[Eeglablist] ICs with identical topographies

Ronald Phlypo Ronald.Phlypo at ugent.be
Tue Sep 13 00:16:49 PDT 2011


Dear Max,

just my small input concerning your question "How could frequencies that 
do not exist in the input be created by the ICA?"

Take a 2 channel, 2 source artificial scenario, wherein source 2 
contains a positive component that is negative in source 1, say with 
respective weights +1 and -1 in observation 1. It can easily be seen 
that in the observation 1 that component (frequency component if you 
want -- I use component to not interfere with what is called a source, a 
source here can consist out of multiple components) will not be visible, 
whereas if one excludes either one of the sources in the reconstruction 
the component does become visible (either positively or negatively). 
These "components" are not always physiologically realistic, but they do 
form a valid "mathematical" decomposition.

Now, you might wonder why this is more often the case with frequency 
components than it is with temporal components? This is because the 
temporal waveform of such components resemble gaussian noise (temporal 
structure not taken into account) and will thus be invisible to the ICA 
algorithm. This can be explained as the source and the source + gaussian 
component having the same behaviour when submitted to an ICA algorithm.

There exist some works that do, e.g., posterior wavelet aided cleaning 
of the sources (e.g., 
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5713804&tag=1 
<http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5713804&tag=1>), 
or some works that try to get a best trade-off between temporal and 
spectral independence (see, e.g., http://si.utia.cas.cz/pubs/TNN08.pdf). 
Another hybrid approach used in ECG data may be found in 
http://personales.gan.upv.es/jjrieta/pub_whole.htm (see Castells, F., 
Rieta, J. J., Millet, J. and Zarzoso, V. "Spatiotemporal Blind Source 
Separation Approach to Atrial Activity Estimation in Atrial 
Tachyarrhythmias", IEEE Transactions on Biomedical Engineering, vol. 52, 
no. 2, pp. 258-267, Feb. 2005. ).

Good luck,

Ronald

Le 07/09/2011 19:26, Maximilien Chaumon a écrit :
> Hi eeglabbers,
>
> I am still having issues with ICA returning extremely similar (but not 
> identical) topographies. The gui (although I run a version updated a 
> few days ago) does not popup any suggestion to reduce the rank. I only 
> get this warning (Warning: fixing rank computation inconsistency (68 
> vs 69) most likely because running under Linux 64-bit MatlabAttempting 
> to convert data matrix to double precision for more accurate ICA 
> results.) I still get 69 components in the end.
>
> Here's some more info:
> the rank of the data is 68. The svd drops abruptly close to zero at 
> the last value. I have 69 electrodes (64 heancap + 3EOG+2mastoids). I 
> guess there's a gel bridge somewhere. Although correlations between 
> all electrodes don't reach .95.
> When I let the ICA run with default options, I still get these two 
> components 
> <http://oszilla.hgs.hu-berlin.de/public/2P3like_components.png> 
> (always 'P3 like' components, this was reproduced in other subjects). 
> Their frequency profiles are too good to be true, with low noise and a 
> peak at 10Hz, another one around 20Hz, see the figure. I would leave 
> them alone if they were not spoiling all my data: As I remove the 
> components, when I click this 'singles' button, which shows me the 
> trial by trial time course. I get what is shown on the right of the 
> figure 
> <http://oszilla.hgs.hu-berlin.de/public/2P3like_components.png>. High 
> frequency bursts appearing every now and then, usually at times where 
> there is high variability across channels.
> Removing both components resolves the issue, but I loose a rather 
> important part of the data.
> Here is the spectopo at 60Hz 
> <http://oszilla.hgs.hu-berlin.de/public/2P3spectopo.png>. There is a 
> strong artefact here. The two components show a high power at all 
> frequencies.
> How could frequencies that do not exist in the input be created by the 
> ICA? I filter my data, before ICA below 45Hz.
>
> I tried running fastica, asking for 68 components, no such artifact 
> appears but the decomposition looks much less nice, at least with the 
> parameters I've used.
>
> So in the end, my question is:
> How can I run an ICA without trouble if the rank of the data is not 
> equal to the number of electrodes? How can I identify potentially gel 
> bridged electrodes?
>
> Many thanks,
> Max
>
>
> 2011/8/27 Arnaud Delorme <arno at ucsd.edu <mailto:arno at ucsd.edu>>
>
>     Regarding the matrix rank, we have recently realized that the rank
>     function (and other rank function we had programmed) are not fully
>     reliable which is probably with Max observes the component he
>     observes. The runica function should automatically decrease the
>     rank of the input data matrix. However, sometimes it does not use
>     the correct rank. We have modified the runica GUI so that if the
>     matrix is not full rank, it now pops up a new window suggesting to
>     the user a rank reduced value. This value may be adjusted by the
>     user based on prior knowledge. For instance, if you have removed 5
>     components from the data, you would reduce the rank by 5 (and
>     overwrite the rank that is automatically detected if it is not
>     correct).
>
>     Arno
>
>     On Aug 23, 2011, at 10:11 PM, John J.B. Allen wrote:
>
>>     Max
>>
>>     I have observed that when the data are not full rank.   You can
>>     test the rank of your data by reshaping your epoched data to a 2D
>>     matrix, and running the rank command, like this:
>>
>>     rank(reshape(EEG.data,EEG.nbchan,EEG.trials*EEG.pnts))
>>
>>     When I did this, your rank is 63, but you have 69 channels,
>>     indicating that some channels are linearly dependent on others.
>>      I think this is the source of your problem, and if you remove
>>     those channels before running ICA, you should no longer see this
>>     issue.
>>
>>     Best
>>
>>     John
>>
>>
>>
>>
>>     On Tue, Aug 23, 2011 at 07:24, Maximilien Chaumon
>>     <maximilien.chaumon at gmail.com
>>     <mailto:maximilien.chaumon at gmail.com>> wrote:
>>
>>         Hi eeglabbers,
>>
>>         I sometimes get ICs with extremely similar topographies and
>>         time courses, like on this slide
>>         <http://oszilla.hgs.hu-berlin.de/public/Similar_ICs.PNG>.
>>         I know that ICA returns independent components.
>>         Does that not mean that they should not look the same?
>>         I know the components are independent in a statistical sense,
>>         which is not the same as uncorrelated, but still. I'm a bit
>>         surprised. What do these two components mean if they cancel
>>         one another? well, do they?
>>
>>         Sorry if my question is naive, but what is happening?
>>
>>         The data is here
>>         <http://oszilla.hgs.hu-berlin.de/public/Similar_ICs.mat>.
>>
>>         Best,
>>         Max
>>
>>         _______________________________________________
>>         Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>>         To unsubscribe, send an empty email to
>>         eeglablist-unsubscribe at sccn.ucsd.edu
>>         <mailto: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
>>         <mailto:eeglablist-request 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
>>     <mailto: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
>>     <mailto:eeglablist-request 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
> For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu

-- 

*Ronald Phlypo*
/Vision and Brain Signal Processing (ViBS) Research Group/
/Universités de Grenoble / CNRS UMR 5216/
961 Rue de la Houille Blanche
BP 46
38402 Saint Martin d'Hères
FRANCE

tel: +33 (0)4 76 82 62 47
fax: +33 (0)4 76 57 47 90

http://www.gipsa-lab.inpg.fr/~ronald.phlypo/ 
<http://www.gipsa-lab.inpg.fr/%7Eronald.phlypo/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20110913/4b7c2310/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Ronald_Phlypo.vcf
Type: text/x-vcard
Size: 679 bytes
Desc: not available
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20110913/4b7c2310/attachment.vcf>


More information about the eeglablist mailing list