Hello everyone,
I’m currently using an FEM head model to compute a leadfield matrix within the volume. Here’s a brief outline of my pipeline:
- Use T1+T2 charm in SIMNIBS for head modeling.
- Manually correct mis-labeled voxels, then remesh.
- In Brainstorm, use “Import anatomy folder” to import the T1, T2, and the FEM model.
- Convert DWI to DTI to import the B1000, 64D DWI images.
- Merge layers (into 5 layers).
- Compute FEM tensors.
- In the “Functional data” tab, within the individual channel file, select Compute head model → MRI volume, EEG, DUNEuro FEM.
When I inspect the results by selecting “View EEG leadfield sensitivity (3D MRI),” I see the image below.
It appears that four voxels in the occipital region have abnormally high values, which I believe indicates an error leading to incorrect calculations.
If I apply an EEG leadfield exclusion zone of about 20 mm, those voxels are removed from the leadfield, but several additional marginal cortical voxels are also removed.
I’d appreciate any suggestions on which part of my pipeline might require modifications or if there are alternative solutions.
Thank you!
This is figure after apply an EEG leadfield exclusion zone of about 20 mm
Hi @bong516612
It's great to hear that you were able to reproduce most of the steps with your data.
Regarding the exclusion zone, it is mainly recommended for iEEG cases.
However, you can still use it for EEG—especially if you're working with high-density EEG data, as this ensures a more homogeneous exclusion zone around brain areas.
Another possible solution is to try a different FEM source model. In the Duneuro FEM panel, I assume you selected the Venant option; you could try switching to the Partial Integration method and check whether the lead field becomes more stable.
You can also experiment with the definition of the source space. For instance, you can generate a smaller cortex envelope and use it to define a new volume source space grid.
To do this: [check the attached video]
- In the Anatomy panel, right-click on the cortex file → select Copy (Ctrl+C), then Paste (Ctrl+V).
- Right-click on the new cortex copy and choose Align Manually on..., then select the original cortex file.
- A panel will appear, allowing you to manipulate the cortex surface. Use the Resize option to make it slightly smaller (with right click of the mouse), and then move it slightly on +Z direction.
- Once satisfied with the results, save the changes [ DO NOT APPLY ALL SURFACE ==> click NO] and use this new cortex as the default before starting the process to generate the new volume grid.
By the way, @Raymundo.Cassani, it would be great to have a "Duplicate" option when right-clicking on a node. That could be very useful for both anatomical and functional items.
