Can I used preprocessed T1w data as input in ASL prep?

Hi,

I want to used T1w data that was already corrected for inhomogeneites (using this: GitHub - CoBrALab/iterativeN3), and I also have my own population average, created with this: GitHub - CoBrALab/optimized_antsMultivariateTemplateConstruction: A re-implementation of antsMultivariateTemplateConstruction2.sh using optimized image pyramid scale-space and qbatch support

Can I use those files and skip ASLprep Structural Preprocessing? I wish to only get the CBF in native space and then register them to my population average template by concatenating ASL to T1w transforms plus T1w-Population average transforms (that I got when creating my template).

Best regards,
-Cindy

ASLPrep can ingress preprocessed anatomical data from fMRIPrep + Freesurfer, and it should be able to derivatives from other BIDS apps, but I haven’t tested with any. If iterativeN3 is a BIDS app that produces a BIDS derivative dataset, then I’m happy to look into ingressing preprocessed anatomical data from it. I can’t guarantee that a given BIDS App will produce all of the anatomical derivatives that ASLPrep needs though.

If iterativeN3 isn’t a BIDS App, and is just producing a modified version of the T1w, you could try copying it into your raw BIDS dataset with a unique name, like including a rec-corr entity in the filename to distinguish it from the actual raw T1w. Then you would create a BIDS filter file (see here) and pass it along with --bids-filter-file. However, you should be aware that the preprocessed anatomical image may not interact well with the anatomical processing ASLPrep does.

If your processed T1w is still in the same space as your original T1w, then you could probably just run ASLPrep with your original T1w and concatenate the transforms across pipelines.

You can include your population average as an output space if you want. There’s some information about how to do this in fMRIPrep’s documentation.

Thank you Taylor,

If the structural preprocessing of fmriprep/aslprep creates its own population average, why does a standard normalization to an MNI space occur? Or do you know if it also produces measures in this population average in addition to the MNI space?

If not, what is the purpose of the population average?

This is just an existential question, I will use ASL without trying to enter my own preprocessed images for now,

Thank you in advance,´

-Cindy

Neither fMRIPrep nor ASLPrep creates a population average. What I meant was that you can pretend an existing population average (i.e., the one you got from your other workflow) is a standard-space template and select that as a target with --output-spaces. The fMRIPrep documentation I linked to covers how to use a custom template (which your population average would be in this scenario).

I don’t work with population averages. I assume that they serve as custom templates where you can be more confident that no subjects are getting warped too extremely to fit a template from another population.