Hi Basile et al. We will make this three-vertex count optional in the future, as I agree that it complicates interpretation of matrix1/3 output.
Cheers Saad On 5 Jun 2014, at 12:30, Glasser, Matthew <[email protected]<mailto:[email protected]>> wrote: You could do that but because white is the counting surface and pial is the stopping surface it is unlikely to be necessary. So basically you don’t like the fact that when a surface triangle is hit, all three vertices counts are incremented instead of the nearest vertex? I suppose that’s not such a good idea if one were connection strengths of different kinds of structures. I’ll ask them about it this afternoon. Peace, Matt. From: basile pinsard <[email protected]<mailto:[email protected]>> Date: Thursday, June 5, 2014 at 11:42 AM To: Matt Glasser <[email protected]<mailto:[email protected]>> Cc: Donna Dierker <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] gifti labels to volume ok, do you have applied dilation to the mask before inverting it? can't there be side effect between aliased volume mask and wm/gm surfaces? However I wonder if in the case it misses the surface for stopping condition and crosses over if the fiber is counted as endpoint in the matrix... the second point is about normalization of the matrix when summing over regions of interest (to get some kind of more quantitative value). When a fiber hits a triangle all of the 3 vertex entries in the matrices (let say rows) to the other endpoint of the fiber have their count increased of 1. so it will count each fiber either 4*3/2= 6 times if one end is on a vertex and the other in a volume ROI or 6*5/2=15 times if both endpoints are surfaces. So dividing by this numbers the fibers count when summing over ROIS is necessary to get sensible results. However the diagonal value will carry over-estimated self-connecting fibers as pairs of vertices in the same triangle have their count increment each time a fiber hits this triangle. This can be filtered if you remove every matrix entry of neighboring vertices on the surface. hope this is clear enough, you can also see this thread here: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=fsl;a98283e4.1405 cheers basile On Thu, Jun 5, 2014 at 12:26 PM, Glasser, Matthew <[email protected]<mailto:[email protected]>> wrote: I’ve used the 32k mesh. Also I restrict the tracking mask not to allow stuff outside the brain (inverse of HCP distributed wmparc file minus the ventricles), so even a few fibers missed don’t go far (--mask= option). Not sure I understand your other point. Can you have another try at explaining it? From: basile pinsard <[email protected]<mailto:[email protected]>> Date: Thursday, June 5, 2014 at 11:19 AM To: Matt Glasser <[email protected]<mailto:[email protected]>> Cc: Donna Dierker <[email protected]<mailto:[email protected]>>, "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] gifti labels to volume Hi Matt, Even if it is marginal a certain number of fibers are in the cortical and ROIs gray matter as assessed by fdt_paths.nii, I think I already mentioned this to Saad on fsl list. Have you been using the fsaverage32k surface as stop/targets or the full resolution (which I believe would produce huge matrix compared to the already huge matrix when using fsaverage32k) ? Also I had a hard time to understand that for each fiber terminating on surfaces at it's endpoints all combinatorial pair of vertices of triangles which are hit are incremented. Cheers basile On Thu, Jun 5, 2014 at 9:45 AM, Glasser, Matthew <[email protected]<mailto:[email protected]>> wrote: I’m a bit surprised at these issues with probtrackx (I’ve not had problems with stop surfaces for example), perhaps you should ask for help/clarification on the FSL list? It certainly would seem to be the path of least resistance/most accurate to use FSL for tractography since it already works with HCP data (and in the future we will be providing fiber orientation results for FSL as well). Also if there are important features missing, I think we would all like to know about that. Peace, Matt. From: basile pinsard <[email protected]<mailto:[email protected]>> Date: Thursday, June 5, 2014 at 7:42 AM To: Donna Dierker <[email protected]<mailto:[email protected]>> Cc: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] gifti