[Eeglablist] Script for artifact rejection by ICLabel

Makoto Miyakoshi mmiyakoshi at ucsd.edu
Mon Mar 22 16:26:09 PDT 2021

Dear Shingo,

Your code contains another more criterion for rejecting ICs, namely
dipolarity of the IC scalp topos.
Can you please check the residual variance (r.v.) from the dipole fitting
to the IC scalp topos? If many of them show > 15% despite you see a nicely
dipolar scalp topos, something must be wrong--most likely, your head model
setting is incorrect. For example, 90-degrees rotated to the right? This
typically happens when you use BESA 4-shell spherical model.


On Tue, Mar 16, 2021 at 9:49 AM 時本真吾 via eeglablist <
eeglablist at sccn.ucsd.edu> wrote:

> Dear EEGLAB experts,
> I usually reject artifacts by independent components with ICLabel. I
> reject an IC when its probability of the brain activity is lower than 70%
> and its residual variance is greater than 15% following Makoto's
> preprocessing pipeline. This criterion works very well, and I tried to
> apply the same criterion to multiple participants with the batch script
> below. However, the number of remaining ICs is much smaller than the number
> of ICs selected by the eye inspection with the same criterion.
> % Perform IC rejection using ICLabel scores and r.v. from dipole fitting.
> EEG       = IClabel(EEG, 'default');
> brainIdx  = find(EEG.etc.ic_classification.ICLabel.classifications(:,1) >=
> 0.7);
> rvList    = [EEG.dipfit.model.rv];
> goodRvIdx = find(rvList < 0.15);
> goodIcIdx = intersect(brainIdx, goodRvIdx);
> EEG = pop_subcomp(EEG, goodIcIdx, 0, 1);
> EEG.etc.ic_classification.ICLabel.classifications =
> EEG.etc.ic_classification.ICLabel.classifications(goodIcIdx,:);
> I would like to automate the artifact rejection by ICLabel to avoid human
> errors. Could anyone give me suggestions for the difference in the humber
> of remaining ICs? Thank you in advance.
> Sincerely,
> ******************************************
> Shingo Tokimoto, Ph.D.
> in Linguistics and Psychology
> Department of Foreign Languages
> Mejiro University
> 4-31-1, Naka-Ochiai, Shinjuku, Tokyo,
> 161-8539, Japan
> tokimoto at mejiro.ac.jp
> ******************************************
> _______________________________________________
> 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

More information about the eeglablist mailing list