Dear Miranda,
> As I understand, FNIRT currently is not designed to register images
> with
> different contrasts. However, I would like to test FNIRT to see
> how it
> would do in registering a T1 (1x1x1) image to a PD (1x1x2) image.
> I am not
> an expert on all of the options that one is able to specify when
> running
> this program. Any suggestions as to how to go about doing this, or
> specific
> command-line options/syntaxes that might be successful? Thanks in
> advance!
I have not tried this, so I cannot say how well it would work. I
would expect it to struggle a bit.
If I were you I think I would start out by trying the "regular"
T1_2_MNI152_2mm.cnf, but changing the sub-sampling so that it is 2
for the last step (to avoid using the full resolution of the PD
image). I would then have a look at the results of that.
If that doesn't work (which I suspect might be the case) I would try
to supply a starting estimate for the non-linear intensity mapping.
This would simply be a text-file with with as many numbers as the
order of the intensity-mapping (the --intorder parameter). For this I
would set all values to zero except for the first order (linear)
term, which I would set to -1 (since the order of the intensities is
"reversed" in PD compared to in T1).
My question though would be why you want to do this? Are these two
images from the same subject? In that case you would want to use
FLIRT instead.
Good luck Jesper
|