Mri binarize

If you have data from different time points, as with the booth data, and want to make comparisons of ROI's across time points you'll want to be looking at the same voxels. Here's how to do that with freesurfer.

At this point you should have conducted group stats (mri_glmfit) and pulled out whatever ROI's are of interest. Now you'll need to convert your group average .annot label file to a volume, so that we can use the binarize function.

For T1 & T2

mri_label2vol --annot T1_rh.200functional_subclusters.annot --temp T1.mgz --o rh.T1_200funclust.mgz --subject fsaverage --hemi rh --reg reg.mni152.2mm.dat
mri_label2vol --annot T1_lh.200functional_subclusters.annot --temp T1.mgz --o lh.T1_200funclust.mgz --subject fsaverage --hemi lh --reg reg.mni152.2mm.dat

mri_label2vol --annot T2_rh.200functional_subclusters.annot --temp T1.mgz --o rh.T2_200funclust.mgz --subject fsaverage --hemi rh --reg reg.mni152.2mm.dat
mri_label2vol --annot T2_lh.200functional_subclusters.annot --temp T1.mgz --o lh.T2_200funclust.mgz --subject fsaverage --hemi lh --reg reg.mni152.2mm.dat

Questions: Does it make sense to use fsaverage's T1.mgz as a template, vs the brainmask? Registery reg.mni152.2mm.dat vs. reg.2mm.mni152.dat ??

Now that you have volumes, use the mri_binarize function to set voxels of min values (0.1 seems to grab all voxels) to 1 for T1 and 2 for T2 (specified by binval).

For Left and Right Hemi

mri_binarize --i lh.T1_200funclust.mgz --min 0.1 --binval 1 --o lh.T1_bin.mgz
mri_binarize --i lh.T2_200funclust.mgz --min 0.1 --binval 2 --o lh.T2_bin.mgz

mri_binarize --i rh.T1_200funclust.mgz --min 0.1 --binval 1 --o rh.T1_bin.mgz
mri_binarize --i rh.T2_200funclust.mgz --min 0.1 --binval 2 --o rh.T2_bin.mgz

Use fscalc to add T1 and T2 for left and right hemi's. This will cause voxels only in T1 to be one, only in T2 to be 2, and voxels in both to be 3.

fscalc lh.T1_bin.mgz add lh.T2_bin.mgz --o lh.T1andT2.mgz 
fscalc rh.T1_bin.mgz add rh.T2_bin.mgz --o rh.T1andT2.mgz 

The venn diagram is cool, but for the analysis you just want the voxels that are coded as 3. When converting the volume back to a label file, use mri_cor2label and specify labelid as 3.

mri_cor2label
--i lh.T1andT2.mgz
--id 3
--l lh.T1_T2_200subclust.annot

mri_cor2label
--i rh.T1andT2.mgz
--id 3
--l rh.T1_T2_200subclust.annot

This also has options for writing label volume in file and interpreting input volume as surface overlay, which may be more appropriate?

Then we convert our new volumes back to label files (can also use mri_cor2label), for both left and right hemi's.

mris_seg2annot 
--seg lh.T1andT2.mgz 
--ctab T1_T2_200subclust_lh_CLUT.txt
--s fsaverage
--h lh
--o lh.T1_T2_200subclust.annot

And, finally apply the new annot file to each subject, for Left and Right Hemi (to run this template as is, you'll have to move the new .annot files into the fsaverage label folder)

mri_surf2surf \
--srcsubject fsaverage \
--trgsubject FS_T1_501 \
--hemi lh \
--sval-annot $SUBJECTS_DIR/fsaverage/label/lh.T1_T2_200subclust.annot \
--tval $SUBJECTS_DIR/FS_T1_501/label/lh.T1_T2_200subclust.annot

mri_surf2surf \
--srcsubject fsaverage \
--trgsubject FS_T1_501 \
--hemi rh \
--sval-annot $SUBJECTS_DIR/fsaverage/label/rh.T1_T2_200subclust.annot \
--tval $SUBJECTS_DIR/FS_T1_501/label/rh.T1_T2_200subclust.annot