I want to preprocess Generic Object Decoding dataset which has individual T2w images for each session. I only want to use the session-wise T2w images for preprocessing the functional data. Is it possible to specify the same in fmriprep?
Hi @Satwick,
Yes, by using --bids-filter-files, please read this part of the documentation: FAQ - Frequently Asked Questions — fmriprep version documentation
Best,
Steven
T2w images are used for refining the pial surface in the anatomical workflow. How do you expect them to be used for preprocessing functional data?
I am not concerned with the exact dynamics involved. I just want to make sure whenever fmriprep uses T2w image for a particular functional run it should only use the T2w from the particular session, as implemented by the original paper using SPM5. Dataset includes T1w images as well for each subject so coregistration shall be performed using the same, I guess. I read somewhere T2w is used only in addition to T1w images, hence I just want to make sure that the T2w images that is used shall be from the same session.
I can select particular sessionwise T2w images but given there are 4 sessions of functional runs and one T2w image for each session, I cannot make sure that the sessionwise functional runs will use T2w from the particular session. Even when I filter the sessions, I still have to select all 4 T2w images, but fmriprep probably uses all available T2w images so I cannot selectively map the sessionwise T2w images to the runs within the runs. Only way is to preprocess each session like I intend to mean running each session separately in fmriprep. Is there any other way to do the same without doing the preprocessing for each session separately is what I would like to know.
Note at present. Right now, using a collection of bids-filter-files is the way to do this.
Per-session analysis is on the roadmap.
So, if I do it like this, I will be selecting all 4 T2w files, and form what I understand, fmriprep will use all for for each functional run irrespective of the session number. And if I want to work around this issue, I need to preprocess each session separately.