Hi Mark,
We have just encountered the same problem here, where we get a
segmentation fault on some images and not others. We are running Redhat
8 with a kernel of 2.4.18-14smp. It has 2Gig Ram and 2Gig Swap space.
The FSL source code with the patchM-fsl-3.2-sources.tar.gz
(11-Aug-2005) was compiled.
This is the command line and output --
__________
$ /home/softwares/fsl/bin/flirt -in
/home/data/oddball/DY004/DY004_SPGR_171_brain_f3.nii.gz -ref
/home/softwares/fsl/etc/standard/avg152T1_brain.hdr -out /home/reza/aa
-omat /home/reza/aa.mat -bins 256 -cost corratio -searchrx -90 90
-searchry -90 90 -searchrz -90 90 -dof 12 -interp trilinear
WARNING: Both reference and input images have an sform matrix set
The output image will use the sform from the reference image
Segmentation fault
__________
The problem images are 256 x 256 x 99.
Interestingly, we do not get the fault when we run the same process on
OSX.
Let me know if you'd like me to post the image and the fslreport output
to our ftp site.
Best,
Robin
________________________
Robin Goldman, Ph.D.
Hatch Center for MR Research
Columbia University
710 W. 168th Street, NIB-1
New York, NY 10032
(212) 342-0867
On Sep 20, 2005, at 5:11 AM, Mark Jenkinson wrote:
> Hi Alex,
>
> People have experienced this before and we made a
> patch which should fix it. If you haven't installed
> this patch, then do this following the instructions
> below. If you have already installed this patch
> then please give us details of the computer that
> you are using plus its RAM, swap, etc. as it may
> be a memory issue. You can run fslerrorreport to
> get most of this information.
>
> All the best,
> Mark
>
> P.S.
> To get the patch EITHER re-download FSL from
> scratch, or just take the appropriate small patch file from
> http://www.fmrib.ox.ac.uk/fsldownloads/patches/3.2
> and then:
> cd $FSLDIR/..
> gunzip [your downloaded patch file]
> tar xvf [the uncompressed patch file]
>
> P.P.S. Your command line looks fine and I don't know why
> verbose would not work - if it still doesn't, please send the
> command line you tried.
>
>
> Alex Fornito wrote:
>
>> Hi,
>> I've been trying to register some 3T T1s (512 x 124 x 400, voxdim
>> 0.64 x 1.7
>> x 0.64) to a resampled version of the Colin 27 (148 x 184 x 144,
>> voxdim 1 x
>> 1 x 1) using 6 dof. FLIRT works fine on some images, but on most I
>> get a
>> segmentation fault. Wandering if anyone has encountered anythin
>> similar and
>> knows why? The command line is below ( I tried to run -verbose to
>> give you
>> more detail, but I got a message saying it lacked arguments if I
>> included
>> -verbose 1 after flirt in the command line - maybe I'm doing it
>> wrong?)
>> Thanks,
>> Alex
>>
>> /usr/local/fsl/bin/flirt -in
>> /data/nelson/work/kerrie/cingulate/orig_and_prep_images/
>> ab_rc_brain.hdr -ref
>> /data/nelson/work/kerrie/cingulate/MNI_111_TS_reorient_256.hdr -out
>> /data/nelson/work/kerrie/cingulate/flirt/ab_brain2mni.hdr -omat
>> /data/nelson/work/kerrie/cingulate/flirt/ab_brain2mni.mat -bins 256
>> -cost
>> corratio -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -dof 6
>> -interp
>> trilinear
>> Segmentation fault
>>
>
|