[Eeglablist] removing ICA components

Makoto Miyakoshi mmiyakoshi at ucsd.edu
Fri May 16 08:52:37 PDT 2014


Dear Luisa,

Update EEGLAB to the latest version. I remember we had a temporary issue of
head plot a few years ago, and this may be it...

I'm worried about the blank ERP image. It should be unrelated to the scalp
map issue.

As a general advice, you should record as many channels as possible (up to
128). Even if your interest is in the motor cortex only, you should have
whole scalp coverage for good ICA results. Remember, cortical patch
activity always create an equivalent dipole that has 8-shape projection
pattern. ICA not only catches the projection of the short pathway (which
you assume 'the source is right *under* the channel') but also catches the
polarity-inversed long pathway on the otherside of the head, which is why
IC scalp maps have positives and negatives. In other word, if you have nice
focal positivity near motor cortex, that should be accompanied by the huge
surrounding negativity because of dipolar (i.e. 8-shaped) projection, and
both are equally important.

Makoto


2014-05-14 22:27 GMT-07:00 Luisa Roeder <l.roeder at qut.edu.au>:

>  Thanks, Steve, for your reply.
>
> Yes, that is what I thought re the ICA amplitude per trial plot…
>
> Re the head maps, I thought there may be another analysis approach; e.g.
> instead of rejecting components based on maps doing it somehow differently…
>
>
>
> Cheers
>
> Luisa
>
>
>
> *Luisa Roeder* *| PhD researcher | Movement Neuroscience Program*
>
> Institute of Health and Biomedical Innovation & School of Exercise and
> Nutrition Science
>
> Queensland University of Technology | 60 Musk Ave, Kelvin Grove,
> Brisbane, QLD 4059, Australia.
>
> *t:* 07 3138 6428 |* m:* 0427 469 478 | *e: l.roeder at qut.edu.au
> <l.roeder at qut.edu.au> *
>
> *w:* *www.ihbi.qut.edu.au/about/researchover/injury/movementNeuroscience.jsp
> <http://www.ihbi.qut.edu.au/about/researchover/injury/movementNeuroscience.jsp>
> *
>
> CRICOS No. 00213J
>
>
>
>
>
> *From:* politzerahless at gmail.com [mailto:politzerahless at gmail.com] *On
> Behalf Of *Stephen Politzer-Ahles
> *Sent:* Thursday, 15 May 2014 1:17 PM
> *To:* Luisa Roeder
> *Cc:* eeglablist at sccn.ucsd.edu
> *Subject:* Re: [Eeglablist] removing ICA components
>
>
>
> Hello Luisa,
>
>
>
> As for the top right panel of that plot, what that plots is the ICA's
> amplitude per trial, and the ERP. So I imagine you can't get those when
> you're not using epoched data (although I haven't tried myself).
>
>
>
> I'm not sure about the issue with the head maps; my hunch is that it might
> not be possible to get good topoplots without more channels, but hopefully
> someone else more familiar with that issue can give you some more input.
>
>
>
> Best,
>
> Steve
>
>
>
>
> Stephen Politzer-Ahles
> New York University, Abu Dhabi
> Neuroscience of Language Lab
> http://www.nyu.edu/projects/politzer-ahles/
>
>
>
> On Thu, May 15, 2014 at 3:37 AM, Luisa Roeder <l.roeder at qut.edu.au> wrote:
>
> Hi all,
>
>
>
> I have been following the EEGLAB discussion list for a few weeks now and
> have been working through most of the tutorials and wiki pages but I am
> still a newbie to MATLAB, EEGLAB and analysing EEG signals in general.
> Hence, I have basic, newbie questions which I hope you can help me with.
>
>
>
> I am trying to follow the analysis steps that are outlined in this paper:
> Petersen, T.H., M. Willerslev-Olsen, B.A. Conway, and J.B. Nielsen, *The
> motor cortex drives the muscles during walking in human subjects.* The
> Journal of physiology, 2012. *590*(Pt 10): p. 2443-52.
>
>
> http://www.ncbi.nlm.nih.gov/pubmed/?term=1.%09Petersen%2C+T.H.%2C+M.+Willerslev-Olsen%2C+B.A.+Conway%2C+and+J.B.+Nielsen%2C+The+motor+cortex+drives+the+muscles+during+walking+in+human+subjects.+The+Journal+of+physiology%2C+2012.+590%28Pt+10%29%3A+p.+2443-52.
>
>
>
>
> At the moment I am looking at a continuous dataset that contains 6 EEG and
> 6 EMG channels and that was recorded during a sustained muscle contraction
> of the leg. Eventually I also want to look at gait data but I am not there
> yet.
>
> As outlined in the paper, I am trying to perform ICA (runica) on the EEG
> channels. After I ran the ICA, I want to remove unwanted components (in the
> gui) via Tools->reject data using ICA-> reject components by map. However,
> since I only recorded 6 channels (C3, Cz, C4, F3, Fz, F4) the head maps I
> receive only cover the region between the 6 electrodes but not the whole
> head. Also, when clicking on one component the top right panel that is
> supposed to show the activity of the component is empty.
>
> Here is a link to 2 images:
> https://www.dropbox.com/sh/t5b8kmb9b9ce93e/AAC5VyeSEQbob0PTSiSZ61Ita
>
>
>
> This does not provide sufficient information to be able to judge whether
> to keep or to discard a component.
>
> What can I do to fix the problem? Is there another/better/more appropriate
> way to remove artefacts from my data than this approach? Or is the only way
> recording more channels (32+?) so that I can plot sufficient head maps?
>
> Currently I use MATLAB R2012a and EEGLAB v10.2.5.8b.
>
>
>
> Thanks so much in advance for your help!!
>
> Kind regards
>
> Luisa
>
>
>
> *Luisa Roeder* *| PhD researcher | Movement Neuroscience Program*
>
> Institute of Health and Biomedical Innovation & School of Exercise and
> Nutrition Science
>
> Queensland University of Technology | 60 Musk Ave, Kelvin Grove,
> Brisbane, QLD 4059, Australia.
>
> *t:* +61 7 3138 6428 |* e: l.roeder at qut.edu.au <l.roeder at qut.edu.au> *
>
> *w:* *www.ihbi.qut.edu.au/about/researchover/injury/movementNeuroscience.jsp
> <http://www.ihbi.qut.edu.au/about/researchover/injury/movementNeuroscience.jsp>
> *
>
> CRICOS No. 00213J
>
>
>
>
> _______________________________________________
> 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
>
>
>
> _______________________________________________
> 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/20140516/a0ccb7b7/attachment.html>


More information about the eeglablist mailing list