Hi Jack,
Jack Grinband wrote:
>You were exactly right. The voxel sizes were wrongly specified during
>reconstruction. But, why is that important? That is, during normal registration we
>have brains of different shapes and sizes, and flirt scales them all to the standard
>brain. It seems to me that all flirt would have to do is scale this brain down by 15% in
>one or more dimensions. Clearly, I'm missing something about how flirt works.
>
>
Normally it isn't important, but flirt only really checks for relatively
small scaling differences,
and 15% can sometimes be too much for it. Things are just easier (and
hence work better)
if the images are calibrated correctly.
>I'm confused. The input to fugue is
>-i <unwarped-image> -p <unwrapped phase map> -d <swell-to-asym-ratio> -w <
>epi-like-result>
>
>Are you saying that the <unwrapped phase map> is really two concatenated
>images? I thought the phase map is the result of calculating the phase difference.
>
>
Yep, all prelude does is unwrap phase images, it doesn't subtract them.
Hence the input to fugue requires the two phase images so that it can
find the difference,
scale them, regularise them, etc. So the only useful phase maps or
field maps to look at
are outputs from fugue, not from prelude. And if you have separate
phase images then
they need to be combined before going into fugue (until I get around to
allowing a more
flexible input). Sorry for the confusion.
All the best,
Mark
|