A probabilistic atlas of the human thalamic nuclei

[Asa Borzabadi]

Dear experts,

May I ask if the probabilistic atlas of human thalamic nuclei provided by Iglesias et al. (2018) has an available counterpart in fsaverage (or MNI) space? 

Regards,

Asa Farahani

[Glasser, Matt]

That might be a better question for the FreeSurfer list.  We have not compiled these new atlases for the HCP data at this point.

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/CA%2Bs6gv%3Dq6RoY2zF2UOEBP77LQEHOq%2BmpQTSjxH%2BDXV9GVskL8Q%40mail.gmail.com.

 

---

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.

[Asa Borzabadi]

Thank you so much.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/B0CC39AC-6600-4ABC-962D-7C5AB8B05CD4%40wustl.edu.

[Harms, Michael]



Is this a volume (mgz) file that FS spits out for each subject, which contains a set of thalamic nuclei labels customized to the anatomy of each subject?

 

If so, adding the relevant image volume name to this ‘for’ loop in FreeSurfer2CaretConvertAndRegisterNonlinear.sh may be all that you need to do:

 

#Convert FreeSurfer Volumes

for Image in wmparc aparc.a2009s+aseg aparc+aseg ; do

    if [ -e "$FreeSurferFolder"/mri/"$Image".mgz ] ; then

        mri_convert -rt nearest -rl "$T1wFolder"/"$T1wImage".nii.gz "$FreeSurferFolder"/mri/"$Image".mgz "$T1wFolder"/"$Image"_1mm.nii.gz

        applywarp --rel --interp=nn -i "$T1wFolder"/"$Image"_1mm.nii.gz -r "$AtlasSpaceFolder"/"$AtlasSpaceT1wImage" --premat=$FSLDIR/etc/flirtsch/ident.mat -o "$T1wFolder"/"$Image".nii.gz

        applywarp --rel --interp=nn -i "$T1wFolder"/"$Image"_1mm.nii.gz -r "$AtlasSpaceFolder"/"$AtlasSpaceT1wImage" -w "$AtlasTransform" -o "$AtlasSpaceFolder"/"$Image".nii.gz

        ${CARET7DIR}/wb_command -volume-label-import "$T1wFolder"/"$Image".nii.gz "$FreeSurferLabels" "$T1wFolder"/"$Image".nii.gz -drop-unused-labels

        ${CARET7DIR}/wb_command -volume-label-import "$AtlasSpaceFolder"/"$Image".nii.gz "$FreeSurferLabels" "$AtlasSpaceFolder"/"$Image".nii.gz -drop-unused-labels

    fi

done

 

cheers,

-MH

 

-- 

Michael Harms, Ph.D.

-----------------------------------------------------------

Associate Professor of Psychiatry

Washington University School of Medicine

Department of Psychiatry, Box 8134

660 South Euclid Ave.                        Tel: 314-747-6173

St. Louis, MO  63110                          Email: mha...@wustl.edu

 

From: Asa Borzabadi <asa.bo...@gmail.com>

Reply-To: "hcp-...@humanconnectome.org" <hcp-...@humanconnectome.org>
Date: Monday, May 2, 2022 at 2:36 PM
To: "hcp-...@humanconnectome.org" <hcp-...@humanconnectome.org>

Subject: Re: [hcp-users] A probabilistic atlas of the human thalamic nuclei

 

* External Email - Caution *

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CA%2Bs6gvnAd%2B98uaBMVGDkxKX7GXWsPQGWqfjdQp1cRgertUyq3A%40mail.gmail.com.

[Asa Borzabadi]

Hi,

Thank you for your continuous support and help. 

I just want to make sure I am in the right direction.

I did the thalamic segmentation for a subject using the Iglesias et al. method ( method developed in FreeSurfer version 7+). It culminated in giving us a .mgz file which included the intra-thalamic labels. Then, we decided to study the voxels of specific thalamic nuclei within our functional data ( 2mm isotropic); to get the thalamic labels in the CIFTI space, these are the steps we followed: 

mri_convert -rt nearest -rl ${SubjectDir}/${Subjectid}/T1w/T1w_acpc_dc_restore.nii.gz ${SubjectDir}/${Subjectid}/T1w/${Subjectid}/mri/ThalamicNuclei.v13.T1.FSvoxelSpace.mgz ThalamicNuclei.v13.T1.FSvoxelSpace.nii.gz

