Scaling subject with Equinus

[Axel Koussou]

Dear OpenSim experts,

Dealing with biomechanical data from subjects with cerebral palsy, some of whom present equinus, I have trouble to correctly scale my model for those subjects.

Indeed, before any OpenSim workflow (IK, ID, and so on...), I have to scale my model to my subject. However, this procedure does not perform well for the subjects with equinus.
I understand that this might be because my model's default pose (typical standing position with no equinus) is different from the subject's position during his calibration test (standing with equinus).

Thus, I would like to know if you have any advice to improve this scaling procedure for subjects with equinus?
I know I can change the my model's default pose to match those subjects, but among all the subjects with equinus, standing positions are not the same. Some have a more pronounced ankle-plantarflexed position, others also have a knee-flexed position.
So, if I want my scaling to be right, I would often have to adjust the default pose of my model, which is annoying.

Thanks in advance,

Regards

[Najoua Assila]

Hi,

How do you know that your scaling is failing ?

To clarify, the scaling tool performs first a scaling, than a marker placement. The first step is dependant on the distances between your markers, usually the ones on the bony landmarks. This step shouldn't be problematic just because your subject position is different than that of the model, as in theory, your landmarks have minimal soft tissue artefact. The second step, might be more problematic, so an iterative approach (scaling-IK) might be needed to find the optimal marker weights that don't create any bias due to the participant's pose. It should be noted though, that an "implicit IK" is run for this step, and thus your default model pose, shouldn't be too problematic, if the joint coordinate differences aren't extreme.
You can find more information on the scale tool troubleshooting guide https://simtk-confluence.stanford.edu/d ... leshooting

Hope this helps,

Najoua

[Axel Koussou]

Hi,

Thank you for your return.

Hereinafter you will find a little example to answer to your question.
In a subject with equinus only at the left side, the scaling result does not seem correct. That make me think that the scaling procedure is failing :

For this subject, we recorded a static calibration trial. However, because of his equinus, his right foot was correctly on the ground, whereas his left foot was plantarflexed.

When I perform the scaling procedure for this subject, using Rajagopal's model in his default pose (typical standing position with no equinus), you can see that the left calcaneus seems totally disproportionate.

Scaling_Equinus1.PNG (58.95 KiB) Viewed 453 times

While, the right calcaneus is more appropriate.

Scaling_Equinus2.PNG (52.21 KiB) Viewed 453 times

I know that bones are only for visualisation, but this scaling error also affects my MTU-lengths. There is length difference (not negligible) between my right and my left Gastroc Lateralis-Tendon units.

Do you think that I am wrong to thing that the scaling procedure is failing?

Thanks in advance,

Regards

PS : My scaling procedure for the foot is based on the one adviced by OpenSim community (In this video https://www.youtube.com/watch?v=ZG7wzvQC6eU from 23:00).

[John Davis]

Are you projecting markers onto the floor and using them for scaling the foot? That might cause the strange distortions in the calcaneus in your first image.

[Axel Koussou]

Hi John,

Thanks for you answer.

Indeed, I do use projected marker ont the floor to scale the foot.
However, I have already tried to base my scaling procedure direclty on the anatomical markers of the foot, not the projected ones, and it seemed that I still had the same problem.

I will give an other try.

Regards.

[Najoua Assila]

Hi,

Using the projection method improves your chances of having a flat foot at the end of the scaling, supposing it is in contact with the ground. Thus you can only use this approach correctly for the Z direction.
Supposing you have no additional data, you could try to use the average of Heel-Medial and Heel-Lateral for the Y direction, and that of the Metacarpal-Medial, Metacarpal-Lateral and Metacarpal-Heel for the X direction.

Hope it helps,

Najoua