Hi Carien,

To start with, there's nothing to worry about here.

We don't use pixdim[4] for anything in FSL, so it is safe
to ignore it.  I'm not sure why fslmaths is changing it 
though, or why you see a difference between the two
platforms unless there is a small version difference.

Anyway, if this is a 2008 version of FSL then it is likely
that this odd behaviour may have disappeared by now.
So I would just ignore it all as your images and analyses
will be fine.

Bonus points for actually following the instructions and
checking the images though!

All the best,
	Mark


On 24 Feb 2011, at 15:38, Carien van Reekum wrote:

> Hi Mark,
> 
> Whilst I did find a similar error message reported in the archive, our
> situation is a little different from the one reported earlier, so here
> goes:
> 
> We have been processing functional MRI data on two systems, MacOSX and
> Linux (running Ubuntu). FSL software versions are both 4.1 (2008).
> Setting up feat, the preprocessing steps were identical and involve B0
> corrections (one processing stream involved FNIRT, and one FLIRT at
> the registration step, which is the only difference in the processing
> path). During the fieldmap correction step, when invoking fslmaths,
> the processing run under Linux gives us the following error, that the
> Mac OSX processing does not give us:
> 
> WARNING:: Inconsistent orientations for individual images in pipeline!
>          Will use voxel-based orientation which is probably incorrect
> - *PLEASE CHECK*!
> 
> The images printed in the report.html of the unwarping step do seem
> fine to me in both cases.
> 
> Upon checking the header information of the fslmaths input files,
> there do seem to be differences in pixel dimensions, which may well
> underlie the error/warning reporting, but I am not sure why that is an
> issue on one system but not the other (suggesting software
> differences?)?
> 
> Relatedly, if we need to worry about these pixeldimension differences,
> how best to fix these?
> 
> Below the header information of the different files (processed under
> Linux for the most part), in case this is useful information. Happy to
> send the log-file along if useful.
> 
> thanks in advance for any pointers,
> Carien
> 
> fslinfo ../../../fieldmap/ar04_fm_rad_per_sec.nii.gz
> data_type      FLOAT32
> dim1           64
> dim2           64
> dim3           30
> dim4           1
> datatype       16
> pixdim1        3.0000000000
> pixdim2        3.0000000000
> pixdim3        5.0000042915
> pixdim4        0.4880000055
> cal_max        1275.8230
> cal_min        -1277.0702
> file_type      NIFTI-1+
> 
> fslinfo FM_UD_fmap.nii.gz
> data_type      FLOAT32
> dim1           64
> dim2           64
> dim3           30
> dim4           1
> datatype       16
> pixdim1        3.0000000000
> pixdim2        3.0000000000
> pixdim3        5.0000042915
> pixdim4        1.0000000000
> cal_max        0.0000
> cal_min        0.0000
> file_type      NIFTI-1+
> 
> fslinfo ../../../fieldmap/T1_ar04_fm_rad_per_sec.nii.gz
> data_type      FLOAT32
> dim1           64
> dim2           64
> dim3           30
> dim4           1
> datatype       16
> pixdim1        3.0000000000
> pixdim2        3.0000000000
> pixdim3        5.0000042915
> pixdim4        0.0000000000
> cal_max        0.0000
> cal_min        0.0000
> file_type      NIFTI-1+
> 
> fslinfo FM_UD_fmap_mag.nii.gz
> data_type      FLOAT32
> dim1           64
> dim2           64
> dim3           30
> dim4           1
> datatype       16
> pixdim1        3.0000000000
> pixdim2        3.0000000000
> pixdim3        5.0000042915
> pixdim4        0.0000000000
> cal_max        0.0000
> cal_min        0.0000
> file_type      NIFTI-1+
> 
> fslinfo FM_UD_fmap_mag_brain_mask
> data_type      INT32
> dim1           64
> dim2           64
> dim3           30
> dim4           1
> datatype       8
> pixdim1        3.0000000000
> pixdim2        3.0000000000
> pixdim3        5.0000042915
> pixdim4        1.0000000000
> cal_max        0.0000
> cal_min        0.0000
> file_type      NIFTI-1+
> 
> fslinfo FM_UD_fmap_mag_brain_mask_inv
> data_type      FLOAT32
> dim1           64
> dim2           64
> dim3           30
> dim4           1
> datatype       16
> pixdim1        3.0000000000
> pixdim2        3.0000000000
> pixdim3        5.0000042915
> pixdim4        0.4880000055
> cal_max        0.0000
> cal_min        0.0000
> file_type      NIFTI-1+
> 
> Why is the 4rth pixeldim more than half in brain_mask_inv from the
> brain_mask when all that is done is /usr/local/fsl/bin/fslmaths
> FM_UD_fmap -abs -bin -mas FM_UD_fmap_mag_brain_mask -mul -1 -add 1
> -bin FM_UD_fmap_mag_brain_mask_inv  ??
> 
> Finally, the example func has been copied into a file called EF_D_etc
> and has the following info:
> 
> fslinfo EF_D_example_func.nii.gz
> data_type      FLOAT32
> dim1           64
> dim2           64
> dim3           30
> dim4           1
> datatype       16
> pixdim1        3.0000000000
> pixdim2        3.0000000000
> pixdim3        5.0000000000
> pixdim4        2.0000000000
> cal_max        0.0000
> cal_min        0.0000
> file_type      NIFTI-1+
> 
> 
> -- 
> Centre for Integrative Neuroscience and Neurodynamics
> School of Psychology and Clinical Language Sciences
> University of Reading
> Earley Gate, Whiteknights
> P.O. Box 238
> Reading RG6 6AL, UK
> Ph.  +44 (0)118 378 5556
> Fax +44 (0)118 378 6715
> [log in to unmask]
> http://www.personal.reading.ac.uk/~sxs07cv/BECLAB/Carien_van_Reekum.html
> http://beclab.org.uk/
>