On May 27, 2015, at 3:16 AM, Stuart Oldham <[email protected]> wrote:
> Hi Donna, > > That is correct that we want the source parcellation mapped on the subjects > surfaces. I believe we are wanting it in MNINonlinear stereotaxic space. > > Should I be using the fs_LR standard mesh then? I'm not sure what resolution your volumes will be, but try it both ways (project from native, project from fs_LR surfaces), and subtract one result from the other to see how big the diffs are. I would expect small. The fs_LR way is easier, because it is a standard mesh, which means you resample from fsaverage (hem-specific) mesh to fs_LR (hems in register) mesh once. Then you can use the same fs_LR parcellation to project from each subject's anatomical surface (i.e., parcellation remains constant, while surface projection substrate varies across subjects). With native, you must resample the parcellation onto each subject/hem's mesh. It isn't clear to me, based on your goals, whether you should use the white, midthickness, or pial as your substrate. The FS -> fs_LR resampling is tricky to understand. See this script: https://github.com/Washington-University/Pipelines/blob/master/PostFreeSurfer/scripts/FreeSurfer2CaretConvertAndRegisterNonlinear.sh scroll down to this line and below: ${CARET7DIR}/wb_command -surface-sphere-project-unproject "$AtlasSpaceFolder"/"$NativeFolder"/"$Subject"."$Hemisphere".sphere.reg.native.surf.gii "$AtlasSpaceFolder"/fsaverage/"$Subject"."$Hemisphere".sphere."$HighResMesh"k_fs_"$Hemisphere".surf.gii "$AtlasSpaceFolder"/fsaverage/"$Subject"."$Hemisphere".def_sphere."$HighResMesh"k_fs_"$Hemisphere".surf.gii "$AtlasSpaceFolder"/"$NativeFolder"/"$Subject"."$Hemisphere".sphere.reg.reg_LR.native.surf.gii What you are doing is different, though, because your parcellation is already on the fsaverage hem-specific/intermediate mesh. > With the mri_label2vol command I should be only producing a single output > volume. I now input the annotation file into this command, bypassing the need > for labels (the computer I am using also has trouble loading in all of the > labels). You might need to mris_convert GIFTI to freesurfer and name per expected conventions to get this to work. > I have a pre-existing script that was written in my lab to create random > parcellations and I am attempting to adapt that. So I do the above and I will > get a label volume for each hemisphere. Then these two label volumes are > combined into a single image. Subcortical parcellations are then added to > make a whole brain parcellation image which we then use to make connectomes. > > Sorry if anything is unclear or not elaborated on enough, I am very new to > all of this. > > Regards, > > Stuart > > On 27 May 2015 at 07:01, Donna Dierker <[email protected]> wrote: > Hi Stuart, > > I have fewer answers than questions, but it seems like at the end of this, > you want your source (fsaverage) parcellation mapped onto the HCP subjects' > surfaces, and then projected from surface to volume (subject's orig/magnet > space or MNINonlinear stereotaxic space?). > > The HCP pipeline uses the Freesurfer ?h.sphere.reg to get surface data onto > fs_LR mesh: > > lh: freesurfer_native_mesh -> fsaverage.left -> fs_LR (32k or 164k) mesh > rh: freesurfer_native_mesh -> fsaverage.right -> fs_LR (32k or 164k) mesh > > Note that fsaverage left and right hem meshes are not in register with one > another (across hemispheres -- vertex i on the left is probably nowhere near > vertex i on the right hem). In the fs_LR standard mesh, left and right hems > are in register with one another. > > It sounds like you have an annotation on the fsaverage mesh. It seems in > principle possible to get your annotation, which is a single, atlas-like > parcellation that will be used across HCP subjects, onto the fs_LR mesh, and > then you could use the subjects' fs_LR white, midthickness, or pial surface > to project the annotation/label to volume somehow, but I don't know how those > freesurfer commands work or what they need. You can mris_convert from GIFTI > to Freesurfer format, if needed. > > It might help to zoom out a bit more, though: What do you plan to do with > the resulting individual label volumes? > > Donna > > > On May 24, 2015, at 8:45 PM, Stuart Oldham <[email protected]> wrote: > > > Hi all, > > > > I am trying to create a random parcellation for the HCP data. I have > > created the parcellations first using the fsaverage subject and am trying > > to map these onto the HCP subjects. > > > > First I have map the annotation from the source (fsaverage) to the subject > > using mri_surf2surf (for each hemisphere individually). Usually this just > > uses a sphere.reg file but the HCP data doesn't have the file in this > > format. Is the {subject}.{hemisphere}.sphere.reg.native.surf.gii (located > > under MNINonLinear/Native) the file I should be using here? > > > > Next I turn the annotations into labels using mri_annotation2label. Which > > surface file do I need to specify it to run (is it the same as above)? > > > > Then I run mri_label2vols. Again which surface file should I be inputting? > > Should it now be the {subject}.{hemisphere}.white.native.surf.gii? Also is > > there a way with this command I can specify it to find a thickness file > > (i.e. {subject}.{hemisphere}.thickness.native.surf.gii) as by default it is > > looking for an lh or rh .thickness file. > > > > Thanks in advance! > > > > _______________________________________________ > > HCP-Users mailing list > > [email protected] > > http://lists.humanconnectome.org/mailman/listinfo/hcp-users > > > > _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
