Loading GIFTI to MATLAB with HCP command

[mason wells]

Hi Experts,

I am using the HCP command gifti=(filename) to load BOLD data projected to the fsaverage surface into MATLAB. However, the command is taking almost 17 minutes to run when using 2021b and close to an hour with 2018b. Has anyone else ran into this issue, or is there any suggestions for speeding this up?

Any tips would be great.

Best,

Mason

[Glasser, Matt]

GIFTI is not the best format for fMRI timeseries data, particularly at high mesh resolution.  I would recommend using CIFTI, which should load a bit faster.

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/ac0fa0f4-6c2b-4a76-8621-19f2d2396146n%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.

[mason wells]

Thanks, Matt. Unfortunately, I am working with an existing dataset and the space they are using is fsaverage in GIFTI format (the data was processed using fMRIPrep). I only need to load the data in MATLAB to perform some exploratory vertex-to-vertex correlations and so need the data in the standard vertex space. Are there any other GIFTI loading commands that HCP use that might load the data faster?

[Glasser, Matt]

That’s too bad then.  My understanding is that fMRIPrep can also produce CIFTI files, but the folks who developed fMRIPrep may have been a bit more focused on volume-based analysis.  Although we produce some GIFTI native mesh timeseries, they are only an intermediate file and only are used in wb_command, which doesn’t have as much of a slowdown.  If you don’t like the speed of the GIFTI toolbox, you can complain to its developer or improve it yourself. 

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/e04f4e28-01d6-4c0e-a425-4177a5cdad9dn%40humanconnectome.org.

[Patterson, Dianne K - (dkp)]

fMRIprep can definitely produce CIFTI files and can do so in both standard and native spaces.

I am surprised that it is possible to get just the GIFTI files. Here's what I get from fMRIprep.  Note that the fMRI data is in standard space by default, unless you ask for native space to be provided as well (which I have here):

├── anat

│   ├── sub-219_ses-itbs_desc-aparcaseg_dseg.nii.gz

│   ├── sub-219_ses-itbs_desc-aseg_dseg.nii.gz

│   ├── sub-219_ses-itbs_desc-brain_mask.json

│   ├── sub-219_ses-itbs_desc-brain_mask.nii.gz

│   ├── sub-219_ses-itbs_desc-preproc_T1w.json

│   ├── sub-219_ses-itbs_desc-preproc_T1w.nii.gz

│   ├── sub-219_ses-itbs_dseg.nii.gz

│   ├── sub-219_ses-itbs_from-MNI152NLin2009cAsym_to-T1w_mode-image_xfm.h5

│   ├── sub-219_ses-itbs_from-MNI152NLin6Asym_to-T1w_mode-image_xfm.h5

│   ├── sub-219_ses-itbs_from-T1w_to-MNI152NLin2009cAsym_mode-image_xfm.h5

│   ├── sub-219_ses-itbs_from-T1w_to-MNI152NLin6Asym_mode-image_xfm.h5

│   ├── sub-219_ses-itbs_from-T1w_to-fsnative_mode-image_xfm.txt

│   ├── sub-219_ses-itbs_from-fsnative_to-T1w_mode-image_xfm.txt

│   ├── sub-219_ses-itbs_hemi-L_inflated.surf.gii

│   ├── sub-219_ses-itbs_hemi-L_midthickness.surf.gii

│   ├── sub-219_ses-itbs_hemi-L_pial.surf.gii

│   ├── sub-219_ses-itbs_hemi-L_smoothwm.surf.gii

│   ├── sub-219_ses-itbs_hemi-R_inflated.surf.gii

│   ├── sub-219_ses-itbs_hemi-R_midthickness.surf.gii

│   ├── sub-219_ses-itbs_hemi-R_pial.surf.gii

│   ├── sub-219_ses-itbs_hemi-R_smoothwm.surf.gii

│   ├── sub-219_ses-itbs_label-CSF_probseg.nii.gz

│   ├── sub-219_ses-itbs_label-GM_probseg.nii.gz

│   └── sub-219_ses-itbs_label-WM_probseg.nii.gz

└── func

    ├── sub-219_ses-itbs_task-rest_run-1_AROMAnoiseICs.csv

    ├── sub-219_ses-itbs_task-rest_run-1_desc-MELODIC_mixing.tsv

    ├── sub-219_ses-itbs_task-rest_run-1_desc-confounds_timeseries.json

    ├── sub-219_ses-itbs_task-rest_run-1_desc-confounds_timeseries.tsv

    ├── sub-219_ses-itbs_task-rest_run-1_from-T1w_to-scanner_mode-image_xfm.txt

    ├── sub-219_ses-itbs_task-rest_run-1_from-scanner_to-T1w_mode-image_xfm.txt

    ├── sub-219_ses-itbs_task-rest_run-1_space-MNI152NLin6Asym_desc-smoothAROMAnonaggr_bold.json

