iEEG Simulation with MEG Data

Discussions
1

Hello Brainstorm team!

I have been trying to simulate iEEG data using MEG recordings and am not sure if my output is correct or if I am missing something. I performed the following steps after pre-processing:

  • Created the depth electrode implantation scheme in Brainstorm using the post-implantation MRI
  • Generated a 3mm volumetric DUNEuro FEM SEEG head model in the implantation folder
  • Generate a 3mm volumetric DUNEuro FEM MEG head model in the MEG data folder
  • Computed MEG sources (full results, minimum norm, current density, unconstrained)
  • Copied and pasted the full MEG results to the implantation folder and simulated SEEG activity
  • Applied bipolar montage to simulated SEEG activity

I don’t get any error or pop-up messages, but there are quite a few bipolar traces that appear “flat” in comparison to others (please see attached image). I have checked, and they are not actually zero but just very small (~10 orders of magnitude smaller).

When I z-score the data, you can appreciate the wave forms of all the traces (please see attached image). However, I don’t think I should use z-scored data in the next step of my study.

Is this output with “flat” bipolar traces typical for this method? I wanted to seek your guidance before progressing to the next steps.

Any help you can provide would be deeply appreciated!

Ellie

simulation_bipolar
simulation_bipolar1597×742 438 KB

simulation_bipolar_Z_score
simulation_bipolar_Z_score1593×743 460 KB

2

Hi @EHill

This is an interesting approach, thanks for detailing all the steps.

One possible issue could be related to the forward model used for the SEEG computation. In particular, when computing the head model for SEEG, very high leadfield (dipole) values can sometimes occur in regions very close to the SEEG contacts. This is a known effect, and it is usually recommended to refine the FEM mesh locally around the contacts to mitigate it.

As a first check, you could try applying an exclusion zone to the SEEG volume grid:

  • Right-click the leadfield and apply an exclusion zone around the contacts (e.g., start with ~5 mm).

  • Apply the same exclusion zone to the MEG leadfield as well, to keep both models consistent.

  • Then recompute the inverse solution and rerun the simulation.

This approach removes sources located too close to the SEEG contacts (within the 5mm) while preserving the same grid structure across modalities, which may help stabilize the results.

Please let me know if this improve the results.

A+