Brain atlas projection from fsLR to MNI space

[Xiaoxuan Yan]

Hi,

Thanks for processing my request. I am a Ph.D. student from the National University of Singapore and I am currently working on a brain parcellation project. I am using HCP data to general my cortical parcellation. 

I am looking for some script that could help me project a brain atlas from fsLR (32k) to MNI space (surface mesh only), and also from MNI to fsLR. Just wondering if such a script is already available on your side?

Thank you and hope to hear from you soon :)

Best regards

Xiaoxuan

[Coalson, Timothy Scott (S&T-Student)]

I'm not entirely clear on what you want - HCP data provides individual subject surface files that use a nonlinear volume registration to align with the MNI coordinate system (if this is what you mean by "surface mesh only"), in the MNINonLinear folder.

If you mean projecting surface data to voxels, you can do this in individual subjects with wb_command -metric-to-volume-mapping.  From individual volume data to individual surfaces can be done with -volume-to-surface-mapping.  We generally recommend the ribbon mapping method.

However, we do not recommend mapping human data from a group-average volume file to the surface, or mapping human surface data to a group-average volume space, because current volume-based registrations don't entirely overlap cortex with cortex across subjects, due to significant folding differences across human subjects (and volume registrations typically don't try to use data functionally-relevant data for alignment).  We have a paper on this topic:

https://www.pnas.org/content/115/27/E6356

Tim

---

From: Xiaoxuan Yan <xiaoxu...@gmail.com>
Sent: Monday, August 17, 2020 9:48 PM
To: hcp-...@humanconnectome.org <hcp-...@humanconnectome.org>
Subject: [hcp-users] Brain atlas projection from fsLR to MNI space

 

--
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/CAGhFCVXPtHeRd98sZJCmpUUykkW9NNomjFJ5zG_duYDtv5WbcA%40mail.gmail.com.

[Jeiran Choupan]

Dear Xiaoxuan,

I am not sure if my solution would be related to what you are interested to working on, but I will explain the analysis that I am doing in Matlab, and maybe you can modify it for your own purpose.

I needed to use an atlas that was in T1w Native space for my fMRI results that is in fs32k.

What I did was to find the corresponding closest euclidean points between these two surfaces on their sphere space (specifically sphere space was used to minimize the contribution of sulci and gyri). For each hemisphere I use the following files from the MNINonlinear folder:

sphere_fs32_R = gifti([char(k) '.R.sphere.32k_fs_LR.surf.gii']); %from /path to your FREESURFER_SUBJECTS_DIR/SUBJECT_ID/MNINonLinear/fsaverage_LR32k

sphere_native_R = gifti([char(k) '.R.sphere.reg.reg_LR.native.surf.gii']); % from /path to your FREESURFER_SUBJECTS_DIR/SUBJECT_ID/MNINonLinear/Native

MNI_32_R  = gifti([char(k) '.R.inflated.32k_fs_LR.surf.gii']);% from /path to your FREESURFER_SUBJECTS_DIR/SUBJECT_ID/MNINonLinear/fsaverage_LR32k

%% Right Hemisphere Euclidean distance between the two spaces (FS32K and Native)

 sphere_fs32_v_R = sphere_fs32_R.vertices;

 sphere_native_v_R = sphere_native_R.vertices;

 

% Euclidean distance

for j = 1:size(sphere_fs32_v_R,1)

    [~,D_Ind_Sphere_R(j,1)] = min(sqrt((sphere_native_v_R(:,1) - sphere_fs32_v_R(j,1)) .^2 + ...

        (sphere_native_v_R(:,2) - sphere_fs32_v_R(j,2)) .^2 + ...

        (sphere_native_v_R(:,3) - sphere_fs32_v_R(j,3)) .^2 ));

end

The `D_Ind_Sphere_R` index corresponds to the indices in fs32 space that correspond to my atlas of interest in Native space.

Hope this is helpful. Let me know if you have more questions.

Best,

Jeiran 

Jeiran Choupan, PhD, MSCS

Research Scientist

NICR-Group (http://nicr.ini.usc.edu)
USC Stevens Neuroimaging and Informatics Institute

Keck School of Medicine 
University of Southern California

[Glasser, Matthew]

If you need surfaces on the standard 32k_fs_LR mesh or the subject’s native mesh in their physical volume space, these are already provided:

 

${StudyFolder}/${Subject}/T1w/fsaverage_LR32k

${StudyFolder}/${Subject}/T1w/Native

 

We provide well-tested tools for resampling between different meshes in wb_command as well.

Matt.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/55EAA7A9-1BB5-42AB-8362-C923D3B67027%40usc.edu.

 

---

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.

[Coalson, Timothy Scott (S&T-Student)]

Specifically, wb_command -metric-resample, -label-resample, or -surface-resample use the same concept of "what sphere vertices on that other sphere are near this sphere vertex", and use that information to do a weighted average or mode of the input data, with the aim of minimizing smoothing and avoiding the jaggedness you get from nearest-neighbor logic.

Tim

---

From: Glasser, Matthew <glas...@wustl.edu>
Sent: Tuesday, August 18, 2020 6:59 AM

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/707661AC-4691-4E35-9748-9259940F66F4%40wustl.edu.

[Jeiran Choupan]

Dear HCP users,

I am trying to use wb_command -X-resample for a MGZ file that is in `MNINonLinear/Native` space (130k X 1 double). 

I learned that the wb_command resample only reads the gifti files.

Do you know how I can convert the MGZ to gifti?

Looking forward to your kind response

Best,

-Jeiran 

Jeiran Choupan, PhD, MSCS

Research Scientist

NICR-Group (http://nicr.ini.usc.edu)
USC Stevens Neuroimaging and Informatics Institute

Keck School of Medicine 
University of Southern California

[Glasser, Matthew]

Perhaps with mris_convert or mri_convert.  You could google search or ask on the FreeSurfer list.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/E69061B4-2B80-41C1-AA22-BF59616259F0%40usc.edu.