[Eeglablist] question about ica decomposition of ictal iEEG signal to compute component topography

[Makoto Miyakoshi]

Hi Anne-Cecile,

I can't see your attachment. Please upload them to a server and give us the
URL.

> I am supposed to retrieve the unmixing matrix by multiplying icaweights
by icasphere, isn't it?

Correct. The unmixing matrix is EEG.icaweights*EEG.icasphere. This
operation is not commutative (AB ~= BA), so pay attention to the order.

My comment: iEEG are highly independent across electrodes. EEG.icawinv (or
IC topography) should show a single-channel spot in most of the components.
See the following paper. Notice how opposite the authors' claims are
compared with the data shown.
Whitmer D, Worrell G, Stead M, Lee IK, Makeig S. (2010).  Utility of
independent component analysis for interpretation of intracranial EEG.
Front Hum Neurosci. 2010 Nov 02; 4 184

> The ictal data is common average referenced. Do you recommend it or shall
i use the raw data before referencing?

Even if you use the raw data, it does not mean you can escape from the
issue of choice of the reference potential. I recommend you empirically
compare multiple reference methods/locations and choose what you like.

Makoto


On Wed, Sep 18, 2024 at 9:45 PM Anne-Cecile Lesage via eeglablist <
eeglablist at sccn.ucsd.edu> wrote:

> Dear eeglab team
>
> I ran ica decomposition of ictal iEEG signal to compute component
> topography. The patient is public data.
>
> More details are in the attached file (patient data set, channel
> localization, ictal plots).
> I ran the ica as in the attached screen capture.
> The ICA plots do not look bad but the matrices do not make sense.
> I am supposed to retrieve the unmixing matrix by multiplying icaweights by
> icasphere, isn't it?
> The ictal data is common average referenced. Do you recommend it or shall i
> use the raw data before referencing?
>
> Thank you
> Regards
>
> --
> Anne-Cécile Lesage, PhD
> Research Scientist
> Department of neurosurgery
> University of Texas Medical Branch
> anlesage at utmb.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