Hi - there are a couple of things that you probably want to do differently
here.
a) if you have an image with regular slice separations (eg an image with
15x1mm slices with 2mm spacing between all slices) then you should not pad
the image at all. Instead, you want to simply make the slice "thickness"
in the header 3mm, so that, for example, when you view the image with
the "slices" program, the shape looks correct. this is all you need to do.
b) in general you should use the better quality image (in your case t1) as
the reference and the worse quality (t2) as the input. this should improve
robustness. It is easy to invert the resulting transform and in general is
easy to transform either image to the other with the appropriate
transforms.
for example,
t1=reference
t2=input
flirt -ref t1 -in t2 -out t2_to_t1 -omat t2_to_t1.mat
invert_xfm -matonly -inverse -omat t1_to_t2.mat t2_to_t1.mat
flirt -in t1 -ref t2 -out t1_to_t2 -applyxfm -init t1_to_t2.mat
this transforms t1 to t2's space (t1_to_t2)
thanks, Steve.
ps - if this doesn't help then feel free to email me the images directly
and I'll look into it further - thanks :)
Stephen M. Smith
Head of Image Analysis, FMRIB
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
On Fri, 23 Nov 2001, Martin Kavec wrote:
> Hello to all,
>
> I can not get flirt to align my T2-weighted images to high resolution
> T1-weighted (3D) images (100).
> T2 images are heavily weighted (TE=120ms) high SNR ->OBLIGUE<- images (9
> images) with slice spacing 2.5 mm.
> T1 is 3D nice contrast between GM and WM and 1.3mm slice covering whole head
> (100 images). These images have axial orientation.
> My aim is to get T1-weighted images with exactly the same position as T2
> weigted images.
> According to my colleague's advice, I inserted two zero 1.3mm zero slices
> between each T2 image before alignment. Then in Flirt GUI I choose
> "Input image -> Reference image"
> "Reference image" = T2 volume
> "Input image"= T1 volume
>
> and as model I expect 6/7 model should be fine. However, the results are
> surprising. They look like sagital orientation of T1 volume.
> I think I am doing some principal mistake somewhere.
>
> Many thanks for any help in advance,
>
> Martin
>
|