Vicon Reliability

Daniel Chan Wai Hon Me Me,Thomas Chan Kam Kee, Banny Hui Chi Cheung,
May Leong Bik Sai, Fiona Liang Wai Yun

Alex Tse (PT, PMH), Chris Kirtley (supervisor), Drew Smith & Dora Poon Mei Ying (Hong Kong PolyU Staff)

Gait Laboratory of Princess Margaret Hospital (PMH), Hong Kong (by kind invitation of Dr. SH Yeung, Chief of Service)

Introduction

This final year undergraduate physiotherapy study was the first gait analysis project using the Vicon 370 motion analysis system. At the time of the study the laboratory at the Hong Kong Polytechnic University was not yet completed, and Dr. Yeung of Princess Margaret Hospital kindly offered the students the use of his laboratory.

It was decided that the first task that needed performing was a collection of data on normal individuals, to gather normative data for the Hong Kong population, to assess reliability of the Vicon measures, and (perhaps most importantly!) get expereince with VCM marker placement and operation of the Vicon system.

Methodology

Subjects

A total of 20 healthy volunteers (men and women) were studied. They were divided to two groups, with ages ranging from 15-25 and 45-55 years, respectively.

All subjects denied any history of significant musculoskeletal, neurological or cardiovascular disease, had normal strength and range of motion in the hip, knee and ankle joints (assessed by clinical examination), and provided informed consent. Subjects with obvious abnormality on observational gait analysis were excluded, as were those admitting to discomfort, disability or premature fatigue during walking and/or standing. 

Apparatus

The walking patterns of the subjects were captured and analyzed by a Vicon 370 (version 2.5) three-dimensional motion analysis system (Oxford Metrics Ltd., Oxford, UK). The six cameras had a frame rate of 60 fps and used infrared (IR) light-emitting diode strobes, gen-locked. Lightweight retro-reflective markers were attached to the skin over the following bony landmarks: sacrum (S2), anterior superior iliac spines (ASIS), lateral thigh, knee-joint axes, lateral shank, lateral malleoli and second foot ray. These points form the Vicon Clinical Manager (VCM, or Davis) marker set (Davis, et al, 1981). A Pentium II workstation running the Windows NT operating system was used for data transfer, analysis, and storage. Ground reaction forces during the stance phase were recorded by two strain-gauged force platforms, dimensions 508 x 460 mm (ORS-5, Advanced Medical Technologies, USA). 

Measurement Procedure

Subjects were required to don short trousers and walk barefooted. Anthropometric measurements and marker placement were found to be critically important in the measurement process, and will therefore be described in some detail. 

Anthropometry

a. Height was recorded by a stadiometer;
b. Body weight was measured with an electronic balance;
c. Leg Length was determined supine as the distance between the lowest point of the ASIS to the ipsilateral medial malleolus;
d. Knee Width was determined from measuring the distance between the lateral and medial femoral condyles in standing position with the Knee Alignment Device (KAD).

e. Ankle Width was determined from the transmalleolar distance in the standing position with the KAD. 

Marker Placement

a. Pelvic Markers (LASI and RASI) were attached directly over the lowest points of anterior superior iliac spines using double-sided adhesive tape.

b. Sacral Marker (SACR) was attached mid-way between the skin dimples formed by the posterior superior iliac spines.
c. Knee Joint Axis Determination.

The VCM model is extremely sensitive to errors in location and orientation of the knee-joint axis, and considerable care and practice was needed. Whilst several techniques are described by the manufacturer and their users, the following procedure was eventually found to give optimal results:
i. The KADs were attached to both knee with the subject sitting on a plinth high enough from the floor to allow the legs to swing freely;


ii. The horizontal wands of the KADs were aligned parallel to the floor;
iii. The subject was then asked to actively flex and extend each knee in turn, whilst an observer watched the wand indicating the flexion/extension axis of the KAD.
iv. The location of the KAD was adjusted until the point was found at which the flexion/extension wand showed minimum movement;
v. The position of the KAD was then marked with a ball pen;

vi. The procedure was repeated for the contralateral knee.
vii. The KADs were removed, and the subject asked to stand in the centre of the walkway, whereupon they were reattached;
d. Thigh wand markers (LTHI and RTHI) were aligned with the hip joint centre (greater trochanter) and the flexion/extension axis wand of the KAD with the aid of a full-length mirror, placed at a distance of around 2m lateral to the subject.

