[Eeglablist] ICA and evenly spread noise?

Maarten Mennes Maarten.Mennes at med.kuleuven.ac.be
Tue Apr 19 07:35:57 PDT 2005

hi all,

have been using ICA a while now and noticed that for removing unwanted 
EEG-components it seems to work better when the signals of these 
components are gradually spread over electrodes. For removing for 
instance blinks (which are clearly gradually spread over electrodes) or 
muscle-artifacts that are limited to only one electrode, ICA does the 
job very well.
However, I encountered a number of subjects in my study which had noise 
that was evenly spread across electrodes. This is most likely due to a 
non-optimal head position resulting in non-optimal reference signal from 
our linked mastoid reference electrodes. Running ICA for these subjects 
I'm unable to correct for all of this noise.
The same is true for alphabursts we tend to see in some of our subjects 
across most electrodes. 

Is ICA the best procedure to correct for this kind of "noise" or are 
other procedures better suited and which would that be?


More information about the eeglablist mailing list