Hello, i was wondering if it makes sense to regress out the “noise component” time series from an FSL melodic spatial ICA decomposition (ie, the file created by melodic called “melodic_mix”) using 3dTproject. this file contains a time series associated to each spatial ICA component. i would like to regress out some of the melodic ICA components.
typically, i use the fsl command (fsl_regfilt) for subtracting components from the BOLD signal, but i would also like to remove the global signal and do bandpass filtering in the same step, using 3dTproject (i learned recently that it was not mathematically correct to regress out these components sequentially). i’m aware that it’s not good practice to combine different software packages in a single analysis pipeline, but fsl_regfilt doesn’t allow for bandpass filtering in a single step, whereas 3dTproject does.
if anyone here has experience with fsl’s melodic, do you think what i am proposing sounds correct? thank you,
quick reply - this doesn’t seem to work so well when i include a large number (40+) of noise component time series recovered from melodic. (it regresses out a lot of useful stimulus and resting state activity).
however, if i cluster my noise component time series into ~5 clusters, and average within clusters to produce 5 separate noise regressors, it works quite well to regress out noise while retaining stimulus/resting state activity.
does anyone know why including so many regressors (40+) even though they are all noise (cardiac, motion, scanner drift, etc) removes much of my useful signal as well? it shouldn’t be a dof issue, i have 800 time points per scan.
While I don’t use that software, if regressing out all of the components
results in removing a lot of BOLD signal, then those components must
contain BOLD. Maybe when you average them, some BOLD signal
gets corrupted enough so that its removal is not detrimental.
Of course global signal regression will also remove some BOLD signal.