[Eeglablist] Use of ICA for removing cardiac artefacts

Pandelis Perakakis peraka at ugr.es
Sun Mar 31 06:04:15 PDT 2019

Dear group

I want to use ICA to identify and remove components associated with cardiac
activity. I am using a BrainVision amplifier with 64 EEG electrodes and
also record ECG with the BrainAmp ExG bipolar amplifier.

In most subjects ICA does not isolate a cardiac component. However, when I
average epochs time-locked to the ECG's R wave, a clear ECG pattern emerges
in occipital electrodes which are the ones I would expect to better pick up
the ECG activity. So my data tells me there is still some ECG activity
spread to the EEG electrodes although ICA is not sensitive enough to detect
it and isolate it as an individual component.

I also tried to run ICA including the ECG channel. In that case, ICA
identifies a component associated to ECG activity, but removing this
component merely changes activity in EEG channels and averaging around the
R wave still produces a clear ECG pattern.

*Is this second procedure correct? Does it make sense to run ICA including
channels recording peripheral variables like ECG?*

*Do you recommend other methods to detect and remove ECG activity?* Most
recent articles mention using ICA but with no sufficient detail and also
when you see the figures of the heartbeat evoked potential after supposedly
having removed the cardiac artefact, the ECG pattern is still clearly

Any ideas, recommendations would be appreciated!


*–What you do is not important. It's why you do it that matters*

Pandelis Perakakis, PhD
Associate Professor
Department of Psychology
Universidad Loyola Andalucía

Research Associate
Mind, Brain & Behaviour Research Centre (CIMCYC)
University of Granada

e-mail: pperakakis at uloyola.es | peraka at ugr.es | perakakis at gmail.com
Tel. +34 955 641 600 (Ext. 629)
url: pandelisperakakis.wordpress.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20190331/5991e308/attachment.html>

More information about the eeglablist mailing list