Dear Arun,
If your images are initially fairly close to each other then the easiest
thing to do in the GUI is to select no search as an option.
Otherwise, if you need to use an initial transformation then I'm
afraid you'll have to use the command line - there is no such capability
in the GUI.
As for the matrices, they are not the same as AIR. The matrix relates
mm coordinates where the origin is the bottom, rear, left corner of the
volume.
If you just need to swap some axes then you can run avwswapdim without
the final output argument and it will display the required matrix.
Otherwise,
just play with some and use the -applyxfm option to see what effect they
have until you get it right.
The statistics are calculated originally in native space (you can do a
whole
first level analysis without any registration), and then transformed later,
as
you thought.
All the best,
Mark
On Wednesday, November 6, 2002, at 08:30 PM, Arun Bokde wrote:
> Hi,
>
> I have run a single subject analysis using FEAT GUI starting from epi to
> high resolution registration, normalization and stats. The step in
> which it does the normalization of the subjects's structural scan to
> Talairach space, it found as "best" fit a solution where the scans are
> actually 180 degrees (yaw) off from each other (quite obvious when using
> AFNI to view the images). Anyway, the best way to correct this seems to
> give it an initial transformation and then to register as usual with a
> restricted search space. Question: is there a way to set an initial
> transform within the GUI ? If not, can you give me a pointer to some
> information about how to set up the initial transform matrix ? (is it
> identical to the matrix that AIR uses ?)
>
> When transforming the data to Talairach space, I would like to confirm
> that it is the stats that are transformed, and that the analysis is
> actually done within the subject's native space.
>
> TIA.
>
> Best,
>
> Arun Bokde
|