search

nipreps / fmripost-rapidtide

A BIDS App for running rapidtide on a preprocessed fMRI dataset.
https://fmripost-rapidtide.readthedocs.io
Apache License 2.0
1 stars 0 forks source link

Leverage multi-echo signal? #5

Open tsalo opened 3 weeks ago

tsalo commented 3 weeks ago

I was thinking that multi-echo data could be leveraged in fMRIPost-rapidtide by ingressing outputs from fMRIPost-tedana (once it's up and running). Since the sLFO signal we're trying to isolate and remove is primarily BOLD-based, maybe rapidtide could use tedana's ICA components and their classifications as BOLD or non-BOLD in some way?

Related to https://github.com/ME-ICA/tedana/issues/1071.

bbfrederick commented 2 weeks ago

I don't know enough about what tedana puts out to have a sensible opinion about this, but I'm up for looking at it.

tsalo commented 2 weeks ago

The entire sLFO signal should be BOLD, right? Tedana runs ICA on multi-echo data and classifies the components as either BOLD-based or non-BOLD based on the signal decay over echo times. Maybe fMRIPost-rapidtide could use the non-BOLD components from tedana in its initial denoising step, in addition to the motion parameters? My thinking is that limiting the data going into the regressor refinement step to BOLD data (instead of a mix of BOLD signal, BOLD noise, and non-BOLD noise) might help identify the BOLD noise, if that makes sense.