CMC activation- type of muscle

[Chiara Camaschella]

Hi,
I'm working with the Stanford VA Upper Limb model.
I've added to it the inertial properties for dynamic simulations.
I am simulating an arm-crank exercice and I have problems with CMC results, activations are really very low than the results of SO and warnings appear.
That's an example of my results for some muscles of the model:

CMC.png (16.7 KiB) Viewed 553 times

I've noticed that the muscles in the model are implemented as Schutte's muscles, which are no longer used, can that be the problem? If I had to convert all the muscles into Millard muscles? It would be a complex job?

Thanks in advance

Chiara

[Pierre Kibleur]

Hello Chiara,

I've been having the same issue. Changing the muscle model is not complicated (just a find and replace in the .osim), but OpenSim does not seem to be entirely transparent as to what it does (for instance, the Schutte muscles in your model have custom-defined splines, which I guess are lost forever the next time you'll save the model. Since -at least I hadn't found it- there is no explanation as to how and why those splines were defined by the publishers (what problem were they fixing, one muscle at a time), we are at the risk of losing some knowledge in the process).

However it does seem better to use the Millard model, especially for the upper limb and dynamic simulations, as explained by Millard:
https://www.youtube.com/watch?v=Q1rId1fOkjw&

After that, the muscle activations could be low if the muscles are too strong for the inertial data.

Best,
Pierre

[Elisa Panero]

Hello Chiara,

I've been having the same issue. Changing the muscle model is not complicated (just a find and replace in the .osim), but OpenSim does not seem to be entirely transparent as to what it does (for instance, the Schutte muscles in your model have custom-defined splines, which I guess are lost forever the next time you'll save the model. Since -at least I hadn't found it- there is no explanation as to how and why those splines were defined by the publishers (what problem were they fixing, one muscle at a time), we are at the risk of losing some knowledge in the process).

However it does seem better to use the Millard model, especially for the upper limb and dynamic simulations, as explained by Millard:
https://www.youtube.com/watch?v=Q1rId1fOkjw&

After that, the muscle activations could be low if the muscles are too strong for the inertial data.

Best,
Pierre

Hi Pierre,

my name is Elisa and I am working with Chiara. Please, could you explain what do you mean with the term "splines"? Thanks in advance for your attention.

Elisa

[Pierre Kibleur]

Hi,

I understand it as a discrete set of points given to reproduce a curve, that is being spline interpolated to have continuous data.
For instance, this force length curve defined for the DELT1 tendon in the MoBLaS human arm model

Code: Select all

					<!--Function representing force-length behavior of tendon-->
					<SimmSpline name="tendon_force_length_curve">
						<x> -10 -0.002 -0.001 0 0.00131 0.00281 0.00431 0.00581 0.00731 0.00881 0.0103 0.0118 0.0123 9.2 9.201 9.202 20</x>
						<y> 0 0 0 0 0.0108 0.0257 0.0435 0.0652 0.0915 0.123 0.161 0.208 0.227 345 345 345 345</y>
					</SimmSpline>

Edit: Yes, I've just double checked. Replacing the SchutteDeprecated by MillardEquilibrium removes amongst others the DELT1's custom tendon force length curve. Which as a general thing is good (that makes the models less opaque). But in the transition period I don't know enough

[Elisa Panero]

Hi,

I understand it as a discrete set of points given to reproduce a curve, that is being spline interpolated to have continuous data.
For instance, this force length curve defined for the DELT1 tendon in the MoBLaS human arm model

Code: Select all

					<!--Function representing force-length behavior of tendon-->
					<SimmSpline name="tendon_force_length_curve">
						<x> -10 -0.002 -0.001 0 0.00131 0.00281 0.00431 0.00581 0.00731 0.00881 0.0103 0.0118 0.0123 9.2 9.201 9.202 20</x>
						<y> 0 0 0 0 0.0108 0.0257 0.0435 0.0652 0.0915 0.123 0.161 0.208 0.227 345 345 345 345</y>
					</SimmSpline>

Edit: Yes, I've just double checked. Replacing the SchutteDeprecated by MillardEquilibrium removes amongst others the DELT1's custom tendon force length curve. Which as a general thing is good (that makes the models less opaque). But in the transition period I don't know enough

Thank you. So, let me know if I've understood in the right way: if I change the model from Schutte to Milland I will loose some custom characteristics of muscles?
Sorry, I am not so expert as you can see

[Pierre Kibleur]

No problem. Yes. The issue is that I've just checked, and every single one of the 50MTUs have a custom tendon_force_length_curve... Maybe you can try with the Millard model anyway, it should be more robust. It's just in the new parametrisation of the muscle that some things could need to be tuned, but with knowledge that I don't have.

I'm not an expert either , so again what I write is not groundtruth

[Thomas Uchida]

You might try (1) extracting the x and y points that define the Schutte muscle curves in your original model, then (2) finding the parameters for the Millard2012EquilibriumMuscle curves that match the original curves as closely as possible. The parameterizations of the Millard muscle curves are provided in the API documentation:
- https://simtk.org/api_docs/opensim/api_ ... ml#details
- https://simtk.org/api_docs/opensim/api_ ... ml#details
- https://simtk.org/api_docs/opensim/api_ ... ml#details
- https://simtk.org/api_docs/opensim/api_ ... ml#details