I have my SIFT-filtered tractograms that I would like to use to make connectomes. However, I preprocessed my anatomical data in FSL, and therefore do not have any recon-all FreeSurfer outputs to work with. Thus, I am unsure of what to use for the node parcellation image.
On the other hand, in anticipation of doing a joint structural/functional connectivity analysis, all of my tractograms are in MNI152 space. Would this negate the need to get individual node labels? Could I use something like the image file from this MNI atlas (https://github.com/brainspaces/power264) as my node labels?
Thanks, and let me know if I can clarify further!
Steven
However, I preprocessed my anatomical data in FSL, and therefore do not have any recon-all FreeSurfer outputs to work with.
Does processing anatomical data in FSL preclude one from processing anatomical data in FreeSurfer?
… all of my tractograms are in MNI152 space.
Firstly I want to check how the tractograms arrived there. If tracking was done in native space and then transformed to template space, then it’s all good; but there’s a small chance that they got there through inappropriate means; e.g. by transforming DWI data to template space and doing tracking there.
Assuming that’s all good though, there’s no issue with what you propose. If you think about it, all you really need is spatial correspondence between the streamline endpoints and the parcels. The more typical pipeline for a volume-based parcellation is to register subject data to a template, transform a parcellation from template to subject space, and then assign streamlines to parcels. Here you’re registering subject data to a template, transforming streamlines from subject to template space, and then assigning streamlines to parcels. The differences between the two approaches are likely to be very minimal, and only in the edge cases based on how the streamline-to-node assignment mechanism works. Where the former might have an advantage is if you want to generate connectomes with alternative contrasts; e.g. anything that incorporates streamline length would be better calculated in subject space (though there are clever ways to do such things using the latter approach as well…).
Thanks for the reply! Since I first used these data for TractSeg, I made the tractograms (and requisite FODs, FRFs etc) in MNI Space, although I could backtrack and make these in native space easily enough. Based on my understanding of what you said, it looks like I could either
Make native space tractograms --> convert streamlines to MNI --> track with MNI atlas
Make native space tractograms --> convert atlas to subject spaces --> track with subject space atlas
The reason I raised this question initially is because there’s a reasonable chance that there’s a bad decision being made upstream of this point. Subject-specific data can’t magically be processed in MNI space; there has to be some form of spatial transformation process that moves the subject’s data into MNI space. If this is in any way non-rigid, then it’s not possible to appropriately reorient / modulate the DWI intensities. So in your pipeline you may be “corrupting” your data before you even get to the point of asking your question. (Which is why I often give seemingly strange answers to questions on here: I’m persistently trying to diagnose such from limited information)
Once the DWI data have been “transformed” to FODs, it is then possible to perform a non-rigid spatial transformation of the data. But even then, performing tracking on reoriented FODs is not guaranteed to be equivalent to tracking on native FODs and then transforming the streamlines, see @dchristiaens work here.
So even if you already have your tractograms in MNI space and your parcellation in MNI space, and therefore combining those data together to produce connectomes seems trivial, I would look more critically at how it is that the tractograms happened to find themselves in MNI space.
Beyond that:
Based on my understanding of what you said, it looks like I could either
I would re-phrase those to:
Make native space tractograms → transform streamlines to MNI → assign streamlines to parcels in MNI space
Make native space tractograms → transform atlas to subject spaces → assign streamlines to parcels in subject space
