How to BIDS ASL extensions?

Hi ASL/BIDS experts :smile:,

I would need a little bit of help with an asl acquisition with multiple nifti output and need advice on how I should call these extra files. I checked the bids documentation but I’m really not familiar with asl acquisition or its potential extensions and it’s really not clear to me what to do with these niftis.

The first acquisition would be called sub-XX_asl.nii.gz but the others are MOCOs with different comments
017 → “ImageComments”: “Reference volume for motion correction.”
017 → “SeriesDescription”: “MoCoSeries”,
017 → “ImageType”: [“DERIVED”, “PRIMARY”, “ASL”, “ND”, “NORM”, “MOCO”, “MOSAIC”]

018 → “ImageComments”: “Perfusion Weighted Image (PQ2T)”
018 → “SeriesDescription”: “Perfusion_Weighted”
018 → “ImageType”: [“DERIVED”, “PRIMARY”, “ASL”, “ND”, “NORM”, “MOCO”, “SUB”, “TTEST”, “MOSAIC”]

019 → “ImageComments”: “relCBF (PQ2T)”
019 → “SeriesDescription”: “relCBF”
019 → “ImageType”: [“DERIVED”, “PRIMARY”, “ASL”, “ND”, “RELCBF”, “NORM”, “MOCO”, “SUB”, “TTEST”, “MOSAIC”]

Maybe they should not even been converted with bids naming convention and I can let them go.

Thank you for your help.

So I think that the MoCo is nothing ASL specific and is a version of your data where motion correction was done post-acquisition “on the scanner side”.

A possible thing to do with those if you want to include them in your data set is to the rec entity (for reconstruction): sub-XX_rec-moco_asl.nii.gz.

Usually for those the realignement parameters computed for those images are stored in the dicom header and could also be found in the json file depending on the dicom to nifti converter you have been using. So you could add them to the json side-car file in your datasets (though this is not officially supported / specified by BIDS).

In most instances I have seen though, I think people run the realignment themselves on the original time series with their software of choice.

So whether to keep the MoCo or not is really for you to decide.

The other output… I am not entirely sure. Can try to ping ASL experts.

You do not need derived images for subsequent processing. Tools like FSL BASIL will do a much better (though slower) job of processing the raw data, and should generate better quality derived data. For future acquisitions, you can change the settings on the scanner console to not store derived images.

1 Like

All of these derived data can be stored, alternatively to raw ASL data, in sub-XX_asl.nii.gz if needed. MoCoSeries is saved in the same way as raw ASL. PWI and CBF are also saved under the same filename, but their aslcontext.tsv is then set to either deltam or cbf - see the *_aslcontext.tsv section.

But as Chris says - best is to save the raw data and ignore Moco, PWI or CBF as most pipelines can produce these derivatives as well.

Great !!!
Thank you @Chris_Rorden and @jan-petr !!