I am new to using fMRIPrep to preprocess BOLD resting state data, and I just had a couple of questions about the outputs. I would like to use ICA-AROMA in our preprocessing and output our data into the MNI Infant template space (as provided in TemplateFlow), but I am a little confused as to which output file from fMRIPrep I should take for further analysis.
From what I’ve read, it appears that fMRIPrep performs ICA-AROMA in MNI152NLin6Asym space, even if I specify that I want the data to be output in the MNIInfant template in the –output-spaces option. Is that correct? If so, I am unsure whether the final output from fMRIPrep after ICA-AROMA has been applied is the “*MNI152NLinAsym_desc-smoothAROMAnonaggr_bold.nii.gz” file (in which case, the final, preprocessed data are in MNI adult space) or whether fMRIPrep will take into account that I would like the data to be output in MNI infant space, and does so in the “*MNIInfant_desc-preproc_bold.nii.gz” file (or am I mistaken and this file does NOT have ICA-AROMA applied?). Could anyone please point me to which file I should use, or whether I am misunderstanding anything in the fMRIPrep workflow?
If it is not possible at the moment to have fMRIPrep both apply ICA-AROMA and subsequently output the data into the MNI Infant template space, would it be okay to just warp the denoised data from ICA-AROMA from the MNI adult to infant space? Or should I do something different?
Thank you very much for your time and help, and please let me know if I can clarify anything.
From what I’ve read, it appears that fMRIPrep performs ICA-AROMA in MNI152NLin6Asym space, even if I specify that I want the data to be output in the MNIInfant template in the –output-spaces option. Is that correct?
Yes, when using ICA-AROMA, the output space will be MNI152NLin6Asym , which is equivalent to the FSL MNI152 space. Unfortunately, this is the only output space by ICA-AROMA.
I am unsure whether the final output from fMRIPrep after ICA-AROMA has been applied is the “*MNI152NLinAsym_desc-smoothAROMAnonaggr_bold.nii.gz” file (in which case, the final, preprocessed data are in MNI adult space) or whether fMRIPrep will take into account that I would like the data to be output in MNI infant space, and does so in the “*MNIInfant_desc-preproc_bold.nii.gz” file (or am I mistaken and this file does NOT have ICA-AROMA applied?).
The ICA-AROMA [non-aggressively] denoised data is the *MNI152NLinAsym_desc-smoothAROMAnonaggr_bold.nii.gz file, you will be able to tell because it has smoothAROMAnonaggr in the file name. The corresponding file (*MNIInfant_desc-preproc_bold.nii.gz), has been minimially preprocessed, but has not been ICA-AROMA denoised.
Given ICA-AROMA’s limitation to MNI152NLin6Asym space, you unfortunately won’t be able to perform ICA-AROMA in MNIInfant space.
That being said, for each functional bold acquisition, fmriprep provides a corresponding *desc-confounds_timeseries.tsv file, which contains columns of confounds which can be incorporated into your design matrix or denoising procedure. You’ll only want to select a subset of the columns however.