Multiband image processing

[Linjing95]

Hi,

I am wondering if HALFpipe can be used to preprocess multiband imaging data? For example, I don't see any options to import a single-band reference image.

Thank you!
Linjing

[HippocampusGirl]

Hi @Linjing95,

HALFpipe can be used for multi-band data. Since preprocessing is done with fMRIPrep under the hood, HALFpipe supports all data types that fMRIPrep can process.

The option to import a single-band reference image is currently not available in the user interface. However, you can manually edit the spec.json file in your working directory to provide the information. To do that, simply add an entry to the files array inside the file.

Here is an example for how that entry might look:

        {
            "datatype": "func",
            "path": "/fast/work/groups/ag_walter/leadata/hcp/raw/{sub}/unprocessed/3T/tfMRI_{task}_{ses}/{sub}_3T_tfMRI_{task}_{ses}_SBRef.nii.gz",
            "extension": ".nii.gz",
            "suffix": "sbref"
        },

After editing, restart HALFpipe, select your working directory and choose Run without modification.

[Linjing95]

Hi @Linjing95,

HALFpipe can be used for multi-band data. Since preprocessing is done with fMRIPrep under the hood, HALFpipe supports all data types that fMRIPrep can process.

The option to import a single-band reference image is currently not available in the user interface. However, you can manually edit the spec.json file in your working directory to provide the information. To do that, simply add an entry to the files array inside the file.

Here is an example for how that entry might look:

        {
            "datatype": "func",
            "path": "/fast/work/groups/ag_walter/leadata/hcp/raw/{sub}/unprocessed/3T/tfMRI_{task}_{ses}/{sub}_3T_tfMRI_{task}_{ses}_SBRef.nii.gz",
            "extension": ".nii.gz",
            "suffix": "sbref"
        },

After editing, restart HALFpipe, select your working directory and choose Run without modification.

Hi @HippocampusGirl,

It worked! Are there any other settings I should change aside from adding an SBRef? For instance:

1. Would you still recommend doing slice-timing correction?
2. Should I choose to remove any confounds (e.g., ICA-AROMA, GS, WM, CSF, motion) from the data (my data is task-based fMRI)?

Thank you! I appreciate it!

[HippocampusGirl]

Hi @Linjing95,

If you are combining multiple data sets, then I would use consistent slice-timing options between them. If not, then it's not something that is usually done for fast multi-band data, so I would disable it.

We recommend some settings in the manual at https://bit.ly/tbfmrimanual.

• There seem to be some anecdotal reports that ICA-AROMA does not perform as well with multi-band data, so if you choose to use that I would also add another identical feature without ICA-AROMA for comparison.
• GS, WM and CSF are not commonly used for task-based data.
• Motion regressors are generally not needed when doing ICA-AROMA (since it also uses motion data to determine its regressors). It can be a great option when ICA-AROMA is turned off though!

[Linjing95]

Hi @Linjing95,

If you are combining multiple data sets, then I would use consistent slice-timing options between them. If not, then it's not something that is usually done for fast multi-band data, so I would disable it.

We recommend some settings in the manual at https://bit.ly/tbfmrimanual.

• There seem to be some anecdotal reports that ICA-AROMA does not perform as well with multi-band data, so if you choose to use that I would also add another identical feature without ICA-AROMA for comparison.
• GS, WM and CSF are not commonly used for task-based data.
• Motion regressors are generally not needed when doing ICA-AROMA (since it also uses motion data to determine its regressors). It can be a great option when ICA-AROMA is turned off though!

This is really helpful! Thanks so much!