Dear all,
I am trying to apply my forward model (MNE unconstrained, mixed) to project my different scouts back out to sensor space. My end goal would be to see the simulated 3D sensor topographies to visually assess the topography if only my main scouts were selected.
The main scouts I would like to project are primary auditory cortex left and right. However, when I try to simulate the recordings for both of them together is giving me the following error:
bst_simulation: The selected scout spans over multiple regions of the 'Source model' atlas.
If I simulate them right and left separately it works, but I don't know then how to see a 3D sensor topography including both.
Is there any way I could simulate them together?
Many thanks,
Natàlia.
You need to use two separate scouts instead of only one.
And generate twice the same signal for the two regions.
Thank you very much Francois,
I already had the two separate scouts, so I will simulate the MEG signal for each of the two regions (left and right) separately.
In order to see a joined topography of the two simulations, should I just average the two simulations to obtain one shared one? I am not sure this is the correct approach...
No, you should simulate the recordings for the two scouts simultaneously. You need to have two scouts available and match it with a "matrix" file with
Please post screen captures illustrating why you don't manage to do it.
Here is a screenshot of what I am trying to do. I am convinced I am doing something wrong, but I can't manage to find what it is exactly.
So I have the source file created. What I am doing is I open it, Select the two scouts I want to simulate and then going to Sources>Simulate Recording. In the screenshot you can see the error that it gives me.
If I select only one of the scouts, it simulates the recording without problem, but if I select both sides (Right and Left), it gives me an error.
Regarding my source model atlas, it is a mixed model that includes Brainstem, right and left Thalamus and right and left cortex. Could it be a problem with this, should I merge Right and Left Cortex in my source model?
Many thanks,
Natàlia.
Your matlab console also gives an error that you have more scouts you want to simulate from than scouts. If you have 1 signal, which will be shown in brainstorm as signal name (1x#samples) and you want to simulate that from 2 scouts it will error.
If you want to simulate from 2 scouts you will need two rows in your simulated signal. If you want it for both scouts the same signal you can add: Data(2,: ) = Data(1,: ) at the end of the signal generation code, like the following :
Kind regards,
Steven
Dear Steven,
I am aware of that error and I already solved it. It came out because I tried something that made no sense and, of course, matlab complained.
But you are right and my error comes from the input signal I am trying to simulate. My input signal is a grandaverage of many trials. From this grandaverage, I computed the sources and defined the scouts. What I want to do is simulate the activity recorded in the sensors if only some of the scouts where activated. Which should it be the input signal?
Kind regards,
Natàlia.
Dear Natalia,
My apologies, I didn't see that that was already solved. I honestly don't know a direct answer to that question so I will leave that up to the professionals.
Kind regards,
Steven
Many thanks for answering anyway, at this point any help is appreciated!
The problem comes from the fact that you are using a mixed head model.
These models are complicated to handle, and some features are not fully supported for them.
When calling this function bst_simulation, all the scouts are grouped together. If not all the vertices fall in the same "region" in the "source model atlas", then this is a complex case that is not handled by bst_simulation. It would require a lot of work for very few use cases, so not a priority for the developments at the moment, sorry.
One easy solution would be to use a regular surface head model instead, or a volume head model.
Thank you very much. I wrote to @ecoffey to see if she can help me, as I know she did this particular analysis previously. As soon as I find a solution I'll let you know.
Thanks again.
Francois,
I am trying to implement your solution and I realized that I need to modify the source model so it only contains cortical structures. Is it possible to have two different source models or do I have to delete/modify the one I already have?
If I modify the source model, the created ROIs will not be deleted right?
Many thanks,
Natàlia.
In the anatomy folder, you can select the low-resolution cortex surface instead of the cortex+subcortical, then compute a new forward model, and a new inverse model.
Nothing will be lost, but since the scouts are attached to a given surface, you would have to create again the scouts. Same thing if you decide to work with a volume head model.
https://neuroimage.usc.edu/brainstorm/Tutorials/HeadModel
https://neuroimage.usc.edu/brainstorm/Tutorials/TutVolSource
