maybe sure you specify nearest neighbor interpolation when converting the aseg. You will lose some accuracy with the resampling of course.
cheers Bruce On Fri, 8 Mar 2013, Tal, Assaf wrote:
Dear FreeSurfer community,
I am trying to obtain subcortical segmentation masks for a group of healthy volunteers using FreeSurfer v5.1 running on Ubuntu 12.10. I need the masks to be in the native space of the original 1mm^3-isotropic 256*256*160 T1-weighted images. I found a helpful link which seems to answer my need: http://surfer.nmr.mgh.harvard.edu/fswiki/FsAnat-to-NativeAnat
Running it seems to produce good results when overlaid on top of the original images. However, I couldn't find any mention in either the newsgroups or the wiki as to whether this affects the subcortical masks in any (significant?) way. What sort of differences should I expect between the masks in the "freesurfer space" and the ones converted back to native space? Are there any structures or artifacts I should mindful of?
Thank you!
Assaf
PS Just to make things concrete, assuming I have a patient called Sam, then, after running recon-all, I typed in:
cd $SUBJECTS_DIR/Sam/mri
mri_vol2vol --mov brain.mgz --targ rawavg.mgz --regheader --o brain-in-rawavg.mgz
mri_label2vol --seg aseg.mgz --temp rawavg.mgz --o aseg-in-rawavg.mgz --regheader aseg.mgz
e.g., for cerebral WM: mri_extract_label aseg-in-rawavg.mgz 2 41 cerebralWMNative.mgh
mri_convert cerebralWMNative.mgh cerebralWMNative.img (this last step is used to create ANALYZE format images which is the format my code is already structured to read out)