I am currently using a MoCap model to simulate walking trials with my c3d dataset derived from a motion capture system. As first step I run the RunParameterIdentification operation. The output gives me well aligned markers (AnyBody markers (red) and patient markers (blue)) but they are placed on the wrong place. For example, the first marker on the spine from the top should be the C7 and the second at T3 (see attached picture). Do you know what’s the cause of it or do you have any suggestions about what to do to fix this situation and have markers with a good alignment and position?
My guess would be that the trunk height is getting optimized as well as the vertical position of the marker. Assuming your marker placements were correct (during the trial and in the model), what you would see now is a model with a rather small trunk height. Is that so?
But, I also have my doubts about the location of the markers during the trials. The blue marker on the patient look a little bit strange: It looks to me that the T3 marker is placed above the shoulder markers. That should normally not be the case. Unless, it's just the viewpoint of the image that is making it look like that.
Thank you for your reply, but I think I didn’t explain the problem properly.
From what I see the problem here is not about the reduced height of the trunk. Is about the markers placement, as you can see the markers of C7 and T3 are too close to each other and are placed too high as well. So in my opinion the problem is that the markers are placed in the wrong spot and I don’t know how to place them correctly. Do you have any suggestion?
Thank you in advance!
The blue markers that you see in the model are representing the raw data coming from your c3d file. This is why I thought there is some issue in the marker placement during the experiment, where T3 and C7 markers are located too high on the subject (when you compare to the shoulder markers). Or they are mislabeled as T3 and C7 and should be some other landmarks.
I agree with you that the markers are placed too high, but I think the problem occurred during the experiment. I would really like you to look into your raw data and check if the T3 and C7 markers were placed on the subject correctly. If the issue is in the raw data, then, it can be difficult to fix in the model.
You can work around the issue in the model. In the model, you can look at the MarkerProtocol.any that you can find through LabSpecificData.any. You can read more about it over here. You can also read a little bit more in the class template CreateMarkerDriver about the options available to you regarding marker placement, weights, optimization direction, etc.
So you can move the marker position in the model. It can be placed on a specific segment and you can also provide some offset distance from a reference frame on the segment. (You may want to think about placing the markers on another segment and not T3 and C7). Or, you can change the weight of the specific markers in the optimization problem. Or, you can completely exclude the markers from the model. It's hard to say what's the best way till I have a complete understanding of the problem.
Sorry for my late reply. I tried to do a posture calibration with the X-Ray data I have of the spine. It consists in a sagittal alignment considering the centroids calibration and vertebrae orientation. After it the spine has the patient’s specific shape and at that moment, I run the RunAnalysis (load parameters, marker tracking, inverse dynamic) of my MoCap model which performs a gait. With the beginning of marker tracking the markers play a role on the shape of the spine and the sagittal alignment is no more maintained. I tried to play a bit with markers weight and opt directions in order to have a spinal shape similar to the calibrated one but I’m not sure that set weights = 0 or almost is the right way to work. Do you have any suggestion of what to do in my situation and how can I deal with markers and calibration without altering too much the spine shape but considering the markers for the motion?
I assume that you are working with the thoracic model, since you have the upper spine markers included. I think, the problem with your model is that you probably have one height variable to optimize for the whole spine and many markers (markers on the spine, shoulders, and maybe head) to optimize this one variable. So you will not get a result that all the blue and red ones are on top of each other. However, I think it is smart to use the Xrays as well, as they can help account for marker placement errors.
I suggest that you use the xray data as soft drivers and keep them in the whole marker tracking process. Then you can play with the weights of these soft drivers and also the weights of the markers to get the best out of it.
And if you have also AP xrays, I think you can use them too.
