On Wed, Feb 20, 2019 at 8:03 AM CHAUMON Maximilien <
maximilien.chau...@icm-institute.org> wrote:
> Hello,
>
> I'm looking at fine changes in MEG forward leadfields and would like to
> use the 164k meshes in each subject (I know 164k vertices are overkill, but
> I need this high res rendering for one of my figures). I'm interested in
> the actual original location of individual vertices in the brain in
> subject-specific ACPC-aligned headcoordinate space. I don't want any non
> linear spatial transformation applied to the mesh.
> So I would like to use the 164k mesh with coordinates without the
> nonlinear transformation that (as far as I understand) was applied to all
> the 164k_fs_LR files. Is there an easy way to revert the non linear
> transformation?
>
Only the surface files under the MNINonLinear folder have had any nonlinear
anatomical warp applied. The surface files under T1w all line up with the
distortion corrected, rigidly aligned T1w image (we don't really keep
scanner space around, and we often call this distortion corrected rigid
alignment space "native volume space"). Surface registration/resampling
does not deform the anatomy, it just tiles the same contour in 3D space
with a new set of triangles.
Note that averaging the coordinates of vertices across subjects will change
their location, and this will affect geometry and "foldedness", to a degree
depending on what registration was used. As long as you stick with
individual surfaces, you don't need to worry about this.
> I would then use the file
> {Subject}.{Hemi}.midthickness.164k_fs_LR.surf.gii and apply the inverse
> transformation,
>
> Alternatively, is there a way to easily downsample
> {Subject}.{Hemi}.midthickness.native.surf.gii to 164k vertices? is this
> then in subject-specific ACPC-aligned headcoordinate space? how could I
> move to that space?
>
This is what the existing 164k surfaces in the T1w folder already are.
Native mesh is commonly ~130k for HCP scans, so 164k is actually a small
upsampling. We recommend the MSMAll versions, as the same vertex number
across subjects is more often in the same area than for other registrations.
For reference (since we have already dealt with this resampling for you),
the recommended sphere to use for resampling from native mesh to any fs_LR
mesh is "MNINonLinear/Native/100307.L.sphere.MSMAll.native.surf.gii" (and
the R version, of course). Ignore the fact that it is in the MNINonLinear
folder, sphere surfaces don't have a volume space, that is just where it
got put. The standard spheres for fs_LR are in the pipelines
under global/templates/standard_mesh_atlases/, read the readme file. The
command to do surface resampling is wb_command -surface-resample.
> Does that make sense?
>
> Many thanks,
> Max
>
> _______________________________________________
> HCP-Users mailing list
> HCP-Users@humanconnectome.org
> http://lists.humanconnectome.org/mailman/listinfo/hcp-users
>
_______________________________________________
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users
