Hi I am new to diffusion imaging and have a very basic question. I have run functional connectivity analyses using a parcellation scheme. From this, I have a small number of regions that are related to a clinical measure. Is it possible to use these regions as seeds? If so, can you guide me in right direction? My goal is to determine if the abnormalities in tracts from these regions differ based on severity of clinical measure. Thank you.
Welcome @cchin!
There’s a few different ways that this could be done. But given your functional connectivity analysis is based on a parcellation scheme, I would personally suggest that rather than attempting to use the identified regions as tractography “seeds”, you should instead generate the whole-brain structural connectome based on the same parcellation as was used in your functional connectivity analysis, and then simply focus your interrogation of the resulting connectome matrix to only consider those edges associated with your prior identified regions.
I never feel comfortable tooting my own horn, but my recent preprint is relevant here also, as it demonstrates why it is beneficial from a methodological perspective to utilise whole-brain tractography even if one is only interested in quantifying specific pathways of interest, as well as how the connectome framework is essentially equivalent to having a large number of pathways of interest.
Cheers
Rob
Hello Rob
This is very useful and your preprint really did help to explain it from a methodological perspective. The parcellation scheme that I am following is the Schaefer Parcellation (https://github.com/ThomasYeoLab/CBIG/tree/master/stable_projects/brain_parcellation/Schaefer2018_LocalGlobal/Parcellations/FreeSurfer5.3). From my understanding, I will need to run FreeSurfer on all my subjects using this parcellation rather than the generic freesurfer scheme. Is this correct? Then using this result, I can do whole brain tractrography? I apologize for the very basic qyestions.
Do you have a recommended walkthrough guide for newbies?
Thank you,
Cherise.
Hi Cherise,
From my understanding, I will need to run FreeSurfer on all my subjects using this parcellation rather than the generic freesurfer scheme. Is this correct?
If the Schaefer parcellation is based on one of the default FreeSurfer templates to which surface-based registration is performed during the default recon-all pipeline, then my understanding is that you should not need to re-run the entire pipeline. There should be individual FreeSurfer commands that can be used firstly to transform that particular parcellation from the template to your individual subject’s surface representation, and then to generate a parcellation voxel image from those data. From my own code, I think you want either mri_ca_label and mri_label2vol or mri_surf2surf and mri_aparc2aseg, depending on the format in which the parcellation is provided. But I’m not a FreeSurfer guy…
Then using this result, I can do whole brain tractrography?
The set of MRtrix3 commands used for structural connectivity estimation are intentionally designed such that the generation of a whole-brain tractogram is independent of any specific brain parcellation. So you can do whole-brain tractography without any parcellation, or you can change parcellations after generating the tractogram, or you can use the same whole-brain tractogram with multiple parcellations to generate multiple connectome matrices.
Do you have a recommended walkthrough guide for newbies?
The BATMAN tutorial is the best we have right now. The example there uses the HCPMMP1.0 atlas mapped to fsaverage space, I don’t know if the same commands will apply directly you your parcellation.
Cheers
Rob
1 Like