    ├── sub-219_ses-itbs_task-rest_run-1_space-MNI152NLin6Asym_desc-smoothAROMAnonaggr_bold.nii.gz

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsLR_den-91k_bold.dtseries.json

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsLR_den-91k_bold.dtseries.nii

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsLR_den-91k_bold.json

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsaverage_hemi-L_bold.func.gii

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsaverage_hemi-L_bold.json

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsaverage_hemi-R_bold.func.gii

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsaverage_hemi-R_bold.json

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsnative_hemi-L_bold.func.gii

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsnative_hemi-L_bold.json

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsnative_hemi-R_bold.func.gii

    ├── sub-219_ses-itbs_task-rest_run-1_space-fsnative_hemi-R_bold.json

    ├── sub-219_ses-itbs_task-rest_run-2_AROMAnoiseICs.csv

    ├── sub-219_ses-itbs_task-rest_run-2_desc-MELODIC_mixing.tsv

    ├── sub-219_ses-itbs_task-rest_run-2_desc-confounds_timeseries.json

    ├── sub-219_ses-itbs_task-rest_run-2_desc-confounds_timeseries.tsv

    ├── sub-219_ses-itbs_task-rest_run-2_from-T1w_to-scanner_mode-image_xfm.txt

    ├── sub-219_ses-itbs_task-rest_run-2_from-scanner_to-T1w_mode-image_xfm.txt

    ├── sub-219_ses-itbs_task-rest_run-2_space-MNI152NLin6Asym_desc-smoothAROMAnonaggr_bold.json

    ├── sub-219_ses-itbs_task-rest_run-2_space-MNI152NLin6Asym_desc-smoothAROMAnonaggr_bold.nii.gz

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsLR_den-91k_bold.dtseries.json

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsLR_den-91k_bold.dtseries.nii

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsLR_den-91k_bold.json

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsaverage_hemi-L_bold.func.gii

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsaverage_hemi-L_bold.json

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsaverage_hemi-R_bold.func.gii

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsaverage_hemi-R_bold.json

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsnative_hemi-L_bold.func.gii

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsnative_hemi-L_bold.json

    ├── sub-219_ses-itbs_task-rest_run-2_space-fsnative_hemi-R_bold.func.gii

    └── sub-219_ses-itbs_task-rest_run-2_space-fsnative_hemi-R_bold.json

Dianne Patterson, Ph.D

RII Neuroimaging Staff Scientist

Program Coordinator: Neuroimaging Methods Certificate 

Cognitive Science GIDP affiliate faculty member

d...@arizona.edu

https://profiles.arizona.edu/person/dkp

Office: Speech, Language and Hearing Sciences, Room 314

---

From: Glasser, Matt <glas...@wustl.edu>
Sent: Thursday, September 22, 2022 9:20 AM
To: hcp-...@humanconnectome.org <hcp-...@humanconnectome.org>
Subject: [EXT]Re: [hcp-users] Loading GIFTI to MATLAB with HCP command

 

External Email

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/EBDD6473-E504-4CDE-8FE3-5D688ECCEF1A%40wustl.edu.

[mason wells]

Hi Dianne,

Thanks for the reply. The issue isn't with fMRIPrep, I have other datasets that I have output in bunch of other spaces, including CIFTI. My issue is that with this one particularly dataset I only have the processed outputs and these do not included CIFTI, only fsaverage GIFTI. If I was processing the data myself, you're correct, I would certainly output in CIFTI space. 

Cheers,

Mason

[Coalson, Timothy Scott (S&T-Student)]

Have you compiled the mex file for zstream in the gifti toolbox?  It should be faster than the java that it uses when the mex isn't compiled, though I don't know how much faster overall.

As an alternative, you could use one of the wb_command -cifti-create commands to make a cifti file from them, so that you don't have to read from gifti every time you want to use them.

Tim

---

From: mason wells <mason...@hotmail.co.uk>
Sent: Thursday, September 22, 2022 11:52 AM
To: HCP-Users <hcp-...@humanconnectome.org>
Cc: Patterson, Dianne K - (dkp) <d...@arizona.edu>

Subject: Re: [hcp-users] Loading GIFTI to MATLAB with HCP command

 

WARNING: This message has originated from an External Source. This may be a phishing expedition that can result in unauthorized access to our IT System. Please use proper judgment and caution when opening attachments, clicking links, or responding to this email.

To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/710b752c-0a95-4d02-879a-6745b328d3can%40humanconnectome.org.

[江亚]

I encountered a similar issue as yours, and my solution was to use the built-in gifti toolkit in SPM, rather than other versions.

[Glasser, Matthew]

I wouldn’t recommend putting fMRI data on a 164k mesh.  That is massively oversampling the data.  I would use the CIFTI dense timeseries that we create, which sample the data at their acquisition resolution.

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/954bd4da-daa7-416b-9d14-fd88cc5aeb64n%40humanconnectome.org.