[Prabhjot Kaur]

Hello Matt,

We are using workbench to process our  restFMRI data. We are using Cole anticevic atlas for parcellations. 

When trying to generate connectivity maps using wb-view, we are not getting connectivity matrix. What we can understand with our limited experience is that the label names are not been carried forward by parcels when applying  cifti-correlation command after cifti-parcellate. When checked using file info, the generated pconn.nii file is showing 'Maps with labelTable as false'. 

We have tried it with v2.1.0 also along with v1.4.2. 

Please guide with step-by-step process on how to go about this.

Prabhjot Kaur

PhD Scholar (IIT, Delhi) 

Senior Technical Officer (DRDO, Ministry of Defence, Govt of India)

91-9810904950

[Glasser, Matthew]

Maybe you should start with showing the commands you have used (with the latest version of Connectome Workbench).

Matt.

--
You received this message because you are subscribed to the Google Groups "HCP-Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hcp-users+...@humanconnectome.org.
To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANTmu3kMVQ6KM0AimgBmYrCHOJ3CW4Qnr0oUdq6qx1ht_YqPvQ%40mail.gmail.com.

 

---

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.

[Prabhjot Kaur]

Dear Mathew, 

I am using the following command line for generating pconn.nii file.

wb_command -cifti-parcellate sub-H002_bold.ptseries.nii \
           CortexSubcortex_ColeAnticevic_NetPartition_wSubcorGSR_parcels_LR.dlabel.nii \
           COLUMN sub-H002_bold_parcellated.ptseries.nii - method MEAN

wb_command -cifti-correlation sub-H002_bold_parcellated.ptseries.nii \
           sub-H002_bold.pconn.nii -fisher-z
The .pconn.nii file is created successfully and contains the correlation matrix, but the label tables from the atlas are missing (label table shows as false in Workbench).

Prabhjot Kaur

Delhi

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/4D46B2A6-E241-4E48-962A-9066852CE4DF%40wustl.edu.

[Glasser, Matthew]

That is expected behavior.  Label tables are found only in label files.

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANTmu3m%3D9a4Xx7xQJTTPAgX%3DX6%3D%3DbU4y6JHU1ui2oks4E8Zqfg%40mail.gmail.com.

[Tim Coalson]

To elaborate, -cifti-parcellate uses the information from the label table to make the parcel definitions, but the vertices and voxels that participate in a parcel are encoded directly into the cifti XML, without using the table of "label colors, names, and keys" (which is only needed for displaying the extent of the labels when there isn't any associated area-averaged data in the file).  ptseries, pscalar, and pconn, etc. all contain parcel-summarized data values that can be colored with continuous palettes, so they don't need to know about label colors.

Tim

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/EA55B79A-D49A-4505-9545-9AFB31892AF6%40wustl.edu.

[Prabhjot Kaur]

Hello Tim,

Understood that Pconn or ptseries is not supposed to retain those labels. But when I am trying to open this pconn.nii file in wb_view to generate connectome, it is not showing the connectivity matrix. As per my understanding, when I click on any voxel, it should show its connections in the form of connectivity matrix. Guide me to generate those connectivity maps or heatmaps.

Thanks

Prabhjot Kaur

Delhi

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU5CeCC7h0hbDx0B%3DhyutoAOCbFMNJ%3DiTiuHSasJvmeO4g%40mail.gmail.com.

[Tim Coalson]

You need to click on a vertex or voxel that is actually used by a parcel for the pconn layer to show anything, even when enabled.  Also make sure that both "load" and "layer" checkboxes in the connectivity tab are enabled (I think they should be by default).  You can load your dlabel file in another layer to show you where clicking will do something, and either turn the label layer off or change it to outline mode so you can see the pconn data through it.

Tim

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANTmu3%3DFqLH1vmAD3O9g_dh5xUiq00oX2AcJ%2BT6Gv-nWfDDEGQ%40mail.gmail.com.

