View Full Version : Summary of Pertinent Responses - Pronation/Supination withPlug-In Gait

05-02-2010, 05:54 AM
Thanks to all who provided feedback regarding their experiences with
pronation/supination measures with Plug-in Gait. These are not all the
responses I received but the ones that most directly addressed the problem
and where the problems might lie. Other responses regarding the need to
change the model we are using and how to optimally do that were helpful but
I didn’t include them here for the sake of brevity.

My summary of the situation:

1. The problem we are having is the same as identified by Jacki in the
first email below – an inverted inversion/eversion graph.
2. Even when we initially assumed the shape of the curve was correct
(misinterpreting the sign convention for PIG), our values were smaller than
expected, the same problem identified by Jerrod in the second email below.
I do not know if it is possible the Jerrod’s curves are shaped like Jacki’s
and mine and they are assuming the opposite sign convention or rather they
are getting correctly shaped curves with small numbers.
3. There were several other emails (not included below) from people in
various labs saying something like: We’ve been discussing this in our lab as
well because we aren’t sure what’s going on so let us know what you find
4. If you only read one of the emails below, make it the fourth one.
Lasse describes a possible mechanism for why the processing completed by
NEXUS is faulty. I have not been able to confirm with our rep. at VICON
whether this method of establishing the axes for the ankle are still in use
or whether they have been improved now that modern optical capture systems
are capable of detecting such fine differences. Based on my next point, I
suspect Lasse’s explanation is a fairly accurate explanation of the problem
I, Jacki, Jerrod and others are seeing.
5. A friend of mine at another lab saw my question on the listserv and
volunteered to analyze some of my data. He processed the trajectory data as
captured by Nexus using one of his own Visual 3D models based on a Helen
Hayes set. The graphs for the ankle components were just about textbook
perfect including curves that were all shaped as expected and values that
were at least reasonable if not spot-on as expected.

In no way is any of this meant to be disparaging of Vicon as they have been
great to us and our primary contact person there has been extremely
knowledgeable and helpful. However, my conclusion from all of this is that
the problems I and others have encountered appear to be related to problems
with how Nexus models and processes outputs for the ankle and not related to
the limitations of the PiG model. That is, the model is capable of providing
measures and meaningful data beyond what Nexus will produce and in fact, the
data for the ankle are inaccurately represented as something they are not.
It would actually be more helpful if the output omitted the incorrect ankle
components and classified them as “not available” or something similar than
to call incorrect data something they are not.



Cris Stickley, PhD, ATC, CSCS

Associate Program Director, Post-Certification Athletic Training Master’s

Director, Biomechanics & Gait Lab

Dept. of Kinesiology and Rehabilitation Science

Dept. of Anatomy, Biochemistry, Physiology & Reproductive Biology

University of Hawaii, Manoa


Go Buckeyes!!!


We have noticed the same thing but had assumed that we were misunderstanding
the PiG sign convention. However, after verifying with Vicon that inversion
is positive and eversion is negative, we are seeing that during stance the
ankle is shown as inverting after heel strike and everting towards toe off
instead. Does this sound like what you are observing? We haven't found a
solution yet since we just noticed it! I would be very interested if you
find out what is going on.


Jacki Watt

Biomechanical Engineer, Gait Laboratory

Physical Medicine and Rehabilitation

University of Virginia

(434) 982-1422



Hello Cris,

Our lab has been having this discussion lately as well. Our values for
pronation/supination have been very small compared to what we had expected
during both walking and running gait. I would be very interested if you
hear of any responses to the reasoning behind this issue. If we figure
anything out, I will let you know what we find.

Take care, and good luck.

Jerrod Braman



The plug in gait model, in Vicons own words "Fudges the data" for pronatory
movements. I think it estimates the rotation from the axial position of the
foot segment compare to the alignment and position of the shank segment,
hence I would strongly recommend the use of the oxford foot model for any
research or clinical application. I have seen the oxford foot model used
with relative ease accurately by podiatrists with very limited biomechanical

Hope this helps.


Andy Greenhalgh


Having worked for Vicon for many years, I can possibly shed some light on
the issue.

In the static trial, the Conventional Gait Model (and therefore Plug-in
Gait) uses three points to establish the foot coordinate system: the ankle
joint centre (AJC) from the tibia segment, the heel marker and the toe
marker. The HEE-TOE line defines the long axis of the foot segment, but for
historical reasons - mainly that the motion capture systems back in the 90s
had trouble with the heel and ankle markers being in close proximity - the
heel marker was taken off for the dynamic trial. Therefore, the Conventional
Gait Model establishes the foot coordinate system based on only two points -
AJC and TOE - adjusted for the static foot offset angles.

In other words: even if the heel marker is left on during the dynamic,
CGM/Plug-In Gait DOES NOT use this marker for dynamic processing.

Plug-in Gait does, for technical reasons, establish a 3D segment for the
foot prior to the calculation of the ankle kinematics. However, since no
mediolateral axis for the foot can be defined from markers, it simply uses
the ankle flex-ext axis. This means that ONLY the long axis of Plug-in
Gait's foot is aligned with a true anatomical foot axis. The "mediolateral
axis", established as mentioned above for technical reasons, is the axis
that is mutually perpendicular to both the long axis and the axis defined by
the cross product of the long axis vector and the AJC-ANK vector.

This again means that no meaningful inversion/eversion angles are calculated
(or defined) by the CGM or Plug-in Gait, and the angles output by Plug-in
Gait's second ankle kinematic component contain only cross-talk.

If more detailed kinematics for the foot is required, I recommend trying a
detailed foot model such as the Oxford Foot Model. Alternatively, if such a
multi-segmented foot model is too detailed or complicated, I guess you could
add a marker or two to establish an anatomical mediolateral foot axis, and
write a relatively simple BodyBuilder model to calculate proper 3D ankle

Best regards,

Lasse Roren

Prophysics AG

Zurich, Switzerland


Hi Cris,

Just as a remark, but there may be more issues involved: is it not because
the markers on the foot are, because of the size of this limb, almost placed
in line? This then makes the rotation around this line very difficult to
calculate/measure, and strongly increases the noise in the measurements. To
increase precision you should place some markers in the plane of rotation
(more or less left-right) as far apart as possible.

Best regards,

Jan Veneman

Departamento Biorobótica - Unidad Salud y Calidad de Vida Biorobotika Saila
- Osasun eta Bizi Kalitate Unitatea Biorobotics Department - Health and
Quality of Life Unit

Email: jveneman@fatronik.com

FATRONIK - Tecnalia

Paseo Mikeletegi 7 - Parque Tecnológico

E-20009 Donostia - San Sebastián

Telf: (+34) 943 005500

Fax: (+34) 943 005511

Webs: www.fatronik.com - www.tecnalia.es