Dear list,

I have a problems running FSL on Fedora Core 2 and Debian (sarge) boxes.
In both cases I'm trying to run the precompiled RH9 binaries
(compilation from source causes problems for flame and mcflirt, will
post details later).

I have a simple 16-minute dataset (block design) and 4 EVs. Everything
runs smoothly through prestats (including MELODIC), and seems to pass
through stats too. However, the first stage of post-stats, contrast_mgr,
causes a segfault :

In FSL 3.2b:
Saving results...

/usr/local/fsl/bin/contrast_mgr -f design.fts stats design.con
** ERROR: nifti_image_read(stats/sigmasquareds): can't open header file
** ERROR: nifti_image_open(stats/sigmasquareds): bad header info
Error: failed to open file stats/sigmasquareds
Error:: FslGetDim: Null pointer passed for FSLIO

In FSL 3.1:
Saving results...
Unthresholding Series... Completed
Saving image... Completed
Thresholding Series... Completed
Unthresholding Series...
/usr/local/fsl/bin/contrast_mgr -f design.fts stats design.con
AvwOpen: Error while opening header!: No such file or directory

The file stats/sigmasquareds doesn't exist and I can't figure out why. I
don't think it's a memory issue, as has been discussed recently, since
the nii.gz is only 65 MB. At some stage of the analysis I get two
massive core dumps
-rw-------  1 cjb petadm 316022784 Dec  2 15:50 core.25436
-rw-------  1 cjb petadm 794587136 Dec  2 15:54 core.27381

I've placed the .feat directory (minus the core dumps) tar.gzipped on

(btw, it's SE data...) If anyone has time to check it out, I'd be


Christopher Bailey <[log in to unmask]>
PET Centre and
Center for Functionally Integrative Neuroscience
Aarhus University Hospital, Denmark