Maybe SPM should stop supporting ANALYZE 7.5 format, and use only NIfTI. This
would prevent any confusion about which are the left and right sides of the
If you are using SPM5, then it does not flip the data during spatial
normalisation. This flag simply indicates that indexing of your data uses a
left-handed system. The stuff about radiological and neurological
orientation is a bit of a distraction, as this only indicates the orientation
of a single slice. For example, you could rotate the volume by 180 degrees
and have the left and right sides of the axial and coronal views reversed,
but the handedness would remain unchanged (ie a rotation would not flip the
If you have certain types of DICOM files, then the DICOM import of SPM5 may be
used to convert your data to NIFTI format. If SPM5 can't deal with it, then
here are a couple of sites that contain links to DICOM to NIfTI converters:
Providing you only use SPM5 or any other NIfTI compatible software, working
with your data in this way should not lead to any ambiguity about left and
right sides (unless there are bugs in the code that may cause the behaviour
to be otherwise). NIfTI files may end .nii, or they may be .img and .hdr, as
in the old ANALYZE format. If the files are NIfTI format, then the flip
setting in spm_defaults.m is ignored, as this information is available in the
However, if you use ANALYZE format images, then the flip setting may be used -
depending on whether the images have a .mat file associated with them or not.
If there is a .mat, then the flip setting in spm_defaults.m is ignored
because the required information is in the .mat file. If there is no .mat
file for your ANALYZE format images, then the handedness needs to be
determined from the spm_defaults.m file.
I WOULD SUGGEST THAT THE defaults.analyze.flip SHOULD NEVER BE CHANGED. It is
the responsibility of the user to ensure that the orientation of your images
is as described http://www.fil.ion.ucl.ac.uk/spm/software/spm2/#Compatability
prior to doing anything else in SPM2 or SPM5. Provided this is satisfied,
then the orientations should remain the same. However, if you take images
generated or modified using SPM5 (ie NIfTI format images), and try to do
something with them in SPM2 (not NIfTI format), then the orientation flag in
the NIfTI headers will be ignored, and the image orientations may become
messed up again.
On Thursday 03 July 2008 22:46, Manish Dalwani wrote:
> Hi SPM'ers,
> I checked the spm_defaults.m file in SPM5 and the defaults.analyze.flip = 1
> i.e. it flips during the normalization. I used the dartel tools to warp,
> segment and modulate the images into GM/WM.
> If my images were in radiological format (my anatomicals), would the
> results now be in neurological format because of flip being turned on? I
> had it turned off in SPM2 and wouldn't bother because I knew my final
> results matched my initial input. I am just not sure how many times the
> images flip when it's turned on!
> Kindly advise!!!
> Manish Dalwani
> Sr. PRA
> University of Colorado at Denver