Print

Print


Pia,

FSL has utilities called "avwhd" (FSL 3.3 or before) and "fslhd" (FSL after
3.3) that may be helpful to you in determining exactly what is in an Analyze
or NIFTI header generated by any software.  How the files are viewed in SPM
and FSL is a separate question though.  Sometimes the qform information in
the file header may be different between FSL and SPM.  In addition to
avwhd/fslhd, FSL has numerous other "avw" and "fsl" command-line routines
that can be useful in reading or modifying file headers.

Kathy Pearson
UAB Psychology

-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
Behalf Of Pia Rotshtein
Sent: Tuesday, October 07, 2008 1:04 PM
To: [log in to unmask]
Subject: Re: [SPM] FSL 2 SPM

Hi John,
Thanks for the replay.
The thing is that we have being using SPM for co-registration and
nevertheless we get that off-set, which is not viewed when presenting
with FSL (even tough the images are co-registered in SPM).
Which made us think that there is something hidden in the *.hdr, that
when presenting with SPM shift the image, or vice versa.
Given that these images have been processed by FSL first (to generate
the FA maps of DTI), we thought it is something that FSL add to the
header. We just not sure where this thing maybe hidden.
I guess I should ask the FSL people ;o).

Does that make any sense?

Thanks

Pia


-----Original Message-----
From: John Ashburner [mailto:[log in to unmask]] 
Sent: 07 October 2008 18:48
To: Pia Rotshtein; [log in to unmask]
Subject: Re: [SPM] FSL 2 SPM

Hi Pia,
I'm not too sure of the details of how FSLview displays images, but in 
general, the coregistration in SPM is more robust if given reasonably
good 
initial starting estimates.  Sometimes though, even this does not help -

particularly if the anatomical images were collected at high field and 
contain large intensity inhomogeneities.  One thing that you could try
is to 
segment the anatomical image first, so that a bias corrected version is 
generated.  Coregistration can then be done with the bias corrected
version.  
Note that if you do this, the coregistration should have the bias
corrected 
anatomical as the target and the FA image as the source (with any other 
images that are in alignment with the FA specified as "other").  This
should 
allow the original *_seg_sn.mat to be used for spatially normalising the
FA.

All the best,
-John

On Tuesday 07 October 2008 10:04, Pia Rotshtein wrote:
> Hi Tom,
>
> Thanks for the reply,
>
> We exhausted few options, converting to *.nii, *.img, *.hdr
>
> It just seem like the origin register by SPM are different
(spm_vol.m).
>
> We figure there is something else hidden in the hdr, that FSL reads
and
> write but SPM do not but could not find what.
>
>
>
> Will look at the discussion in the FSL list
>
>
>
> Cheers
>
>
>
> Pia
>
>
>
>
>
>
>
> ________________________________
>
> From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]]
> On Behalf Of Thomas Nichols
> Sent: 06 October 2008 22:43
> To:
> Subject: Re: [SPM] FSL 2 SPM
>
>
>
> Hi Pia,
>
> Have you double-checked that all of the images are true NifTI images
> (and not Analyze)?  (fslinfo will tell you).  If you've got a mix of
> Analyze and NifTI you'll have trouble.
>
> OTW, you're might be having qform-sform nightmares.  See a thread on
the
> FSL list "MNI Space in MRIcro and FSL" from 2007.
>
> -Tom
>
> On Mon, Oct 6, 2008 at 7:22 PM, Pia Rotshtein <[log in to unmask]>
> wrote:
>
> Hi everyone,
>
> Our aim is to co-register FA maps to T1 images.
>
>
>
> We have tried with both SPM and FSL.
>
> FSL seem to be doing a good job, and when check with FSLview it is
> perfect, but when checked with SPM5 check-reg it seems to be offset
> (mostly on the y,z dimensions).
>
> Similarly, when SPM is co-registering, when we check with FSLview it
is
> perfect, but with SPM check-reg it is offset (mostly on the y,z
> dimensions).
>
>
>
> We have read that the scales are different in FSL-nifti and SPM-nifti
> but we could not understand how to go about it.
>
> We want to end up with both images co-registered in SPM so we could
> apply the normalization parameters of the T1 to the FA images.
>
>
>
> Any advice would be appreciated
>
>
>
> Thanks
>
>
>
> Pia (& Magda)
>
> ____________________________________________
> Thomas Nichols, PhD
> Director, Modelling & Genetics
> GlaxoSmithKline Clinical Imaging Centre
>
> Senior Research Fellow
> Oxford University FMRIB Centre