[VoxBo] normalization question

[Nina Hsu]

Hello,
I'm also having some normalization issues.  For many of my subjects, the
Anatomical.cub and Functional.cub files both show origins roughly around the
AC.  However, the nAnatomical.cub and nFunctional.cub files both show
distinct origins that are way off (the nAnatomical.cub origin is somewhere
in the frontal lobes, and the nFunctional.cub origin is on the edge of a
ventricle).  The origins for the .tes files are all pretty similar, but
still nowhere near the AC.  There appears to be some systematicity to this
(for all of the subjects I've looked at, these origins are appearing in the
same place).

Strangely, when overlaying the maps on top of each other and setting a
threshold to 0, the maps look pretty well aligned (in other words, the
functional activity doesn't extend outside of the brain, and these maps
don't look as off as the origin misplacements would make them seem to be).
 These files have all been made/preprocessed with Geoff's preprocessing
script.

Any ideas for what might be going on?  Could this be a problem with the
normalization?  The relevant lines (I think) from the preprocessing script
are copied below.

Thanks in advance for your help!
Nina

runonce shellcommand "setorigin -m Anatomy/Anatomical.cub
Anatomy/Functional.cub"
runonce shellcommand "setorigin -m Anatomy/Functional.cub
Anatomy/Anatomical.cub"

runonce spm2_norm_calc $DIR/Anatomy/ $paramfile Anatomical.cub $refvol $xa
$ya $za $bb_anat
runonce spm2_norm_apply3d $DIR/Anatomy $DIR/Anatomy/$paramfile $anat
nAnatomical.cub $xa $ya $za $bb_anat
runonce spm2_norm_apply3d $DIR/Anatomy $DIR/Anatomy/$paramfile $func
nFunctional.cub $xf $yf $zf $bb_func
runonce spm2_norm_apply3d $DIR/Anatomy $DIR/Anatomy/$paramfile $anat
nDisplay.cub $xa $ya $zf $bb_disp

shellcommand "setorigin -c Anatomy/Functional.cub $scan/$scan.tes
spm2_norm_apply4d $DIR/$scan $DIR/Anatomy/$paramfile $scan.tes $scan.tes $xf
$yf $zf $bb_func
shellcommand "vbsmooth -vox 3 3 3 -o $scan/$scan.tes $scan/$scan.tes
threshold $scan/$scan.tes $scan/$scan.tes $thresh
calcgs $scan/$scan.tes
calcps $scan/$scan.tes

On Thu, Mar 13, 2008 at 10:01 AM, Daniel Y Kimberg <
kimberg at mail.med.upenn.edu> wrote:

> Ranjani wrote:
> >  I'm trying to normalize my tes files using a version of the bigprep
> script that
> >  creates both normalized and non-normalized tes files.
> >
> >  My question is this:  Should the origin of the normalized tes files be
> in the
> >  same location as the nFunctional.cub and nAnatomical. cub files?
>  Because when
> >  I check the origin of my nrunX.tes files, the origin is off (at the
> edge of
> >  a ventricle) rather than in the middle of the AC.
>
> Ultimately, yes -- the normalized tes file should end up with the
> standard origin.  One thing to check is that your tes files and your
> Functional.cub have the same dimensions.  This can occasionally mess
> things up.  I'm also not sure if the value for adir in your dataset
> already has a trailing slash, but you might want to change
> $adir$paramfile to $adir/$paramfile, just to be safe.  Have you
> checked visually to make sure the tes files were actually getting
> normalized?
>
> >  I checked the origin of my nAnatomical.cub and nFunctional.cub files
> and they
> >  both look correct (in the middle of the AC).  And my Anatomical.cub and
> >  Functional.cub files seem to have the origin in the same place, though
> not in
> >  the AC (because I never specified for it to be in the AC for these
> files in my
> >  pre-processing script).  And the origin of the nrunX.tes files doesn't
> look
> >  like it's in the same place as the Anatomical.cub and Functional.cubfiles -
> >  though the script is copying the origin from the Functional.cub file
> onto the
> >  nrunX.tes files.
>
> Sounds like everything's reasonably normal leading up to the SPM2
> normalization.  Incidentally, the origin of the nAnatomical and
> nFunctional should actually be a slice or two superior to the anterior
> commisure.  For whatever reason, that's the standard SPM origin.
>
> >  So, I'm wondering if the origin of the nrunX.tes files should match the
> origin
> >  of the nFunctional.cub (that is, should the origin of the nrunX.tesfiles be in
> >  the middle of the AC)?  If so, I can modify the script to copy the
> origin from
> >  nFunctional.cub rather than Functional.cub.
>
> It should but if it's not happening automatically, I'd worry that the
> normalization didn't work properly.
>
> dan
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.voxbo.org/pipermail/voxbo-general/attachments/20080313/db8f5f6c/attachment.html