[Eeglablist] Removing EOG artefact using ICA and converting back to EEG channels

[Marco Congedo]

Dear Aleksandra,

  regardless the number of component retained
the inverse transformation will gibe you back the original number of EEG
channels. Even if you keep just one component, this is true.

 The model

x=A s

with BSS solution

s=A+ x

is inversed by

x'= A A+ x

You see that x' has the same dimension as the original x,
regardless if you set to zero some columns of A (and corresponding rows of
A+) or if you consider just a subset of columns of A (and corresponding
columns of A+).

Marco

-- 
Marco Congedo, PHD
France Télécom, R&D, InoVallée (Grenoble), FRANCE
Tel: +33 4 76764339
http://www.novatecheeg.com/Congedo/MC_Home.html


> Hello,
>
> I would like to remove eye movement artefacts from EEG signals. I saw
> that ICA is very efficient in it. However I would like to work on EEG
> data not on ICA components. I have 64 channels EEG. So when I perform
> ICA, EOG will be one component which I can discard. However if I now
> want to perform inverse transformation I have a problem: 63 ICA
> components and 64 EEG channels. Does anybody have a suggestion how to
> solve the problem? Would it help adding a dummy component of a constant
> value around 0, and then performing inverse ICA?
>
> Thanks a lot in advance!
>
> Aleksandra Vuckovic
>
> _______________________________________________
> eeglablist mailing list eeglablist at sccn.ucsd.edu
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
>