labels to volume Many thanks for the answers @Timothy I know that probtrackx2 support surfaces but I did not manage to make it work without hectic results (fiber crossing the surfaces used as stop, combinatorial counting of endpoints ... ), even if FSL seems the most able to process multi-shell data. @Donna I already tried the caret_command method but got stuck to converting the gifti scalars to a metric/paint format, I will give it a try with the command you mentioned, but still this implies a lot of conversion steps. However I wrote a python function that for each voxel in the cortical ribbon finds the closest vertex on the surface and assigns the latter's label,. The voxels which are close to the surface (white) should be properly assigned the right label which is ok for me as I am using this as target for tractography. Cheers basile On Thu, Jun 5, 2014 at 3:45 AM, Donna Dierker <[email protected]<mailto:[email protected]>> wrote: As far as I know, caret5's multi-fiducial mapping is volume->surface rahter than surface->volume. The caret_command method works, but you'll need to convert HCP surf.gii -> caret5 coord/topo (caret_command -file-convert -sc …). And your label.gii or func.gii will need to be converted, too (caret_command -file-convert -format-convert BINARY …). On Jun 4, 2014, at 7:22 PM, Timothy Coalson <[email protected]<mailto:[email protected]>> wrote: > Workbench currently doesn't have a way to do that. caret5 has a way to do > this (caret_command -surface-to-volume, and there is another method called > multi-fiducial mapping, though I'm not sure how to invoke it), though you may > need to do some file conversion. > > On a side note, FSL's probtrackx does support surface data, and is generally > what we advocate for tractography. > > Tim > > > > On Mon, Jun 2, 2014 at 8:50 AM, basile pinsard > <[email protected]<mailto:[email protected]>> wrote: > Hi hcp experts, > > I was wondering if anybody knew a way to map back to volume the values in a > surface gifti from HCP data. The main interest is to create ROIs on the > surface using this topological constraint and then send it back to volume for > let say a tractography software which does not support surface. > > mri_surf2vol won't work as it requires the freesurfer structure (which can > certainly be restored by converting all the HCP files). > > Many thanks > > -- > Basile Pinsard > PhD candidate > Laboratoire d'Imagerie Biomédicale, UMR S 1146 / UMR 7371, Sorbonne > Universités, UPMC, INSERM, CNRS > Unité de Neuroimagerie Fonctionnelle, CRIUGM, Université de Montréal > _______________________________________________ > HCP-Users mailing list > [email protected]<mailto:[email protected]> > http://lists.humanconnectome.org/mailman/listinfo/hcp-users > > > _______________________________________________ > HCP-Users mailing list > [email protected]<mailto:[email protected]> > http://lists.humanconnectome.org/mailman/listinfo/hcp-users > -- Basile Pinsard PhD candidate Laboratoire d'Imagerie Biomédicale, UMR S 1146 / UMR 7371, Sorbonne Universités, UPMC, INSERM, CNRS Unité de Neuroimagerie Fonctionnelle, CRIUGM, Université de Montréal _______________________________________________ HCP-Users mailing list [email protected]<mailto:[email protected]> 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. -- Basile Pinsard PhD candidate Laboratoire d'Imagerie Biomédicale, UMR S 1146 / UMR 7371, Sorbonne Universités, UPMC, INSERM, CNRS Unité de Neuroimagerie Fonctionnelle, CRIUGM, Université de Montréal ________________________________ 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. -- Basile Pinsard PhD candidate Laboratoire d'Imagerie Biomédicale, UMR S 1146 / UMR 7371, Sorbonne Universités, UPMC, INSERM, CNRS Unité de Neuroimagerie Fonctionnelle, CRIUGM, Université de Montréal ________________________________ 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 [email protected]<mailto:[email protected]> http://lists.humanconnectome.org/mailman/listinfo/hcp-users -- Saad Jbabdi University of Oxford, FMRIB Centre JR Hospital, Headington, OX3 9DU, UK (+44)1865-222466 (fax 717) www.ndcn.ox.ac.uk/team/researchers/saad-jbabdi<http://www.ndcn.ox.ac.uk/team/researchers/saad-jbabdi> _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
