Aparc2009rank regions

AFNI Message Board
1

Recently I’m working on ROI parcellation with aparc.a2009s+aseg_rank.nii file. (by using 3dcalc function)

It worked well with Hippocampus, Amygdala, and many other regions but for pulvinar (thalamus-pulvinar) and lateral geniculate

it didn’t work though they are in the list. (https://afni.nimh.nih.gov/pub/dist/edu/data/CD.expanded/AFNI_demos/FATCAT_DEMO/SUMA/aparc.a2009s+aseg_rank.niml.lt.log)
Can’t we use the regions with rank ‘#’ from the link above?

How can I use the two brain area with SUMA?

Thanks.

2

Hello,

Yes, the rank method, which was developed for the purposes of
plotting the parcellation as an overlay, had a flaw when used for
comparisons across subjects: if subjects were missing different
ROIs, then the ranking would subsequently assign different values
to higher-valued ROIs. That method has been replaced be a
fixed-value reordering. Please see this thread https://afni.nimh.nih.gov/afni/community/board/read.php?1,153419,153419#msg-153419 .

  • rick
3

Hello,

Thank you for your kind reply.

Refer to the thread you linked, I’ve run @SUMA_renumber_FS and got many of aparc.a2009s+aseg_REN files successfully.

After that, I ran

3dcalc -a aparc.a2009s+aseg_REN_all.nii’’ -expr ‘step(a)’ -prefix Pulvinar

but got nothing on the result overlay, same as before.

Should I use something else to point the region other than the name itself? (label, rank, or sth?)

And what do the ‘#’ rank region have in common? from (https://afni.nimh.nih.gov/pub/dist/edu/data/CD.expanded/AFNI_demos/FATCAT_DEMO/SUMA/aparc.a2009s+aseg_rank.niml.lt.log)

Left-Hippocampus, having rank 13 label 17 according to the old aparc_rank file is well done

but as I mentioned above, Thalamus-pulvinar , having rank # label 8012 doesn’t work.

How may I extract Thalamus-pulvinar with aparc.a2009s+aseg_REN?

Thank you and sorry for this long question.

-Joonyoung

4

Hi-

As far as I can tell, that region name does not exist in the FreeSurfer (FS) parcellation.

I grepped through the labeltable for anything with the word “thalamus”, and this is what was returned:


3dinfo -labeltable aparc.a2009s+aseg_REN_all.nii.gz | grep -i thalamus
 "6" "Left-Thalamus-Proper"
 "25" "Right-Thalamus-Proper"

Doing the same with “pulvinar” as the region returned nothing.

Using that notation with a region that is in the labeltable did give me mask of that region output


3dcalc -a aparc.a2009s+aseg_REN_all.nii.gz'<Left-Thalamus-Proper>' -expr 'a' -prefix TEST

–pt

5

Hello,

Thank you for your effort.

I think I have to find some other way or just use the one you’ve found (thalamus-proper).

Thank you!

-Joonyoung

6

Hi-

Here is a link to the FreeSurfer look-up table (LUT), which I assume they keep updated:
https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/AnatomicalROI/FreeSurferColorLUT

It should list all the regions they could possibly have. In reality, it seems a much shorter list actually gets returned by either the aparc+aseg or aparc+aseg2009 parcellations. I have tried to make a list of these and classify them by approx. tissue type (NB: the FS people gave me a list of some of what they consider ventricles and CSF, but there were a couple things that they hadn’t included as non-GM which I really think should be; they neither confirmed nor denied these regions, so please interpret as you please). The lists used by AFNI in the “REN” volumes can be found in the source code. For example, they could be listed by doing the following in a terminal (that is in tcsh):


set locafni = `which afni`
set dirafni = `dirname $locafni`
ls $dirafni/afni_fs*txt

–pt