#16 in that FAQ is not at all relevant to this situation. If you have FreeSurfer label files, convert them to GIFTI with mris_convert and then use wb_command -label-resample to get them onto the 32k mesh.
Matt. From: <hcp-users-boun...@humanconnectome.org<mailto:hcp-users-boun...@humanconnectome.org>> on behalf of Timothy Coalson <tsc...@mst.edu<mailto:tsc...@mst.edu>> Date: Monday, October 1, 2018 at 7:08 PM To: Jacob Miller <jacob_mil...@berkeley.edu<mailto:jacob_mil...@berkeley.edu>> Cc: hcp-users <hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>> Subject: Re: [HCP-Users] Resample *.dtseries.nii to native surface space You can use wb_command -surface-closest-vertex to get a single vertex for each coordinate you want to find, just use the applicable surface in the T1w folder (since freesurfer doesn't generate midthickness, you may want to use pial or white surfaces, depending on how you defined the points). You can use the Window->Identify... menu item in wb_view to select a vertex by its number to get a dynconn or dconn layer to display the correlation map for it. If you want all the correlations between a bunch of single vertices, that is actually a little tricky, because that isn't something we generally do. The easiest way is probably to use matlab (BEWARE: workbench vertex indices are 0-based, matlab's indices are all 1-based) to make gifti metric files (not cifti, because the indices don't match between gifti and cifti) with each result vertex having a different nonzero integer, and zero elsewhere, turn those into cifti dlabel files via -metric-label-import and -cifti-create-dense-from-template (make sure not to reuse integers across hemispheres, or the next step will average the cross-hemisphere points together). Then you can use -cifti-parcellate on the dtseries and then use -cifti-correlate. However, you might consider making ROI patches around your points, to reduce the effects of noise on your correlation values (-surface-geodesic-rois can do this). The extreme of this is to use a full parcellation, which uses the same steps, once you have a dlabel file containing the parcellation. Tim On Mon, Oct 1, 2018 at 3:57 PM, Jacob Miller <jacob_mil...@berkeley.edu<mailto:jacob_mil...@berkeley.edu>> wrote: Thanks for your responses! I didn't quite realize the upsampling involved here, and that is certainly not what I'm hoping to do with these data. My goal is to analyze the resting-state data in reference to some (cortical) anatomical landmarks I have defined on a few subjects' original freesurfer meshes, comparing the connectivity of these landmarks within each subject. I created these labels in Freesurfer on the subjects' native surfaces. As you point out, it may be worth mapping these to the 32k mesh. If that's the case, is it possible to map a coordinate (and vertex number) from the original native inflated surface, for example, to a subject's 32k surface, and how would one do this? Otherwise, I may also just use the steps outlined in question 16 here (https://wiki.humanconnectome.org/display/PublicData/HCP+Users+FAQ#HCPUsersFAQ-16.Ca) to get the resting-state data back into native volume space. Thanks again! Jacob On Mon, Oct 1, 2018 at 1:29 PM Timothy Coalson <tsc...@mst.edu<mailto:tsc...@mst.edu>> wrote: As a final side note, if you have a good reason to use the native freesurfer mesh, then to get the data back onto the vertices it originally came from (our pipelines map to the native mesh surfaces and then downsample to 32k, last I checked), you need to use a registered native mesh sphere, not the original sphere. Resampling from the 91282 grayordinates file will also result in slightly more blurring than you would get from mapping from volume to native mesh surfaces, but since the fMRI resolution is similar to our 91282 grayordinates (by design, we chose the 32k sphere resolution because of this data), this effect should be small. Tim On Mon, Oct 1, 2018 at 3:22 PM, Timothy Coalson <tsc...@mst.edu<mailto:tsc...@mst.edu>> wrote: I would also question what you hope to accomplish by using the native freesurfer mesh - the fMRI data doesn't have that kind of resolution, and the result of resampling to it means you can't compare across subjects without resampling again later. Surface resampling is also unlike volume resampling, in that it doesn't move any of the subjects data in anatomical space (the T1w folder contains surfaces with acpc rigid aligned anatomical locations - you probably just want to use the 32k midthicknesses from there, as opposed to the ones in MNINonLinear) - think of it as changing the sampling "grid" (mesh), not as a spatial warp. The trick of surface registration is that we can give the same mesh of samples different coordinates per-subject - we use the index in the mesh for the purpose of matching functional identity as best as we can, but the coordinates remain wherever that point was in the individual's anatomy. Even the coordinates of the individual MNINonLinear surfaces are different, both because the volume alignment doesn't fully match gray matter across subjects, and from when our estimate of functional location disagrees with a folding-based alignment (since aligning T1w images basically tries to align folding). FYI, the error message you are getting is because you specified the 164k spheres, while the 91282 grayordinate cifti data exists on the 32k mesh (so you would need to use the 32k spheres). Tim On Fri, Sep 28, 2018 at 8:20 PM, Jacob Miller <jacob_mil...@berkeley.edu<mailto:jacob_mil...@berkeley.edu>> wrote: Hi all, I am interested in performing a resting state analysis in native surface space, using the surface cifti *.dtseries files. I am trying to resample these files to an individual subject's native surface space using -cifti-resample, but am running into trouble. When I run the following code: L_template='standard.L.sphere.164k_fs_LR.surf.gii' R_template='standard.R.sphere.164k_fs_LR.surf.gii' wb_command -cifti-resample rfMRI_REST1_LR_Atlas_hp2000_clean.dtseries.nii COLUMN ${SUB}.curvature.native.dscalar.nii COLUMN BARYCENTRIC CUBIC ${SUB}.native.dtseries.nii \ -left-spheres ${L_template} ${SUB}.L.sphere.native.surf.gii \ -right-spheres ${R_template} ${SUB}.R.sphere.native.surf.gii When I run this command, I get this note: ERROR: left current sphere doesn't match input cifti Is this the best command to use to resample the .*dtseries.nii to the native surface? If so, what is the correct current_sphere option to use? Also, the command asks for a cifti_template - what should I use here for the best native surface cifti template? I picked the curvature cifti because I wasn't sure. Please let me know what you think is the best way to get the resting-state surface data into native space! Thanks again, Jacob -- Jacob Miller Graduate Student, D'Esposito Lab Helen Wills Neuroscience Institute (HWNI) University of California, Berkeley https://despolab.berkeley.edu/ _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org<mailto:HCP-Users@humanconnectome.org> http://lists.humanconnectome.org/mailman/listinfo/hcp-users -- Jacob Miller Graduate Student, D'Esposito Lab Helen Wills Neuroscience Institute (HWNI) University of California, Berkeley https://despolab.berkeley.edu/ _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org<mailto:HCP-Users@humanconnectome.org> http://lists.humanconnectome.org/mailman/listinfo/hcp-users ________________________________ 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. _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users
