Is the variation in the shape of inflated surfaces common across subjects?

AFNI Message Board
1

Hello,
I would like to ask about surface-based analysis using SUMA and FreeSurfer.
I would like to perform GLM analyses on the standard inflated surface to check activation locations across subjects. If I understand correctly, I can do this analysis by adding the tlrc and surf blocks to the pipeline based on the example 8 on the afni_proc.py website. I think I also need to specify the standard-space spec file, such as std.141.${subj_id}_?h.spec , in the -surf_spec option.

I ran my script and it seemed to be completed without any errors, but I’ve noticed that the shapes of the inflated brains still differ across subjects, even though I included the tlrc block. For example, the outlines of the inflated surfaces and the positions of sulci and gyri vary between subjects.

Is it common to see this variation in the shapes of inflated surfaces across subjects?

2

Howdy-

To do the processing on the surface with afni_proc.py, you would include the surf block, yes, but not the tlrc block. The latter block is for sending data to a volumetric standard space; the surface processing is in lieu of doing that. And yes, Ex. 8 in the AP examples is a good one to look at for surface-based analysis.

When you have run FreeSurfer's recon-all on a subject anatomical to estimate meshes and parcellations, you would then run AFNI's @SUMA_Make_Spec_FS to convert the *.mgz volumetric files to *.nii* format and, quite importantly, the meshes are not only converted to GIFTI format but they are standardized in the process. That is, if recon-all has done its job well, index/node 5000 in all std.141.*.lh.* surfaces should correspond to the exact same anatomical location across subjects (and every subject has the same number of nodes). So, in this way, using SUMA's standardized surfaces helps you prepare for group analysis.

Note that the shape of the brain meshes will still the quirky, heterogeneous set of bumps and wiggles that defines an individual. That geometric aspect of the mesh is maintained. So, sub-001's mesh looks different than sub-002's mesh. That is OK. But the topology of the meshes match. So, again, node #123 in each mesh should correspond to the same location. This is discussed in detail in the AFNI Academy lecture series, for the SUMA playlist here, esp. starting around 5:30. You can get+open the suma.pdf handout that includes this via running the following (or just getting the AFNI Bootcamp download):

afni_open -aw suma.pdf

Note that this recent paper also presents+discusses various processing with AP:

  • Reynolds RC, Glen DR, Chen G, Saad ZS, Cox RW, Taylor PA (2024). Processing, evaluating and understanding FMRI data with afni_proc.py. Imaging Neuroscience 2:1-52. https://doi.org/10.1162/imag_a_00347

... and might be worth looking at for what the blocks do. There is an associated GitHub repo of scripts, which is linked here.

--pt

3

Thank you so much for your detailed reply and for sharing very useful information.
I understand that I do not need the tlrc block for surface-based analysis.
Also, I will check out AFNI's resources you recommended to better understand surface-based analysis, especially geometry and topology aspect of surface-based analysis.