Hello,<div><br></div><div> neither averaging weights nor averaging source time series is a valid approach,</div><div>and this is true in general, not just for coherence analysis.</div><div>Note that ICA weights and time series have arbitrary scaling,</div>
<div>thus averaging them makes no sense.</div><div>Either you perform coherence analysis for pairs of components,</div><div>or, if you are inetrested in coherence of clusters of components,</div><div>you have to use multivariate coherence measures.</div>
<div><br></div><div>See the following paper for multivariate coherence definitions:</div><div><a href="http://arxiv.org/abs/0711.1455">http://arxiv.org/abs/0711.1455</a> </div><div><br></div><div>A paper on BSS coherence analysis can be found here:</div>
<div><a href="http://hal.archives-ouvertes.fr/index.php?action_todo=search&view_this_doc=hal-00423717&version=1&halsid=jbd4j01nu66tdhdoaoc37feqt6">http://hal.archives-ouvertes.fr/index.php?action_todo=search&view_this_doc=hal-00423717&version=1&halsid=jbd4j01nu66tdhdoaoc37feqt6</a> </div>
<div><br></div><div>A software for computing (group) BSS and pairwise (instantaneous and lagged) coherence as described in the latter paper is available here:</div><div><a href="https://sites.google.com/site/marcocongedo/software/nica">https://sites.google.com/site/marcocongedo/software/nica</a>
</div><div><br></div><div>Hope this helps,</div><div><br></div><div><p class="MsoNormal">_______________________________________________________________</p>
<p class="MsoNormal">Marco CONGEDO, </p>
<p class="MsoNormal"> </p>
<p class="MsoNormal">Research Scientist,</p>
<p class="MsoNormal">Centre National de la Recherche Scientifique (cnrs) and Grenoble
University.</p>
<p class="MsoNormal"> </p>
<p class="MsoNormal"><span lang="EN-US">Team ViBS
(Vision and Brain Signal Processing)</span></p>
<p class="MsoNormal">GIPSA-lab (Grenoble Images Parole Signal Automatique)</p>
<p class="MsoNormal"> </p>
<p class="MsoNormal">11 rue des Mathématiques</p>
<p class="MsoNormal">Domaine universitaire - BP 46 -</p>
<p class="MsoNormal">38402, Grenoble, France.</p>
<p class="MsoNormal"> </p>
<p class="MsoNormal">tel: +33 (0)4 76 82 62 52 </p>
<p class="MsoNormal">fax: +33 (0)4 76 57 47 90</p>
<p class="MsoNormal"> </p>
<p class="MsoNormal"><a href="http://sites.google.com/site/marcocongedo">http://sites.google.com/site/marcocongedo</a></p>
<p class="MsoNormal">_______________________________________________________________</p>
<p class="MsoNormal"> </p></div><div><br></div><div><br><br><div class="gmail_quote">On Wed, Mar 7, 2012 at 3:24 AM, Agatha Lenartowicz <span dir="ltr"><<a href="mailto:alenarto@ucla.edu">alenarto@ucla.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
Dear Susann. The coherence bit seems like you have some feedback on. However note that how you combine ICs is not trivial. Imagine you have 4 occipital components: left, right, dorsal and ventral topography. You want to combine them bc they all show similar alpha responses. One approach is to average the time series. Another - that you propose - is to average the mixing weights and retrieve new activations. In my example an average of the mixing weights would place the topography centrally -- where none of the original had weighted. So my new activations would be weighted heavily by locations that weakly contributed to the uncombined ICs. This is all to say - keep an eye on what your combining is doing to your data. I've struggled with this - have no clear solution - but am favoring very simple averaging where a single best IC is not available. Agatha<br>
<br>
<br>
<br>
Sent from my phone.<br>
<br>
On Mar 4, 2012, at 14:29, Susann Sgorzaly <<a href="mailto:susann.sgorzaly@st.ovgu.de">susann.sgorzaly@st.ovgu.de</a>> wrote:<br>
<br>
> Dear All,<br>
><br>
> I am working on my master thesis on EEG data and would like to do an<br>
> coherence analysis on ICA components. So my question is, has anyone<br>
> experience how to analyse it best?<br>
><br>
> My approach was:<br>
> (1) Run ICA and identify task-related components.<br>
> (2) If there are more than one component: average weights of these<br>
> components and recalculate activation matrix<br>
> (3) Run coherence analysis on this new component with all other ICA<br>
> components<br>
> (4) Perform a cluster analysis on those ICA components which are most<br>
> coherent with the averaged component off step (2) to see if they<br>
> have a similar topography.<br>
><br>
> Is there a better way to perform coherence analysis on ICA components?<br>
><br>
> Thanks<br>
><br>
> Susann<br>
> _______________________________________________<br>
> Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>
> To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu">eeglablist-unsubscribe@sccn.ucsd.edu</a><br>
> For digest mode, send an email with the subject "set digest mime" to <a href="mailto:eeglablist-request@sccn.ucsd.edu">eeglablist-request@sccn.ucsd.edu</a><br>
<br>
_______________________________________________<br>
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu">eeglablist-unsubscribe@sccn.ucsd.edu</a><br>
For digest mode, send an email with the subject "set digest mime" to <a href="mailto:eeglablist-request@sccn.ucsd.edu">eeglablist-request@sccn.ucsd.edu</a><br>
</blockquote></div><br><br clear="all"><div><br></div>-- <br>Marco Congedo<br><a href="http://sites.google.com/site/marcocongedo">http://sites.google.com/site/marcocongedo</a><br>
</div>