Hi FSL experts,
What I'm trying to do:
use python to make a .tsv with all confounds of interest in it before I add it to my regression model.
My Question:
I have a question about how fsl_regfilt implements its non-aggressive partial regression. Namely, How would I replicate it where I could derive the estimates from the non-aggressive mode of fsl_regfilt.
What I've tried:
Linear regression
y_hat = m*x + b
so in order to get y_resid, I should be able to do:
y_resid = y - y_hat for each timepoint in the fmri sequence. I do This process for each noise component found in ICA_AROMA.
This did not replicate the result from fsl_regfilt when using output from ICA_AROMA. I'm wondering if there is a basic mathematical error I am committing before I start debugging my implementation.
Thanks!
James
|