[Prabhjot Kaur]

Hello Tim

Thanks for providing some clarity. As I can see, on adding pconn.nii files and dlabel.nii files, my layers box is not checked and is disabled in the connectivity tab.

But when I loaded dtseries.nii instead of pconn.nii, it generated an empty dynconn.nii file with both load and layer boxes checked in the connectivity tab.

Please guide.

Regards

Prabhjot Kaur

Delhi, India

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU77Pxw_rE3Yk-sO9y_1SC8k-JOQZYLDBnb%3DJPDYdSAhtA%40mail.gmail.com.

[Tim Coalson]

For files that already contain a connectivity matrix, they are already available as a layer, so that checkbox wouldn't do anything, so we disable it, but disabling it may prevent it from appearing to be checked, so that is fine.  Make sure the load checkbox is checked.  Then, make sure the pconn layer in the overlays tab is also enabled, click inside one of your parcel labels in the 3D view, and turn off any layers above the pconn to see the connectivity data. 

The "layer" checkbox is basically for disabling the dynconn feature per-file when you don't want to do correlations in wb_view, in order to make the list of files in the overlay selection shorter and easier to work with. 

Tim

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANTmu3kV0pfK%2B0D6Y3FMctucPFWQpYQarC_r%2BrSidcEx01WMPQ%40mail.gmail.com.

[Prabhjot Kaur]

Hello tim

I am still struggling with the issue.... 

I am hereby sharing the snippet of the wb_view window that is opening with pconn.nii and dlabel.nii files. Please have a look and guide as we are not getting an option to open 3d view.

Prabhjot Kaur

Delhi

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU7we4QmL6tHeCmR8-t_enyfmDMQ-5nYNDk_ffRok3%3Dz-A%40mail.gmail.com.

[Glasser, Matthew]

You also need to open some surface files (.surf.gii) to view things on surface models.

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANTmu3nXQ3ZCJ3T-QtDHgKGZfHXwOqp7Ndf%2BjiBxUx7dWq1nTQ%40mail.gmail.com.

[Prabhjot Kaur]

Hello Tim

 Writing you after long. 

While adding surf.gii file, we got an error where fMRIprep midthickness files have higher resolution with 122k vertices, whereas our pconn.nii file is 32k. so we used some open source DiedrichsenLab fs_LR_32 /fs_LR.32k.R.midthickness.surf.gii files. 

 Now some colored maps are generated (file attached) , plz guide how can we generate conncetivity maps/ heat maps for our selected networks.

Prabhjot Kaur

Technical Officer 'C'

Institute of Nuclear Medicine & Allied Sciences

Defence Research & Development Organization 

Ministry of Defence 

Government of India

Delhi

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/F37F5382-B0FD-4D0D-BDA3-B7B0B3A43CDD%40wustl.edu.

[Prabhjot Kaur]

Hello Tim 

I have one more query.

We are using fewer networks  ( only 3 or 4) as defined in CAB-NP atlas. How can we generate connectivity maps for those only.

Prabhjot Kaur

Technical Officer 'C'

Institute of Nuclear Medicine & Allied Sciences

Defence Research & Development Organization 

Ministry of Defence 

Government of India

Delhi

[Tim Coalson]

One possibility is to make a new .dlabel.nii file that represents networks instead of parcels, -cifti-parcellate the dtseries with that, and then use -cifti-cross-correlation to the dtseries or the parcel-based ptseries, depending on whether you want network-to-vertex or network-to-parcel correlations (or just -cifti-correlation for network-to-network correlations).

parcellated correlation files aren't that big, I would just generate the whole thing and then ignore the networks you don't care about.

Tim

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANTmu3nxpTgkA4B6WSqAN_WyaRehN%3DV4Mt_3SNNiUhoamNPdfQ%40mail.gmail.com.

