Hi,
Don't worry - all of our internal code goes through a nifti library,
which also can read
Analyze, due to nifti being backward compatible. So the fact that the
errors mention
nifti only means that it is in this part of the library that the error
occurs, and not that
FSL thinks that either the input or the output is Analyze.
However, the errors are serious and need attention.
What it is basically complaining about is that it cannot open some
image files.
In particular (from the bit you've included here):
> Cannot open volume /ifs/tmp/pipelnvr/2008March18_12h27m58s818ms/
> FSF2Pipe.FEATInputDirectories_0-0.feat/reg_standard/reg/highres for
> reading!
Can you open this image in fslview?
That is the first thing to check.
Also, does it happen for all the images in this fslmerge command or
just this one?
And is your output image (bg_image) getting created or not?
All the best,
Mark
On 19 Mar 2008, at 18:28, Arash Payan wrote:
> Hi all,
> I'm trying to run feat on a Linux box here at my lab. I'm doing a
> higher level analysis, but it fails early on due to some nifti
> conversion errors. The thing is, I've set the FSLOUTPUTTYPE to be
> Analyze and there's no nifti data anywhere in my inputs. The
> following is some of the error text being produced in the
> report_log.html:
>
> /usr/local/fsl-4.0.3_64bit/bin/featregapply /ifs/tmp/pipelnvr/
> 2008March18_12h27m58s818ms/FSF2Pipe.FEATInputDirectories_0-0.feat
>
> /usr/local/fsl-4.0.3_64bit/bin/featregapply /ifs/tmp/pipelnvr/
> 2008March18_12h27m58s818ms/FSF2Pipe.FEATInputDirectories_0-1.feat
>
> /usr/local/fsl-4.0.3_64bit/bin/featregapply /ifs/tmp/pipelnvr/
> 2008March18_12h27m58s818ms/FSF2Pipe.FEATInputDirectories_0-2.feat
>
> /usr/local/fsl-4.0.3_64bit/bin/featregapply /ifs/tmp/pipelnvr/
> 2008March18_12h27m58s818ms/FSF2Pipe.FEATInputDirectories_0-3.feat
>
> /usr/local/fsl-4.0.3_64bit/bin/fslmerge -t bg_image /ifs/tmp/
> pipelnvr/2008March18_12h27m58s818ms/
> FSF2Pipe.FEATInputDirectories_0-0.feat/reg_standard/reg/highres /ifs/
> tmp/pipelnvr/2008March18_12h27m58s818ms/
> FSF2Pipe.FEATInputDirectories_0-1.feat/reg_standard/reg/highres /ifs/
> tmp/pipelnvr/2008March18_12h27m58s818ms/
> FSF2Pipe.FEATInputDirectories_0-2.feat/reg_standard/reg/highres /ifs/
> tmp/pipelnvr/2008March18_12h27m58s818ms/
> FSF2Pipe.FEATInputDirectories_0-3.feat/reg_standard/reg/highres
> ** ERROR: nifti_convert_nhdr2nim: bad dim[0]
> ** ERROR (nifti_image_read): cannot create nifti image from header '/
> ifs/tmp/pipelnvr/2008March18_12h27m58s818ms/
> FSF2Pipe.FEATInputDirectories_0-0.feat/reg_standard/reg/highres.hdr'
> ** ERROR: nifti_image_open(/ifs/tmp/pipelnvr/
> 2008March18_12h27m58s818ms/FSF2Pipe.FEATInputDirectories_0-0.feat/
> reg_standard/reg/highres): bad header info
> Error: failed to open file /ifs/tmp/pipelnvr/
> 2008March18_12h27m58s818ms/FSF2Pipe.FEATInputDirectories_0-0.feat/
> reg_standard/reg/highres
> Cannot open volume /ifs/tmp/pipelnvr/2008March18_12h27m58s818ms/
> FSF2Pipe.FEATInputDirectories_0-0.feat/reg_standard/reg/highres for
> reading!
>
>
> Going into the directory mentioned above, I see two files,
> highres.img and highres.hdr, which are not Nifti. Why feat insists
> on interpreting them as such, I have no clue.
>
> For clarification, I'm executing feat from the following script:
> export FSLDIR="/usr/local/fsl-4.0.3_64bit"
> . $FSLDIR/etc/fslconf/fsl.sh
> export FSLOUTPUTTYPE="ANALYZE"
> echo $FSLOUTPUTTYPE
> $FSLDIR/bin/feat converted.fsf
>
> Also, fsl.sh sets $FSLOUTPUTTYPE to be ANALYZE, but I set it again
> just in case before calling feat.
>
> Any help with regards to this would be fantastic. Everybody in our
> lab is super excited to get started with feat on v4.
>
> Thanks,
> Arash
>
|