Possible Normalization Problem

AFNI Message Board
1

afni_proc.py
-subj_id s${subject}
-script proc.Rests${subject}.tcsh
-dsets s${subject}/func/orig/REST.nii
-copy_anat s${subject}/anat/orig/t1_MPRAGE.nii
-blocks despike align tlrc volreg mask blur regress
-volreg_align_to third
-volreg_align_e2a
-volreg_tlrc_warp
-tlrc_base MNI_avg152T1+tlrc
-mask_segment_anat yes
-mask_segment_erode yes
-blur_size 6.0
-regress_censor_motion 0.2
-regress_bandpass 0.01 0.08
-regress_apply_mot_types demean deriv
-regress_run_clustsim no
-regress_est_blur_errts

I am interested in the residuals (errts file) but 3dinfo has full_mask.s1+tlrc marked as MNI and errts.s1.tproject+tlrc as TLRC. I am assuming that this is some sort of mistake given that volreg_tlrc_warp should have applied the transformation to the EPI. What might be the cause of this mismatch?

Thanks

2

Hmmm, I see only MNI space on all results for an almost identical
command.

Would you verify this by going to the results directory and running:

3dinfo -space -av_space -prefix *.HEAD

Also, what does “afni -ver” show?

  • rick
3

Thanks for your response,

3dinfo -space -av_space for my errts file gives me

TLRC +tlrc errts.s1.tproject

I forgot to mention that I had to make some changes to my tcsh script. Between the mask making stage and the blurring stage I inserted a stage that calls on matlab code the implements CompCor. The output is then plugged into the blurring stage. The only other change made involved changing the cenmode mode from ZERO to NTRP. Maybe something happens when jumping back to AFNI from the CompCor script?

4

That certainly seems to be where you are losing it. That
3dinfo command should make it clear, I would expect.

If the MATLAB processing drops the space, maybe you
could add it back in using “3drefit -space”.

  • rick
5

Thanks for the suggestion, I can try this but could my issue simply be a problem with the label? Maybe it is still in MNI but it has been labelled TLRC. How could it go from one space to another without a transform being applied to it? Are there any other methods of testing which template space it is in?

6

Yes, this is surely just a label of the space in the dataset.
It seems very unlikely that there is any transformation to
worry about.

  • rick
7

Okay thanks for your help