ds114_test1_freesurfer/fsaverage/mri.2mm/README

fsaverage/mri.2mm

These are volumes sampled into a 2mm space. The primary purpose of
these is to support volume-based fMRI analysis as performed in
FS-FAST. This 2mm space is the space that is used by default in
FS-FAST to perform "talairach" group analysis (ie, in the mni305
space). The raw functional data are sampled into this space when
preproc-sess is run with the -mni305 flag. When a group analysis is
done, the output space will be this 2mm space. It is possible to use a
1mm space, but this can create huge files.

These files can be used to render the statistical results on a volume
(though this is not really necessary as they can be rendered directly
on the 1mm volumes). Also, the aseg.mgz created can be used to
generate labels/masks of subcortical structures directly in the group
average space. These can be used with mri_volcluster.

These files were created with the following commands:

# Resample each of these volumes using trilin
foreach vol (orig brain brainmask mni305.cor T1)
  mri_vol2vol --mov ../mri/$vol.mgz --s fsaverage --tal \
  --o $vol.mgz --no-save-reg
end

# Resample the aseg 
mri_label2vol --seg ../mri/aseg.mgz --temp orig.mgz \
  --regheader ../mri/orig.mgz --o aseg.mgz

Note: we don't want to do aparc+aseg because the surface-based 
labels should really be used in a surface-based analysis.

# checks
tkmedit -f T1.mgz -aux brain.mgz -seg ./aseg.mgz
tkregister2 --mov ./orig.mgz --s fsaverage --regheader --reg junk

#-----------------------------------------------------------

Creation of subcortical mask. A challenge for doing analyses in three
ROIs (left hemi, righ hemi, subcortical) is making the ROIs mutally
exclusive while not excluding any voxels. This is particularly hard in
this analysis because a cortical voxel in one subject can map to a
subcortical voxel in another when using the transform to MNI305
space. 

A map of the subcortical structures for each of the buckner 40 where
mapped into the mni305 2mm space. A probability map was then
created. This was repeated for the cortical ribbon and for cerebellum
by itself. 

1. An initial mask was created taking voxels with any 
subcortical origins in the native space as long as they had fewer than
80% cortex. 

2. Cerebellum is a special case because it is so close to fusiform
that it is easy to get bleed over. For this case, a mask was made of
any cerebellum voxels that had at least 20% cortex. Any voxels in this
mask were then excluded from the mask from step 1 to create a new
mask.

3. Little islands were excluded by running connected components and
taking the largest cluster. 

4. This mask was then dialted by one voxel then eroded by one voxel to
remove holes and make the final mask less jagged.


# ----- Below For Version 5 --------------------#
Everything above is for 5.1 and higher.

The mask is created based on mri/subcort.prob.mgz which was created
with make_average_subcort. subcort.prob.mgz is the raw probability of
a voxel being in a subcortical gray matter structure based on the
Buckner40. Inevitably, this mask will include more than just
subcortial gray matter structures, but we would rather have the mask
be too big than too small.

# Resample into 2mm space
mri_vol2vol --mov ../mri/subcort.prob.mgz \
  --s fsaverage --tal --o subcort.prob.mgz \
  --no-save-reg

Threshold at .05 (meaning that at least 5% of the subjects at 
a voxel must have a subcortical label). Also dilate by 2 to
expand the mask.

mri_binarize --i subcort.prob.mgz --min .05 \
  --dilate 2 --o subcort.mask2.mgz

Erode by 1. The net result of dilating by 2 then eroding
by 1 is that holes are filled in and the edges are 
a little smoother.

mri_binarize --i subcort.mask2.mgz --min .5 \
  --erode 1 --o subcort.mask.mgz

rm subcort.mask2.mgz

#-----------------------------------------------------------

Creation of registration matrix between the 2mm fsaverage/mni305 space
and the 2mm mni152 space. This can be used to convert data in the 2mm
fsaverage/mni305 space into the mni152 2mm space.

set d = $SUBJECTS_DIR/fsaverage

# Create registration matrix between the full 256^3, 1mm3 
# volume and the 2mm space (simple regheader)
tkregister2 --mov $d/mri.2mm/brain.mgz --targ $d/mri/brain.mgz \
  --regheader --reg $d/mri.2mm/reg.2mm.dat --noedit

# Registration between the fsaverage/mni305 subject (full 256^3, 1mm3)
# and the 2mm mni152 space.  This was created by hand.
# $FREESURFER_HOME/average/mni152.register.dat 

# Compute the registration between the 2mm fsaverage space and the
# mni152 2mm space by concatenating the two matrices above.
mri_matrix_multiply -im $d/mri.2mm/reg.2mm.dat \
  -iim $FREESURFER_HOME/average/mni152.register.dat \
  -om $d/mri.2mm/reg.2mm.mni152.dat

# Check the registration
tkregister2 --mov $d/mri.2mm/brain.mgz \
  --targ $FSLDIR/data/standard/MNI152_T1_2mm.nii.gz \
   --reg $d/mri.2mm/reg.2mm.mni152.dat

# To view without reslicing
tkmedit -f $FSLDIR/data/standard/MNI152_T1_2mm.nii.gz \
  -overlay sig.nii -reg reg.2mm.mni152.dat

# To do the conversion/reslicing, run something like
mri_vol2vol --mov sig.nii --reg reg.2mm.mni152.dat \
  --targ $FSLDIR/data/standard/MNI152_T1_2mm.nii.gz \
  --o sig.mni152.nii

# To view with reslicing
tkmedit -f $FSLDIR/data/standard/MNI152_T1_2mm.nii.gz \
  -overlay sig.mni152.nii