Sorry, I think I gave you some incorrect info. Try this instead:
tkregister2 --targ bold/004/fmc.bhdr --mov bold/009/fmc.bhdr --identity --reg /tmp/junk.reg
I just added the --identity to tkregister2. I tried to look at your data to verify that this was the problem, but it is sooooooooooo
Let me know how it goes.
doug
On Tue, 7 Mar 2006 raij@nmr.mgh.harvard.edu wrote:
Hi Doug,
I tried the technique you suggested for re-sampling functional volumes from different sessions but from the same subject (and scanner) into the same space, using mc-sess.
To clarify, here we have a situation where the same brain is scanned in the same scanner on more than one occasion (in this particular example, the subjects were taken out of the scanner half way through the lengthy functional recordings for 20 minutes to rest). Obviously the head position and rotation are almost always quite different in session 1 than session 2. It would make building contrasts and combining results across sessions (but within subjects) much more straigthforward if the functional volumes from the two "sessions" could be re-sampled into the same space.
I put all my runs (2 sessions, 3 runs each = total of 6 runs) from the same subject and scanner in one bold directory, ran mc-sess, and got confusing results.
When I check the alignment with tkregister2, the result looks very bad - the volumes from different sessions are clearly not in the same space (1-3 cm difference).
However if I ignore the fact that the runs do not appear to be aligned, and run spmregister-sess just once for all runs (hence there is only one register.dat file) and then check the registration with tkregister-sess, runs from both sessions seem to be similarly and well aligned with the hires anatomicals. Obviously the mis-alignment from mc-sess should be inherited to all following steps, so I must be doing something wrong. I wonder where the mistake is? Am I misusing the tkregister 2 command (see below)? Or is this a bug?
*** Here are the specifics: ***
Session 1 = runs 004, 005, 006 Session 2 = runs 009, 010, 011
on machine ai
source /usr/local/freesurfer/nmr-dev-env setenv SUBJECTS_DIR /space/cognito/5/users/raij/subjects_mri setenv SUBJECT avml19
cd /space/cognito/21/users/raij/avml_fmri_mctogether/avml19_session_BRISI13_ISI1TR_BERT
# MOTION CORRECT mc-sess -sf sessid -df sesspar
# SPATIAL SMOOTHING AND INORM, OUTSTEM = ifmcsm6 spatialsmooth-sess -sf sessid -df sesspar -fwhm 6 -i fmc -o ifmcsm6
# CHECK ALIGNMENT OF RUN 004 VS. 009 WITH TKREGISTER2 - LOOKS BAD tkregister2 --targ bold/004/fmc.bhdr --mov bold/009/fmc.bhdr --regheader --reg /tmp/junk.reg
# RUN SPMREGISTER ONCE FOR ALL RUNS TOGETHER DESPITE THAT ALIGMENT APPEARS BAD spmregister-sess -sf sessid -df sesspar -funcstem fmc
# CHECK REGISTRATIONS (ONLY ONE bold/register.dat FILE) - LOOKS OK FOR BOTH, AND VERY SIMILAR FOR RUNS 004 AND 009??? tkregister-sess -sf sessid -df sesspar -funcstem fmc -run 004 & tkregister-sess -sf sessid -df sesspar -funcstem fmc -run 009 &
Also, the defaults for spmregister-sess and tkregister-sess funcstems are "f". I hope, because my setup does not overwrite the mc-sess output with the output of spatialsmooth-sess, that I am correct in using the "fmc" stem instead of "f"? If I have understood corectly, this has the advantage that I can register motion corrected but not spatially smoothed functional data, which might(?) at least theoretically give better results than registering functionals without motion correction... though the advantage might be too small to be relevant compared to less than perfect registration accuracy. Any thoughts regarding this? I tried the registration with the stem "f" as well, and the problems remained the same, so at least this does not seem to be the cause.
Thanks,
Tommi
hi doug
i think i have actually already done what you suggest as my bold directory consists of all the retinotopy runs (in subdirs 001-010). this is the 'ls' of the bold directory:
001/ 003/ 005/ 007/ 009/ register.dat same-sess-reg.dat seq.info 002/ 004/ 006/ 008/ 010/ rtopy/ scripts/
i ran the motion correction from the study directory as:
mc-sess -sf sessid -df sesspar
and it completed successfully.
-so will that have aligned all the runs together? if so, what can explain the fact that the field sign map is not continuous when overlaid on the cortex and shows gaps in the blue/yellow colouring?
thanks!
jane
In message 43D5258C.3030906@nmr.mgh.harvard.edu Doug Greve greve@nmr.mgh.harvard.edu writes:
You can try putting all the runs into a single bold directory. The motion correction will (do it's best to) align them.
doug
Jane Aspell wrote:
Hi
My question was asked last year by Tommi Raij (see email pasted below)
but I
couldn't find a reply to it in the archives. I am analysing retinotopy
data
that was collected over several sessions but I cannot see how to
register
scans from multiple sessions together, and to the subject's anatomical
scan.
Although Fs-Fast appears to combine the data from multiple sessions
during the
retinotopy analysis it only seems to be possible to use one
registration file
(register.dat) when overlaying the results onto the anatomical. I would
have
thought that registering scans from multiple sessions together would be
a
feature that many people would require. Is there anyway to do it? Tommi
gives
some possible methods in his email below. When I do try and overlay my data onto a surface I see 'gaps' in the blue/yellow colouring of the fieldsign map overlaid onto the cortical
surface.
I wonder if this could be due to the inaccuracy of the registration
given that
I only used the register.dat from the registration of the first
session.
hope that makes sense! i'd be grateful for any help with this!
Jane
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Tommi Raij, M.D., Ph.D. MGH/MIT/HMS Athinoula A. Martinos Center for Biomedical Imaging Bldg 149, 13th St Charlestown, MA 02129 U.S.A.
raij@nmr.mgh.harvard.edu