[Germán González Torres]

Good evening everyone, I would like to ask for your help for a project we are working on, it is about RSFN (Neuropsychological Functional Symbolic Resonance), for this I need complete templates of the cortical and subcortical brain

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU4u1jv2UP%2Bo56UkSEmM%2BtqtGkPMKExac%3DQ23Z%3DYemKH4Q%40mail.gmail.com.

--

[Tim Coalson]

It isn't clear what you are asking for, as "template" can mean a few different things.  What information do you expect the template to contain?

Tim

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CAD5gnwVg_K_1vp-cb2e%2BH%2Byy2gkTEwi660yrmkJYcA4hJDru8g%40mail.gmail.com.

[Prabhjot Kaur]

Hello Tim, 

Plz guide how to modify the dlabel.nii file. Since, the diabetes file will contain parcels info in ascending order, do our chosen parcel list be in the same irder or it can be as per network affiliation.

Also, since we already have ptseries and pconn series, can we pick our parcels from there directly. We have already tried cifti merge command. Ptseries.nii contain info as 718×fmri volume, but it needs parcels in columns.

Fyi,  We are using wb v1.4.2

Thanks 

Prabhjot Kaur

Prabhjot Kaur

Technical Officer 'C'

Institute of Nuclear Medicine & Allied Sciences

Defence Research & Development Organization 

Ministry of Defence 

Government of India

Delhi

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU6v4TrkoyHpXM7aFGjvPumW3Snwy8va90s%2BQmVZM8pyDQ%40mail.gmail.com.

[Tim Coalson]

718 seems large for number of parcels.

Averaging data within a network would be simplest by making a new dlabel file, which is a bit different than modifying an existing one.  If the networks are specifically defined in terms of areas/parcels, then you can use the existing dlabel to build the network ROIs.  This would look something like the following (NOTE: -index treats any integer argument as a numeric index rather than a text name, so if you have a parcel named just "86" and it isn't the 86th smallest key value (ignoring the "???" key) in the original dlabel file, you might want to edit the parcel name to "86_ROI" or similar):

wb_command -cifti-all-labels-to-rois myparcels.dlabel.nii 1 parcelrois.dscalar.nii

wb_command -cifti-merge rightvisualparcels.dscalar.nii -cifti parcelrois.dscalar.nii -index R_V1 -index R_V2 -index R_V3 ...

wb_command -cifti-reduce rightvisualparcels.dscalar.nii MAX rightvisualparcels_combined.dscalar.nii

<same for other networks>

wb_command -cifti-merge networkrois.dscalar.nii -cifti rightvisualparcels_combined.dscalar.nii -cifti rightsensorimotorparcels_combined.dscalar.nii ...

wb_command -cifti-reduce networkrois.dscalar.nii INDEXMAX networkrois_index.dscalar.nii

wb_command -cifti-reduce networkrois.dscalar.nii MAX networkrois_max.dscalar.nii

wb_command -cifti-math 'index * (max > 0)' networkrois_index_masked.dscalar.nii -var index networkrois_index.dscalar.nii -var max networkrois_max.dscalar.nii

wb_command -cifti-label-import networkrois_index_masked.dscalar.nii "" networks.dlabel.nii

If you want to have the network names in your dlabel file (instead of LABEL_1, etc and random colors), create a text file as described in the -cifti-label-import help and pass its filename instead of the empty quotes in the last command.  If you really have 718 parcels, though, this strategy may be tedious, and you should think about how to easily get binary ROIs representing each network.

If you don't actually want to combine the data across parcels, you can instead just drop the parcels that aren't of interest by -cifti-label-export-table, deleting the pairs of lines that represent the labels you don't want, and doing -cifti-label-import with the -discard-others option.  Don't edit the key values here, that would put the label names on the wrong part of the brain.

The order that -cifti-parcellate uses for its output is the label key values, sorted ascending.  You can use -cifti-label-modify-keys to change the keys while keeping the label names on the correct locations.

Tim

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANTmu3%3DhCSyqUDWuTvC040vDKHuJqyX9SEKzvVTgha%3DtAC_dUg%40mail.gmail.com.