[caret-users] SureFit memory failure

[Donna Dierker]

Hi Swann,

With the regular Volume: SureFit Operations (Segmentation), I'm pretty 
sure you still need to crop to a left or right hemisphere.  So before 
launching segmentation, you'd still need to Volume: Resize underlay 
volume to crop it.  (I say pretty sure, because John Harwell is always 
making enhancements, and sometimes I'm surprised to find cool features 
by accident.)

Back in May, though, John did add some automatic segmentation features 
that looked very promising, based on initial results, but which no one 
to my knowledge has tested extensively.  Perhaps you can be the person 
to do so. ;-)

He had a shell script loop through case directories like so:

pushd ./case01/LEFT_HEM/
run_seg.csh
popd

pushd ./case01/RIGHT_HEM/
run_seg.csh
popd

pushd ./case42/LEFT_HEM/
run_seg.sh
popd

pushd ./case42/RIGHT_HEM/
run_seg.sh
popd

Then in each of these left and right hem subdirs, there was a tiny 
script like this:

#!/bin/csh
caret_command -volume-segment-stereo-space \
   -ecv \
   human.case42.L.spec \
   Human_Buckner_Case42+orig.HEAD

This is what the usage says:

               THIS IS EXPERIMENTAL BUT IT WILL PRODUCE A SURFACE

               The input spec file must contain the stereotaxic space 
and the
               structure (LEFT or RIGHT) for this command to operate.  A 
spec
               is created by running this program with the 
"-spec-file-create"
               command.

So, you can either crop manually; run caret_command 
-volume-segment-stereo-space on it (crossing your fingers); or simply 
use the old Colin surfaces, which are still available in sumsdb.

Why are you segmenting Colin, anyway?  There already are several 
surface-based versions of it.

Donna

On 11/08/2007 04:45 AM, swann pichon wrote:
> Dear all,
> I encounter an (out of) memory error during the determination of the 
> inner boundaries of a 1^3 human T1 (colin27). I launched Surefit onto 
> the T1 Colin27 (both hemisphere at the same time) available here : 
> http://imaging.mrc-cbu.cam.ac.uk/downloads/Colin/
> I Used the following parameters
> o/utput format to NIFTI
> gray: 81
> white: 105
> //
> //hindbrain off
> /
> /Automatic err correction : off
> Correct all topological ... off
> Generat inlated off
> very inflated off
> elipsoidal : off
> everything else is on
>
> /
> I am running winXPsp2, P43.4 GhZ with 2G° RAM & 2000 to 8000M° max of 
> swap space.
>
> How much RAM is needed to run Surefit in this context ?
>
> Thanks a lot for your comments,
> All the best,
>
> Swann
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> caret-users mailing list
> caret-users at brainvis.wustl.edu
> http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>