotation but poor (∼0.2) for hip and knee rotation. Subtraction of means again improved repeatability significantly to good (>0.8) for most segments though knee ab/adduction, hip rotation and knee rotation showed less improvement. Discussion Very good agreement in spatio-temporal parameters demonstrates consistency in the underlying measurement protocols at the laboratories. Kinematic repeatability follows a predictable trend down through the trial-session-laboratory hierarchy. Reduced repeatability in frontal and transverse plane knee and hip kinematics may be related to use of a knee alignment device (KAD) at only one of the laboratories.

Reference [1] Growney E, et al. Repeated measures of adult normal walking using a video tracking system. Gait and Posture 1997;6:147–62.

doi:10.1016/j.gaitpost.2009.08.023 O20 Repeatability of a new anatomically based protocol for gait analysis in adult subjects Maria Grazia Benedetti 1,∗ , Stefano Cavazza 2 , Giovanni Ferraresi 2 , Mario Manca 2 , Pia Marchi 2 , Emanuele Zanaga 2 , Alberto Leardini 1 1

Movement Analysis Laboratory, Istituto Ortopedico Rizzoli, Bologna, Italy 2 Dipartimento della Riabilitazione San Giorgio, Ferrara, Italy Variability in gait analysis patterns of motion can be introduced by the examiner during marker positioning. The purpose of the present study was to estimate the inter-trial variation along with intra-observer and inter-observer errors for the recently proposed Total3DGait [1] gait analysis protocol, by using a standard method [2].