[VoxBo] idl, vbview, and single-user VoxBo
Daniel Y Kimberg
kimberg at mail.med.upenn.edu
Wed Dec 5 11:30:53 EST 2007
Katharina Spalek wrote:
> Desperately, I'm now trying to do the analyses with single-user VoxBo,
> using my copy of the data. Some things are trickier than at the CfN,
> because I have neither an IDL nor a matlab-license. However, I believe
> that Dan tried to replace the functions of the IDL voxel surfer with
> vbview, and hence, I was wondering if I can do the things I have to do
> without IDL.
>
> Here's what I need to do and how I would have done it with the voxel
> surfer: I want to save functional regions of interest based on the
> activation of all experimental conditions compared to all control
> conditions. In IDL, I would have: Loaded an nDisplay.cub. Overlayed it
> with a foo.prm file (from an rfx-analysis). Thresholded it with the t
> for p < .05, as calculated in a permutation analysis. Selected all
> contiguous voxels within a region and saved them as a mask. Done the
> same for the next region of contiguous voxels. (And afterwards, run the
> vbdumpstats command to look at each participant's specific values within
> these regions of interest).
I'll assume you already have your permutation threshold, so the real
issue is how to convert your stat map into a bunch of masks. This is
not incredibly elegant in vbview, but I believe it works fine in the
most recently released version. Once you've thresholded the stat map,
there's a "find all regions" button that will add a bunch of regions
to a list. Each should have a little checkbox, which you should
check. Then, click the "Copy Checked Regions as Masks" button. Then
you should have all your regions in the list of masks. You might want
to edit their names to give them intuitive names. Finally, use the
"Save Separate Masks" button, and you should be asked to save each
mask separately.
> Is it possible by now to do this in vbview? My first hurdle is trying to
> overlay the nDisplay.cub with the foo.prm, vbview complains that it is
> not possible to load the statistical overlay.
> If this has been discussed previously, just lead me to the correct
> thread and I'll find the necessary information in the archives.
The first thing to check is that you're using the menu option to load
a glm and not the one to load an overlay. Second, you should make
sure the prm file and the nDisplay.cub have the same dimensions. If
not, then you need to make sure your tes files were properly
normalized. If neither of those solves the problem, I might need to
see the headers from your nDisplay and prm files. Alternatively, you
can pre-generate your stat map using vbstatmap, and then just load it
as an overlay.
I hope this at least gets you over the first few hurdles. There is
unfortunately still functionality that's only in the IDL code, but
this should be do-able.
dan
More information about the voxbo-general
mailing list