Fetal atlases

62 views
Skip to first unread message

Gabriela Epihova

unread,
Jul 14, 2025, 10:46:55 AM7/14/25
to DSI Studio
Hi Frank,

Would it be possible to add these fetal atlases (regional segmentation files) to DSI studio please?: http://crl.med.harvard.edu/research/fetal_brain_atlas/
https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/WE9JVR

I can download them and put them in the atlas dsi folder but when i load them to the gui reconstructed file the alignment is off. 

My goal is to run whole brain tractography on fetal scan of different GA (20-25 poste gestational weeks and get a connectivity matrix parcellated with the fetal atlas that corresponds to the GA of each subject. 

Best,
Gabriela

Frank Yeh

unread,
Jul 14, 2025, 11:11:57 AM7/14/25
to gabriel...@gmail.com, DSI Studio
Hi Gabriela,

The atlas needs deformation to align. I will see if I can add them.

Best,
Frank
> --
> You received this message because you are subscribed to the Google Groups "DSI Studio" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/dsi-studio/75deeded-73b5-46fc-b575-fc4be82e4639n%40googlegroups.com.
Message has been deleted

Gabriela Epihova

unread,
Jul 14, 2025, 11:33:39 AM7/14/25
to DSI Studio
Thank you!

Frank Yeh

unread,
Jul 14, 2025, 12:34:34 PM7/14/25
to gabriel...@gmail.com, DSI Studio
Hi Gabriela,

You may try the following step:

1. Locate the 'atlas' folder in the dsi studio folder (MacOS->right
click to see package content)
2. Create a folder named 'human_ga28' (if you want the ga 28wk as the template)
3. Put fa and md nifti file into the folder and rename them as
human_ga28.QA.nii.gz and human_ga28.ISO.nii.gz, respectively.
4. Also, put the ga28 parcellation file in the same folder. You may
check other folder to see the label text file format.
5. Restart DSI Studio, and load GQI-reconstructed FIB file at SteP T3
Fiber Tracking.
6. At step T2a, select human_ga28 and see if the normalization and the
atlas works.

If there is any issue in any of the steps, please let me know.

Best,
Frank

Gabriela Epihova

unread,
Jul 14, 2025, 2:32:15 PM7/14/25
to DSI Studio
Hi Frank,

Thank you for these steps. First, to try these I used this atlas: https://gin.g-node.org/kcl_cdb/fetal_brain_mri_atlas 
as it was the only one I could find that has both parcellations as well as fa/md. It's not ideal as it doesn't have regions, only tissue, but just wanted to try to see if it would work. 

I did everything but when I create a subfolder (human_ga25_atlas) in the atlas folder (screenshot attached)  it doesn't allow me to choose the atlas int he fiber tracking stage (screenshot). I think it's because there is no .fib file in the folder. When I put all of the human_ga25_atlas files (parcellation, iso, qa and txt) in the neonate atlas folder, it detects it but it doesn't align the fetal atlas correctly (screenshot). 

Any other ideas of how to align the atlas template space to the subjects' space?

Best,
Gabriela
dsi atlas folder.png
dsi.png
Screenshot dsi fetal.png

Frank Yeh

unread,
Jul 14, 2025, 2:51:17 PM7/14/25
to gabriel...@gmail.com, DSI Studio
> I did everything but when I create a subfolder (human_ga25_atlas) in the atlas folder (screenshot attached) it doesn't allow me to choose the atlas int he fiber tracking stage (screenshot).

You also need to put the ga28-based parcellations (.nii.gz) into the
human_ga25_atlas folder.
Something like https://gin.g-node.org/kcl_cdb/fetal_brain_mri_atlas/src/master/parcellations
> To view this discussion visit https://groups.google.com/d/msgid/dsi-studio/eab61cec-8f2f-4915-8180-81971bdf772bn%40googlegroups.com.

Gabriela Epihova

unread,
Jul 14, 2025, 3:09:38 PM7/14/25
to DSI Studio
I have but I've called it humna_ga25_atlas - this is the parcellation file
So I have 4 files :the parcellation, the fa (QA), the md (ISO) and the text file. Do I need anything else like T1/T2/ODF?

Best,
Gabriela

Frank Yeh

unread,
Jul 14, 2025, 3:11:41 PM7/14/25
to gabriel...@gmail.com, DSI Studio
No, there is no need for t1/t2/odf.

Perhaps rename the humna_ga25_atlas.nii.gz into a different name and
see if it shows up.
> To view this discussion visit https://groups.google.com/d/msgid/dsi-studio/6b4b70c2-b687-4b0d-92ae-6ed9d3c80279n%40googlegroups.com.

Gabriela Epihova

unread,
Jul 14, 2025, 3:27:07 PM7/14/25
to DSI Studio
Ah!!! This has worked, when I renamed the parcellation and txt file something different than the qa/iso files it mapped nicely (attched)!
Thank you so much Frank!

Best,
Gabriela

worked.png

Frank Yeh

unread,
Jul 14, 2025, 3:30:17 PM7/14/25
to gabriel...@gmail.com, DSI Studio
Glad it worked!
I would also try upsampling the DWI to 1-mm at Step T2 (top menu resample to isotropic -> 1mm) to see if it minimizes voxelization error.
Best,
Frank


Reply all
Reply to author
Forward
0 new messages