I would agree that center of gravity doesn't seem all that useful a measure of a parcel, which is why parcellated cifti files contain the information of the entire ROI of each parcel. Note that the center of gravity may lie significantly outside the surface if the ROI is large and covers significant curvature, so the center of gravity of a good area definition may end up in an anatomically implausible location. Also note that different subjects will have the same area at different coordinates, despite nonlinear volume registration, for reasons such as subject variability of folding, and that many areas' locations are somewhat variable with respect to the anatomical features that volume registration uses.
However, it is possible to calculate the center of gravity of the surface parcels by using -surface-coordinates-to-metric, separating your parcellation's label file into gifti label files with -cifti-separate, making those into rois with -gifti-all-labels-to-rois, then finally using -metric-weighted-stats to generate the mean of the coordinates in each roi, weighting by surface area (maybe take cortical thickness into account too). For the volume, there isn't an easy way to do this in workbench, as it doesn't have a command to put the coordinates of voxels into a volume file. We generally don't use center of gravity, which is one reason we haven't focused on making these easy to do. Tim On Sat, Jul 18, 2015 at 11:38 AM, Donna Dierker <[email protected]> wrote: > Hi Eran, > > No easy answers leap to mind, but a few questions/thoughts: > > * You imported the surface, but were unable to import the parcels? Or the > link between the two wasn't evident? The vertex index/number is the link > between the two. The surf.gii has the x,y,z coords (and it matters whether > you choose white, midthickness, or pial), and the label file has the > vertex-parcel assignment. Depending on your reader/conversion programs, it > might be easier to convert from cifti to gifti to import into matlab, > though there is a toolbox for reading cifti in matlab. > > * This sort of thing was trivial in caret_command > (surface-region-of-interest-selection and surface-roi-statistical-report), > but I can't find equivalent functionality in wb_command, after looking at > the wb_command list commands output and searching the full help output for > words like report, information, and coordinate. Maybe Matt or Tim can help > me find it, if it is there. > > * I confess I wonder what your broader goals are, because I look at > parcellations like these: > > https://surfer.nmr.mgh.harvard.edu/fswiki/CorticalParcellation_Yeo2011 > > … and I think some of those parcels are pretty big/extended -- elongated > and subject to considerable gyrification. A center of gravity seems less > useful to me in this context, though for the subcortical structures it > seems more useful. > > Those are the thoughts that come to mind. > > Donna > > > On Jul 17, 2015, at 2:30 PM, "Dayan, Eran (NIH/NINDS) [F]" < > [email protected]> wrote: > > > Hi Donna > > > > Just to follow up on my previous question. I am indeed trying to compute > center of gravity (in MNI or talairach) for all cortical and subcortical > parcels from which volume measurements were extracted. I was able to import > the various surf.gii files in MATLAB, but it isn't helping me much as I > can't register the actual parcels onto this imported image and therefore > can't compute what I need. Are you aware, by any chance, of any other way > this can be done? > > > > > > Many thanks > > > > E. > > > > -----Original Message----- > > From: Dayan, Eran (NIH/NINDS) [F] > > Sent: Wednesday, July 01, 2015 12:51 PM > > To: 'Donna Dierker' > > Subject: RE: [HCP-Users] Freesurfer parcellation - list of coordinates > > > > Hi Donna > > > > Thanks! I will have a look right away. Yes, the idea is to compute > center of gravity. But if all coordinates are available for each parcel > then it's pretty easy (e.g., through MATLAB) > > > > Best > > > > Eran > > > > -----Original Message----- > > From: Donna Dierker [mailto:[email protected]] > > Sent: Wednesday, July 01, 2015 12:17 PM > > To: Dayan, Eran (NIH/NINDS) [F] > > Cc: [email protected] > > Subject: Re: [HCP-Users] Freesurfer parcellation - list of coordinates > > > > Hi Eran, > > > > An HCP subject's MNINonLinear subdirectory has files like: > > > > *.surf.gii - vertex coordinate and neighbor relationships > *aparc*dlabel.nii - vertex parcel assignments > > > > So for any vertex, its coords are in *surf.gii and its parcels are in > *aparc*dlabel.nii. > > > > This is for the cortex. For the subcortical structures, you have > aparc+aseg.nii.gz. > > > > Are you trying to compute a center of gravity for the parcels which span > multiple vertices/voxels? > > > > Donna > > > > > > On Jun 30, 2015, at 12:20 PM, "Dayan, Eran (NIH/NINDS) [F]" < > [email protected]> wrote: > > > >> > >> Was wondering if at some point along the Freesurfer pipeline, the > (MNI/Talairach) coordinates of the parcellated cortical and subcortical > structures were saved? If so, how can these be accessed? > >> > >> Many thanks! > >> > >> Eran > > > > _______________________________________________ > > 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 > _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
