Nayyereh,
It's possible that the voxel count for thalamus is different because aseg.stats is derived from aseg.mgz whereas you're using aparc+aseg.mgz - theoretically it shouldn't have a huge difference. How big of a difference are you noticing?
You could try mri_convert with -rl option to reslice the .mgz volumes to nifti, bearing in mind the different space you need to transform into step-by-step (Freesurfer -> Highres Structural space -> DTI space)
As long as the seed is in the correct space probtrackx should run fine (even if it's not isotropic).
I suppose you could use the white matter mask as some constraint mask to keep the tract only in white matter.
hope this helps
Danny
________________________________________
From: FSL - FMRIB's Software Library [[log in to unmask]] On Behalf Of Nayyereh Ghazi [[log in to unmask]]
Sent: Saturday, February 06, 2016 11:57 PM
To: [log in to unmask]
Subject: [FSL] seeds for probtrackx
Dear Experts,
I want to create some cortical and subcortical masks to use as seed regions in fsl’s Probtrackx and have faced some questions that ask you help me with them.
To extract subcortical regions I have used the below commands (e.g. for thalamus_r):
{
mri_label2vol \
--seg $SUBJECT_DIR/${subject}/mri/aparc+aseg.mgz \
--fillthresh 0 > /dev/null \
--identity \
--temp $SUBJECT_DIR/${subject}/mri/orig.mgz \
--o ${datadir}/aparc+aseg.nii.gz \
mri_binarize --i ${datadir}/aparc+aseg.nii.gz --match 49 --o ${datadir}/thalamus_r.nii.gz
}
I noticed that the number of voxels of the created masks in this way are different from which is written in aseg.stats file. Could you please explain the reason and the way to make these numbers as close as possible.
For cortical regions I used both the above and below commands (e.g. for rh.superiortemporal):
{
mri_annotation2label --subject ${subject} \
--hemi rh \
--annotation $SUBJECTS_DIR/${subject}/label/rh.aparc.annot \
--outdir $SUBJECTS_DIR/${subject}/label \
--surface white
mri_label2vol \
--label rh.superiortemporal.label \
--temp $SUBJECT_DIR/${subject}/mri/orig.mgz \
--identity \
--o rh.superiortemporal.nii.gz \
--fillthresh 0 > /dev/null
}
Again I have noticed difference in number of voxels of the results from two methods for cortical regions and also a difference from aparc.stats’s numbers. My final goal in to use the masks in probtrackx to find the connection between subcortical structures and temporal lobe cortical regions and I wonder which method is more accurate to create cortical masks.
I also wonder if it is necessary to have images with isotropic voxel size for tractography? (Both for seed masks and the images we use to create xfm matrices)
My other question regards obtaining white matter tracts. In order to have just white matter fiber tracts, using wm waypoint masks is enough or we have to add some wm voxels to seed region masks to ensure the created tracts are in wm.
Your help is greatly appreciated.
Wishes,
Nayyereh
|