[Eeglablist] Analyse spectrale & ICA

[Arnaud Delorme]

>
>
> I have downloaded the EEGLAB toolbox and in fact, I want to do an ICA 
> on a file with the results of the spectral analysis of an EEG signal. 
> I wanted to do it for only one channel and 6 frequency bands, and I 
> tried o make it with a matrix of one row (for one channel) and 
> 6*Number_of_Epochs columns, but EEGLAB says me that there are not 
> enough values...Is it really possible with EEGLAB to do an ICA after a 
> spectral analysis?
>
It is possible, but it is probably much easier to do it from the command 
line using the runica function. First, run the phasecoher function on 
each trial at one frequency of interest. Then, reorganize the output 
data to have an array with channels in the first dimension, time points 
in the second dimension and trials in the last dimension.  Then run ICA 
on the absolte value of the time-frequency array.
Jorn Anemuller in our lab uses uses this technique with a small 
variation: he uses complex Infomax ICA on the complex spectral estimates 
values instead of using standard Infomax on the spectral amplitude (and 
he gets nice results, see J. Anemüller, T. J. Sejnowski, and S. Makeig. 
Complex independent component analysis of frequency-domain 
electroencephalographic data. /Neural Networks/, 16:1311-1323, 2003. [ 
pdf 
<http://sccn.ucsd.edu/%7Ejorn/webpage/bib/pdf/AnemullerSM_NeurNetw_2003_arxiv.pdf> 
| journal <http://dx.doi.org/10.1016/j.neunet.2003.08.003> ] ).

Best

Arno
-- 

*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>
*To think upon*:

    The ultimate most holy form of theory is action.

        /Nikos Kazantzakis/