Data cleaning for microstate analysis using ICA

Hi Mike,

In the microstate analysis, papers usually do artifact cleaning using ICA on a high-pass filtered signal, e.g. after applying an 1-Hz high-pass filter to raw resting-state EEG. Then, they apply a narrower band-pass filter, e.g. 2-20Hz, to the clean signal for microstate analysis. I could not see a reason why they didn’t do artifact cleaning on the 2-20Hz signal instead. I ask this because we could reduce brain information by removing artifact ICA components that might not be a problem in the 2-20Hz-band-pass signal. I would appreciate if you could please elucidate me. Thank you in advance :).


True – if you plan on analyzing only only 2-20 Hz activity, there isn’t a good reason to include >20 Hz activity in the ICA.