View Full Version : Re: Marker sets and models for the kinematic analysis of lowerlimb amputees

04-08-2010, 04:43 PM
JK wrote
>"I was concerned about the appropriateness of the use of several
>of the common models and methods when analysing amputee function,..."
>And "I imagine it may be possible to achieve more reliable/valid
>results (and a system that may be applied more universally) with
>joint centers determined functionally,..."
>And "I suspect that my idea of a universally applied base marker
>set and model may be unrealistic, even given an allowance for
>additional markers/segments for specific analyses,..."

There are alternatives to the traditional minimalist markers sets you mention that involve expanded numbers of markers, are generalized in there design and so can be applied to varying applications and body segments, and are considerably more accurate. I have been using them for about 15 years (when I first started my postgraduate studies in 3D motion analysis).

A rather long explanation follows:

Minimalist markers sets originated in the 70s and 80s (Hayes, Cleveland, Davis, Vaughan) and were a necessity of the time due to available technology; primarily the need to track individual marker paths in 2D camera views, but also due to limited camera numbers (2-4) with relatively poor resolution.

A good summary of these, including intra-examiner reliability is McGinley et al. [1]. When looking at just the intra-examiner reliability, the average 2*95% CI for flexion/extension, abduction/adduction, and internal/external rotation are approximately: [Only the average errors across the gait cycle are reported and not the peak values, which may be 1.5 to 2.0 times higher, see data below].

Hip: 5, 5, 10 degs.
Knee: 7, 5, 12 degs.
Ankle: 6, 10, 7 degs.

If a minimum clinical difference (MCD) is chosen for each rotation:

Hip: 8, 5, 5 degs.
Knee: 5, 5, 5 degs.
Ankle: 7, 5, 5 degs.

For abd-adduction, internal-external and knee flexion during stance the respective ROMs are typically between 10-15 degrees. For ankle and hip flex-extension the ROMs are approximately 25 and 45 degrees respectively.

We can see that minimalist marker sets have been consistently shown to be unreliable for 3D gait assessment (although defining a MCD is not necessary to see this). These concerns have been expressed for a long time and surface occasionally on the Biomech-L archives, with the view that these marker sets although acceptable in flexion and extension are of questionable reliability and potentially unusable for internal/external and abduction/adduction movements.

Thankfully modern technology has improved and it is not uncommon to use 70 to 80 markers in full body models. (Rowing, gymnastics, cricket fast bowling, soccer penalty kick)

The theory behind expanded marker sets are that marker sets involve 4 to 6 markers per segment which may include virtual joint centers defined from proximal and/or distal segments. Importantly common joint centers defined from distal and proximal segments are not constrained to coincide. The locations of segment markers are only critical in a sense that:

i) At least four markers per segment [2-3], these may be real or virtual;
ii) Markers distribution along at least two axes [4];
iii) A mixture of marker placed on soft tissue and over boney prominences,
iv) Avoidance of sites known to exhibit large skin movement (proximal thigh, Greater Trochanter [5,6], and over calf musculature [6]);
v) Least squares methods rather than direct methods [2-3], even with three markers [2].
[Preferably using the method of Veldpaus et al.(1988) [3].

With segment axes reconstructed from a least squares method, joint anatomical rotations are then calculated based on two key steps; calculation of a Cardan rotation sequence corresponding to the notation of ordered anatomical rotations (flexion/extension, abd/adduction and int/external rotation) and defining a reliable starting or reference position from which the rotations are defined. [The later is not as easy as it sounds as the standing position is not zero joint rotation. Individual standing position, variations in standing position, misplacement of anatomical landmarks [7], and errors in measuring true joint rotations in the standing position can introduce considerable errors in the form of an offset in the curves of anatomical joint rotations. This remains the largest and most difficult source of error in inter and intra examiner gait assessment reliability, and as yet no suitable solution.]

In a reliability study using 4 testers that each tested 11 subjects on three separate occasions, the inta-tester 2*95% CI for each segment (peak values in brackets) were:

Hip: 2.6(5.2), 1.6(3.1), 2.4(4.0) degs
Knee: 2.4(4.7), 2.3(3.7), 2.0(3.2) degs
Ankle: 1.9(3.3), 1.7(3.0), 2.2(3.9) degs

The inter-tester 2*95% CI for each segment (peak values in brackets) were:

Hip: 2.4(4.9), 1.7(3.0), 2.8(5.3) degs
Knee: 2.8(5.9), 2.4(4.1), 2.9(5.0) degs
Ankle: 2.1(3.7), 1.8(3.3), 3.0(5.1) degs

Compared to the minimum clinical difference improvements in inter-tester reliability are still required, possibilities include multiple static calibrations with calibration markers removed and replaced within a trial, and artifact compensation using known marker displacements [8-10].

