Wu-Minn HCP Data

[Kangrim Cho]

Hello HCP team,

I am looking up for Wu-Minn HCP data, and came up with a few questions about it.

1. Can I know the parameters of MR data acquisition? (ex. TR, TE)

2. Can I know the specifics about the 'dtseries' data? (MRI) I found that the size of data is much larger than other open access data that I have seen before such as ADNI. I expected that this data would be the ROI signals per 116 regions based on AAL template, but it seems to be a different one--the size of one of the data was (300, 91282). Can I ask for the specifics about the preprocessing procedure?

Regarding the questions, I looked up for the manuals and appendices about the data but I could not find the proper information. If there are any data sources that I can look into, please let me know about them.

Thank you.

Best regards,

Kangrim Cho

[Glasser, Matt]

1. https://www.humanconnectome.org/hcp-protocols
2. dtseries data are in CIFTI grayordinates.  The correct file to use would be ${StudyFolder}/${Subject}/MNINonLinear/Results/${fMRIName}/${fMRIName}_Atlas_MSMAll_hp2000_clean.dtseries.nii, which has been precisely functionally aligned across subjects and denoised with spatial ICA.  This could be parcellated using the HCP’s multi-modal parcellation which is a much more functionally relevant parcellation than AAL (https://balsa.wustl.edu/file/87B9N) using Connectome Workbench command wb_command -cifti-parcellate and then turned into a connectome with wb_command -cifti-correlation.

 

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 on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/e742e099-17ed-4edd-b088-2af59eb9fe72n%40humanconnectome.org.

 

---

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.

[Kangrim Cho]

Hello, 

Thank you for your kind response.

I just have one more question about the dataset.

Can I know the file name that I can check the ROI signals from the dataset? Is that the file that you mentioned in the second response?

Appreciate your help.

Regards,

Kangrim Cho

2022년 11월 22일 화요일 오후 7시 52분 13초 UTC+9에 glas...@wustl.edu님이 작성:

[Tim Coalson]

wb_command -cifti-parcellate averages its input file within each parcel (ROI) of the label file (atlas).  So, when the input is the BOLD dtseries, the output will be the average timecourse within each parcel.  I don't think we have made the MMP-parcellated timeseries available for download, so you need to download the dtseries files and run -cifti-parcellate yourself.

Tim

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/4f6384c6-f0d4-44bb-aec3-8d67b258a7d2n%40humanconnectome.org.

[Kangrim Cho]

Hello,

Thank you for your reply.

I understood that I should parcellate the dtseries data by myself using connectome workbench if I want the ROI signals. Is my understanding correct?

Also, I found the shape of the "${StudyFolder}/${Subject}/MNINonLinear/Results/${fMRIName}/${fMRIName}_Atlas_MSMAll_hp2000_clean.dtseries.nii" file to be (1200, 99216).

Then, how can I understand the size of this data? Are they already parcelled into 1200 different regions? 

Best regards,

Kangrim Cho

2022년 11월 30일 (수) 오전 5:34, Tim Coalson <tim.c...@gmail.com>님이 작성:

You received this message because you are subscribed to a topic in the Google Groups "HCP-Users" group.
To unsubscribe from this topic, visit https://groups.google.com/a/humanconnectome.org/d/topic/hcp-users/D9RtudfTiRA/unsubscribe.
To unsubscribe from this group and all its topics, send an email to hcp-users+...@humanconnectome.org.
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU4oJUFY-%3DD_tBLJkLhfUp9f5jtxnGhUZr0%2BVRDdE8KiSQ%40mail.gmail.com.

--

Kangrim Josephine Cho

B.A. in Quantitative Risk Management,

Cognitive Science | Yonsei University'22

E | kangr...@yonsei.ac.kr

[Tim Coalson]

99216 doesn't sound like the right number, our standard space is 91282.  Did you perhaps use ft_read_cifti to load the data?  The ft_ functions modify the data matrix so it isn't the original size.  We recommend our cifti-matlab v2 code if you want to use cifti data in matlab, available here:

https://github.com/Washington-University/cifti-matlab

If you didn't use ft_read_cifti (or matlab at all), I'd like to know the details of how you arrived at your 99216 number.

1200 is the number of timepoints in that run (which I expect was a resting state run).  The ~90k dimension is spatial locations (vertices and voxels).  After parcellating with the MMP 1.0, you should get a 360 x 1200 matrix for that run (or if you use a version with subcortical parcels taken from another source, because the MMP 1.0 is cortex-only, you will get somewhat more than 360).

Tim

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANpO2_vO6FCfL4QCaQYqVRxQhyDTRKQLbsX5OoOOhwQBQ7B-uA%40mail.gmail.com.

[Kangrim Cho]

Hello,

Thank you for your response.

I first opened the file through python just to check the size of it, and it gave 99216. But I got 91282 with the given code through matlab.

I am trying to parcellate the data with the MMP as you have explained based on Connectome Workbench, but I could not find the guideline of how to do it. I checked the tutorial file, but it seems that the tutorial does not include the MMP parcellation procedure. Would you provide any information or guidelines that I can refer to?

Much appreciated.

Best wishes,

Kangrim Cho 

2022년 12월 1일 (목) 오전 5:26, Tim Coalson <tim.c...@gmail.com>님이 작성:

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANm3QU5Scqke0oHF67XwZNpMW6RnzkLQhJWBzjUwS2q81%2BH32w%40mail.gmail.com.

[Tim Coalson]

What python library did you use to load the cifti file?  If nibabel is doing this to cifti files, we need to talk to them about it.

The parcellation command is wb_command -cifti-parcellate.  Run it without arguments to get the help info, or look here:

https://www.humanconnectome.org/software/workbench-command/-cifti-parcellate

We recommend using the midthickness vertex areas as spatial weights when parcellating surface data.

Tim

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CANpO2_verG8oLwRL8W13ic%2BdNasJaxAb-dQF0gjeZUATQz9Tqg%40mail.gmail.com.