[VoxBo] Area of activation expands after GLM

[Daniel Y Kimberg]

Joon, hi.  The first thing I'd suggest is that you make sure you're
looking at the same volume the GLM thought it was looking at.  In the
GLM directory there should be a .sub file that lists the tes files.
Those files should have a brain volume very similar to what's in the
prm file (not exactly the same when there's head movement, since we
use an intersection mask, but the prm should never include more voxels
than you see in a given tes file).  If you look at the two and they
still don't match, I'd suggest re-running just that GLM, in case there
was some issue with multiple jobs working with the same files at the
same time.

Normally you tend to get this kind of effect with either explicit
smoothing or the bit of smoothing that you get with resampling (e.g.,
in realignment).  But the prm file mask should still generally
resemble what you see in the tes files you're modeling, except when
there's lots of head movement.

> This is contrary to my understanding that glm calculation is voxel-based
> and therefore the activation area should not be expanded or contracted.
>
> Am I missunderstanding here or is something wrong with my glm calculation (I
> used vbmakeglm)?
> Or is there something beyond?

No, it's all voxel-based and the prm mask should be exactly the
intersection mask of your tes files' masks.  If that's not the case,
then something's gone wrong, probably something trivial (or you may be
looking at the wrong tes files).

dan