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