Overlap parcellation - Freesurfer

23 Nov 2016


      Hello,
I’d like to ask couple of questions regarding parcellation:
First, we used  aparc.DKTatalas40.annot  implemented in Freesurfer;
1) we used the command mris_make_face_parcellation to have a first parcellation:
mri_annotation2label --subject ${subjid1} --hemi lh --annotation ${SUBJECTS_DIR}/${subjid1}/label/lh.aparc.DKTatalas40.annot   --outdir ${SUBJECTS_DIR}/${subjid1}/new_label/lh
mri_annotation2label --subject ${subjid1} --hemi rh --annotation ${SUBJECTS_DIR}/${subjid1}/label/rh.
aparc.DKTatalas40.annot --outdir ${SUBJECTS_DIR}/${subjid1}/new_label/rh
mri_convert ${SUBJECTS_DIR}/${subjid1}/func/splitted_0000.nii.gz ${SUBJECTS_DIR}/${subjid1}/func/splitted_0000.mgz
mri_convert $d/1mm_low/aseg.mgz -o $d/aseg.mgz --reslice_like ${SUBJECTS_DIR}/${subjid1}/func/splitted_0000.mgz --resample_type nearest
mris_volmask_novtk --label_left_white 2 --label_left_ribbon 3 --label_right_white 41 --label_right_ribbon 42 --save_ribbon --save_distance $subjid1
tkregister2 --mov ${SUBJECTS_DIR}/${subjid1}/mri/T1.mgz --noedit --s ${subjid1} --regheader --reg ${SUBJECTS_DIR}/${subjid1}/mri/register.dat
tkregister2 --mov ${SUBJECTS_DIR}/${subjid1}/func/splitted_0000.mgz --noedit --s ${subjid1} --regheader --reg ${SUBJECTS_DIR}/${subjid1}/mri/register_functional1.dat
P.S. splitted_0000.mgz  is a volume from fMRI data.
mri_label2vol --label ${f} --temp ${SUBJECTS_DIR}/${subjid1}/mri/T1.mgz --subject ${subjid1} --hemi ${lh_rh} --surf white  --fillthresh 0.5 --proj frac 0 1 0.01 --o ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}.nii.gz $register_command
mri_binarize --dilate 1 --erode 1 --i ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}.nii.gz --o ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}.nii.gz --min 1
mris_calc -o ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}_ribb_ref.nii.gz ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}.nii.gz mul ${SUBJECTS_DIR}/${subjid1}/mri/${lh_rh}.ribbon.mgz
Is this the correct way to make the parcellation or I would need to do something else?
I also saw that most of the mask volumes created are overlapped each other in the low resolution (fMRI data) (fabout 10%) Is there any way to overcome the issue?
2)Then to obtain a finer parc, we have tried another method based on ico (icosahedron) that used as input fsaverage to create file.annot using mri_surf2surf that resampled one CorticalSurface onto another.This is the order of commands we used
mris_make_face_parcellation ${SUBJECTS_DIR}/${subjid2ref}/surf/lh.sphere.reg ${FREESURFER_HOME}/lib/bem/${ic}.tri ${SUBJECTS_DIR}/${subjid2ref}/label/lh.${ic}.annot
mris_make_face_parcellation ${SUBJECTS_DIR}/${subjid2ref}/surf/rh.sphere.reg ${FREESURFER_HOME}/lib/bem/${ic}.tri ${SUBJECTS_DIR}/${subjid2ref}/label/rh.${ic}.annot
mri_surf2surf --srcsubject ${subjid2ref} --trgsubject ${subjid1} --hemi lh --sval-annot ${SUBJECTS_DIR}/${subjid2ref}/label/lh.${ic}.annot --tval ${SUBJECTS_DIR}/${subjid1}/label/lh.${ic}.${subjid2ref}2${subjid1}.annot
mri_surf2surf --srcsubject ${subjid2ref} --trgsubject ${subjid1} --hemi rh --sval-annot ${SUBJECTS_DIR}/${subjid2ref}/label/rh.${ic}.annot --tval ${SUBJECTS_DIR}/${subjid1}/label/rh.${ic}.${subjid2ref}2${subjid1}.annot
mri_annotation2label --subject ${subjid1} --hemi lh --annotation ${SUBJECTS_DIR}/${subjid1}/label/ rh.${ic}.${subjid2ref}2${subjid1}.annot --outdir ${SUBJECTS_DIR}/${subjid1}/new_label/lh
mri_binarize --dilate 1 --erode 1 --i ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}.nii.gz --o ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}.nii.gz --min 1
mris_calc -o ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}_ribb_ref.nii.gz ${SUBJECTS_DIR}/${subjid1}/new_mri_vol_highres/${lh_rh}/${filename}.nii.gz mul ${SUBJECTS_DIR}/${subjid1}/mri/${lh_rh}.ribbon.mgz
Is this also a  correct way to make the parcellation or I would need to do something else?
I also saw that, also in this case, most of the mask volumes created are overlapped each other in the low resolution (fMRI data) (about 10%) Is there any way to overcome the issue?
Thank you very much.