Problem in subdividing atlas

Hello all,

I attempted to subdivide an ROI into 3 segments (in this case the inferior temporal gyrus from the Desikan-Killiany Atlas). In the right hemisphere, this ROI is parcellated into 3 subdivisions however in the left hemisphere, it seems to generate 4 subdivisions for some reason. The first subdivision (red trace on surface rendering) is rather small compared to the remaining 3, which are homologous to the 3 created for the right hemisphere. I hope the attached file helps articulate this better. Curiosuly, I do not see this issue when subdividing the same (or any) ROI from the Desikan-Killiany atlas in any other participants.

Would anyone have any input on this?

Thanks in advance,
Roozbeh

Hello,

You got this using the menu “Atlas > Subdivide selected scouts > [Fixed]”?
There is probably some topological cases that are not handled correctly by this function.
If you can reproduce this, maybe after re-importing this anatomy in a different subject, you can send me the cortex surface file and I’ll have a look at it.
(right-click on the cortex surface > File > Show in file explorer, upload the file tess_cortex_pial_low.mat somewhere, and send me the link in a separate email).

Francois

Thank you Francois. To answer your question, yes this was from running [B]Atlas > Subdivide selected scouts > [Fixed][/B]. I will go ahead and send you the link to the file.

Thanks,
Roozbeh

Hello,
I added an test to make sure that the number of subdivisions does not go over the requested number.
It will not solve all the issues with this function, but it fixes your bogus subdivision.
Thanks for reporting this bug.
Francois

Dear Francois and all,

I have also encountered a similar problem using the menu “Atlas > Subdivide selected scouts”.
I use the Colin27 anatomy, and the AAL atlas from the DPARSF templates (see here). I try subdividing the Temporal Lobe into a posterior and anterior part (similar problems arise when I try this on other scouts as well).

The original scout is 1276 vertices big, and its area is 214.99 cm2.
If I choose “Atlas > Subdivide selected scouts > [ Vertices ]” and enter 638, I get three scouts, with 1259, 16 and 1 vertices.
If I choose “Atlas > Subdivide selected scouts > [ Fixed ]” and enter 2, I also get three scouts, with 1259, 16 and 1 vertices.

When I try doing a similar subdivision on the Desikan-Killiany atlas, it works perfectly fine.

I should add that Desikan-Killiany looks much better than my own atlas, when interpolated on the cortex (see attached image).
Could it be that using the AAL atlas from DPARSF is a bad idea? It seems like the anatomical image is of the right size, but perhaps the AAL isn’t well fitted to the Colin27 image?

Many thanks,
best,

Roey


Hello,

How did you generate this atlas? With all those discontinuities, it doesn’t look like a surface atlas.
Was it a volume atlas that you imported as a list of scouts? If so, you cannot expect precise results, you should try to find a surface-based version of it.

The clustering algorithm we use for subdividing the regions is not very robust, if one region is contains some regions that are not connected, it will not be able to perform correctly.

Cheers,
Francois

Dear Francois,

I see. This was indeed a volume atlas that I used, and not a surface one. I’ll try a different path then…

Thanks for your quick and helpful reply!
Cheers,

Roey

Hello Francois,
I'm subdividing the selected scouts for each subject by clicking on the GUI. I just wonder if I can do it batchly using a script or something else.
Thank you in advance!

Duo

The function to subdivide the scouts is available in panel_scout.m:

It can be called with the syntax: panel_scout('SubdivideScouts', isAllScouts, Method, param)
However it is not easy to call from a script because it requires the surface to be loaded in the interface and the correct atlas selected manually.