Hi, Gustaf-
Ah, OK, I think that reveals what is happening: your anatomical dataset has obliquity, and 3dSkullstrip is removing the obliquity. This is something that has been on my "to do" list for a while to change---it shouldn't do that.
We typically recommend deobliquing anatomical data before processing with FMRI (but leaving in the obliquity in EPI data), but this might be a different application. Particularly with visualization of oblique data, there is a tradeoff of ignoring obliquity or of blurring/regridding the data slightly. If you are interested, we have adjunct_deob_around_origin to deoblique in a way that preserves the location of the coordinate origin of the dataset, while not blurring at all.
One way to proceed currently (before I finally do get my act together to update the program---it's a bit complicated because it has several output places, woe is me---is discussed here. Basically:
- Since all the other 4 attributes match, you should be able to "reattach" the obliquity to the output dataset from the input one.
- We will use 3drefit, which will change the header in place, so we start with a copy in case we are unhappy with some result:
3dcopy <dset out> <dset out2>
3drefit \
-atrcopy <dset in> IJK_TO_DICOM_REAL \
<dset out2>
- You can verify that things went well with:
3dinfo -same_all_grid -prefix <dset in> <dset out2>
How does that seem?
--pt