Am interested in the choice of 61N/cm2 for the muscle specific tension in OpenSim models (I personally have been using the gait2392 and E.Arnold's Lower Limb 2010 model). I see from her paper (Arnold 2009) that this value comes from Delp's PhD work (Delp 1990 p.34). If I understand it correctly the value of 61N/cm2 is generated by comparing the PCSA data from Wickiewicz (1983) to moment curves measured from younger subjects. I've got a copy of Wickiewicz (1983) but I can't find any reference to the ages of the three cadavers used. Do anyone know how old these cadavers were or where this is reported? Also, 20 years on, what are peoples thoughts on the 61N/cm2 value of specific tension that is used in these models? Do we feel that it is still appropriate in the light of more recent publications of muscle morphology?
Choice of Muscle Specific Tension
-
Edith Arnold
- Posts: 44
- Joined: Fri Apr 06, 2007 2:07 pm
RE: Choice of Muscle Specific Tension
Check out this follow-up paper to Wickiewicz 1983
Wickiewicz, Roy, Powell, Perrine, and Edgerton. "Muscle architecture and force-velocity relationships in humans" J Appl Physiol Respirat Environ Exercise Physiol 57(2); 435-443, 1984. p. 441
"If human skeletal muscle has the same force potential as other mammalian muscle...then the forces observed for all four muscle groups were about twice the predicted value...This might be expected because the CSA's were derived from cadaver materials obtained form older subjects and were undoubtedly reduced in volume by the fixative solutions. In contrast, the torque measurements were obtained from young, healthy subjects."
61 N/cm^2 is high, and I know some people don't like it for all the reasons you mentioned. Some prefer to think about it as applying the multiplier to the PCSA (to account for the loss of muscle mass that happens in the elderly) or to F_Max, to account for the fact that it's hard to get people to truly fully activate their muscles.
Mathematically this is a trivial difference and when I built the model I thought of it as a change to the specific tension (not surprising since I'm Prof. Delp's student), but it does change the physiological meaning of the model.
F_max = PCSA * (30 N/cm^2*2)
compared to
F_max = (PCSA*2)*30N/cm^2
Wickiewicz, Roy, Powell, Perrine, and Edgerton. "Muscle architecture and force-velocity relationships in humans" J Appl Physiol Respirat Environ Exercise Physiol 57(2); 435-443, 1984. p. 441
"If human skeletal muscle has the same force potential as other mammalian muscle...then the forces observed for all four muscle groups were about twice the predicted value...This might be expected because the CSA's were derived from cadaver materials obtained form older subjects and were undoubtedly reduced in volume by the fixative solutions. In contrast, the torque measurements were obtained from young, healthy subjects."
61 N/cm^2 is high, and I know some people don't like it for all the reasons you mentioned. Some prefer to think about it as applying the multiplier to the PCSA (to account for the loss of muscle mass that happens in the elderly) or to F_Max, to account for the fact that it's hard to get people to truly fully activate their muscles.
Mathematically this is a trivial difference and when I built the model I thought of it as a change to the specific tension (not surprising since I'm Prof. Delp's student), but it does change the physiological meaning of the model.
F_max = PCSA * (30 N/cm^2*2)
compared to
F_max = (PCSA*2)*30N/cm^2
-
Andrew Lewis
- Posts: 18
- Joined: Wed Oct 22, 2008 5:58 am
RE: Choice of Muscle Specific Tension
Hi Edith,
Thanks for your reply and paper reference.
Maybe I am getting confused with this issue but in my mind the fact that the value of 61N/cm2 was calculated from comparing elderly (and hence small muscled) cadaveric data of PCSA to younger data for moment curves shows that it must be too high. It might have been a reasonable value to use in the original models where other muscle parameters were generated from the elderly cadaveric data but now we can easily get PCSA data from MRI scans, it seems unreasonable to continue to use 61N/cm2 to convert PCSA to max. isometric force parameters in the muscle model.
I appreciate that, when using physiologically valid muscle parameters, the models currently do not seem strong enough to perform motor tasks such as walking or jumping, and so in order to simply "get the simulations to run" it is necessary to augment the muscle strengths. Also, as you mentioned, MVC experiments may underestimate max muscle strength somewhat but can this justify jumping from c.30N/cm2 (as commonly reported for human muscle) all the way up to 61N/cm2? I am concerned that using the 61N/cm2 specific tension value to justify this muscle strength augmentation is ignoring the issue at hand - namely that there is something wrong with the models as a whole...
