again mne_surf2_bem problem

Dear all,

I am in the process of aligning MRI-MEG, and for some of my participants
(3 out of 22), I failed to create the output file with mne_surf2bem

below is the message error

Hi Elisabeth,

Dear all,

I am in the process of aligning MRI-MEG, and for some of my
participants (3 out of 22), I failed to create the output file with
mne_surf2bem

below is the message error

*****************
Running mne_surf2bem on l18
mne_surf2bem version 1.4 compiled at Apr 5 2007 16:04:15
input file # 1 : /imaging/ef02/lexpro/subject/0327/surf/
lh.smseghead / id = 4 / sigma N/A
output file : /imaging/ef02/lexpro/subject/0327/bem/0327-
head.fif
New quad file : nvert = 193696 nquad = 193927
      Triangle and vertex normals and neighboring triangles...[done]
      Vertex neighbors...Too many neighbors for vertex 94524.
******************

One of my colleague here told me that it happened also for her and
then she tried to re-run the mkheadsurf command
and then the mne_surf2_bem

I tryied that with no luck
I also try to re-run the last step of recon-all -autorecon3 on
freesurfer to be sure that the output was not corrupted, again no
success

Any advices would be appreciated

MNE software is much quite picky about the correct topology of the
surfaces and the mkheadsurf script is not so careful about it because
tksurfer can work around these defects much better than MNE can.

A solution to this problem is to get the latest MNE Matlab toolbox
from the download web site. There is a function mne_reduce_surface,
which can reduce the number of triangles on a surface and in the
process the topology defects are (often) fixed. I have used 20000
triangles on the reduced surface with good results:

mne_reduce_surface('lh.smseghead',20000,'lh.smseghead.red');

Naturally, you need to be in the surf directory of the subject to do
this and use the 'red' surface as input to mne_reduce_surface.

- Matti