Closed bkossows closed 1 year ago
I think there's some epistemological confusion here. Please allow me to clarify:
subject-specific label files are typically annotations that leverage the subject's measured data in order to customize the parcellation scheme -- also known as precision functional mapping this leads to different ROIs and differently shaped systems from subject to subject. Examples can be found here:
https://doi.org/10.1101/2022.01.12.475422
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5568802/
https://www.cell.com/neuron/pdf/S0896-6273(17)30613-X.pdf
In general, template label files that are applied consistently across multiple subjects are often derived a priori from independent data averaged across many subjects (not a literal average, of course). These label files are the same for each subject, and enable comparisons of the same ROI across mutliple subject -- at the cost of precision. These aren't actually specifically labeled for each subject, but in some pipelines, may be provided as a convenience.
As long as such label files are properly mapped to the same standard space (for ABCC data, this would be the MNI 2006 Nonlinear asymmetric atlas in volume, and freesurfer L/R 164/32k for the surface), the cifti parcellate command you have above should work just fine.
I'm curious why Rahul Desikan's (זכרונו לברכה) atlas?
We do have a collection of general ROI sets that can be used for this purpose, here's a link to a folder containing them:
https://drive.google.com/drive/folders/1XUNzIGblYroDvqcNb-GAXNRAfGlZGjek?usp=share_link
Finally, I would be cautious about using myelin maps in general and do some EDA on the stability first. Some normalization transforms to the data may be warranted.
Thank you for the prompt response. I appreciate the atlases you sent me to obtain the relevant statistics. However, I still have reservations about the redundancy of the DK parcellation for the specific participants. When I examined the output of the HCP-ABCD pipeline, I noticed a significant difference between "fsaverage" and the individual parcellation (for fsLR32k). I did not enable MPM registration for my data. Do you think that may have been the cause? I would like to use the ABCD myelin data as a reference for my dataset. Could you please confirm if the file "sub-#/ses-#/anat/sub-#_ses-#_space-fsLR32k_myelinmap.dscalar.nii" corresponds to "#.MyelinMap.32k_fs_LR.dscalar.nii" or the bias-corrected "#.MyelinMap_BC.32k_fs_LR.dscalar.nii" from the HCP ABCD pipeline? With kind regards,
The above link takes me to the NHP version of our pipeline -- we also have human child/adult and human infant . Given the repository, I was under the impression that the child/adult pipeline was being used -- is that correct or am I mistaken?
No worries either way, just want to make sure we're on the same page :)
Ok, I think I understand now -- you wanted to use the freesurfer naive bayes classified parcellations based on Rahul's manually drawn labels. We don't routinely pull these through to the derivatives -- though they are present in the processed output itself (prior to running file-mapper above).
The "fsaverage" will indeed be different from "fsLR32k". The fsaverage refers to freesurfer's fsaverage template as opposed to the freesurfer L/R resampled to 32k vertices.
Yes the correspondence should be the same to the 32k_fs_LR.dscalar.nii -- we have a list of all the correspondences here, if you're interested :)
Thank you again for the clarification. We used human child/adult. So, you're saying that the pipeline we ran is not the same one used for the NDA ABCD Collection?
the NDA ABCD collection used the human child/adult. If you're running the pipeline yourself, one can pull this in assuming one has not run the customclean stage.
Happy to walk this through in a quick zoom meeting if you'd like -- feel free to contact me at feczk001@umn.edu
Eric, thanks a lot for these clarifications!
I would like to extract averaged myelin values from the aparc and apar2009s parcellation. I've found dscalar.nii files in the repository, but there are no dlabel.nii files for the respective parcellation. I am not sure, but it seems like the individual parcellations are required to run the following command. Am I correct?
wb_command -cifti-parcellate xxx.dscalar.nii xxx.aparc.a2009s.32k_fs_LR.dlabel.nii COLUMN output.pscalar.nii