[Eeglablist] ICA for ECG!!

Arnaud Delorme arno at salk.edu
Tue Mar 8 09:01:21 PST 2005


Dear Manjunatha,

>I am currently recording ECG, BP, and respiration signals from human subjects exposed to different particulate matters(eg. Environmental Tobacco). I would like to know how to implement ICA on these recorded signals? Is it feasible to apply ICA on these kinds of data sets. I will analyse these data in time freq. domain parameters.
>
It is hard to believe that the signal from the sensors you are refering 
to can be linearly decomposed into a mixture of sources. You can try to 
use ICA, but it probably will not be very efficient at separating common 
sources to all your sensors. Also you seem to have only 3 sensors, and 
probably more than 3 sources, which is another reason why ICA may not be 
efficient.
If I were you, I would directly correlate spectral decompositions 
between your different sensors (using coherence measures for instance...).

Best

Arno




More information about the eeglablist mailing list