I try to remove EOG artifacts from my EEG signals (I am using a EPOC+ for recording and MNE Python for processing).

Therefore I've recorded test data with blinks every 5 seconds. The blinks can be seen clearly in the EEG data. The data have been recorded with 128Hz, bandpass filtered (1-30Hz) and resampled (64Hz).

Raw EEG data

After running the ICA ('fastica') the blinks clearly show up in 3 source signals (IC #0, #1 and #2). I have the same behavior for different amounts of ICs.

enter image description here

ICA topo map

I think the blinks are to strong and that's why they spread in more than one source signal.

I wonder if there is a way to let the blinks only show up in 1 source signal?

  • $\begingroup$ This look like you could further low-pass filter this and then apply a threshold to the absolute value? (I have no idea about EEG, though) $\endgroup$ – Marcus Müller Jan 16 '17 at 15:56
  • $\begingroup$ What would be the effect of further low-pass filtering? $\endgroup$ – ppasler Jan 16 '17 at 18:07
  • $\begingroup$ I'd assume it'd practically eradicate all the fast, small-amplitude noise, and your peaks would still stand out, though being rounder, and you might be able to find a threshold between these and your "normal" signal. By the way, adding a fairly low-cutoff-frequency high-pass might also help with removing any bias, but I'd do that in a second step. $\endgroup$ – Marcus Müller Jan 16 '17 at 19:18
  • $\begingroup$ @MarcusMüller As I am interested in the frequency range from 1-30Hz, I cannot cut off more frequencies to remove noise. $\endgroup$ – ppasler Feb 20 '17 at 8:18
  • $\begingroup$ You'd still keep the unfiltered data and use the filtered only for detecting the position of blinks. But then again, I'm really no expert in EEG $\endgroup$ – Marcus Müller Feb 20 '17 at 8:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.