Conventional Gait Model Calculations

[Mathew Schwartz]

Hi Arnaud and team,

A couple years ago we were looking into BTK for joint kinematics, but I could not find the ability, so we worked on making a python implementation ourselves.  I noticed from the google group you have started to implement some options so I just released ours under MIT license.  https://github.com/cadop/pyCGM

The code is written in python, and meant to be easy for non-programmers to follow along how it is programmed, so as a goal it is (in my mind) pretty different than BTK, but maybe the algorithms can help speed your development along?

I look forward to the future MOKKA!

Mat Schwartz

[Arnaud Barré]

HI Mathew,

Congratulations for your project. The code for the conventional gait model in BTK (currently called PluginGait in the code) is already implemented and verified against original Vicon results for the whole body (greater differences are around 5e-4mm for the computation of some segments poses). The only differences are the origin for the feet and the head compared to Vicon. I did not find the good way to compute them but this does not influence the results of the joint kinematics (or kinetics).

Do you plan to plan to implement other biomechanical models?

Regards,

Arnaud

> --
> You received this message because you are subscribed to the Google Groups "BTK Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to btk-users+...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Mathew Schwartz]

Hi Arnaud,

Testing against Vicon and the sample data of ROM with 'Colin', we are around 5.7E-05 for segment position and 1E-05 for angles.  We too are not sure about foot, since most sources state it as the toe marker, but that is not the case in Plugingait.  Are you computing the progression frame?  We have had an issue with the global pelvis location since different files can change the world coordinates, and we see now way of knowing what they are.

Right now we want to implement the basic options of kinematic models (currently not interested in kinetics for our work).  So the direct method for conventional gait, and next a simple model using marker clusters.  Our purpose is largerly for people to get a boiler code and then modify it or integrate into their own projects.

Thank you,

Mat

[Arnaud Barré]

Hi Mathew,

I have some code in development to compute the global angles of the pelvis, head, thorax, and the feet. This works correctly when the direction of progression is around the global X axis, but a user reported differences compared to Vicon results when the direction of progression is along another axis. I am thinking to a robust way to compute the direction of progression especially when there are turns in the trial.

Arnaud

[Mathew Schwartz]

Hi Arnaud,

Yes we have the same problem.  The angles are correct, as long as it matches vicons global, which seems to be non-obvious.  We have done trials with the same person standing in different directions with no motion, and the global axis changes.  If I remember, there was some vicon documentation that mentioned progression angle being a summation of 80mm of motion in the middle of the trial, but cant remember exactly. 

Please keep me updated if you solve this issue, I will do the same.

Mat