Hello Freesurfers,
Using FS 4.5, I am attempting to do an ROI analysis based on activations
seen in the group contrast. I'd like these ROI's to be a sphere around the
peak voxel of each activation. Using a matlab program written by a
colleague, I was able to generate spherical masks around the peak voxel in
talairach functional space (i.e. no registration file is needed to view
these masks overlayed on fsaverage and the masks have dimension 76 x 76 x 93
and voxel size 2 x 2 x 2 mm). I'd like to transfer these masks to native
functional space (the same space as each subject's beta.nii file) so that I
might extract the timecourse activity in matlab, and I was wondering 1) if
what I list below is a valid way to do this transformation, and 2) if there
is an alternative/superior way to accomplish this.
After a lot of experimenting, the following series of commands appears to
work:
1. mri_convert --input_volume group_sphere_mask.nii.gz --output_volume
output1.nii.gz --ait Subject/mri/transforms/
talairach.xfm
--this gives me an error message:
ERROR MSGS -- INFO: Trying to get the source volume information from the
transform name
corRead(): can't open file Subject/mri/transforms/../orig/COR-.info
INFO: Failed to find Subject/mri/transforms/../orig as a source volume.
The inverse c_(ras) may not be valid.
2. mri_vol2vol --mov output1.nii.gz --targ Subject/bold/004/f.nii
--regheader --o output2.nii.gz --interp nearest
--this gives me a file with the right dimensions, but the values in the mask
are no longer = 1, so then I use
3. mri_binarize -i output2.nii.gz -o output3.nii.gz --min .02
--where the min is a somewhat arbitrary threshold that seems to capture most
of the voxels that were in the mask
When I overlay this file on the subjects anatomical using: tkmedit Subject
orig.mgz -overlay output3.nii.gz --reg Subject/bold/register.dat
I get a mask which seems to be in the right place, but is no longer truly
spherical and has the occasional voxel dropped from from the mask
First and foremost, I would just like to know if this is valid, but I would
also be eager to learn about any methods which might preserve the spherical
shape and/or prevent any voxel dropout (and that don't give me error
messages).
Thanks,
Sam Cartmell
Research Assistant
Visual Cognitive Neuroscience Laboratory, Dept Psychology
Yale University