Extraction of metrics in all tracts

[Kaonashi22]

Can we add a loop to extract metrics (MT, T1, diffusion) in all regions and tracts in the "info_label.txt" file?

[jcohenadad]

Yes, we can certainly to that. I will implement it after we merge #8

[Kaonashi22]

Can we update the script to extract the metrics in all tracts ?

[jcohenadad]

sure-- i'll try my best to do it today

[jcohenadad]

how do you want the results? aggregated into a single CSV file, or one CSV file per tract?

[Kaonashi22]

Thank you! A csv file per tract would be more convenient

…

________________________________ From: Julien Cohen-Adad ***@***.***> Sent: May 1, 2024 08:40 To: sct-pipeline/spine-park ***@***.***> Cc: Lydia Chougar, Dr ***@***.***>; Author ***@***.***> Subject: Re: [sct-pipeline/spine-park] Extraction of metrics in all tracts (Issue #10) how do you want the results? aggregated into a single CSV file, or one CSV file per tract? — Reply to this email directly, view it on GitHub<#10 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BFCFJYUBF27CZEGW7CD2ORDZADPETAVCNFSM6AAAAABEAH4QY6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOBYGQYDSMRUGU>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[jcohenadad]

Another consideration: some tracts are veeeeeery small (ie: less than a mm2 per axial slice). When considering tracts in isolation, metrics will be more prone to noise, hence if there is a pathology-related effect, it would possibly be non-statistically significant. Whereas when aggregating across tracts, the sensitivity to detect changes statistically will be higher. So my recommendation would be to aggregate tracts instead of studying them in isolation. Aggregation should follow hypotheses in terms of the tracts that are expected to be affected (eg: all descending tracts?)

[Kaonashi22]

I'm particularly interested in the intermedio-lateral columns, located in the lateral horns (in green). I'll look at it more in detail.

[jcohenadad]

could you suggest which tracts to aggregate from the info_label.txt

[Kaonashi22]

The region that I'm interested (intermedio-lateral columns) should be located in these GM areas (~16 voxels in the axial plane for each side):
32, GM left intermediate zone, PAM50_atlas_32.nii.gz
33, GM right intermediate zone, PAM50_atlas_33.nii.gz
Do you think if it can be studied given the size?

WE can can also look at the GM globally
52, gray matter, 30:35

These could be "control regions":
_51, white matter, 0:29
53, dorsal columns, 0:3
54, lateral funiculi, 4:13
55, ventral funiculi, 14:29

30, GM left ventral horn, PAM50_atlas_30.nii.gz
31, GM right ventral horn, PAM50_atlas_31.nii.gz

34, GM left dorsal horn, PAM50_atlas_34.nii.gz
35, GM right dorsal horn, PAM50_atlas_35.nii.gz

If not too small
4, WM left lateral corticospinal tract, PAM50_atlas_04.nii.gz
5, WM right lateral corticospinal tract, PAM50_atlas_05.nii.gz_

[jcohenadad]

ok great, this is very helpful. Just to clarify, the regions you listed per paragraph (eg: [32, 33] vs. [32], [33]): are you suggesting to merge them? (again, the pros in merging is the better robustness to noise)

[Kaonashi22]

I think we can merge right and left (eg: 32 and 33)

[jcohenadad]

Good, so I'll produce the following extraction strategies (one extraction per bullet point):

• 32,33
• 52
• 51
• 53:55
• 30,31
• 34,35
• 4,5

[Kaonashi22]

we can merge 54 and 55, but I would keep 53 separately unless you think it's too small
Sounds good for the rest

[Kaonashi22]

I'm also thinking it could be interesting to merge all ascending vs descending tracts