Source points omitted with mne_forward_solution

Hi MNE users,

I am getting some source points omitted when running mne_forward_model and
I was wondering if anyone knows the possible reasons for this to be
happening. I run the mne_watershead_bem to get the BEM model but also used
seglab to get a inner_skull.tri file.

Somehow the sources are crossing the inner_skull boundary. Where do you
recommend me to focus in order to understand what really is causing this
problem? In my further analysis I will need the sources neighbors so
missing sources is a problem.

- Here is the code:

mne_forward_solution --src
$SUBJECTS_DIR/$SUBJECT/bem/structural_mri-ico-5-src.fif --mri
$SUBJECTS_DIR/$SUBJECT/mri/T1-neuromag/sets/COR-asalazar-090406-142236.fif
--meas $SUBJECTS_DIR/$SUBJECT/eeg_mri/base/dec_base_raw-ave.fif --bem
$SUBJECTS_DIR/$SUBJECT/bem/structural_mri-5120-5120-5120-bem-sol.fif --fwd
dec_base_raw-ave-ico-5-fwd.fif --eeg --mindist 5 --fixed

- I am getting this 30 sources omitted:

Source spaces are in head coordinates.
Checking that the sources are inside the inner skull and at least 5.0
mm away (will take a few...)
14 source space points omitted because they are outside the inner skull
surface.
715 source space points omitted because of the 5.0-mm distance limit.
16 source space points omitted because they are outside the inner skull
surface.
788 source space points omitted because of the 5.0-mm distance limit.
Thank you for waiting.

Thank you in advance for any tips or ideas,

Hi Andres,

It is because of the --mindist 5 option. This means that any points
closer than 5 mm to inner skull surface are omitted. The rationale
behind this is (or used to be) that the BEM forward solution gets
inaccurate when the sources are close to the inner skull surface. With
the now standard 5120-triangle surfaces it is most probably safe to
omit the --mindist 5 option.

- Matti

Thank you Matti,

Unfortunately, even when I do not use the --mindist option I still get
these sources omitted.
I want to be sure the whole pipeline of process I have done are correct,
to be confident the inner_skull, outer_skull and outer_skin files are
correct (although they were checked on tkmedit) and that the sources are
well placed (this was checked using MRILab).

Can the lack of these omitted sources affect a source localization results?

Also, in case anyone has done this before, does MNE has any tool to
convert the sources (*.pnt files) to a format tkmedit can read in order to
plot the three boundaries (inner_skin, outer_skull, inner_skull) + the
sources? I want to see the sources that are crossing the boundary (in case
this can be visible)

Thanks again for you help,

-- Andr?s

Hi Andres,

It is because of the --mindist 5 option. This means that any points

closer than 5 mm to inner skull surface are omitted. The rationale
behind this is (or used to be) that the BEM forward solution gets
inaccurate when the sources are close to the inner skull surface. With
the now standard 5120-triangle surfaces it is most probably safe to omit
the --mindist 5 option.

- Matti

Hi MNE users,
I am getting some source points omitted when running
mne_forward_model and
I was wondering if anyone knows the possible reasons for this to be

happening. I run the mne_watershead_bem to get the BEM model but also
used

seglab to get a inner_skull.tri file.
Somehow the sources are crossing the inner_skull boundary. Where do you
recommend me to focus in order to understand what really is causing this
problem? In my further analysis I will need the sources neighbors so

missing sources is a problem.

- Here is the code:
mne_forward_solution --src
$SUBJECTS_DIR/$SUBJECT/bem/structural_mri-ico-5-src.fif --mri
$SUBJECTS_DIR/$SUBJECT/mri/T1-neuromag/sets/COR-
asalazar-090406-142236.fif
--meas $SUBJECTS_DIR/$SUBJECT/eeg_mri/base/dec_base_raw-ave.fif --bem

$SUBJECTS_DIR/$SUBJECT/bem/structural_mri-5120-5120-5120-bem-sol.fif
--fwd

The remaining omitted points are outside the inner skull and nothing
can be done about that. Omission should not be a serious problem.

You can plot the BEM surface vertices in MRIlab as well, see manual
Section 3.7.

- Matti

Good evening MNE users,

I want to create a set of simulations of EEG data on a specific label in
the cortex.
I have been informed of the importance of the localization of this label
and how this influences in some way the results.

Could anyone recommend me specific zones I can focused on or give me some
tips on choosing them (like avoiding grooves or similar), to allow me to
have reliable EEG simulated signals for source localization?

Thank you all for your help.