Re: [HCP-Users] ROIs and Betas from Cifti Data

[Harms, Michael]

As an aside, if the end-game of the analysis is a permutation approach (e.g,. PALM) applied to GLM betas, then I wouldn’t expect it to make much difference in terms of power to detect an effect if you parcellate before fitting the Level 1 GLM, or if you simply average the betas from the dense maps within each parcel (and then use those average betas as input to the permutation testing).  The latter approach would allow one to simply run -cifti-parcellate on the dense task maps that HCP has already pre-computed. cheers, -MH --  Michael Harms, Ph.D. ----------------------------------------------------------- Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave. Tel: 314-747-6173 St. Louis, MO  63110 Email: mha...@wustl.edu From: <hcp-users-boun...@humanconnectome.org> on behalf of "Glasser, Matthew" <glass...@wustl.edu> Date: Tuesday, September 20, 2016 at 10:06 AM To: "Michael F.W. Dreyfuss" <mid2...@med.cornell.edu>, NEUROSCIENCE tim <tsc...@mst.edu> Cc: "hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org> Subject: Re: [HCP-Users] ROIs and Betas from Cifti Data As Tim mentions, it sounds like you might want to use a parcellated analysis, as this will be more sensitive/powerful and you’ll know exactly what areas you are finding.  The HCP’s multi-modal parcellation is available here: https://balsa.wustl.edu/study/show/RVVG   Also, the HCP’s task analysis pipeline will allow you to parcellate before fitting the GLM, rather than afterwards to get the addition SNR benefits from averaging across a parcel.   Peace, Matt. From: <hcp-users-boun...@humanconnectome.org> on behalf of "Michael F.W. Dreyfuss" <mid2...@med.cornell.edu> Date: Monday, September 19, 2016 at 9:40 PM To: Timothy Coalson <tsc...@mst.edu> Cc: "Michael F.W. Dreyfuss" <mid2...@med.cornell.edu>, "hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org> Subject: Re: [HCP-Users] ROIs and Betas from Cifti Data Thank you, Are there any examples available for how to use this, possibly? I have been trying to figure these out, but there are a lot of options and  I am just not able to decipher how to use these from the help page alone. The errors I am getting also do not clarify what I need to do to get the output I am looking for. Before using this kind of multi-band data I had been using afni. To give an example of what I would want to do in terms of afni commands (if that’s any help), I would have saved all ROIs and then used 3dmaskave to extract mean beta weights for a given GLM beta for each subject and then I would relate those beta weights so subject’s behavior in R or another stats package.  Definitely agree that there’s not much meaning to a peak coordinate per se. I’m just trying to figure out how to report the clusters I am finding. In previous reports we would typically focus on broadmann areas or more general regional nomenclature (i.e. vmPFC, mid temporal lobe, etc.). Some of the clusters I’m finding also cover large areas from motor to visual cortex, so I am trying to consider good ways to report that. At this point I would prefer to use TFCE or some other thresholding method to identify contiguous swaths of volumetric and surface activation. Thank you very much again, Michael On Sep 19, 2016, at 6:41 PM, Timothy Coalson <tsc...@mst.edu> wrote: On Mon, Sep 19, 2016 at 4:51 PM, Michael F.W. Dreyfuss <mid2...@med.cornell.edu> wrote: Thank you, How can I turn the ROIs into a label file? You can use -cifti-find-clusters if you just want spatial contiguity to define where ROIs should be considered separate, then use -cifti-label-import to make them into a dlabel file. Also, how can I simply get a list of the ROIs with some information like cluster extent and peak voxel to be able to identify what part(s) of the brain each ROI is covering? A single coordinate isn't a faithful representation of the cluster.  You can make a figure showing the clusters displayed on the brain (for instance, choose two of: beta maps, significance outlines, area outlines), and hopefully also provide the unthresholded beta and z maps for others to use. You can get cluster extent info with -cifti-weighted-stats. If the question you want to ask is "which areas are involved", you could do a parcellated analysis instead of a cluster analysis. Thank you, Michael From: Timothy Coalson <tsc...@mst.edu> Sent: Monday, September 19, 2016 4:48:28 PM To: Michael F.W. Dreyfuss Cc: hcp-users@humanconnectome.org Subject: Re: [HCP-Users] ROIs and Betas from Cifti Data   The wb_command -cifti-weighted-stats command with -mean is probably what you want (outputs a number to the command line), though you'll need to have each ROI as a separate file and run it separately for each of them.  Alternatively, if you turned the ROIs into a label file, you could get -cifti-parcellate to make a file where each parcel contains the answer for an ROI. Tim On Mon, Sep 19, 2016 at 3:26 PM, Michael F.W. Dreyfuss <mid2...@med.cornell.edu> wrote: Hello, I have run palm with TFCE on my group level data successfully for a task based fMRI study (yay!), and I would like to be able to identify ROIs from my cifti data (both surface and volume). I then want to extract subject level beta weights for a given condition from those ROIs to relate those betas to behavior (offline). Are there simple ways to: 1) identify regions implicated on the group level and 2) extract subject-level beta weights from them, such as with wb_command? Thank you, Michael _______________________________________________ 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   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 HCP-Users@humanconnectome.org 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. _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users