[Mne_analysis] 2.5.2 OS X version, beginner's comments

Yury Petrov y.petrov at neu.edu
Fri Jul 13 18:41:34 EDT 2007
Search archives:

Hi guys,

I am just starting with MNE suite and loved it so far. A couple of  
comments regarding my experience with MNE 2.5.2 on Intel OS X 10.4.10:

Head model preparation:

2. When using mne_surf2bem with .tri meshes I had to specify the  
inner skull radial shift after the outer skin mesh file and vice  
versa, i.e. it looks like the shift options are somehow swapped  
between --id 1 and --id 4. The outer skull shift option works fine.

3. The outer skull mesh in the above routine has to be specified with  
--id 3 instead of --id 2, contrary to what the manual says.

4. Although the mne_surf2bem seems to check for inter-surface min  
distances and containment it did not detect intersections of surfaces  
imported from BrainSuite. Also, it reported de-facto negative  
distances as positive in those instances when the intersections  
occurred.

5. mne_prepare_bem_model exits with an error when head surfaces have  
"flaky" faces, i.e. faces which collapse on one another to form a  
double-faced triangle. This is to be expected, of course, but it  
would be nice if mne_surf2bem checked for such topological defects.  
Such a check is really straightforward: look for vertices which  
belong to < 3 faces.

6. Although displaying electrodes as disks of varying diameter works  
great for alignment purposes, it would be nice to have an option of  
viewing them as small spheres after the alignment is done. Also, see  
the 'EEG electrodes' comment below.


Data analysis:

1. On opening, the width of the mne_analyze window is ~20% larger  
than my screen width (1400).

2. Electrical and magnetic fields could not be displayed for scalp  
meshes imported from BrainSuite. This feature works fine for scalps  
produced with the FreeSurfer watershed algorithm. The message in "A  
problem appeared..." popup window reads: "The source surface has a  
matching number of  triangles but ordering is wrong". As far as I  
could see the problem was not with the cortical (source?) surface,  
but with the scalp surface.

3. What is the functionality of the 'EEG electrodes' Viewer/ 
Options... checkbox in case of EEG-only electrodes? This checkbox  
becomes inactivated in the Viewer/Options menu when I use my own .fif  
file with averaged EEG-only data. I used data.info.chs( ch_num ).kind  
= FIFF.FIFFV_EEG_CH when filling in the Matlab 'ch' structure and  
data.info.dig( ch_num ).kind = FIFF.FIFFV_POINT_EEG when filling in  
the 'dig' structure. Did I miss something else, that tells  
mne_analyze that I am using EEG electrodes? After all the fuss with  
alignment I'd love to see them!

4. The magnetic field lines reappear after a 'Rescale' button press  
even if the 'MEG field map' checkbox is left unchecked.

5. Reload button does not reload surfaces in a consistent fashion. I  
have to restart mne_analyze to really update all surfaces.


Wishes:

1. Be able to translate the picture in the Viewer window.

2. Some way to choose my custom sensor layout as default (why not  
make it default by default?).

3. Auto-scaling of signals in the topographic view (it doesn't work  
for EEG-only data).

4. Electrical potential calculation on other surfaces besides the  
scalp, in particular on the inner skull, aka dura imaging.

5. It would be nice if the mne_forward_solution routine 'knew' about  
the $SUBJECTS_DIR and $SUBJECT variables, the same way that  
mne_do_forward_solution knows about them.


Thank you,

-----------------------------------
Yury Petrov
Assistant Professor
Department of Psychology
Northeastern University
Boston, MA 02115

Phone: 617 373-6054
--------------------------------------





More information about the Mne_analysis mailing list