Soft Tissue Artifact (STA) open source dataset

[Vivek Karmarkar]

Hello,

I was wondering if anyone knew where it might be possible to find open source datasets containing Soft Tissue Artifact (STA) data for skin markers placed on the thigh segment? I'm specifically interested in time histories of these markers? Various marker configurations on the thigh are also of interest to me

Kind Regards and Thank you,
Vivek

[Edmund Cramp]

There were a few studies done in the early days of motion capture data collection where the researchers had markers placed on their thighs for one trial and then had a wand inserted into their thigh bones with a marker on top so that the skeletal movement could be compared to the skin movement. I remember talking to one of the subjects at the NIH, but that was a long time ago - he said that it had been an interesting study and had only needed a few minutes of medical work to get the wands attached, and then removed once the data had been collected.

[Vivek Karmarkar]

Dear Edmund,

Thanks for your prompt response and pointing me in the right direction

Kind Regards,
Vivek

[Edmund Cramp]

There was an article in Gait & Posture, Volume 24, Issue 2, October 2006 (pages 152-164) titled "Effects of Skin movement artifact on knee kinematics during gait and cutting motions measured in vivo" that might be helpful. Another earlier Gait and Posture article in Volume 11, February 2000 (pages 38-45) titled "Comparison of Surface mounted markers and attachments methods in estimating tibial rotations during walking: an in vivo study" describes the results of some easy measurement research, documenting a study in a gait lab that I had helped set up originally many years earlier.

Thanks for your question, re-reading both of these was very nice. It would be fascinating to see these studies repeated in the current "markerless" data collection environment.

[Ton van den Bogert]

Hi Vivek,

A good recent article is the review and standardization proposal by Andrea Cereatti et al. (2017): https://www.sciencedirect.com/scienc...21929017301008

This may point you to available data. Our 1994 data from running with bone pins may be available there. If not, I can dig it up and share through Biomch-L.

Here is a video (actually 16 mm film was used) from those experiments: https://www.youtube.com/watch?v=nf6jkyNgkwE

Ton van den Bogert

[Chris Kirtley]

Even those bone pins seem to buzz around a lot, Ton? I guess you just filter it out.

[Vivek Karmarkar]

Dear Ton van den Bogert,



The article by Cereatti et al. is exactly the sort of thing I was looking for. STA marker time histories and all in a .mat file. I honestly cannot ask for more. Thanks a million!

Table 1 in the paper mentions running data from the 1997 paper titled: "Effect of skin movement on the analysis of skeletal knee joint motion during running". I did not come across a 1994 article involving running and bone pins. I'd be more than happy to receive this data from you on Biomch-L

The video is fantastic but personally running with bone pins seems a little scary although the research and data analysis part of it is fun!



Thanks again and Kind Regards,
Vivek

[Vivek Karmarkar]

Dear Edmund,

Thanks a lot for sharing the links to those papers! I'm currently carrying out a simulation study so from that perspective, it is "markerless" since we don't have an experimental setup with markers but I do introduce markers in my simulation. It's always nice to get a different perspective on the problem

Thanks again and Kind Regards,
Vivek

[Edmund Cramp]

Vivek, you might want to look at some of Chris Kirtley's work, he just said, "Even those bone pins seem to buzz around a lot, Edi? I guess you just filter it out." - I think that his point is that putting a marker wand into a bone can be seen as generating reliable and accurate data, but in fact that's not the case, the wand can be thought to be "attached" to the bone but could wiggle a little if it's not deeply screwed in and the wand might be moved by the skin and muscles too - the result could be slightly different artifacts in the data that we thought was accurate, but wasn't perfect.
My view about data collections (as a result of many early years of motion capture system installations) is that it's always best to start out by trying to prove that the data is bad, once you fail to prove that it's bad then you can trust the data to a degree because you have thought about, and looked at, all the potential problems. For example if you see artifact and "... just filter it out" you need to return to the original unfiltered data view to determine the source of the artifacts - filtering data and saying that it looks good don't guarantee that it's perfect; filtering marker data to make it look good doesn't guarantee that the marker was in the correct location.
I'm not saying that markerless data is any better, it's a different environment so it will just have different problems.

[Ton van den Bogert]

Vivek,

You found the correct data. The experiments were done in 1994 but the paper is dated 1997. We also did walking experiments,but that data is unfortunately lost.

Chris and Ed: Indeed bone pins are not perfectly rigid either. They probably vibrated at heelstrike, but hopefully the amplitude was small. I don't recall seeing evidence of vibration in the raw data. The movie cameras ran at about 200 fps so we may not even see the vibration if the frequency was >100 Hz. If anyone remembers tuning forks, they vibrate at 440 Hz. And then it can't be filtered out either.

Ed is right about looking critically at raw data. In one subject, there were strange differences between trials during running (during walking it was fine). We eventually concluded that the pin must have rotated axially, due to the iliotibial band rubbing on the pin. That subject's data was not used.

Ton

[Vivek Karmarkar]

Edmund,
Thanks a lot for mentioning all those great points. I'll keep this in mind while looking at the raw data. As of now, I'm not collecting any new STA data. I'll be using the data from Ton's paper. We need a noise model in our simulation based on real STA data so as long as there aren't any crazy outliers, we should be good. I'm currently focused on walking, so hopefully it'll be less of an issue now. I'll have to be meticulous when I looked at activities like running!

Ton,
Thanks for clarification regarding the data as well as about your input regarding bone pins. I'm currently carrying out simulations and I've looked at data for walking from the paper you mentioned and it does indeed look great and will be sufficient for our simulation. Thanks a lot once again!

Kind Regards and Thanks.
Vivek