Hi again, Many thanks Matt, Carl and Tim for such quick replies! A few follow-up questions are in-line below:
<http://about.me/matthewliptrot> I am a novice at handling surfaces, and at using the Connectome Workbench/commandline, so apologies if the following is obvious. We want to be able to perform connectivity analysis of subjects (both HCP and our own) using diffusion tractography. In order to directly compare the resultant (voxel-to-voxel) connectivity matrices across subjects, we would like to have a one-to-one correspondence between the tractography seed voxels in each subject. I presume that the best way to do this would be to first select the desired set of (inter-subject matched) seed points upon the aligned surface meshes already provided (BIG thanks!) by the HCP and then convert them to the (nearest) voxel? (I realise that this may be a surjective mapping). Are there functions/commands within the Connectome Workbench that can do this? Or are there more sensible approaches to achieving this? We don't seed cortex tractography from voxels, we seed from surfaces. FSL tractography supports this: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FDT/UserGuide#Using_surfaces Great! Qn 1: Is the recommended approach to perform tractography seeded from the ’32K registered surface mesh in native volume space’ (the final surface generated by the PostFreeSurfer pipeline)? Qn 2: Are these surfaces stored in what FSL calls the ‘caret’ convention? Then we use workbench commands to convert the outputs as needed for viewing in workbench, and to stitch the two cortex pieces and any volume pieces together into one file, etc. Also, in [1], it states that the ‘standard grayordinate space’ (in CIFTI format) is used for achieving bijective inter-subject correspondence of the WM surface vertices and the subcortical voxels. However, I’m a bit confused how this bijective correspondence is maintained during the resampling from the 164K to the 32K meshes i.e. does the adaptive barycentric surface resampling method guarantee that the 32K-to-32K mappings are also bijective? The resampling of an individual surface from one atlas mesh to another uses the atlas spheres, not the subject spheres, so you get the same mesh and vertex correspondence every time. Also, we do surface resampling with barycentric, not adaptive barycentric area resampling, because "signal loss" doesn't really apply to coordinate data - barycentric resampling doesn't add smoothness when downsampling, so it stays closer to the original anatomical contours in areas with curvature. Thanks for the clarification. Qn 3: Do you have any references/citations for the barycentric resampling methods that were implemented? Cheers, M@ -- Matthew George Liptrot <http://about.me/matthewliptrot> Department of Computer Science University of Copenhagen & Section for Cognitive Systems Department of Applied Mathematics and Computer Science Technical University of Denmark http://about.me/matthewliptrot _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
