AFNI Message Board

Dear AFNI users-

We are very pleased to announce that the new AFNI Message Board framework is up! Please join us at:

https://discuss.afni.nimh.nih.gov

Existing user accounts have been migrated, so returning users can login by requesting a password reset. New users can create accounts, as well, through a standard account creation process. Please note that these setup emails might initially go to spam folders (esp. for NIH users!), so please check those locations in the beginning.

The current Message Board discussion threads have been migrated to the new framework. The current Message Board will remain visible, but read-only, for a little while.

Sincerely, AFNI HQ

History of AFNI updates  

|
Anonymous User
August 31, 2012 02:09AM
Daniel,

Thank you very much for your reply. I assumed that ANTS wants the affine transformation matrix as the manual mentions that the first nine values represent "the matrix component (rotation, scaling, shearing all composed together)." So, I did the following: I re-ordered the values in the matrix to match the ANTS requirements, passed the adapted matrix to ANTS, manually tried out different centers of rotation and compared the results of the ANTS transformation with the aligned EPI from align_epi_anat.py.

As this did not really give good results, I finally resorted to writing a script that tried different centers of rotation and automatically evaluated the overlap of the aligned EPI generated by align_epi_anat.py and the aligned EPI generated by ANTS. After trying out > 20,000 different possibilities, it seems that it's not possible to achieve a decent match. Some results are close, but even the closest results have a mismatch of approximately 170,000 non-overlapping 2x2x2 mm voxels.

I'm currently trying to understand why that is and was wondering whether deobliquing might play a role in this. As my original anatomical and EPI images were not deobliqued, I now used 3dWarp to deoblique them and then repeated the alignment process with align_epi_anat.py. Interestingly, I now do get a much better match when I pass the resorted matrix to ANTS (the two aligned images are not identical, but very, very close). Therefore, I was wondering whether align_epi_anat.py internally always deobliques the images before registration and generates an affine transformation matrix that is only valid for deobliqued images? Does this make sense?

In addition, I found that the values that I need to enter as center of rotation in ANTS to get the good match seem to be in no obvious way related to the center of the image... Any insights on this would be very much appreciated.

Thanks again,
Michael

P.S. I used the following command to generate the transformation matrix: align_epi_anat.py -anat <ANAT> -epi <EPI> -epi_base 0 -epi2anat -suffix _align -deoblique off -giant_move -epi_strip 3dAutomask -cost lpc -anat_has_skull no -master_epi <ANAT>
Subject Author Posted

Convert affine transformation for ANTS

Anonymous User August 27, 2012 08:21AM

Re: Convert affine transformation for ANTS

Daniel Glen August 27, 2012 11:47AM

Re: Convert affine transformation for ANTS

Anonymous User August 31, 2012 02:09AM

Re: Convert affine transformation for ANTS

Daniel Glen August 31, 2012 07:16AM