This arrangement allowed the observer a parallax-free view whilst enabling simultaneous adjustment of the thigh wand; 
e. Ankle-joint markers (LANK and RANK) were attached directly over the lateral malleoli.
f. Shank wand markers (LTIB and RTIB) were aligned with the ankle-joint markers and the flexion/extension wands, also aided by the mirror;
g. Forefoot markers (LTOE and RTOE) were attached to the second metatarsal heads, after asking the subject to flex the toes in order to facilitate identification;
h. Heel markers (LHEE and RHEE) were attached over the os calcis at the same height as the forefoot marker (as determined by a Vernier caliper).
 

Static Trial

The subject was requested to stand quietly on the force platform (centre of walkway) whilst several seconds of video data were recorded. 

Dynamic Trial

The KADs were removed, and replaced, half a marker-width posteriorly, by a standard marker, and the os calcis marker removed. The half-marker readjustment was found to compensate for soft-tissue movement on standing. Adequate rehearsal was permitted in walking on the walkway to ensure a clean foot strike on the force platform, and encourage a natural gait pattern.  Subjects were asked to walk at a self-selected natural velocity barefoot, with gazing forwards in the plane of progression.
 
 

Knee Flexion/Extension Offset

For a normal subject, the knee should be almost full extension (0°) immediately prior to foot contact, and should extend to around 8° during late stance (Winter, 1991). In early trials, it was found that if the KADs were placed directly on the points determined by the method described above, there would be a flexion offset of greater than 5 degrees in the F/E curve. This phenomenon is caused by movement of knee musculature on standing. It could be argued that the markers should therefore be attached with the subject erect, but this is rather uncomfortable for both the subject and the operator, especially if the subject is a patient with difficulty in standing for prolonged periods. An alternative approach was therefore used, in which the knee axis marker was attached somewhat posterior to the point of attachment of the KAD.  The optimal distance for this adjustment was determined empirically, by noting the flexion offset when the marker was placed, in turn, a whole-marker diameter, and half a marker diameter posterior to the KAD attachment point. The former caused an extension offset to appear in the F/E curve, whilst a half-marker diameter was found to result in minimal offset. The half-marker distance was therefore used in all subsequent studies.

Knee Varus/Valgus Artefact

Anatomical constraints in the normal knee prevent significant varus/valgus motion during normal gait. However, in initial trials placing markers according to the manner described in the VCM manual, the Knee Varus/Valgus curves also showed large artefacts. These artefacts occur whenever the axis of the KAD is rotated away from the true axis, causing some of the sagittal-plane motion of the knee to be falsely recorded as frontal-plane motion (see discussion on CGA list). This crosstalk effect is particularly pronounced when large motions occur in the sagittal plane (i.e. in swing phase). Following correspondence with several experienced Vicon users, the method described above was arrived at, which resulted in almost no varus/valgus artefact during stance, and only slight deviation during swing.

Following marker placement, the subject walked repeatedly down the walkway in which two AMTI force platforms were embedded. Three-dimensional trajectory reconstruction (AMAS) was performed by Vicon 370 (version 2.5) software to derive marker kinematics. Following labeling by the operator, these co-ordinates were integrated with ground reaction forces (recorded simultaneously via a 16-bit analog-to-digital converter) into a link-segment model by the VCM to derive joint moments and powers by standard inverse dynamics (Winter, 1991). 

Results

Averaged Curves (all subjects) All.gcd

The dotted lines indicate +/- one standard deviation.


Reliability

The graph summarises the Coefficients of Variation (CV) for all VCM variables in intrasession, intersession and inter-subject measurements in both the young and old age groups. These are comparable to thise reported previuosly (Kadaba et al, 1989; Winter, 1991; Eng & Winter, 1995). The Excel file is here.

Intra-class Correlation Coefficients (ICC1,k)

ICCs were calculated on the maximum and minimum values of each variable. An ICC > 0.8 is generally regarded as evidence of reliability. Note, however, that ICCs can be misleading if the between subject variation (MSB) is very small or very large. This is probably the reason for the unexpectedly low ICCs for OppositeFootContact and StepTime. In general, the rest of the ICCs are consistent with the CVs above.

Summary


Good Reliability Moderate Reliability Poor Reliability
All temporal-distance parameters Pelvic tilt & obliquity Pelvic rotation
All sagittal joint angles All sagittal joint powers All transverse joint powers
All sagittal joint moments Ankle & knee frontal powers
Foot progression angle Foot frontal moments
Hip frontal moment Hip frontal joint power
Foot, knee & hip transverse moments

Comparison with Previously Published Data

Sagittal-plane Coefficients of Variation (%)

Joint
 
Hip
 
 
Knee
 
 
Ankle
 
 Study
Angle
Moment
Power
Angle
Moment
Power
Angle
Moment
Power
Winter 2D
52
140 (121)
221 (170)
23
135 (108)
157 (127)
72
42 (32)
100 (71)
Winter 3D
 
37
70
 
70
79
 
28
49
Vicon*
6
24
46
11
35
38
22
14
30
Dr. Cho~
23
83
109
26
82
80
49
128
150
Dr.Selber'
37
    
40
      
89
 Young#
22
40
56
20
49
66
35
18
51
Old#
20
46
62
27
63
83
39
19
50
All#
53
65
106
31
76
126
56
28
83

Winter 2D moment and power values appear to have been overestimated - corrected values in parentheses.
*Data supplied by Oxford Metrics with Vicon 370 (N=3).
~Data from Dr. Cho's 6 yo female Korean children (N=4).
'Data from Dr. Selber's 8-16 yo Brazilian children (N=44).
#Inter-subject data from this study (N=19).


Discussion

Main differences between PolyU & Vicon 'normal' curves

The subjects in this study walked with a gait velocity somewhat slower than expected, with a shortened stride. This is likely due to the small size of the laboratory and close proximity of the force platforms. Of course, the joint kinematics and kinetics are therefore likely to be abnormal, and as expected they are indicative of a slower gait velocity (Kirtley et al, 1985): The knee flexion offset and valgus/varus artefact are indicative of malalignment of the KAD and care clearly needs to be taken to improve the marker placement technique.

Reliability

Notwithstanding the subnormal gait velocity in this study, the reliability coefficients obtained may be reasonably expected to reflect the variation observed in a more natural gait. Most clinically important measures showed CVs in the range 0 - 100%. However, pelvic rotation, foot, ankle and knee frontal-plane joint moments, and all tranverse-plane joint powers were very unreliable (CVs > 150 %).

In general, young subjects were more consistent than older subjects, with the exception of foot progression angle.

Not surprisingly, measurements repeated in the same session (i.e. using the same marker placement) were more repeatable than those conducted in separate sessions (i.e. with markers removed then reattached), and intra-subject reliability was better than inter-subject reliability.

References

Davis, RB III, Ounpuu, S, Tyburski, D, and Gage, JR (1991). A gait data collection and reduction technique. Human Movement Sciences 10, 575-587.

Eng J, Winter D (1995). Kinetic analysis of the lower limb during walking: What information can be gained from a three-dimensional model? Journal of Biomechanics, 28:6, 753-758

Kirtley C, Whittle MW & Jefferson RJ (1985) Influence of Walking Speed on Gait Parameters Journal of Biomedical Engineering 7(4): 282-8.

Kirtley C (2002) Sensitivity of the Modified Helen Hayes Model to Marker Placement Errors. Seventh International Symposium on the 3-D Analysis of Human
Movement, Newcastle, UK, July 10-12.

Kadaba M P, et al. (1989). Repeatability of Kinematic, Kinetic and Electromyographic Data in Normal Adult Gait. Journal of Orthopaedic Research, 7:6, 849-860.

Winter DA (1991) The biomechanics and motor control of human gait: normal, elderly and pathological. University of Waterloo press, Ontario. 

Acknowledgements

Dr. SH Yeung was very kind in allowing us to use his gait laboratory at Princess Margaret Hospital, where Alex Tse helped with supervision. We'd also like to thank Richard Baker (Belfast, Northern Ireland), Michael Orendurff (Portland Shriner's) and Jeremy Linskell (Dundee, Scotland) for their painstaking email help and advice on marker placement technique. Finally, thanks to the staff of PMH physiotherapy department and our student colleagues who were the subjects for this study!

Back to CGA

This page is maintained by Dr. Chris Kirtley

Last updated on 24/10/98