I have been asked to use FreeSurfer to create a good brain mask for use in another analysis stream (DTI and other things). I have chosen to do this by running recon-all on the structural scan, then taking the (aseg+aparc) and using FSL to dilate and erode to fill holes. This I binarize and rotate back into the same orientation as the original structural volume, and finally strip the padding added by recon-all, thus making the mask match the original T1 so it can be overlaid properly.
The only problem is that, besides rotating and padding, recon-all also resamples to exactly 1.0 isovoxel, whereas the original T1 was 1.0 x 1.0 x 1.1 mm. Thus, I need to find a way to undo this resampling. Best I can determine this happens in the recon-all stream near the beginning where is seems to create some Talariach transform.
Any ideas how to undo this xform, or can you suggest a better way to accomplish my goal.
-Jeff Eriksen