I am interested in using a customized forward model for EEG source
reconstruction (for example the FEM model available here: http://neuralengr.com/nyhead/)
Does MNE has an utility to create the forward object (as delivered
by mne.read_forward_solution, for example) when I have all the information
regarding lead field matrix, dipole positions etc...)
or I have to have to create this object by hand? (which would be rather
inconvenient, since it seems to be a huge object)
Any ideas about the best way of achieving this are super welcome, thank you!
Does MNE has an utility to create the forward object (as delivered
by mne.read_forward_solution, for example) when I have all the information
regarding lead field matrix, dipole positions etc...)
The Forward object is a thin wrapper around the standard Python dict:
In this sense, you can just create and populate the Forward object in the
same way as a dict.
or I have to have to create this object by hand? (which would be rather
inconvenient, since it seems to be a huge object)
Yep, you'll need to do it by hand. Because the Forward class is so simple,
there probably isn't a great way to simplify this process, beyond providing
documentation. The variables in the Forward class are all there for either
inversion on bookkeeping.
Can you look through the docs, and if you don't see what you need, open an
MNE issue? We can improve the documentation for what all the variables mean
as we go.
At some point we could also try to add support for OpenMEEG
<Issues · mne-tools/mne-python · GitHub, which would give you
more options for forward models, too.