Hi,
To run bbregister outside of the recon stream, I tried: setenv SUBJECTS_DIR /autofs/space/brownie_003/users/gcoombs/SUBJECTS_DIR cd $SUBJECTS_DIR bbregister --mov mri/nu.mgz --int ../mni305.cor.mgz --reg mri/transforms/talairachTEST.xfm --init-spm --t1 --s amac_TEST --tmp ./TEMP
but when I tried to view it with:
tkregister2 --mov amac_TEST/mri/nu.mgz --fstal --reg amac_TEST/mri/transforms/talairachTEST.xfm --surf orig --s amac_TEST
It looks the same as it did before, and I am concerned that the bad tal registration will affect the localization of fMRI data and ICV measurements. Does anyone have any advice on how to fix the registration?
Thanks, Garth
--
On Tue, 16 Mar 2010, Garth Coombs wrote:
Hi,
I recently ran into a problem with some of my tal registrations during the recon-all process (see below message to Martino's Tech). In order to fix the problem, it was suggested that we run bbregister outside of the recon-all stream and then feed that registration into the subsequent recon stages. Does this make sense, and how would I go about doing this? I'm not sure what the --mov file should be, or if there is a way to specify fstal as the reference structural.
Thanks for your help!
-Garth
Hi Doug & FreeSurfers,
I recently re-ran a large number of recons and have been checking to make sure they all ran correctly, and I have come up with a couple of questions about the talairach transform/recon checking.
For some of my recons, the green line outlining the white matter on the individual brain extends slightly further than the cortex surface of the fstal brain. While it looks relatively well aligned (e.g. ventricles, brain stem), the fstal brain looks slightly too small for the subject's brain. Additionally, the green outline tends to extend into the skull of fstal in the sagittal plane. An example is:
setenv SUBJECTS_DIR /autofs/space/brownie_003/users/gcoombs/SUBJECTS_DIR/ tkregister2 --mgz --fstal --surf orig --s amac_anat
or
tkregister2 --mgz --fstal --surf orig --s amra_anat
In order to fix these talairach transforms, it seems like the best option would be to scale the brain; however, we will be doing a cortical thickness analysis on this group, and we do not want to scale the brain if it will somehow affect the outcomes. Is this something we should worry about?
Another question we have is about general recon quality control: when checking that the recons ran smoothly, what should we look for? I look to make sure that the pial and white matter surface lines don't include dura or any other matter and that they follow the structures correctly, skull stripping, and intensity normalization, but is there anything else that is particularly important?
Any advice would be greatly appreciated!
Thanks, Garth