Global signal correction (gscontrol_raw and gscontrol_mmix)

[kuanghan]

Hi,

I am supporting the preprocessing of multi-echo fMRI data here at Mount Sinai (used to work with Prantik here :-)) We have preprocessed quite a lot of data using meica.py from this repository, and now we are having questions about what exactly was done by meica.py and tedana.py in terms of global signal correction. We saw that there are two functions in tedana.py that seem to do something with global signal: gscontrol_raw and gscontrol_mmix. We know that gscontrol_mmix is minimal image regression (with confirmation from Prantik), but we are wondering is gscontrol_raw is GSR or not?

Also, we see that gscontrol_mmix is done as the very last step in tedana.py so it shouldn't affect the ICA convergence, but gscontrol_raw happens before ICA. Does turning off gscontrol_raw affect the convergence of ICA?

I also just very recently found out about the new tedana package that you are developing. Would you recommend calling the new tedana from meica.py? Or should we change our workflow completely and not use meica.py at all?

Sorry I'm not super familiar with functional data myself, but I'm building an image processing pipeline to integrate processing of multiple modalities here. So any help/insights you can provide will be greatly appreciated.

Thank you very much!

Kuang