Hi Shane,
Regarding your 3dWarp command, the default interpolation
method is linear. Nothing will necessarily be perfect,
but consider -NN, just to see what you think. Perhaps
it would be possible to add wsinc5 as another optional
method.
1. Yes, we would call this the template. It is mostly
used for alignment, and really defines the space. One
would hope to have a good atlas in this same space.
2. The coordinate system is more for defining what some
coordinate actually means, and is often interchanged
with orientation. For example, does 13, -21, 47 mean
13 right, 21 posterior and 47 superior (LPI orientation
would imply that) or does it mean 13 posterior, 21
superior and 47 left (ASL)?
Also some packages define this orientation name using
the positive directions, while in AFNI the negatives
are used (as with the DICOM standard). It is very
muddy.
Any of the 48 cardinal orientations (or even 96, if
positive and negative directions are included) could
be used to describe some results.
An atlas will be in some orientation.
3. Yes, though the coordinates could be specified in
any cardinal orientation. The afni GUI defaults to
viewing in RAI, though perhaps the default reporting
orientation is now LPI, since it is more common.
Reprocessing the data may not be necessary, but it is
the most accurate way to go. Sure, either TT_avg152T1
or TT_N27 should be fine for the template.
Daniel Glen has a couple of posts discussing ways to
transform from MNI to TLRC space, though. They are
worth a read, and are
here.
However, I think you might want to think harder about
which atlas seems appropriate. Is the Talairach atlas
what you really want to use, or might there be a useful
alternative?
- rick
