Check how the spm_flip_analyze_images.m file is set in your two SPM
versions. My best guess is that they differ.
When working with ANALYZE (rather than NIfTI) format images, it may be
a good idea to ensure the orientations are set correctly prior to
doing anything at all with the data. As soon as you change the
position of an image (eg via a coregistration or a realignment) then
the headers will be converted to NIfTI format, with the orientation
set as it has been defined in the spm_flip_analyze_images.m file.
Best regards,
-John
On 6 February 2012 13:40, Joao Pereira <[log in to unmask]> wrote:
> Dear John,
>
> Thanks for the advice. I have looked at both the original and segmented
> versions in check reg, in both versions, and the left-right flip is always
> there (I am using patients with a clearly lateralised pathology to assess
> this as well.) The SPM5 segments are the correct ones. Do you have any idea
> of hos this can be corrected, to force the correct orientation in both
> versions?
>
> Thank you!
>
> Best,
>
> Joao
>
> On 2 February 2012 19:30, John Ashburner <[log in to unmask]> wrote:
>>
>> Maybe you could compare the original images and the results together.
>> I would expect check reg to treat these as if they both have the same
>> orientations. Try this using both spm versions.
>>
>> It's possible that the flip settings differ between the two SPM
>> versions. If so, then spm5 will assume that non-NIfTI data have one
>> orientation, whereas spm8 will assume they have the other. Both
>> versions will write out the segmented results in NIfTI format, which
>> encodes left-right orientations in the header.
>>
>> Best regards,
>> -John
>>
>>
>> On 2 February 2012 18:02, Joao Pereira <[log in to unmask]> wrote:
>> > Hi John,
>> >
>> > Thanks for the quick reply.
>> >
>> > I had segmented them in SPM5 and now I am using SPM8 to do the same
>> > thing.
>> > Comparing the results of both in checkreg I can see the flip. Also, I
>> > noticed this effect only after checking the glass brain, after
>> > performing an
>> > analysis, and the atrophy locations were also clearly flipped, so it
>> > seems
>> > like a real issue.
>> > Best,
>> >
>> > Joćo
>> > On 2 February 2012 17:56, John Ashburner <[log in to unmask]> wrote:
>> >>
>> >> How do you know they are flipped? The important thing is how SPM (and
>> >> other NIfTI compliant software) view the image orientations. Try
>> >> displaying the original image and the segmented images together using
>> >> CheckReg.
>> >>
>> >> Bye for now,
>> >> -John
>> >>
>> >> On 2 February 2012 11:40, Joao Pereira <[log in to unmask]>
>> >> wrote:
>> >> >> Dear all,
>> >> >>
>> >> >> I have an odd problem with SPM8 when analyzing old structural 1.5T
>> >> >> analyze
>> >> >> images I have stored. After segmentation, the results are flipped
>> >> >> left
>> >> >> to
>> >> >> right. I have tried converting the images to nii, saving them in
>> >> >> SPM,
>> >> >> in
>> >> >> FSL, even changing header orientation data, but the results are
>> >> >> always
>> >> >> flipped!
>> >> >>
>> >> >> Any suggestions on how to solve this will be greatly appreciated.
>> >> >>
>> >> >> Best,
>> >> >>
>> >> >> Joćo Pereira
>> >> >
>> >> >
>> >
>> >
>
>
|