simtk.org

Hi Everyone,

I am just looking for some feedback on the best practices for inserting new muscles into a generic model.

My problem is that I have data on muscles of the foot from four different cadavers. I am trying to figure the best practice for inputting these muscles into a model such as the Gait2392 model. Here are some options:

1. Use only data from one cadaver, scale the model, input all the muscle parameters from the one cadaver the scaling is based off of.

2. Take the average values from the four cadavers, scale the model based on average values (height, mass), and then input the average muscle parameters from the four cadavers.

3. Use average values for four cadavers for all muscle parameters except the muscle path points, input path points of muscles based on other anatomical literature (such as bony landmarks), rather than coordinate data from the literature.

I might be leaning to option 3 to start because options 1 and 2 give some results that do not look physiological. I could always optimize the muscle paths after if I have known moment arms of these muscles, but I am having trouble finding moment arm values for these muscles in the literature.

Note that I only have mass, height, foot width, foot length from each cadaver. Any tips for effectively scaling the model just based on those variables?

Any insight would be much appreciated.

Thank you!

Mike

Hi Mike,

Could you remind me what muscle parameter data you have? Typically, we think of intrinsic muscle properties (i.e. max isometric force, tendon slack length) as muscle parameters, and everything else as mass or geometric properties. If your interested in having four different subject-specific foot models, it would be best to scale/set muscle parameters for each foot and not do any averaging. If you want a generic model for your simulations, averaging may be appropriate.

I might be leaning to option 3 to start because options 1 and 2 give some results that do not look physiological.

What exactly do you mean by not looking physiological?

Note that I only have mass, height, foot width, foot length from each cadaver.

You can create pairs of virtual markers on your foot model that represent where each measurement was taken (i.e. heel to big toe), and then assign manual scaling factors based on your measurements.

Hi Nick,

Thanks for your response!

I have tendon resting length, Pcsa, optimal fascicle length, pennation angle for the muscle parameters and I have coordinates from a cadaver for origin insertion and via points.

Probably would be best to have an average model then because I want to have a generic model.

The attachments do not look physiological because the insertions are nowhere near a bony landmark. It could be a scaling issue though. What are your thoughts? Would the anatomical landmarks be more appropriate?

I can go ahead scaling the model from the foot using the virtual markers. Thanks for that. Any advice for scaling the rest of the model (gait2393) based on height and mass?


Thanks!

Mike

It probably makes sense to adjust origin/insertion locations based on your cadaver data after you've scaled the foot, that way you don't have to deal with the model scaling moving these locations into undesirable locations.

I'm not sure if placing origin/insertion locations near anatomical landmarks is the best option. I would use the cadaver data as a guide for deciding on those locations. Depending on how much your trust the accuracy of your data set, you could make slight adjustments to the origin/insertion locations to the adjust moment arms as well. As for moment arm data, you could try referencing other models if you can't find anything in the literature. Another option is to include them as another variable in your optimizations, but you may run into trouble deciding on how to bound them and your problems being underdefined and not converging.

Only having height and mass will making scaling the rest of the model tricky. A rough (but likely bad) assumption is to apply the scale factor from the difference between the height of the model and your data (again using virtual markers) to all the bodies in the model. You do have a shot at scaling the muscle volumes (and subsequently max isometric force) using the Handsfield et al. 2014 paper: "Relationships of 35 lower limb muscles to height and body mass quantified using MRI". Although, to get to max isometric force, you'll need a good estimate of optimal fiber length, which you won't have unfortunately.

Hopefully this is somewhat helpful and you can piece something together using your data set. Let me know if you have any other questions.

Hi Nick,

It probably makes sense to adjust origin/insertion locations based on your cadaver data after you've scaled the foot, that way you don't have to deal with the model scaling moving these locations into undesirable locations.

I'm not sure if placing origin/insertion locations near anatomical landmarks is the best option. I would use the cadaver data as a guide for deciding on those locations. Depending on how much your trust the accuracy of your data set, you could make slight adjustments to the origin/insertion locations to the adjust moment arms as well.

Thanks for your input. I inserted all the muscles into the model using the cadaver data. I think I will then adjust (minimally) the origin, via points, and insertions to bony landmarks that I believe the muscle should be attached. Any major criticisms or does this approach seem suitable?

As for moment arm data, you could try referencing other models if you can't find anything in the literature. Another option is to include them as another variable in your optimizations, but you may run into trouble deciding on how to bound them and your problems being underdefined and not converging.

Are you referring to other foot models? If so, do you have any models to refer me to? Thanks for the suggestion about the optimization of the moment arms. I could try that down the line.

Only having height and mass will making scaling the rest of the model tricky. A rough (but likely bad) assumption is to apply the scale factor from the difference between the height of the model and your data (again using virtual markers) to all the bodies in the model. You do have a shot at scaling the muscle volumes (and subsequently max isometric force) using the Handsfield et al. 2014 paper: "Relationships of 35 lower limb muscles to height and body mass quantified using MRI". Although, to get to max isometric force, you'll need a good estimate of optimal fiber length, which you won't have unfortunately.

Is it too big of an assumption to use optimal fascicle length? I am thinking that method may be better than scaling based on just height and mass. What are your thoughts? I will check out the reference you sent. Thanks!

Also, for max isometric force, I have used by multiplying PCSA and a set specific tension similar to Arnold et al. 2010. I can adjust the specific tension down the road, but I want to have something to start.

Also, for the muscle parameters and geometry, I additionally have fascicle length, muscle volume, tension fraction, ratio of average sarcomere length to optimal sarcomere length, normalized fascicle length.

Sorry for all the questions, but thanks a lot for all your help!

Mike