How to create a sphere file to compute source space?

✉️ Mailing List Archive (read-only)
1

Hello !

I would like to create a forward solution using source space
(mne.setup_source_space).

I?ve created the surface mesh from Brainvisa but I don?t have the sphere
file (.sphere) to compute a source space.

How can I create it in MNE ? I?m thinking to use mne.make_sphere_model but
I?m not sure how it works.

Any help or suggestion would be appreciated!

Thanks.

Alexandre F
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20160502/6f68fe03/attachment.html

2

hi,

I would like to create a forward solution using source space
(mne.setup_source_space).

I?ve created the surface mesh from Brainvisa but I don?t have the sphere
file (.sphere) to compute a source space.

you cannot generate a source space with MNE for a surface not produced
by freesurfer.
at least not easily without hacking the MNE data structures.

How can I create it in MNE ? I?m thinking to use mne.make_sphere_model but
I?m not sure how it works.

the make_sphere_model is for the forward model and is unrelated with
the freesurfer .sphere files that are used for source space
subsampling.

I really recommend you use freesurfer to generate your cortical
surfaces. Unless it does not work it will make your life much easier
with MNE.

HTH
Alex

3

Dear Alexandre,
thanks for your feedback. What we are trying to do with Alexandre
(Fabre), who posted the first email, is to create a source space using
the cortical meshes from Brainvisa.
Today, we managed to create a .sphere file in Brainvisa that maps each
node of the Lwhite mesh onto the sphere. This seems equivalent to the
procedure done in Freesurfer. Both have approximately 50k nodes per
hemisphere. What we are tying to deal now is to create the source space by :
1) downsample the mesh to 10k or 5k nodes
2) avoid some nodes which are not on the cortical sheet. For example, in
Brainvisa some nodes are in the cingular pole, where we don't want
source reconstruction from. We do have a "texture" for each node of the
mesh that indicates the brain region according to MarsAtlas (the atlas
we use in Brainvisa).

To do so, we were thinking to use mne.setup_source_space. Any suggestion
as you said to "hack" this functiona are welcome.
If it is not possible, I guess we will have to modify the function.
Thanks a lot,
Andrea

4

hi Andrea,

thinking about it what you can do is to setup a discrete source space
using the locations and normals for your cortical mesh.

See setup_volume_source_space and the pos parameter.
You'll have a source space that you can then use for forward/inverse modeling.

Let me know it this works.

Alex