Using gaitapplication2 and GaitUniMiamiRightLegTD for simulation of left leg

Hello, I have managed to understand how to use the GaitUniMiami application for gait that has full body data. Right now I am trying to use data from a local gait lab, the only problem is that the data has been recorded for the left and right leg individually. I have modified a version of GaitUniMiamiTDRightLeg to fit a left leg, the markers have been set up, I also managed to get gaitapplication2 to run on the dataset, the only problem is that I have bad output of the movement, the leg tends to go too much to the left as if the walking individual was leaning on the right one, this is not conclusive with the recorded data, the markers themselves can be seen walking in a straight line (MarkerModelStudy). I am wondering if the problem is caused by the solver in gaitapplication2, since all markers have been positioned as close as possible to reality. The gaitapplication2 can solve both study types (POS and SCALE), with a tollerance of 1e-8.

Hi MihaiV

Sorry about the slow reply

Is it correctly understood that if you run the model with the markers visble on the human model and at the same time display the free floating markers, there is a significant difference?

If the gaitapplciation did converge and you see significant differences between the markers it sounds as a problem model setup problem in relation to available data.

You say that the recordings have been done individually for the right and left leg, i assume that it is the within the same trial?

When you run the gaitapplication2 is it running on the combined data or on the separate sets?. I am asking because you write that you have created the model based on the GaitUniMiamiRightLeg model ?

The gaitapplication2 should run on both legs simulatenously so in the
GaitStudy.txtr file for the analysis type? it should be
MODELTYPE GAITALLUNKNOWN

but i guess this is already the case?

Sorry about all the questions i just need to make sure i understand the model.

Best regards
Søren

I will answer to each question in sequence so you understand what I am doing with the model.

1)“Is it correctly understood that if you run the model with the markers visble on the human model and at the same time display the free floating markers, there is a significant difference?”

When I run the model there is a very big difference between what the markers in the recording show and the position of the segments. The spine is curved to one side and also the foot tends to go very much to the right.

2)“You say that the recordings have been done individually for the right and left leg, i assume that it is the within the same trial?”

The recordings have been done individually as in one recording for the right leg and then switching the markers to the left and doing one more run. For example, I have data for the right leg that comprises of markers for the sacrum region, asis, thigh, knee and tibia with 3 markers for the leg - ankle, toe, calcaneous - the leg data is for one leg at a time, it is not synchronized.

3)“When you run the gaitapplication2 is it running on the combined data or on the separate sets?. I am asking because you write that you have created the model based on the GaitUniMiamiRightLeg model ?”

As stated before, the sets are different, and I was trying to modify the GaitUniMiamiRightLeg into something that could be called GaitUniMiamiLeftLeg for the left leg recordings. -separate sets-

4)"The gaitapplication2 should run on both legs simulatenously so in the GaitStudy.txtr file for the analysis type? it should be MODELTYPE GAITALLUNKNOWNbut i guess this is already the case?"

I have tried using MODELTYPE [FONT=Verdana]RIGHTLEGTWOREVANKLE for the modified application that has only the left leg.[/FONT]
[FONT=Verdana][/FONT]
[FONT=Verdana]I figured that if I specify segments as required and have the markers themselves attached to the left side I would get good results since the solver optimises the movement regardless of what the marker positions where in the trial. [/FONT]
[FONT=Verdana][/FONT]
[FONT=Verdana]My question was if it would be possible to use the gaitapplication2 for such a model (transform GaitUniMiamiTDRightLeg so it works on left leg data).[/FONT]
[FONT=Verdana][/FONT]
[FONT=Verdana]Thank you for your reply,[/FONT]
[FONT=Verdana][/FONT]
[FONT=Verdana]Mihai.[/FONT]

Hi Mihai

I think i understand what you are doing better now, here are some comments:

I do not think it will be very dfficult to combine measurements from different trials into one model. This will lead to multiple syncronization problems both interms of motion but also the loading.

The gaitapplication contains hardcoded kinematic models of the lower extremity, the models i know is available: is the lower extremity model with two legs and the right leg model.

It is not possible to apply a left leg dataset to the rightleg model, this will not work.

So with the data you have available i would recommend to run a rightleg model only.

The fortcomming release of Ver.4.1 will make this type of problems much much simpler to solve. It can all be done inside AnyBody, and the methods used to solve the overdeterminancy can be applied to any of the availble body models.

Best regards
Søren

Thank you for the info and the support. Now I understand this problem better and will continue using only the right leg model.

That sounds good

I read my last message again and it should ofcourse have been:

“I think it will be very dfficult to combine measurements from different trials into one model. This will lead to multiple syncronization problems both interms of motion but also the loading.”

But i think you understood this already.

Best regards
Søren