How good are added DoFs?

[Markus Kurz]

Hej,

First of all, i want to give a big shoutout to this really helpful community here.

I added 2 rotational degrees of freedom to the knee (for knee adduction and tibia rotation relativ to the knee) as well as 2 rotational degrees of freedom to the ankle (to get the technical axes for flexion, eversion, and rotation). I furthermore locked the subtalar joint (by giving it a range from 0 to 0 ).
What i want to do with this model is using it for IK and ID, to get joint angles and moments for runners.

My Question is:
Is something like this even "legit" to do, meaning can i even get plausible results out of that, or am I missing some constraints or other things i have to add to the model in order to make it a good model? Or are there full body models out there already having all 3 rotational degrees of freedom within the knee and ankle joint?

Help is appreciated.

Kind regards,
Markus

[Ton van den Bogert]

Markus,

If you want joint angles and moments, there is no fundamental problem with adding those degrees of freedom. The only price you pay is that the inverse kinematics is less robust, so may be more noisy and more sensitive to soft tissue artifact. It may actually be more accurate to assume, for instance, that the kneeab/adduction is zero, instead of trying to measure it and have a lot of error.

If you don't add those degrees of freedom, you won't get the corresponding joint angles, but you can still get the corresponding joint moments! A knee joint with 1 DOF can still report 6-DOF joint reaction loads in Opensim.

If you are going to estimate muscle forces, you should not add those degrees of freedom. They have been left out because those degrees of freedom are not controlled by muscles. For instance, if you make a moment arm plot in opensim, for any knee muscle, and plot the moment arm for internal/external rotation, you see that the moment arms are extremely small. Therefore, if your movement has any internal/external rotation moment, the estimated muscle forces will be huge. Those degrees of freedom are controlled by articular contact and ligaments, and that is not represented well in an opensim model. So it's best not to have those degrees of freedom.

I don't remember a reference where they looked at that, but as an example of huge and unrealistic muscle forces, you can look at:

Burdett RG (1982) Forces predicted at the ankle during running. Med Sci Sports Exerc 14(4):308-16.

This is a consequence of having degrees of freedom that are not mostly controlled by muscles.

Ton van den Bogert

[Markus Kurz]

Hej Ton,

a really big thanks for the quick answer. So far we are not getting really big values for the moments. But I am still interested in your proposal of using only 1 DoF during IK and than still having 6 DoFs for the moments. It would be really great if you could give me a quick explanation on that. So far, if I am using a 1 DoF Model and then doing inverse dynamics I am getting of course only the moment for that plane. So help would be really appreciated.

Best Regards
Markus

[Ton van den Bogert]

Markus,

The magnitude of the moment is not affected by the extra DOFs, though they may get noisier. So your moments are not systematically affected.

It is only the muscle forces that will become unrealistically high, because the moment arms for some of those DOFs can be very small.

I don't do inverse dynamics with Opensim (I have other tools), so I can't give a specific example of how to get the 6DOF loads.

Opensim has a "joint reaction load" analysis tool that will give you the 6DOF loads (3D force and 3D moment) for any joint. If you want the net or resultant joint loads (without muscle contributions), you will have to remove the muscles from the model. Unless there is another tool that can tell you the resultant loads.

Without muscles, this is really just a multibody dynamics analysis and it may be available in Simbody without you having to remove muscles from the model.

Ton

[Axel Koussou]

Hi Ton,

Sorry to reopening this channel forum, but I am also interested in this kind of questions.

More exactly, I would like to know how legit are my muscle-tendon length after adding external/internal rotation and abduction/adduction DoF to the knee.
For example, my musculotendinous complex of the semitendinosus will be affected but I don't know how legit is my relation between this length and the added DoF.

Could you enlighten me?

Thanks in advance,

Regards

[Ton van den Bogert]

In principle, there is no problem with muscle length calculations if you have additional degrees of freedom.

My only concern would be that the inverse kinematic analysis is less robust, but I don't think it would make a big difference.

If you want to know for sure, you can do it both ways: a 1-dof knee and a 3-dof knee. if the difference is small enough (as I expect), you can trust both.

This is always a dilemma in modeling. An error is made by assuming zero abduction and rotation. An error is also made when estimating those degrees of freedom from skin markers. It's hard to say which error is larger!

Ton

[Axel Koussou]

Hi Ton,

Thank you for your answer !

I will definitely try what you suggest.

Regards