Hi,
I'm still unclear on several issues:
- If you view your images with FSLView then the q/sform makes
*no difference* to how the images look. So are you using a
different viewer? Or are you changing something besides the
q/sform?
- I did not ask if the q/sform was zero itself - just if the *code*
was
zero or not. Was the *qform code* or *sform code* zero?
- Why couldn't you convert the FIRST results back into MNI space?
The transformation matrices for this are already generated so
it is fairly easy to apply them to the output.
- What does "completely destroyed" mean? Are you sure that your
display range is set correctly - as if not the image can look black
but still contain useful values. Use fslstats with -R to check the
actual range.
- What is the exact flirt command that you run? And what are the
dimensions of the image that you start with?
- How are you creating your initial mask and is it aligned with the
MNI? If not, then applying an identity matrix with flirt is not
what you should be doing.
All the best,
Mark
On 8 Jan 2009, at 18:17, siamak wrote:
> Hi Mark,
>
> I mean copying Q/S-forms will stretch or reorient the resulting
> image to what I
> don't expect when I compare it to the origibnal image. The Q/S-forms
> of the
> result of fslhd command were not zero. Anyway, since I have not the
> matrix of
> transformation, I can not change one image space to the other space.
> So, I
> changed my pipeline. Now my problem is to change a sample mask to
> the MNI
> space. I've used flirt command to change the orientation and grid
> size of a
> brain image to MNI space before, but I've not used it yet to change
> the
> orientation and grid of a sample mask with 35 different integer
> values to MNI
> space. When I apply flirt with reference of MNI and identity matrix
> as init to
> my mask, the orientation and grid changes to MNI, but when I view
> the mask I
> see that it is not the previous mask and completely destroyed. Do
> you have
> any idea?
>
> Best,
> Siamka
>
|