I am using mne_analyze to align MEG/MRI coordinates in accordance with the
instructions at http://www.martinos.org/mne/manual/sampledata.html#chdijbig.
However I am having an issue with some of the way my inflated surfaces
appear when displayed in the viewer. I've attached two examples of some
particularly bad ones, and the third attachment is an example of what the
inflated surface looks like ordinarily. This third image is the same
surface as the first one that looks like a block. At what point in
processing is my workflow going wrong? Thanks in advance.
I assume this problem happens with your own data not for with the sample
data set, correct? Did you check the FreeSurfer segmentation? It looks
to me like FreeSurfer failed for some reason. Here are some instructions
on how to check the FS output:
The BEM surface segmentation using mne_watershed_bem has gone wrong. There is most probably a way the avoid this. There are new options --gcatlas and --preflood <number> for mne_watershed_bem in the nightly build.
I do not recall which one of them helps, I hope someone at the Martinos can help you.
Using the -gcatlas option with mne_watershed_bem successfully resolved this
issue; thanks for the helpful responses. Also, I just wanted to suggest
packaging libgfortran.so.1 with new MNE installations because it seems to
cause issues with many users who have newer versions of the library already
installed. Thanks again!
I alsoe head a few cases with quite distorted skulls so played a bit with the gcaatlas option.
Could it be that the watershed routine expects a particular version of the atlas:
...
Mode: Atlas analysis
Mode: T1 normalized volume
Mode: Use the information of atlas (default parms, --help for details)
GCAread: could not open GCA /localdata/freesurfer/average/RB_all_withskull_2007-08-08.gca for reading
No such file or directory
...
It seems I've got to link the exisiting RB_all_withskull_2008-03-26.gca agains the expected name.
Or it should also be possible to simply pass the gcaatlas name to --atlas on invoking mne_watershed_bem, right?