I am trying to plot a mesh of points from a sourcespace (e.g.,
$SUBJECTS_DIR/$SUBJECT/bem/$SUBJECT-ico-4-src.fif) together with the points
in the label file. (the reason is because we made some custom labels using
a script, and they turned out looking patchy–see http://imgur.com/bUAuaJH--so I want to check whether these labels really
cover the whole regions they're supposed to, or whether there are
sources/vertices that should be in those labels but aren't.) However, when
I do this, the locations of the vertices from the label file don't quite
line up with vertices from the source space (see http://i.imgur.com/IGlKB9T.png and http://i.imgur.com/CUj6cF7.png for
comparisons; the green vertices from the label file are in more or less the
right place [what's plotted here is Brodmann area 46], but don't quite line
up with the black vertices from the source space), which I was surprised
by, since I thought the locations of vertices in a label file came directly
from the source space. What I was doing was reading the -ico-4-src.fif file
into MATLAB using the FieldTrip toolbox, and reading a label file in
directly, and then plotting them together.
So I am just wondering if there is a different file I should be using to
plot the locations of all the vertices in my sourcespace that labels could
have come from? Otherwise I guess it's a problem with how we created the
labels.
Offhand I see a couple possibilities. If you're referencing by vertex
numbers, you might be hitting off-by-one errors somewhere, since MATLAB
uses 1-based indexing, and surfaces / vertices are 0-index (C) based. If
you upload the source space file and label file somewhere, someone (maybe
me) might be able to track down what's going on. For what it's worth, if
you have a working mne-python install, it uses 0-based indexing, and has
useful functions for plotting brains with labels and the like. By OBO
errors in my experience make usually result in worse errors, so that's
probably not it.
If you're getting the positions directly from the label file, one thing you
might want to check is the positions you get if you take the vertices
referenced in the label file and use those to get positions in the source
space. (In this case, be careful about the indexing...) Label files contain
position values (what I have a feeling you're using to make the plots), but
they also contain vertex numbers that those are meant to correspond to in
source space. You might be hitting a problem where the label locations are
pulled from one surface (e.g., pial), while the source space is defined
over another (e.g., orig). I usually do my source-space manipulations using
vertex numbers (instead of positions) to avoid this issue completely, as it
treats all surfaces (e.g., pial, orig, sphere) equally.
Thanks, that solved it! It was indeed the second issue; plotting my label
by using the vertex numbers instead of the positions from the label file
worked perfectly.
