Scaling of the feet

[Erica Beaucage-Gauvreau]

Hi,

I am having trouble scaling the feet for my model. I used the following markers on my participants (1st MT, 5th MT, Calcaneus):

Picture of feet.JPG (32.17 KiB) Viewed 3266 times

Picture of feet_2.JPG (13.73 KiB) Viewed 3266 times

I am trying to scaled using the guidelines by Jenny Yong in her OpenSim webinar:

Scaling of the feet.JPG (27.94 KiB) Viewed 3266 times

Where the feet are scaled in 3-dimensions:

I project all three markers to the same plane, placing them with y=0 in the calcn body. I also project the ankle joint marker on the same plane, and the midfoot marker between 1st MT and 5th MT on the same plane. I fix all of those markers so they do not move, so I can get the scaling factors. I also used them to get the pose of the feet for the participant during marker registration, but I leave the actual markers (not the projected ones: 1st MT, 5th MT, and calcaneus) unfixed so they can get placed to the right position on the foot, for better tracking. However, I get really large scaling factor in the y-direction of above 2.2! The feet look very large, but yet the experimental are still floating above the bone:

feet_scaled_only.JPG (21.89 KiB) Viewed 3266 times

Scaling.JPG (20.02 KiB) Viewed 3266 times

It looks very wrong, and once I get to IK, my trials don't always solve.

What am I doing wrong? How could I fix it?

Thank you for your help.

Erica

[Jennifer Yong]

Hey Erica,

I have a couple of questions to try to help you debug your scaling of the feet.

1. When you placed your additional model markers to represent the markers projected onto the plane, did you place them at y = 0 in the calcaneus frame or did you calculate where y = 0 in ground is located in the calcaneus frame and place them at that position? Based on your image, it looks like these were placed at y = 0 in ground, but I just wanted to make sure they're projected onto the floor and not just at y = 0 in the calcaneus frame.

2. I just want to confirm that the position of the virtual markers based on your experimental data were placed at y = 0 in the lab frame.

3. In simulations created in our lab, we have also found that we have very tall feet with large scale factors in the y-direction. For example, most of my running subjects have y-axis foot scale factors of 1.2 - 1.6. The foot geometry as visualized in the GUI looks super stretched out, but we generally find this representation to be accurate for a person's foot in shoes. I recognize that your subject is barefoot though, so it does seem as though your foot is scaling a little higher than we'd expect.

Jenny

[Erica Beaucage-Gauvreau]

Hi Jenny,

Thank you very much for this answer.

1. I project the virtual experimental markers to y=0 in the ground.

2. I struggle with placing the virtual model markers correctly at y=0 in the ground. I was initially placing them at y=0 in the calcaneus frame but now realize thanks to your post that it is wrong. However, I am not sure how to find y=0 in the ground frame for the virtual markers (!) ? If I put them at y=0 in the ground frame on the model, then they are not relative to the actual bodies in the model, and since the position of the pelvis will change, then the markers will not be in the right position. I am not sure if I am explaining this clearly? For example, if I put the 5th MT in my markerset at y=0 in the ground body, it ends up being 0.93 in the calc body... I ended up with very flat feet:

feet flat.JPG (13.89 KiB) Viewed 3222 times

3. Do you fix the projected markers onto the ground in your markerset? I think so?

4. Do you leave the "real" marker (i.e. not the projected ones) on the feet unfixed so that they can move to the right spot and be used as tracking markers?

Thank you again,

Erica

[Jennifer Yong]

Hey Erica,

I'm glad we seem to have identified the problem! There is a relatively easy way to figure out where y = 0 in ground is located in the calcaneus frame. Here's my method for doing that.

While displaying your model in a default standing pose in the GUI, add a new model marker. In the Navigator, right-click on Markers and then Add New. Find that marker in the marker list, and click on it to see its properties displayed in the Properties tab. The marker should be in the ground frame (body = ground), and you should be able to set its location to y = 0 in ground. Now, if you change the body that the marker is fixed on to the calcaneus body, OpenSim will automatically express the location of the marker in the calcaneus frame. This means, that the new y-position of the marker should represent the y-coordinate of the floor in the calcaneus frame. Use this value when placing your new model markers on the calcaneus.

You are correct that these projected markers should be on the calcaneus body, not on the ground body.

Also, you are correct that I leave the "real" model markers (not projected model markers) on the feet unfixed so that they can move to the correct positions and be used as tracking markers.

Hopefully, this helps your scaling issues!

Jenny

[Erica Beaucage-Gauvreau]

Hi Jenny,

Thanks for answer. Sorry I think there is something I am missing. I had tried what you mentioned: i.e. placing the projected virtual on the markers at y=0 in ground, and then changing the body to calc to get what it is wrt to that body. However, since those markers are fixed in the calc body, it results into them being wrong placed. When I fix them at y=0, they are placed at the pelvis in the default OpenSim model position:

Fixed markers projected in ground.JPG (31.98 KiB) Viewed 3203 times

Which then results in wrong scaling with the feet in the y-direction:

Fixed markers projected in ground_scaling factor.JPG (16.33 KiB) Viewed 3203 times

as the markers are placed above y=0:

feet flat_2.JPG (18.92 KiB) Viewed 3203 times

Sorry about this confusion. I think I am missing something really simple but somehow can't figure it out!?

Would it be possible for you to share your marker file and scale set up file, maybe that would help?

Cheers, Erica

[Jennifer Yong]

Hey Erica,

I'm happy to post my model marker file and scale set up, but I'm not sure how useful it will be. My scale file looks like most example files that you'd find. And my model marker file is going to be model specific. I've attached them just in case they'll be helpful.

If you're willing to post your data, I'm happy to try and debug the problem. I'd need your experimental marker positions, your model, your model marker set (if that's in a different file), and your current scale setup file.

Jenny

[Erica Beaucage-Gauvreau]

Hi Jenny,

Thanks for the help. I have attached to this email my model, my markerset, and my scale set up.

I will look at your files as well.

Cheers, much appreciated.

Erica

[Jennifer Yong]

Hi Erica,

I'm glad we seemed to figure out the issue! I wanted to make sure that the numbers I mentioned during your call were documented somewhere for your records.

The model when first opened is opened with the mass center of the pelvis located at ground y = 0, but using gait 2392's ground platform, I estimated that if the model's pelvis ty = 0.945, the model appeared to be standing on top of the ground. From that pelvis height, ground = 0 is located at y = -0.0111293 in the calcaneus frame. When I used these values for your projected markers, the y-dimension of the foot scaled to 1.64516, which is slightly bigger than the ranges I found, but likely more reasonable.

As I mentioned, since there's some estimation of when the model is standing on the platform, and this estimate will directly affect the height of the model's foot, feel free to play around with that value if you still want your foot height to be a little bit smaller.

Jenny

[Erica Beaucage-Gauvreau]

Hi Jenny,

Thanks for taking the time to look at my model, as always, much appreciated. Happy we could find the issue.

Cheers,

Erica