Dear Doug, The determinant sign is negative for both example_func and orig.mgz (-73.6 and -1, respecitively).
The files differed also in their format, type, vox2ras-fsl and orientation, but I guess it is ok.
My feeling was that flirt is unable to fit the brain given the maximum angle of rotation 90, so I doubled this using: re-feat2anat --dof 6 --maxangle 180 --feat run1.feat --subject xx
I tried this on several subjects and the registration looks very good: I would not make it better manually.
Can I assume it works now?
Cheers, Aga
-----Ursprüngliche Nachricht----- Von: Douglas N Greve [mailto:greve@nmr.mgh.harvard.edu] Gesendet: Mo 2011-02-07 19:36 An: Burzynska, Aga Cc: 'freesurfer@nmr.mgh.harvard.edu' Betreff: Re: [Freesurfer] reg-feat2anat: a problem!
That is not a problem. By "anatomical" I meant the FreeSurfer anatomical (eg, orig.mgz) doug
Agnieszka Burzynska wrote:
Dear Doug, Thank you a lot! Our feat data is always nifti, but anatomical FSL data is sometimes in the analyze format. Is this already a problem?
I will now check the rest.
Thank you, cheers, Aga
On 2/7/11 4:35 PM, "Douglas N Greve" greve@nmr.mgh.harvard.edu wrote:
Hi Aga, are you using nifti? If analyze, that could be the problem. Do your functionals and anatomicals have a different sign to their determinants (you can check with mri_info). In principle, reg-feat2anat should handle this but in some cases it does not. The 5.1 version will handle it properly. If they do have the same sign, then look through the log file and find the flirt command. Run that command by itself (you may need to run reg-feat2anat with --nocleanup to keep it from deleting the files). If that fails, then contact the FSL people
doug
Agnieszka Burzynska wrote:
Dear all, I am using the command reg-feat2anat to transform functional data pre-processed in FSL to the anatomical image in Freesurfer.
As suggested, I check the registration with reg-feat2anat --feat run2.feat <manual.
I have 3 runs of functional data for each subject. Surprisingly, some runs are registered well but others not (even of the same subject, so using the same reference anatomical images: see the attached image. Upper row: bad registration of one run and lower row: good registration of another run).
I have no problem to manually correct the registrations, it works well, but as I have many subjects, I wonder if there may be a way to change some options of reg-feat2anat to make it work better? Can you see from the attached image what is the problem with my images? (I viewed the functional and anatomical data of a few subjects, some woth good and some with bad registration and see no difference or flip)
Thanks a lot for your help! Aga
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer