issue with brainsuite and bem ico_downsample

✉️ Mailing List Archive (read-only)
1

External Email - Use Caution

?Hi guys,
I'm having some issues with the BEM meshes for my EEG data. I've tried two approaches:
(1) Using freesurfer watershed and trying different preflood levels, but it seems to do a bad job with the outer_skull.surf being misplaced (see attached image)
(2) I've had better success with using Brainsuite and have followed the guidelines; converting it from dfs to surf using mne_convert_surface, and then reducing the number of triangles to 10,000 using mne_reduce_surface. When I try to run mne.make_bem_mode, I get the error:
File "/usr/anaconda3/envs/mne/lib/python3.6/site-packages/mne/bem.py", line 344, in _ico_downsample

raise RuntimeError(bad_msg)

RuntimeError: A surface with 10000 triangles cannot be isomorphic with a subdivided icosahedron.

Any ideas?

Best, Yvonne
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20190404/d09cdfbe/attachment-0001.html
-------------- next part --------------
A non-text attachment was scrubbed...
Name: src_surf_bem.png
Type: image/png
Size: 201287 bytes
Desc: src_surf_bem.png
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20190404/d09cdfbe/attachment-0001.png

2

External Email - Use Caution

hi Yvonne,

I've never tried this route to BEM model generation. Can you do this
procedure with
MNE sample dataset and share with us the generate .surf files so we can
replicate
and hopefully fix the pb?

ALex

3

External Email - Use Caution

The error is in the downsampling step (we should probably improve the
message). Set `ico=None` and it won't try to downsample your surface:

https://martinos.org/mne/stable/generated/mne.make_bem_model.html

It would be good indeed to see how this works with sample data and see if
it affects source localization -- feel free to open an MNE issue about that.

Eric

4

External Email - Use Caution

I opened an MNE issue for the downsampling problem (https://github.com/mne-tools/mne-python/issues/6127)

When I set the ico=None, I run into a new issue with mne.make_bem_solution after mne.make_bem_model. I don't get this issue with the MNE sample dataset but with my acquired T1's. Any ideas? Thanks for your help guys.

bem = mne.make_bem_solution(surfaces, verbose='INFO')

Approximation method : Linear collocation

Three-layer model surfaces loaded.

Computing the linear collocation solution...

Matrix coefficients...

head (5002) -> head (5002) ...

/dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/mne/bem.py:151: RuntimeWarning: divide by zero encountered in double_scalars

miss /= (4.0 * n_memb)

head (5002) -> outer_skull (5002) ...

head (5002) -> inner_skull (5002) ...

outer_skull (5002) -> head (5002) ...

outer_skull (5002) -> outer_skull (5002) ...

outer_skull (5002) -> inner_skull (5002) ...

inner_skull (5002) -> head (5002) ...

inner_skull (5002) -> outer_skull (5002) ...

inner_skull (5002) -> inner_skull (5002) ...

Inverting the coefficient matrix...

Traceback (most recent call last):

File "<ipython-input-18-24943e2f3a65>", line 1, in <module>

bem = mne.make_bem_solution(surfaces, verbose='INFO')

File "</dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/mne/externals/decorator.py:decorator-gen-39>", line 2, in make_bem_solution

File "/dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/mne/utils/_logging.py", line 88, in wrapper

return function(*args, **kwargs)

File "/dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/mne/bem.py", line 326, in make_bem_solution

_fwd_bem_linear_collocation_solution(bem)

File "/dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/mne/bem.py", line 267, in _fwd_bem_linear_collocation_solution

m['solution'] = _fwd_bem_multi_solution(coeff, m['gamma'], nps)

File "/dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/mne/bem.py", line 225, in _fwd_bem_multi_solution

return linalg.inv(solids, overwrite_a=True)

File "/dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/scipy/linalg/basic.py", line 945, in inv

a1 = _asarray_validated(a, check_finite=check_finite)

File "/dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/scipy/_lib/_util.py", line 239, in _asarray_validated

a = toarray(a)

File "/dagher/dagher6/yyau/toolbox/anaconda3/envs/mne/lib/python3.6/site-packages/numpy/lib/function_base.py", line 498, in asarray_chkfinite

"array must not contain infs or NaNs")

ValueError: array must not contain infs or NaNs

5

External Email - Use Caution

That divide by zero is probably the problem and might be fixable. Can you
share the surfaces on DropBox or some other file hosting service?

Eric

6

External Email - Use Caution

Ahh I see that they are in the GitHub issue. I'll take a look next week if
nobody beats me to it.

Eric