Mapping data between FreeSurfer and HCP

[Ziteng Han]

Dear all，

I want to replicate a brain parcellation project of Thomas Yeo (Schafer et al. 2018) using the HCP preprocessed data. The parcellation code needs to be computed in Freesurfer fsaverage6 surface, while the HCP preprocessed fMRI data is in fs_LR space. I have read the HCP official manual about mapping data between FreeSurfer and fs_LR, and it seems that the instructions there are only for metric and label files. 

So, is there a way to convert the HCP preprocessed fMRI data (in 'dtseries.nii' format and about 100 subjects) to fsaverage space? If not, I'll have to modify their code completely.

Best regards,
Ziteng Han

--
School of Life Science
Beijing Institute of Technology
Beijing, China

[Glasser, Matt]

I’m not sure what format FreeSurfer likes to represent timeseries data in, but perhaps Thomas can help us here?

Matt.

--
You received this message because you are subscribed to the Google Groups "HCP-Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hcp-users+...@humanconnectome.org.
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/1acd98ce-4a3b-48be-9f6b-515a4525f2d4n%40humanconnectome.org.

 

---

The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.

[Tim Coalson]

Sorry for not replying sooner.  wb_command -cifti-separate will let you get the surface data as metric files, which you can then resample.  It will also let you get the voxel data for subcortical structures (and cerebellum) as a nifti file with -volume-all.

I don't know if you will need to convert it further into freesurfer formats, that depends on how the code was written.

Tim

--

[Ziteng Han]

Thanks for your reply, Tim and Matt.

Hi, Tim. Do you mean that I can separate CIFTI timeseries data into metric files along the time dimension (e.g., 100 time points in the dtseries.nii, then 100 metric files)?

Best regards,

Ziteng

You received this message because you are subscribed to a topic in the Google Groups "HCP-Users" group.
To unsubscribe from this topic, visit https://groups.google.com/a/humanconnectome.org/d/topic/hcp-users/s2KGPV5EG24/unsubscribe.
To unsubscribe from this group and all its topics, send an email to hcp-users+...@humanconnectome.org.
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU4YwSreVyJrcXa_1oPBA97dp2Fp9z%3D_HjJX_XZQW4OZbA%40mail.gmail.com.

[Tim Coalson]

-cifti-separate will give you 2 metric files, a right cortex and a left cortex, with 100 timepoints in each of them.  If you wanted to then separate the timepoints into separate files (not sure why you would, though), you would need to do a bunch of -metric-merge calls afterwards.

Maybe this would help:

https://www.humanconnectome.org/software/workbench-command/-cifti-help

https://www.humanconnectome.org/software/workbench-command/-gifti-help

Tim

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CAM3xhiNTcFKpx3Nyuh9Ch174Jw773ZzkzhP-CCcCLWty3brGzA%40mail.gmail.com.

[Ziteng Han]

Thank you very much, Tim.

I know what I should do now. First, generating metric files for the left and right hemisphere using the '-cifti-separate' operation. Then, following the instructions in the HCP official manual (9. How do I map data between FreeSurfer and HCP?) to perform mapping from the fs_LR to fsaverage. 

Ps: I would have thought that the metric here (9. How do I map data between FreeSurfer and HCP?) could only be one dimensional, so I need to separate the timepoints into separate files. 

Best regard, 

Ziteng

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU5cMirRDr2fk3UKkDDHNtm-A1ekNkVS7w0S1W8hjdRSbw%40mail.gmail.com.

[Ziteng Han]

Dear Tim and Matt, 

I have a follow-up question about mapping data between FreeSurfer and HCP.  I have followed the HCP official manual and your advices to convert the cortical time cources on fs_LR 32k mesh (dtseries.nii) to FreeSurfer fsaverage6 surface mesh, which can be then put into the Schaefer parcellation pipeline of Thomas Yeo (Schaefer et al. 2018). However, I found that the converted time series data on the fsaverage6 surface has no signal at some cortical vertices (~300 cortical vertices, all 1200 time points with 0 signal) and has signal at some medial wall vertices. The cortical vertices and medial wall vertices were defined by the FreeSurfer (/usr/local/freesurfer/subjects/fsaverage6) and I used a code written by Thomas Yeo to load the mesh into matlab (CBIG_ReadNCAvgMesh.m).

So I'm writing to ask if this error is due to resampling between the FreeSurfer space and HCP space? Or maybe I should define the medial wall vertices of fsaverage6 surface mesh in another way?

P.S. I made this resampling just to run the Thomas's parcellation project and thus generate a sample-specific brain parcellation, with all subsequent analysis going back to the HCP space. 

Thanks for your time and I'll be looking forward to your reply.

Best, 

Ziteng 

[Glasser, Matt]

There might be differences in the definition of the medial wall.  You could dilate those away; however, is there a reason you cannot use the HCP’s multi-modal cortical parcellation?

Matt.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/e800dc0f-7fd5-4348-9665-a95adc3ae461n%40humanconnectome.org.

[Ziteng Han]

Sorry for not replying sooner.

Thanks a lot, Dr. Matt. But I think that methods like dilate and smooth result in a loss of precison in the HCP data, so I decided to change the way Freesurfer defines the medial wall to be the same as HCP. 

I'm currently studying the human connectome of a specific population, and we think that a brain parcellation created from individuals within this specific population may be more representative of the functional organization of this group than the HCP_MMP and can reduce mesurement bias when calculating the connectome due to topography(ref 1 and ref 2). Of course, the hypothesis needs further validation. 

ref 1: https://www.biorxiv.org/content/10.1101/2023.09.12.557193v2.abstract

ref 2: https://www.science.org/doi/full/10.1126/sciadv.abj1812

Best,

Ziteng

[Glasser, Matt]

I see.  We haven’t yet found evidence of that at the group level (there is plenty of individual variability), but we are working on making a similar tool available to folks anyway.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/10893a49-22f4-440f-8624-db7b507890dan%40humanconnectome.org.