[Eeglablist] Questions about AMICA

Makoto Miyakoshi mmiyakoshi at ucsd.edu
Thu Jan 17 17:47:52 PST 2019

Dear Thomas,

   1. If you know how many channels are rejected, using that number to
   determine the rank is probably the safest (as my code you pasted takes the
   number from the result from clean_rawdata() log)
   2. Average referencing WITHOUT the initial reference (which is just flat
   line at 0) will reduce the data rank by one. However, this way of average
   referencing is NOT the gold standard. In fact, when you perform average
   referencing, adding the original reference channel back is a correct
   3. 'Is this approach correct?' I would not say this is THE correct
   approach. Particularly, the reference issue is a longstanding problem in
   EEG research, and I know there is a good alternative (REST by Yao et al.)


On Wed, Jan 16, 2019 at 5:30 PM Thomas mas <thomasmas at gmail.com> wrote:

> Dear all,
> I would like to avoid the issues associated with data rank when running
> I know that interpolating bad channels and average re-referencing are
> common operations that could make the data rank deficient. I also know that
> the Matlab 'rank' function is not always reliable.
> I am asking whether the following approach is correct.
> After interpolating bad channels and re-referencing to the average, I
> estimate data rank by using the following snippet of code:
> % Estimating data rank
> dataRank = EEG.nbchan;
> if strcmp(EEG.ref,'averef')
>     dataRank = dataRank - 1;
> end
> if isfield(EEG.etc,'clean_channel_mask') % using ASR to remove bad channels
>     dataRank = dataRank - length(find(~EEG.etc.clean_channel_mask));
> end
> dataRank = min([rank(double(EEG.data')) dataRank]);
> Next, I enter dataRank in the field named '# PCA Dims' of the
> pop_runamica() window. In this way, I get exactly dataRank independent
> components, keeping all the channels (remember that this step occurs after
> channel interpolation).
> Is this approach correct?
> Thank you in advance.
> All the best,
> Thomas
> _______________________________________________
> 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

Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20190117/605d1f61/attachment.html>

More information about the eeglablist mailing list