Hello,

I'm trying to perform a pet ROI analysis on an image recon from a T1.  
I have to perform the analysis without normalization and partial volume correction.  

I'm having problems with regions with few or no voxels which makes me think I'm doing the analysis completely wrong .  

With this command I obtain negative values   for these regions:

mri_gtmpvc --i ${SUBJECTS_DIR}/PET.nii.gz --reg ${SUBJECTS_DIR}/PET.reg.lta --seg gtmseg100.mgz  --no-rescale --auto-mask 1 .01 --mgx .01 --o $GTM_DIR/gtmpvc_PET.output

While with this I get many counts even for regions without voxels:

mri_gtmpvc --i ${SUBJECTS_DIR}/PET.nii.gz --reg ${SUBJECTS_DIR}/PET.reg.lta --seg gtmseg100.mgz --no-rescale --auto-mask 1.01 --no-tfe --o $GTM_DIR/gtmpvc_PET.output

Could you help me understand my mistake.  
Thank you very much.

Parodi Matteo.

I will insert the commands I used before the ROI analysis:

recon-all -s Subj -i $IMAGED  -all -threads 10

gtmseg --keep-cc  --s Subj --xcerseg --o gtmseg100.mgz

mri_coreg --s Subj --mov ${SUBJECTS_DIR}/PET.nii.gz --reg ${SUBJECTS_DIR}/PET.reg.lta --threads 10