Two observations on reliability:
Larger variations in knee flexion angle were seen during the swing phase with large knee flexion, while the variations in hip flex-extension, although less than that of the knee, were higher at the beginning of stance phase. This may be associated with joint angles being further from the calibration pose, with increased influence of errors in estimates of joint centers and of low frequency skin movement artifact associated with joint movement, or alternatively knee flexion angles during swing or to a lesser extent hip flexion at early stance are naturally more variant than other joint movement during normal gait. This is contrary to many previous studies that report the greatest variability in abduction-adduction and int-external rotation, which most likely reflects inaccuracies in methods rather than joint movements.

Any subjective measurement introduced to the testing procedure decreased between tester reliability of joint angle data. This is despite individual testers, whether novice or experienced, being reliable in their own measurements. It is therefore desirable to minimize the need for palpation of boney landmarks or goniometry based measurements on which segment axes locations and zero point of anatomical rotations are based. Hence the preference for using functional joint centers in 3D movement analyses.

I first used an expanded marker set for a 7 segment, 60 marker, lower body walking model in the mid-90s, with 8 markers per segment [11], and have been using them ever since for leg, pelvis, lumbar, lower and upper trunk, head and arm movements. The 8 markers per segment for gait were excessive as little improvement in reliability was observed beyond 6 markers.

Unfortunately the literature in the area of skin movement artifact and marker set design has been poor. With more resent reviews not fully reflecting current knowledge and modern practices [12-13] or re-presenting previous minimalist markers sets and inheriting their inaccuracies [14-15]. [A Kit Vaughan marker set with two PSIS markers instead of one sacral and an extra foot marker].

The bottom line for me was when I first started to perform gait assessments I could not use the Helen Hayes markers set (and the Ortotrak software that we had at the time) and confidently talk to an Orthopedic surgeon about gait movement patterns when results varied greatly and at times obviously disparate from what we were observing.

JK wrote
>"I'm doing some work on the use of biomechanical models and marker
>sets to study the kinematics of lower limb amputees. My efforts were >initially geared towards selecting an appropriate marker set for
>amputee analysis that could also be applied broadly to all patient
>groups and healthy subjects"

I use the same marker set and principles of marker placement when conducting gait assessments on children with cerebral palsy, but placing markers on AFO's. I have also use the marker set once on a below knee amputee for cycling at different work loads. The methods and marker placement were exactly the same. For your interest the results for the prosthetic ankle joint, which should be constant across all cycling trials, were:

Rotation, mean (Stdev)
Flex/Ext, 89.67 (0.46) degrees
Abd/Add, 1.08 (0.48) degrees
Pron/Sup, -1.09 (0.79) degrees

This will give you more to think about


Dr Allan Carman
Research Fellow
School of Physiotherapy
University of Otago
New Zealand

[1] McGinley, JL, Baker, R, Wolfe, R, Morris, ME (2009) The reliability of three dimensional kinematic gait measurements. Gait and Posture, 29(3), 360-369.
[2] Cappozzo, A, Cappello, A, Della Croce, U, Pensalfini, F (1997) Surface marker cluster design criteria for 3-D bone movement reconstruction, IEEE Trans. Biom. Eng., 44, 1165-1174.
[3] Carman, AB, Milburn, PD (2006) Determining rigid body transformation parameters from ill-conditioned spatial marker coordinates. J. Biomechanics, 39, 1778-1786.
[4] Solderkvist, I, and Wedin, PA (1993) Determining the movement of the skeleton using well-configured markers, J. Biomechanics, 26, 1473-1477.
[5] Cappozzo, A, Catani, F, Leardini, A, Benedetti, MG, Della Croce, U (1996) Position and orientation in space of bones during movement: experimental artifacts, Clinical Biomechancis, 11, 90-100.
[6] Stagni, R, Fantozzi, S, Cappello, A, Leardini, A (2005) Quantification of soft tissue artifact in motion analysis by combining 3D fluoroscopy and sterophotogrammetry: a study on two subjects. Clinical Biomechanics, 20, 320-329.
[7] Della Groce, U, Leardini, A, Chiari, L, Cappozzo, A (2005). Human movement analysis using stereophotogrammetry Part 4: Assessment of anatomical landmark misplacement and its effects on joint rotations. Gait and Posture, 21, 226-237.
[8] Lucchetti, L, Cappozzo, A, Cappello, A, Della Croce, U (1998) Skin movement artefact assessm,ent and compensation in the estimation of knee-joint kinematics, J. Biomechanics, 31, 977-984.
[9] Alexander, EJ, and Andriacchi, T (2001) Correction for deformation in skin-based marker systems, J. Biomechanics, 34, 355-361.
[10] Cappello, A, Stagni, R, Fantozzi, S, Leardini, A (2005) Soft tissue artifact compensation in knee kinematics by double anatomical landmark calibration: performance of a novel method during selected motor tasks, IEEE Trans. Biomed. Eng., 52, 992-998.
[11] Carman, A.B. (2000) The development of methods for improving accuracy and validity of musculoskeletal modeling of the lower limb. PhD Thesis, University of Otago, Dunedin New Zealand.
[12] Chiari, L., Della Groce, U., Leardini, A., Cappozzo, A. (2005). Human movement analysis using stereophotogrammetry Part 2: Instrument errors. Gait and Posture, 21, 197-211.
[13] Leardini, A., Chiari, L., Della Groce, U., Cappozzo, A. (2005). Human movement analysis using stereophotogrammetry Part 3: Soft tissue artifact assessment and compensation. Gait and Posture, 21, 212-225.
[14] Leardini, A., Sawacha, Z., Paolini, G., Ingrosso S., Nativo, R., Benedetti, MG. (2007) A new anatomically based protocol for gait analysis in children. Gait and Posture, 26(4), 560-571.
[15] Benedetti, MG, Cavazza, S, Ferraresi, G, Manca, M, Marchi, P, Zanaga, E, Leardini , A. (2009) Repeatability of a new anatomically based protocol for gait analysis in adult subjects, Gait & Posture, 30 S2, 2009, S14-S15.

