BEM surfaces

Discussions
1

Does anybody knows why am I getting these BEM surfaces? What parameters should I modify so I don’t get that deformation?

I attach the BEM surfaces and the original ones so you can see them and tell me what you think.

Thank you!

Lorena

2

Hi Lorena,

We know that the computation of the BEM surfaces from Brainstorm is not a very stable process. It can sometimes grow the surfaces a little bit outside of the initial head (it’s usually not as bad as what you show though…)

Those initial layers that you show look really nice, but they are way too precise. There is no way a BEM algorithm works with those high-definition surfaces, with the full shape of the skull.

Initially, we developed this tool as a temporary solution, waiting for one of the software specialized in MRI processing (BrainVISA, FreeSurfer, BrainSuite…) to include tools to create BEM-compatible surfaces. I don’t know whether or when this is going to happen…

For now, the two solutions I can offer are:

  1. to play with the parameters of this BEM interface in Brainstorm, starting from minimizing the skull thickness (progressively down to 1mm); if it doesn’t work, you can try larger sets of points for all the layers, hoping that my dirty algorithm gets a little more stable.
  2. to contact the developers of the software you used to generate those nice skull surfaces, and ask them how to use them to generate a BEM model

Cheers,
Francois

3
  1. You can also try to compute an OpenMEEG BEM model using downsampled versions of your initial surfaces, and see what you get…
    It would be amazing if it worked!
4

Hello Francois,

I have already tried computing BEM surfaces with downsampled versions of the initial ones, sadly that didn’t work…
I will try now using a smaller skull thickness, maybe that will work. If it doesn’t, I will try computing the surfaces in other computer with more RAM memory. I think this could work because in the past I could get the BEM surfaces of those same initial surfaces I posted in other computer.
Anyway, I will let you know which solution worked.

Thank you very much for your answer, you have really helped me!

Lorena

5

Hi Francois,

The solution that worked was computing the BEM surfaces in a computer with more RAM memory.

I attach the resulting surfaces, so you can see how they look.

Thank you for your help!

Lorena

6

If you don’t get any “out of memory” error message, it cannot be the amount of RAM that changes the behavior of this algorithm, there has to be something else different between your two computers (or your two databases):

  • the surfaces you used as inputs (the selected scalp / cortex / inner skull surfaces = the ones displayed in green)
  • the matlab version
  • or the brainstorm version
7

Well, the matlab version is newer in the computer with more RAM, everything else is the same. Must be that then what causes those BEM surfaces I obtained first.