Low-frequency drifts and ICA

Hi, I noted that in many applications people use to preliminarly remove low-frequency drifts before applying ICA. I also noted that ICA is not able to pick such low-frequency drifts, but I would like a simple explanation or a reference to motivate this, please.

I am not an ICA expert. I think removing very-low-freuqency drift (e.g., < 0.01 Hz) might be helpful for PCA before ICA. Very-low-freuqency drift might contribute so much, even not the largest, and hence, might survive the PCA, and hence makes subsequent ICA complex. But a very strong (might not be true) hypothesis for doing this is very-low-frequency drift is noise.

Error | Forum of resting-state fMRI

Error

The website encountered an unexpected error. Please try again later.