-----Original Message-----
From: * Biomechanics and Movement Science listserver [mailto:BIOMCH-L@NIC.SURFNET.NL] On Behalf Of DMRC-HSO1 (Kent, Jenny Miss)
Sent: Wednesday, 31 March 2010 5:30 a.m.
Subject: [BIOMCH-L] Marker sets and models for the kinematic analysis of lower limb amputees

Dear Biomech-L subscribers,

I'm doing some work on the use of biomechanical models and marker sets
to study the kinematics of lower limb amputees. My efforts were
initially geared towards selecting an appropriate marker set for amputee
analysis that could also be applied broadly to all patient groups and
healthy subjects (as normative controls) to facilitate comparison, with
both clinical and research applications in mind. This would comprise a
standard configuration that could be built upon for specific
applications that require additional segments/parameters to be analysed,
but would ensure data collection uniformity as far as possible.

A lit search I conducted brought up many studies on individual amputees
or amputee cohorts using a variety of marker sets, although the majority
appeared to be some form of Helen Hayes derivative. On the prosthetic
side markers were commonly placed on mechanical centres of rotation
where obvious (e.g. on single axis joints) or "estimated from the sound

I was concerned about the appropriateness of the use of several of the
common models and methods when analysing amputee function, particularly
when directly comparing the function/movement of prosthetic and sound
limbs. Examples of such applications would be comparing prosthetic gait
to normative data, evaluating limb symmetry (is this appropriate
anyway?! I'm doubtful in most cases, esp when it concerns unilateral
amputees), and comparing components that function differently
mechanically (e.g. 4-bar knee vs single axis knee, SACH foot vs foot
with articulated ankle joint).

I imagine it may be possible to achieve more reliable/valid results (and
a system that may be applied more universally) with joint centres
determined functionally, technical markers for tracking segmental
movement and perhaps the addition of extra markers to monitor relative
movement of the residual limb with respect to the socket (although I
won't even attempt to touch on the issues of placement and treatment of
soft tissue artefact here..). Regarding the selection/development of a
model that will enable prosthetic components that have not been designed
to reproduce natural segment movement to be adequately represented (e.g.
SACH foot, running blades) and that will allow different limb alignments
to be compared (where the neutral condition at joints may be altered by
nature of the study) I am a little lost. I anticipate when I begin to
consider the analysis of joint moments, torques and powers, which
undoubtedly will open yet another can of worms, I will be even more so.

I suspect that my idea of a universally applied base marker set and
model may be unrealistic, even given an allowance for additional
markers/segments for specific analyses, and that with amputee cohorts
kinematic models will have to be more specifically geared towards the
question that the analysis is required to answer.

I would be very grateful if anyone has any thoughts or advice to share
on the subject. I remember there was a very relevant presentation and
discussion at the CMAS UKI annual meeting and conference in Edinburgh
2009 - I am aware that there may be groups specifically looking into it
and I was wondering if there have been any recent developments that I
have not come across. I'll happily post a summary of replies - please
let me know if you'd prefer not to appear in it!

Many thanks

Jenny Kent
Higher Scientific Officer
Centre for Human Performance, Rehabilitation and Sports Medicine
DMRC Headley Court
Epsom, Surrey
e: DMRC-HSO1@mod.uk