Hi Chris,

we actually do use my robust registration tool for it since a while (FS 5.0, Aug 2010), replacing FLIRT which was used before. Here is the citation:

M. Reuter, H.D. Rosas, B. Fischl. Highly Accurate Inverse Consistent Registration: A Robust Approach. NeuroImage 53(4):1181-1196, 2010.

It is a robust rigid registration, using the tukey biweight function on the residuals. The statistical robustness makes it less sensitive to outliers (e.g. differences in jaw, neck, eye regions) by identifying regions with local differences during the registration and reducing their influence on the registration. This leads to very accurate alignment in the regions that do not change (the brain).

Cheers, Martin

On 06/25/2013 12:05 PM, Bruce Fischl wrote:

yes, it's rigid and used to be sum of squares but in the newer versions we 
will be using Martin's robust registration (which is a tukey biweight 
instead of SSE)

On Tue, 25 Jun 2013, Parker, Christopher wrote:

Hi,
I understand that the recon-all script registers and then averages two
anatomical images (after bias field correction) if two are provided in the
"orig" folder, before performing the other parcellation stages. I am writing
a report and it might be good to mentioned what registration is used. I'm
guessing this is a rigid registration using sum of squared differences cost
function?

Best,
Chris

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

-- 
Martin Reuter, Ph.D.
Assistant in Neuroscience - Massachusetts General Hospital
Instructor in Neurology   - Harvard Medical School
MGH / HMS / MIT

A.A.Martinos Center for Biomedical Imaging
149 Thirteenth Street, Suite 2301
Charlestown, MA 02129

Phone: +1-617-724-5652
Email: 
   mreuter@nmr.mgh.harvard.edu
   reuter@mit.edu
Web  : http://reuter.mit.edu