Question about aligning MRI and surfaces created by freesurfer

Discussions
1

Hello,Francois
I created head_outer_skull_surface and LH/RH hemisphere pial surace by using freesurfer. after that,I imported them and t1-MRI together to brainstorm. but I found head/corex surfaces are misaligned with MRI. Can you help me out for this issue?

Many thanks
Jun

2

Hi,Francois
Thanks for you quick reply. I think I already figured it out,here are what I did (1) using outer-skin surface instead of outer-skull surface (2) changed the surface name to “lh_skin” so brainstorm can align it when loading.
everything looks fine right now. but I still have one question related with alignment. Tutorial said "Brainstorm registers them with the MRI using its fiducials points (NAS, LPA, RPA). " and I understand for MRI,I marked fiducial points. but for head surface,I didn’t do that. how braintorm align them together?
also,if I want to average source estimation across subjects and displaying averaing source, should I averge sources directly,or like some posts in forum mentioned, project individual sources to MNI 127 brain first,then average?
Can you help me understanding these issues?
many thanks

Jun Wang

3

Hi Jun,

Indeed you should use the “outer_skin” as the head envelope, when working with FreeSurfer surfaces. But should not have to rename it to get it detected as a “head” surface. Can you tell me the name of the file? I will check if there is a bug somewhere…

The surfaces where extracted using the MRI. When we know what software created the meshes, it is generally possible to guess the transformation MRI coordinates<->Surface coordinates directly from the surface files. So, no need to define the fiducials on the surfaces, we use the ones from the MRI.
You can check the alignment surfaces/MRI as explained in the tutorial pages.

If you want to average the source estimates across different subjects, you have two options:

  • Do the source reconstruction for all the subjects using the same default MNI anatomy; then you can average directly the sources files across subjects
  • Do the source reconstruction on each individual anatomy; but then you need to re-project them on the same default anatomy before averaging them. You can average time series only when they are in the same amount (same number of vertices) and when they belong to dipoles located at the same exact position.

We did not formally compare the two methods yet, and did not have time to fully test the accuracy of the re-projection on the MNI brain. So I cannot prove that the reconstruction on the individual brains + projection is better than the reconstruction on the default brain.
Those questions will probably find answers next year. For now, you can try on a few subjects and see what method gives you the better results.

Let me know if you have further questions.
Francois

4

Thanks,Francois
This is very helpful. and the file name is “lh.testhead_outer_skin_surface”. and I have troubles when loading pohelums file(temp.txt in attachment,you need to rename to temp.dat) exported from neuroscan. When I exported and used “elp” file instead,everything is fine,but elp file doesn’t include surface point I collected.any thought about that? many thanks,Jun

5

Hi Jun,
So you can import properly your file but you just have to rename it from .txt to .dat? Is that correct?
If so, I fixed your problem. I updated the function import_channels.m so that you can select .txt files for the Neuroscan format, and they are read exactly like the .dat files.
Does that solve your problems?
Francois

6

not really,the reason I uploaded ‘temp.txt’ instead of ‘temp.dat’ is the forum only allows me to upload txt file.
my problem is when I loaded “temp.dat”(neuroscan format,64 sensors and sevral surface points) to brainstorm,the sensors are in wrong position. however,when I loaded “temp.elp”(with same 64 sensors but no surface points),everything is fine. dont’ know what happened,seems brainstorm has a problem to read neuroscan dat file format.

thanks

7

Hello!

I am trying to import the scalp layer of the bem model created in Freesurfer during the call to "mri_watershed -surf" but when I import it in Brainstorm using "Import Surface" it is out of alignment with the MRI and the other surfaces. There was no registration sphere generated that matches it and I'm a bit confused about how this works between Freesurfer and Brainstorm. Any guidance would be very much appreciated.

Thanks!
Brendan