Dear Ed,
I am very concerned about this.
If you run:
fslswapdim in -x y z out
then it should issue a big warning about
the fact that this is altering the left/right
order of your images! This effectively will
mean that the registration will match left
with right. However, if you find that this
gives better registrations, then that might
be an indication that your initial left-right
orientation was incorrect. Is there any
way that you can check that independently?
For example, do you have markers in the
images? If not then it is really hard to know
what is correctly labeled as left and right
in the images.
Also, what anatomical axis does the y axis
correspond to? Is it the left-right axis, or
anterior-posterior, or superior-inferior?
I am very concerned that you have mixed up
left and right somewhere, either before FSL or
as part of this processing.
Please let me know the answers to the above
questions.
All the best,
Mark
On 1 Sep 2010, at 11:43, Ed wrote:
> Dear Mark,
>
> I got an email from you on this topic (7 July) suggesting that I
> could either use fslswapdim or fslreorient2std.
> Since I'm using FSL 4.1.4, I applied fslswapdim. (command was:
> fslswapdim in x -y z out)
> The labels of the original stacks were all correct in FSLView, so
> there is no reason why I should
> flip the y axis. But it resolved the registration problems, though.
> Reading your response I get the impression that this is not the
> right way to do. My feeling is that the orientation is not causing
> these problems but fsl_reg.
>
> Ed
>
|