Or have I got the wrong end of the stick, What do you think?
Thanks,
Andy
Thanks for your reply and paper reference.
Maybe I am getting confused with this issue but in my mind the fact that the value of 61N/cm2 was calculated from comparing elderly (and hence small muscled) cadaveric data of PCSA to younger data for moment curves shows that it must be too high. It might have been a reasonable value to use in the original models where other muscle parameters were generated from the elderly cadaveric data but now we can easily get PCSA data from MRI scans, it seems unreasonable to continue to use 61N/cm2 to convert PCSA to max. isometric force parameters in the muscle model.
I appreciate that, when using physiologically valid muscle parameters, the models currently do not seem strong enough to perform motor tasks such as walking or jumping, and so in order to simply "get the simulations to run" it is necessary to augment the muscle strengths. Also, as you mentioned, MVC experiments may underestimate max muscle strength somewhat but can this justify jumping from c.30N/cm2 (as commonly reported for human muscle) all the way up to 61N/cm2? I am concerned that using the 61N/cm2 specific tension value to justify this muscle strength augmentation is ignoring the issue at hand - namely that there is something wrong with the models as a whole...
Or have I got the wrong end of the stick, What do you think?
Thanks,
Andy
-
Edith Arnold
- Posts: 44
- Joined: Fri Apr 06, 2007 2:07 pm
RE: Choice of Muscle Specific Tension
I think using imaging data from young healthy subjects would be a really interesting modification to the model. However, until you can get optimal fiber lengths directly you'd have to weigh whether it was more important for your application to have a consistent set of data from elderly cadavers that needs to be adjusted, or a mix-and-match set that uses elderly cadavers and healthy subjects.
As always, the hard part about modeling is deciding what tools and assumptions you need to answer your research question. In my method I prioritized architectural consistency, but that's probably not right for everyone. Since the powers that be won't grant me a PhD for making lots of different models (:-)) it's better for me to use this one, understanding and respecting its limitations, and move on to use it for a few things. I'd love to see others using it as a stepping stone to make the model that they need.
In these muscle models the f-l and tendon strain curves are normalized to max isometric force, so if you've got the right value for F_max and good moment arms, then there isn't anything wrong with the amount of moment that can be produced (what I think of as the strength). The breakdown is in the assumptions we make to go from pcsa to force, and maybe with pcsa from young healthy people you wouldn't need the same adjustments.
One clarification:
The basic version of the model has F_max = 61N/cm^2 *PCSA
This accounts for the elderly cadaver to young healthy adjustment.
This model is most similar to Delp's original model that agrees with MVC torque curves
The cmc-friendly version has F_max_new = 1.50 *F_max
This accounts for the underestimation of true maximum moment from MVC
This model is most similar to gait_2392 that is strong enough for cmc and comes packaged with OpenSim. It included force multipliers of 1.5-2.5 over Delp's model for various muscle groups
-Edith
As always, the hard part about modeling is deciding what tools and assumptions you need to answer your research question. In my method I prioritized architectural consistency, but that's probably not right for everyone. Since the powers that be won't grant me a PhD for making lots of different models (:-)) it's better for me to use this one, understanding and respecting its limitations, and move on to use it for a few things. I'd love to see others using it as a stepping stone to make the model that they need.
In these muscle models the f-l and tendon strain curves are normalized to max isometric force, so if you've got the right value for F_max and good moment arms, then there isn't anything wrong with the amount of moment that can be produced (what I think of as the strength). The breakdown is in the assumptions we make to go from pcsa to force, and maybe with pcsa from young healthy people you wouldn't need the same adjustments.
One clarification:
The basic version of the model has F_max = 61N/cm^2 *PCSA
This accounts for the elderly cadaver to young healthy adjustment.
This model is most similar to Delp's original model that agrees with MVC torque curves
The cmc-friendly version has F_max_new = 1.50 *F_max
This accounts for the underestimation of true maximum moment from MVC
This model is most similar to gait_2392 that is strong enough for cmc and comes packaged with OpenSim. It included force multipliers of 1.5-2.5 over Delp's model for various muscle groups
-Edith