We have a large number of scans with excellent, visually inspected and manually trimmed brain masks, so we would like to use these masks to assist the FreeSurfer processing. One way we have experimented with to achieve this is to clip the T1 to remove anything non-brain before importing it into the subject directory. In using brain-clipped T1 images as a starting point for FreeSurfer we notice that the Talairach transform checking is quite far off, scaled badly and rotated downward 10+ degrees in pitch. I assume that this is due to the lack of similarity to the moving image, which includes skull. However, even if we don't modify this transform, the final parcellation looks quite good, and very, very similar to that from using a standard, non-clipped T1 as input. Is there any reason not to use a clipped T1 as input into FreeSurfer? Is the Talairach-based information used for segmentation or parcellation, and are we potentially creating problems? Or would it be better to simply use our mask to trim up the images after autorecon1 (coregistering them into the same image space and clipping them with a custom routine)?
Ron Pierson
Department of Psychiatry
Carver College of Medicine
W-278 GH, 200 Hawkins Dr
Iowa City, IA 52242