Dear brainstorm experts,
Hellow,
It seems that brainstorms flips the side of subcortical grid points in visualization or labeling in files, when using a mixed model.
For example, in the "surface" tab, I choose the "left (hemisphere)" button, it shows the left cortical hemisphere and some grid points. I build some volume structures (like that of amygdala grid points) and export that to matlab. I also export a "source localization result" file containing "GridAtlas". when checking the indices of the exported amygdala scout, they belong to "Amygdala R" according to GridAtlas.scouts!.
I had also cared this when mapping the results on MRI. what was appearing in left amygdala in brainstorm visualization, appeared on right side in the MRI.
I also checked the coordinates of the subcortical grid points, what was labeled "amygdala R" appeared on right according to the associated GridLoc ...
It is really confusing ...
Would you please tell if the flipping problem is in visualization or in other parts ...?
Can you please assemble an example to share with us?
Right-click on the subject folder > File > Duplicate subject
In the subject copy, delete all the files that are not necessary to reproduce the observations and to prevent any display error. In the anatomy: keep the reference T1 MRI and the "mixed" cortex surface used for the computation. In the functional data, keep only one folder, with the channel file + head model (mixed) + one recordings file (average or epoch) + the mixed source maps associated with it in which you observe these problems.
Make sure you can still produce all the displays in which you reported errors with this subset of files
Right-click on the subject > File > Export subject
Upload the subject.zip file somewhere and post the download link here
Describe here the exact sequence of operations with the Brainstorm and/or the Matlab interface, so that we can reproduce exactly the same observations on our end, and hopefully identify and fix the bugs you are mentioning.
Include screen captures for each final erroneous observation.
Dear Francois,
Thanks a lot for your detailed guideline. Actually, after posting the above bug, I again, tried to investigate the problem, and I am now almost sure that it is a visualization error as I am describing:
The problem is that since the subcortical structures are inside the brain, we can not see them unless we: 1- choose "struct" optin in the "surface" tab, or 2- change the brain's transparency , or 3- choose only one of the hemispheres, or 4- other method that I am not aware of!
The problem is with the third method. When we choose only right hemisphere, we see it, plus some grids in its left. I thought that those grid points belong to the right subcorticals and are moved out of it, in order to facilitate the activity observation. But now I understood that they actually belong to the left hemisphere, just in their true physical position. I am attaching two figures (showing exactly the same thing, but in one, the right hemisphere's transparency is set to 83%) . I am sorry, may be it was my mistake, but I think visualization of one hemisphere with grid points of the opposite subcortical is confusing and it is better when we choose only one hemisphere to observe, the grid points of the opposite side to be removed.
Indeed, the buttons and resection sliders at the bottom of the "Surface" tab apply only to the selected surface, not the source results for mixed head models.
It might not be a very intuitive behavior, but it is not considered a bug: there is no side flip or incorrectly located grid points.
Since we are not actively promoting the use of these mixed source models (see the warnings at the top of the tutorial: http://neuroimage.usc.edu/brainstorm/Tutorials/DeepAtlas), we might not invest much development time in improving the displays associated with it.
But thank you for reporting this unexpected behavior.
Please keep on sharing all your suggestions for improving the software on this forum.
Thanks a lot.
Actually, I suggest you to invest more on the mixed model, Since the mentioned warning is about mapping the subcortical structures to the MNI, the solution to which, might not be very difficult. On the other hand, it has a very large advantage over the FEM method which has much larger number of unknowns in the eeg reverse problem and we know that the main problem with the eeg reverse is that it is ill-posed. Further more, we get really good results (matching the fmri results) using this mixed models even in the group-level. So, please invest on it!