 

applywarp --rel --interp=nn -i ThalamicNuclei.v13.T1.FSvoxelSpace.nii.gz -r  ${SubjectDir}/${Subjectid}/MNINonLinear/T1w_restore.nii.gz -w  ${SubjectDir}/${Subjectid}/MNINonLinear/xfms/acpc_dc2standard -o ThalamicNuclei.v13.T1.FSvoxelSpace_native.nii.gz

applywarp --interp=nn -i ThalamicNuclei.v13.T1.FSvoxelSpace_native.nii.gz -r  ${SubjectDir}/${Subjectid}/ MNINonLinear/ROIs/Atlas_ROIs.2.nii.gz -o ThalamicNuclei.v13.T1.FSvoxelSpace_Atlas_2mm.nii.gz

wb_command -volume-label-import ThalamicNuclei.v13.T1.FSvoxelSpace_Atlas_2mm.nii.gz FreeSurferColorLUT.txt ThalamicNuclei.v13.T1.FSvoxelSpace_Atlas_2mm.nii.gz -drop-unused-labels

Do you think the approach is errorless when converting mgz data into the HCP-style data? 

Regards,

Asa Farahani

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/92677B55-A615-4CD4-8C52-10AA50E0ECA8%40wustl.edu.

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

Without looking too hard at the transforms used, that looks plausible, though I wouldn't recommend calling the MNINonLinear space version "_native".  The multiple nearest neighbor interpolation steps could make the borders a bit less accurate (you should be able to use the 2mm space as the reference in the first applywarp in order to skip the next applywarp), but if it is using structural resolution until the last interpolation, that may be a small effect.

Tim

---

From: Asa Borzabadi <asa.bo...@gmail.com>
Sent: Monday, August 1, 2022 9:06 AM

To: hcp-...@humanconnectome.org <hcp-...@humanconnectome.org>
Subject: Re: [hcp-users] A probabilistic atlas of the human thalamic nuclei

 

WARNING: This message has originated from an External Source. This may be a phishing expedition that can result in unauthorized access to our IT System. Please use proper judgment and caution when opening attachments, clicking links, or responding to this email.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CA%2Bs6gvnv-QMM%2BjR-XvF3tpNykM%3DyE__7hM1m0E%3DbWk15K5vqfg%40mail.gmail.com.

[Asa Borzabadi]

Thank you so much for the help, and thanks for the correction.

I still have a general question. When viewing the converted intra-thalamic label files (now in MNI152_2mm space) on top of the Atlas.ROIs.2.nii.gz template, there exist voxels outside the boundary of Atlas.ROIs.2.nii.gz template (their existence could possibly be explained by the fact that FreeSurfer7, which does the intra-thalamic segmentation, uses a different approach for thalamic parcellation than FreeSurfer6); in other words, there exist some voxels that are not defined in the standard subthalamic space of HCP. 

(the figure is attached for better illustration, the layout of the template and the resulted intra-thalamic segmentation are plotted, some inconsistent voxels are pointed to with red circles) 

If I do "-cifti-create-dense-from-template", the result would be the same label file, but this time, the voxels outside the boundary are deleted, which is misleading due to the fact that those voxels are informative for my analysis, and I want to still consider them in my intra-thalamic ROI.

I suppose there is no other way but analyzing the data in nifti volume format, am I right?

Regards

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/SN6PR01MB499121B136C9E819E268AFBC819A9%40SN6PR01MB4991.prod.exchangelabs.com.

[Glasser, Matt]

Is this an individual subject or an average?

Matt.

 

From: Asa Borzabadi <asa.bo...@gmail.com>
Reply-To: "hcp-...@humanconnectome.org" <hcp-...@humanconnectome.org>
Date: Wednesday, August 10, 2022 at 2:41 PM
To: "hcp-...@humanconnectome.org" <hcp-...@humanconnectome.org>
Subject: Re: [hcp-users] A probabilistic atlas of the human thalamic nuclei

 

 

Thank you so much for the help, and thanks for the correction.

 

I still have a general question. When viewing the converted intra-thalamic label files (now in MNI152_2mm space) on top of the Atlas.ROIs.2.nii.gz template, there exist voxels outside the boundary of Atlas.ROIs.2.nii.gz template (their existence could possibly be explained by the fact that FreeSurfer7, which does the intra-thalamic segmentation, uses a different approach for thalamic parcellation than FreeSurfer6); in other words, there exist some voxels that are not defined in the standard subthalamic space of HCP. 

(the figure is attached for better illustration, the layout of the template and the resulted intra-thalamic segmentation are plotted, some inconsistent voxels are pointed to with red circles) 

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CA%2Bs6gv%3DyxU15NTFjxT9e-t89riuZAR-BTfgM7FiUkZv4dgM%3DLQ%40mail.gmail.com.

