Question about 3dUnifize outcome

AFNI Message Board
1

Hi all,

I am working on prepping images for preprocessing and running my MPRAGE files through 3dUnifize prior to using 3dSkullStrip.

I can definitely see that intensities have been normalized after 3dUnifize, but I am concerned about some peripheral areas of the white matter that still appear brighter than the rest, specifically areas close to the cortex. I tried adding the -gm option and the difference in those areas of the white matter still existed.

Is this something I should be concerned about? If so, how could I go about improving the images? I will attach photos of the before and after to give you all an idea of what I mean. In this example, the original is on the left and the unifized image is on the right.

I am using Version AFNI_17.2.09 (Aug 22 2017) and this is the line of code I ran: 3dUnifize -prefix 206_MPRAGE_unifize -input 206_anat_+orig.BRIK

Thanks in advance for your help!

Best,
Samantha

Orig_vs_Unifized.png
Orig_vs_Unifized.png1169×538 319 KB

2

I see what you mean. However, I don’t understand what your concern might be in “prepping images for preprocessing”. What follows the preprocessing? If you are going to do quantitative image segmentation (i.e., partition into WM and GM and count voxels), then the effect you see might be a concern. If you are going on to skull stripping or the like, then I don’t think these small brightness patches will matter much.

3

Hi Bob,

Thanks for the input. While we are not doing any segmentation, we are prepping our images to create a template using @toMNI_Awarp and @toMNI_Qwarpar. The template will be used to compare younger and older adult samples. Do you think the small brightness patches will matter in that case?

Also, I noticed after my last message that the 3dunifized images show this bright rimming/halo effect around the outside of the skull - I will attach another photo here. Might this negatively impact the skull stripping and the template creation, and what would cause this effect on the images?

Thanks for your help!
Sam Williams

3dunifize_halo.png

4

I don’t think the small brightness patches will affect the template formation significantly, but I can’t really know.

As far as the “halo” goes, it won’t affect the skull strip in any way. That tiny shell far away from the head will be eliminated almost instantly in the skull stripping process.