FMEG FORWARD Modeling

[tmedani]

Hello,

@Francois

I am starting with new project based on patent technology the OPM -based magnetoencephalography ( MEG), and the team would like to do every thing in brainstorm. I would like first to ask if someone solve the forward model of FMEG here so can get the leadfield matrix between the fetus cortex and electrode located on the abdominal skin of the pregnant women.

I guess you mean "leadfield matrix between the fetus cortex and OPM sensors ", not "electrode", right?
I don't know if the MEG forward models in Brainstorm can be transposed to OPMs. Let me ask the experts, we'll get back to you.
@John_Mosher @Sylvain @tmedani @juangpc @labyt

From my side, I think brainstorm LF computation is possible between the fetus cortex and OPM sensors , the principle is still the same.
However, I'm not sure that we have a model of the OPM sensors (file format and integration points). Also, the full geometry model of the fetus should be constructed outside of brainstorm.

@Francois

Do you have any suggestion how to do it, in fact there Femonum project that can provide the meshes of the fetus, uteus, abdominal skin but they did not provide the MRI, I think it is possible to work with meshes with brinstorm, but the broblem is they did not provide the mesh of the cortex of the fetus
Let say we have all the meshes and we want to import the meshes now to brainstorm, infact we have lot of surfaces more than 5. what is the maximum domain or surfaces openmmeeg in the brainstorm can solve?

All these tissues should not affect much the MEG leadfields. Having only a few surfaces could be enough, maybe the abdominal skin and the fetus head.

From the different compartments, you could create a "fake MRI", similarly to what we did with our MEG phantom tutorials: https://neuroimage.usc.edu/brainstorm/Tutorials/PhantomCtf#Generate_anatomy

Then for the source model, maybe you could use a volume grid as a first attempt:
https://neuroimage.usc.edu/brainstorm/Tutorials/TutVolSource

We currently have new ongoing developments using FEM models, these could be helpful in this context as well. Documentation is not finished, but here is a draft so you can have an idea of what we are doing: https://neuroimage.usc.edu/brainstorm/Duneuro

To add more infos to Francois's answer.
You can use these available meshes and build your complete model and then import the final model to brainstorm. You can also merge/modify and align all these mesh surfaces from the brainstorm interface.

Regarding the cortex mesh of the fetus, from what I know there are not available models, but you can approximate it by one model form here: https://www.pedeheadmod.net/
I
also here is an interesting reference: Effects of sutures and fontanels on MEG and EEG source analysis in a realistic infant head model - PubMed (infant head model and the effect of the fontanels)

When you have all the geometry and channels models available, you will need the conductivities values of the tissues in order to compute the LF.

For the computation, you can use either the Opnemeeg or Duneuro. For openmeeg I think you can use up to 3 or 4 layers (check here: Tutorial — OpenMEEG). With FEM Duneuro, you can use as much as you want, but the recommended number is 5/6 layers. The effects of the tissues are discussed here : A guideline for head volume conductor modeling in EEG and MEG - PubMed

If you want to use the Duneuro, you can easily generate the FEM mesh from the surface with brainstorm.

For the layers to use, it depends, the EEG is more sensitive to the number of layers and theire conductivities than the MEG (refer to the previous Lew paper link).

I hope this may help.
Best
Takfarinas