Inline replies Tim
On Tue, Jun 10, 2014 at 1:24 PM, Ommen, Jurgen <omme0...@stthomas.edu> wrote: > First of all, again - thank you very much for your help. The support is > really great. > > > > Sorry for the ambiguity with the term “network” – let me try to be more > specific in what I try to do. > > I meant a classic graph with a set of nodes and edges, where the nodes > represent voxels (or other very fine brain regions) and the edges represent > the connections between voxels. > > As I currently have the opportunity to analyze very large graphs, I would > like to use a graph which is derived from brain imaging. Therefore I hoped > that I could use the dense connectome files from your HC project to > generate/extract my nodes and edges. > Well, .dconn.nii files are basically adjacency matrices - its just that it is a 91,282 node complete graph, so it may be unwieldy. If you know of groups of brainordinates that you want to be averaged to form a single graph node, you can use -cifti-parcellate after putting that information into a .dlabel.nii file. However, finding good parcellations is still a very active topic. If you are able to deal with a 91k node complete graph, you could convert the .dconn.nii to gifti with -cifti-convert -to-gifti-ext, and read it into matlab with the matlab gifti toolbox (Matt has some matlab functions to streamline this), which may get you closer to having it in a format you can use. > Are there some steps you could recommend to get from the dense connectome > files to a classic graph representation? And do you also offer tools to > generate anatomical graphs of dMRI images? > > > > And when I use –cifti-correlation on the dtseries files, should I first > combine the LR and RL dtseries files with some averaging functions or would > I also get a “correct” output by just using one of them? > Averaging resting state runs is not a good idea. Concatenation after normalizing is a better idea, see http://www.mail-archive.com/hcp-users@humanconnectome.org/msg00444.html and the other messages in the thread. There may be a better method in the future. > Many thanks again for your help, > > Jürgen > > > > > > *From:* Timothy Coalson [mailto:tsc...@mst.edu] > *Sent:* Friday, June 06, 2014 6:41 PM > > *To:* Ommen, Jurgen > *Cc:* hcp-users@humanconnectome.org > *Subject:* Re: [HCP-Users] Brain Connectivity Matrices > > > > First, a slight correction to my previous reply: the command is > -cifti-correlation, but you seem to have figured that out. > > > > The HCP data is in Cifti files, and contain combined surface (for cortex) > and voxel (for subcortical structures and cerebellum) data. The surfaces > used have roughly equivalent spacing to the voxels used, so in some sense > it is at the voxel "scale". However, two thirds of the data is not in > voxels, but in surface vertices. Yes, it is a complete connectivity > matrix, hence the large size. > > > > Connectome Workbench ( > http://www.humanconnectome.org/software/connectome-workbench.html) is the > main tool we use with cifti files, you can visualize the data with the GUI, > or do various operations on it (including extracting the data into other > formats) with wb_command. Matt Glasser has written some matlab functions > that use wb_command -cifti-convert and the matlab gifti toolbox to get the > data into matlab, but I don't think we currently have a good way to do > spatial operations on it in matlab. > > > > I don't know what you mean by "network", and might not be able to help you > with it if I did, but others on the list might. > > > > Tim > > > > On Fri, Jun 6, 2014 at 12:33 PM, Ommen, Jurgen <omme0...@stthomas.edu> > wrote: > > Hi Tim, > > > > Great, thanks for your fast answer. Generating the dense connectome files > works good. > > > > However, I’m not quite sure about the content of the dense connectome > files. As far as I understood it, it is the complete connectivity matrix on > a voxel-based scale. Is this right? This would be exactly what I need. > > > > My question now would be: Which steps or tools are usually used to extract > the information stored in the dense connectome files for further > post-processing? It would be really great if you could just outline the > steps roughly and the tools I could use to get the network. Is there maybe > a Matlab library available which is able to read .dconn.nii files? > > > > Thanks again for your help, > > Jürgen > > > > *From:* Timothy Coalson [mailto:tsc...@mst.edu] > *Sent:* Wednesday, June 04, 2014 7:42 PM > *To:* Ommen, Jurgen > *Cc:* hcp-users@humanconnectome.org > > > *Subject:* Re: [HCP-Users] Brain Connectivity Matrices > > > > I believe we don't include those in the releases, because dense connectome > files are very large (~30GB each). You should be able to generate them by > running wb_command -cifti-correlate on a subject's rfMRI dtseries file. > > > > Tim > > > > > > On Tue, Jun 3, 2014 at 5:55 PM, Ommen, Jurgen <omme0...@stthomas.edu> > wrote: > > Hello everyone, > > > > I’m working on graph theoretic analysis of the human’s brain network. > After studying the documentation of the Q3 release, I’ve only found group > average dense connectome files so far. > > > > I’d like to know if there are any connectivity matrices available for > individual subjects with which I could generate the corresponding > connectome networks. Do you provide this kind of data? Where could I find > it? > > And if not, do you plan to include it in future releases? > > > > > > Thanks for your help in advance and my best regards, > > Jürgen > > > > _______________________________________________ > 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
