[Eeglablist] Strange mirror-image components

Arnaud Delorme arno at ucsd.edu
Thu Jun 9 21:01:08 PDT 2011


Dear Hamish,

you can filp the polarity of ICA components so there is no "miror" components as you call them. You can flip the polarity of a component scalp map as long as you also flip the polarity of the corresponding ICA component activity. For instance, in Matlab notation, after loading a continuous EEGLAB dataset and computing ICA (assuming you used all channels to compute ICA), the data is equal to

EEG.data = EEG.icawinv*EEG.icaact;

you can plot the scalp map of component 1 using

figure; topoplot(EEG.icawinv(1,:), EEG.chanlocs);

if you change the polarity of component 1 scalp map and component 1 activity

EEG.icawinv(:,1) = -EEG.icawinv(:,1);
EEG.icaact(1,:) = -EEG.icaact(1,:);

then the data is still equal to 

EEG.data = EEG.icawinv*EEG.icaact;

So changing both the polarity of the component scalp map and component activity does not have any effect. Note that it is clearly visible in your plots that for the components that have opposite scalp map polarities, they also have ERPs with opposite polarities.

Arno

On May 19, 2011, at 11:57 PM, Hamish INNES-BROWN wrote:

> Hello all, I have just come across some strange ICA components, and I’m wondering if anyone can help explain them. The first four components in more than one of the current batch of subjects are generally these weird mirror images of each other. Below are the first four component maps, and their ERPs….  They all have pretty strong peaks at about 10Hz as well.
>  
> What can it all mean? Hopefully not a tear in the universe (sorry).
>  
>  
> <image001.png>
>  
>  
> <image002.png>
>  
> Hamish Innes-Brown
> Music & Pitch Project
> Bionic Ear Institute
> t: +61 3 9667 7529
> f: +61 3 9667 7518
> e: hinnes-brown at bionicear.org
> w: http://www.bionicear.org/profiles/HamishInnes-Brown.html
> w2: http://musicalbionics.wordpress.com/
>  
> 2nd International Medical Bionics Conference, Phillip Island, 20-23 Nov 2011
> http://www.iceaustralia.com/medicalbionics2011/
>  
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20110609/959edabf/attachment.html>


More information about the eeglablist mailing list