SS Warper - Skull stripping shows deformed results

Dear all,

I ran SS Warper over a dataset of 21 subjects. The results are perfect, except for two subjects. Lets call them subject 1 and 2.

  • Subject 1: in the quality control images of the file the whole head (skull plus brain) looks totally deformed after running SS Warper. Then, I ran skullstripping again but with adding “-deoblique” first, and “-deoblique_refitly” in a third try. Those two additional steps did also not yield better results. The images I added actually stem from the third try, but the results look like that for all skullstripping attempts.

  • Subject 2: the result looks very noisy. I realized that the raw data of this subject also has this noisy “flickering” effect. Is there a way to reduce this effect in AFNI? I am afraid that this effect leads to meaningless results in further computations concerning brain data.

Here is my script for skullstripping:

for subject in Subject1 Subject2 …
-input $directory_raw/anat1to3d+orig
-base MNI152_2009_template_SSW.nii.gz
-deoblique_refitly \ # added for the problematic subject1)
-subid $subject
-odir $directory_sswarper

The first two pictures are from subject 1. The third picture is from subject 2. The picture preview looks messy; please just open them via right click and open in a new tab to take a proper look at them.


Here is an update on subject 1, the subject with the massively misaligned skull and brain.

I ran SSwarper again with the following changes in the script:
-deoblique_refitly # This was already added before, but I left it in just in case
-cost_nl_init nmi # I changed this function from “lpa” (standard if not differently specified) to nmi.

Now the brain (and the skull) looks normal and fine. However, a new problem occured. A big part of the frontal upper lobe was clipped out.
After checking the raw data again, I assume that the problem stems from the fact that the raw anatomical data of this subject lacks precisely this part of the skull and brain. Take a look at the image below. The AFNI window is not artificially cut around the top of the head, but the window really shows the “end” of the data.

A colleague of mine suggested to exlucde this subject; I assume that you agree and that there is no proper way to save this subject?

Beside the problem with subject 1 I am interested in your opinion on the other subject with the noisy data in and around the brain. Where does this noise come from and is it possible to save this subject for further AFNI analyses?