rick reynolds wrote:

> So do you apply any warp/translation/rotation/re-grid
> on the tlrc dataset when it is in your Pittsburgh MRI
> format? If not, it may be best to try to leave the WARP
> matrices as they are.

I don't do any of those operations on this dataset, so it sounds
like my leaving the WARP matrices alone is just fine.

>
>
> In either case, it seems to me that you should be able
> to use the tlrc dataset for -apar, particularly if you
> use the version _before_ conversion to P-MRI format.
>

This does appear to work, as long as I use the _before_-conversion
+tlrc file. The after-conversion version produces all zero's.

>
> As a second possibility, and one you may have to consider
> anyway (to match grids), try using "3dresample -orient RAI"
> (or "-master preMRI+tlrc") on the post-P-MRI dataset. It
> might then be okay to use it as the -apar dataset (that
> would need testing).

Well, this *appears* to work, but the 3d location of the normalized functional data is different! I have no idea why, but it's noticably shifted across the head- the version produced by using the original, unconverted-and-unresampled +tlrc anat file is in the expected location and the version produced by using the converted-and-3dresampled file is not. This is despite the fact that 3dinfo claims their bounding boxes are all the same.

>
> I guess I'm hoping that the RAS orientation is the problem...
>

Yes, cleary. Is the RAS orientation supposed to change the interpretation of the WARP_DATA matrix? If not, I think this may be a bug in adwarp.

-Joel