[Eeglablist] How can I reduce EKG artifact?

[Fuh-Cherng Jeng]

Dear EEGLAB experts,

I have a question about EKG reduction.

I am a Hearing Science person and I deliver sound to the ears to elicit a 
response. For the type of responses I am interested in, the recording 
montage is usually a "bipolar" recording setup: Cz ("positive" recording 
electrode), C7 ("negative" recording electrode), and mid forehead (ground). 
During experiments, participants were asked to rest in a comfortable 
recliner with their eyes closed. In many occasions, I observed EKG artifact 
in the continuous EEG recording. I think it is due to the fact that the 
"negative" electrode is in proximity to neck vessels and therefore picked 
up the heart beating EKG artifact. With an attempt to reduce the EKG 
artifact, I used two "monopolar" recording channels instead of one 
"bipolar" channel.  That is, EEG Channel #1 was Cz and EEG Channel #2 was 
C7.  Both EEG channels were referenced to the mid forehead ground. In order 
to obtain a "good" template of EKG signals, I placed an additional 
monopolar recording electrode on the left fifth intercostal space at 
midclavicular line. This is my EKG channel (referenced to the same mid 
forehead ground).

Semlitsch et al. (1986) published an algorithm to reduce eyeblink artifact. 
Although my recordings contained no eyeblink artifact, I had tried to adopt 
this algorithm to reduce EKG artifact without success. I think it was due 
to the fact that the EKG artifact observed in EEG channels were not only 
attenuated in amplitude but also exhibited a temporal delay (between the 
EEG and EKG recording channels). This temporal delay makes the EKG 
reduction much more complicated. That is, an EKG waveform contains multiple 
peaks (e.g, QRS). If I aligned the peaks of R waves in the EEG and EKG 
channels, the S wave peaks won't be aligned, and vice versa.  My question 
is: What shall I do to reduce the EKG artifact?  Is there any existing 
EEGLAB/MATLAB function or source code I can use?


Any comments or inputs will be greatly appreciated.


Sincerely,
Fuh