Hi Jack,
I've had a quick look at the data.
Your mag image has a nasty slice-by-slice intensity change, and I'm
afraid that
I couldn't download the epi or the mag2spgr.hdr as they were all
zero-length!
Anyway, I had a look at the mag to spgr registration and there is a huge
scale
difference. It was over 15% which is really pretty big. If these were
of the
same individual then something is wrong. I'd suggest double checking the
voxel sizes as that is normally what gets set wrong. So this slice
artefact
plus the scale difference is probably why the registration doesn't work
well.
If in fact only one of the scales is incorrect (say the slice thickness)
then
that would also create a form of distortion that 6 or 7 dof cannot overcome.
As for fugue, it requires a pair of phase images to calculate the phase
difference, which is proportional to the field (it needs a 4D analyze
format - I
keep meaning to add alternative input options, but for the moment use
avwmerge if you have them separately). That's why it is complaining.
All the best,
Mark
Jack Grinband wrote:
>Mark,
>
>
>
>>Don't despair! Although registration with a small FOV is always tricky
>>and it seems that small FOV in z is the worse. But, critically, how big is
>>the FOV in mm for the T2.
>>
>>
>
>88mm in z (it's actually a T1). 16 slices, 5.5mm thick.
>
>
>
>>As for scaling, are these images of different people?
>>If not, then you don't want to be offering it too many scaling options,
>>no more than 7 dof anyway.
>>
>>
>
>Same person (me) on two different days. I used 7 dofs, mutual info. There is a pdf of
>the registered T1 overlaid on the SPGR and you can see significant error in the
>registration.
>
>
>
>>There isn't much more I can say without the images though, so if you
>>can, put them somewhere that I can download ....
>>
>>
>
>www.columbia.edu/~jg2269/
>
>I have one more question. When I ran fugue:
>fugue -i mag -p unwrap -d .164 -w warpedmag
>
>I got the following error: Unwrappedphasemaps must contain at least two volumes!
>
>Is fugue looking for complex numbers?
>
>I really appreciate you looking at these images. thanks a lot,
>
>jack
>
>
|