[Asa Borzabadi]

It is an individual subject.

Thank you so much.

Regards, 

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/0C8111AD-95B4-4C88-A8FD-8BCCA7D9CE19%40wustl.edu.

[Glasser, Matt]

Is it possible that things aren’t quite lined up in the volume?  What did you use for volume registration?

Matt.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CA%2Bs6gvmyMLUNxrMTrLTv9qmNigJcYBjkSthKgUj%2BHU7X94BGDw%40mail.gmail.com.

[Harms, Michael]

 

Isn’t this fairly common?  IIRC, nothing prevents the FS subcortical labels in an individual from mapping outside of the CIFTI atlas labels during the computation of the registration to MNINonLinear space.

 

-- 

Michael Harms, Ph.D.

-----------------------------------------------------------

Associate Professor of Psychiatry

Washington University School of Medicine

Department of Psychiatry, Box 8134

660 South Euclid Ave.                        Tel: 314-747-6173

St. Louis, MO  63110                          Email: mha...@wustl.edu

 

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CC109575-2B2D-4354-8F6D-ABEF661F255A%40wustl.edu.

[Glasser, Matt]

It shouldn’t be off by much.

Matt.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/27C1994E-CD9A-4D13-AB8F-607026D9F1BA%40wustl.edu.

[Harms, Michael]

But it happens to at least some extent with some regularly (because the MNINonLinear registration is not constrained/informed by the FS subcortical labels).  In the provided example image, the “spill-over” outside of the CIFTI atlas indeed wasn’t much.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/0A09413E-4B92-4B73-A727-1F3E06C6F0D6%40wustl.edu.

[Asa Borzabadi]

Dear experts,

Thank you so much for the useful information.

I do not think there exists any volume registration problem. When applying the Iglesias method for intra-thalamic segmentation, the probabilistic atlas, which is in tetrahedral mesh, will fit itself to the structural T1 of the individual subject and use the anatomical landmarks there to get the best result. After that, I converted the resulted mgz label file into nifti using "mri-convert", used the acpc-dc2standard wrapping to have the labels in "T1w_restore" space ( MNI-0.7mm space), and then resampled the results using "applywarp" to get the labels in the  MNI-2mm space. 

According to the conversation you had above, what I released is that when the goal is analyzing the data of an individual subject, if one is interested in studying intra-thalamic regions ( especially those near the boundary of the thalamus, like LGN) the grayordinate space might not be the best shot, as some near-boundary voxels corresponding to the LGN might have been missed during the procedure of defining subcortical areas (even if the number of missing voxels is not great, this can cause problems due to the overall small size of the target nucleus). In this case, the best option could be conventional volume data analysis instead of using the grayordinate space, am I correct?

Regards,

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/8BA8E1DA-0901-471C-BAEE-C274FA791CDC%40wustl.edu.

[Glasser, Matt]

You can certainly analyze data within individual subject defined parcels in individuals, but I take it you want to do a group analysis, or you would have just done that.  I think the question that we are trying to address is whether this new thalamic parcellation on average is including voxels that were not previously included as thalamic greymatter voxels on average by FreeSurfer (and hence are not in the CIFTI standard space) or if this is simply expected small imperfections of non-linear volume-based alignment from subject to subject.  The best way to address that would be to take a large group of non-linearly aligned subjects with that new thalamic parcellation and take the mode of their parcellations and see if there is a difference with the current CIFTI grayordinates space.  If not, it would be sufficient to use parcel constrained resampling between the group and individuals even with minor misalignments (as the resampling will pull data from neighboring slightly misaligned voxels into the group average locations).  If there is a difference, that would suggest a need to modify the CIFTI grayordinates space.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CA%2Bs6gvkWfUP_AfJmgXMz493UCTKOeDbmaCmFLWxMDGFCH4YrUg%40mail.gmail.com.

[Asa Borzabadi]

Thanks. To my understanding in the HCP preprocessing pipeline, the data of the individual subject is nonlinearly registered into the MNI space; then a template of subcortical gray matter structures is used to mask out the subcortical region, while there is nothing preventing the FS subcortical labels in an individual from mapping outside of the CIFTI atlas labels. So, there is a possibility that few voxels belonging to the subcortical region of the individual are located outside the defined template for segregation the subcortical CIFTI area, and this can produce some error.

Yours sincerely,

[Glasser, Matt]

Yes, but I wouldn’t expect any systemic effects and that this would be limited to a voxel from which at least some signal should be recovered when generating the CIFTI timecourse for the neighboring voxel.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/ee47c939-76ad-42e7-ade5-fee6ed513dd0n%40humanconnectome.org.