Format of MATLAB mne_read_stc_file output

✉️ Mailing List Archive (read-only)
1

Hello,

I am looking for some clarification on the format of the output of
mne_read_stc_file.m . For example, when I run it on one of my files (a left
hemisphere .stc file which was computed morphed to an average brain), I get
a file with 10242 vertices:

        tmin: -0.8000
       tstep: 0.0017
    vertices: [10242x1 uint32]
        data: [10242x780 double]

However, when I read the source space that corresponds to this .stc file (a
source space that was made using --ico 4), it only has 2562 source vertices
per hemisphere (from src(1).vertno):

              id: 101
              np: 163842
            ntri: 327680
     coord_frame: 5
              rr: [163842x3 double]
              nn: [163842x3 double]
            tris: [327680x3 int32]
            nuse: 2562
           inuse: [1x163842 int32]
          vertno: [1x2562 int32]
        nuse_tri: 5120
        use_tris: [5120x3 int32]
         nearest: []
    nearest_dist: []
           pinfo: []
            dist: []
      dist_limit: []

So basically I am just not sure what the 10242 vertices read from my .stc
file correspond to, as that number doesn't correspond to anything I see in
the source space. Based on my reading of lines 40-45 of
mne_label_time_courses.m, it looks to me like the first 2562 vertices in
the stc file correspond to the actual vertices from my source space,
whereas the rest are ignored --- is that true? (The reason I am interested
in this is because rather than reading in label time courses, I want to try
reading in just the time courses that correspond to the used vertices in
the source space, and then use the triangle information from src.use_tris
and src.tris to identify source vertices that are adjacent to each other
for the purpose of spatio-temporal clustering. In that case, would
stc.data( source(1).vertno, : ) be the way to get just the timecourses
corresponding to source vertices?)

Thank you,
Steve

Stephen Politzer-Ahles
New York University, Abu Dhabi
Psychology Department
http://files.nyu.edu/spa268/public
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20130811/2b8766b9/attachment.html

2

Hey Steve,

Once you morph the individual subject's data to the average brain
"fsaverage", the source space points in the STC will correspond to those in
fsaverage's source space, which is usually the standard grade-5 decimation
with 10242 vertices per hemisphere. The vertices used in fsaverage are
conveniently just the first 10242 vertices in each hemisphere (1:10242 in
1-based indexing). Read fsaverage's source space and it should make sense.
If on the other hand you morphed to some other average brain (not
fsaverage), the vertices might correspond to some other set of indices, but
they probably still correspond to a grade-5 decimation.

Cheers,
Eric

3

Hi Steve,
   The standard fsaverage surface that is included has 10242 vertices per
hemisphere. The stc files do not retain any information about the
original source-space to which those time-courses belong (since they
are just overlays and you could overlay many different things). They
only have vertex numbers and the corresponding time courses (i.e data).
It is upto to the user to pair the right stc file with the right
surface.

HTH,
Hari