[Eeglablist] Artifacts

[Arnaud Delorme]

>
> Dear all,
> I 've just started to use ICA analysis to remove artifacts from the 
> EEG traces.
> I have two questions:
> 1- Is it correct to analyze only few (8-15) seconds and then remove 
> artifactual components from whole trace (700-1000 sec)?

It is acceptable since ICA extract spatially stationary components (i.e. 
we assume that the component are spatially stable throughout the whole 
data). In practice, unless you cannot do otherwise (i.e. memory 
constraints...), because artifacts are not totally stable it is better 
to use the whole data so ICA will find the most spatially stable artifacts.

> 2- Can I use ICA analysis repetitevely on the same trace? (I saved 
> filtered EEG, but sometimes I had to filtered the filtered EEG to 
> obtain a good trace) 

Do you mean that you want to remove some artifactual component, then 
reapply ICA? When you remove ICA components, the data loose some 
dimensions (try "rank(EEG.data);" under Matlab) and ICA cannot find any 
more as many components as electrodes. In practice, ICA will not 
converge (weight changes will diverge). For more info see 
http://sccn.ucsd.edu/eeglab/maintut/ICA_decomposition.html (note of 
paragraph I.9.5)

Best

Arno

> Thank you.
> Elisa Visani 



-- 

*Arnaud Delorme, Ph.D.*
Computational Neurobiology Lab, Salk Institute
10010 North Torrey Pines Road
La Jolla, CA 92037 USA

*Tel* : /(+1)-858-458-1927 ext 15/
*Fax* : /(+1)-858-458-1847/
*Web page *: www.sccn.ucsd.edu/~arno <http://www.sccn.ucsd.edu/%7Earno>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20031021/16470b6f/attachment.htm