rick reynolds wrote:

> Hi Joel,
> Does the tlrc dataset that you translate to the Pittsburgh RAI
> format undergo any shifts or rotations before you translate
> back to the AFNI format?
> If not, there should be no reason to run adwarp again. Since it
> should already by in tlrc format (you can use 3dresample to go
> from RAS back to RAI, without altering the data), running
> adwarp on it would apply the original transformation a second time.

What we do is convert one anatomy file to AFNI format, Talairach it, and
convert it back to Pittsburgh MRI format. It then serves (or is supposed to
serve!) as a prototype for Talairaching other data (functional data) later
on. Obviously there is no point in Talairaching the same file twice. When
I said that I was using the Talairach'd volume as input to adwarp, I mean that
I was using it as the geometry input, the -apar input. Giving any +orig-space
dataset as the -dpar parameter produces all zero's as output. It's just a particularly
simple example of the problem to use the +orig-space version of the original
anatomical dataset as the -dpar value.

So, should the values of the WARP_DATA field depend on the data order (RAI
vs. RAS)?

Thanks,
-Joel