I collected structure and functuonal mri images for multiple sessions (N). Thus, I have N t1 images, 4XN fieldmap images (2 magnitudes & 2 phases), and NXR bold images where R is the number of runs of a session. In this situation, my goal is to align the bold images on their native t1w by using fmriprep.
However, it is hard for me to find the manual for this multisession alignment of bold signal. It will be really grateful if you can let me know the document for the multisession alignment using fmriprer.
You can use the --output-spaces T1w flag. If you need to treat each session individually, as opposed to doing an average T1w across sessions, you can use a --bids-filter-file, and keep in mind that you would need to have separate --fs-subjects-dir for each session if you want to keep each session separate. Otherwise, if you are fine using an average T1w across sessions, you do not have to worry about using a filter file or separate FS subjects directories.
Thanks very much for your prompt response! I have one more question! If I want to register all bold signals of different sessions to the T1w of the first session, which way would be better?
(1) to register bold signals to T1w of each session, respectively, and then register this “registered bold signals” to the first session’s T1w.
(2) just register bold signals of different days directly to the T1w of first session.
Also, can you inform me how I implement the better way by using fmriprep? I appreciate your help so much!!
In that case, I suppose you can just only use the first session’s T1w by specifying that in a BIDS filter file. But I am curious why this is important to you. What is the difference to you of having it registered to a particular T1 vs just being aligned to any T1? Keep in mind that the average T1 will likely be less noisy (unless the sessions are so far apart that the brains are appreciably different across sessions).
Just to confirm, do you mean that if I organize a multi-session BIDS with T1w for each session and just run fmriprep, fmriprep automatically constructs the averaged T1w of multiple T1w images and coregisters the bold signals to this averaged T1w? Also, do you mean that the other processes such as freesurfer or ANTS would be also implemented by using this averaged T1w?
Yes. Actually, it just occurred to me that the average T1w will be aligned to the first T1w image unless you pass in the --longitudinal flag. Though it may not be perfectly aligned to the raw data. Seems like, if you want things to be closest to the first image, you can not use that flag. Read more in this FAQ: Processing pipeline details — fmriprep version documentation.
Yes, FreeSurfer and ANTs are run on the averged T1w.