[Eeglablist] AMICA to clean eye blinks on 8-channel EEG

[Jason Palmer]

Hi Ke,

8 channels is probably too few for ICA to effectively separate the EEG
sources. When the number of sources (eye blink, Sacramento, scalp muscle,
heartbeat, brain sources, artifacts, line noise, ....) is greater than the
number of channels, ICA will fail to separate  the sources, e.g. the eye
blinks.

When you have too few channels to isolate the eyeblink in a single source
(you can examine the IC activations), then regression techniques, like
Gratton, may be the best option.

Best,
Jason

On Fri, Jul 1, 2022, 2:00 PM liao ke <kekeliao at gmail.com> wrote:

> Dear All,
>
> I have a problem to use AMICA to clean eye blinks on 8-channel EEG and want
> to get your help.
>
> The EEG is continuous data ~10 min. There are many eyeblinks on frontal
> channels like Fp1 & Fpz. AMICA can separate eye blink IC component and Fp1
> & Fpz are well cleaned after removing the eye IC. But it creates many
> eyeblink-like artifacts on other channels like Oz & Pz at the time when the
> original blinks exist. It looks like when removing the blinks, those
> channels subtract the blinks and therefore create blink-like negative
> peaks.
>
> What's the best way to clean the eyeblinks on this continuous EEG data?
> Since we don't want to get any data segment removed, we don't use
> Clean_rawdata() and ASR to remove blinks.
>
> Is there other tool to clean blinks for continuous EEG ? My
> colleague suggests to use Gratton algorithm but it looks only works for
> epoch data.
>
> Thank you very much for your help!
>
> Best,
> Ke
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
>