[Eeglablist] Trouble rejecting ICs via STUDY

Makoto Miyakoshi mmiyakoshi at ucsd.edu
Wed Nov 13 21:00:42 PST 2019


Dear Dan,

If you got confused after reading 'incredibly helpful preprocessing
pipeline', I guess it was not very helpful after all. Sorry for the
confusion.

I have two things to tell you: One bad news, one good news. Overall, good
news!

1. Unfortunately, my plugins for STUDY functions are not compatible with
the latest version of EEGLAB2019 except for std_envtopo(). This is because
the internal structure of STUDY was changed.

2. Actually, I don't use cluster-level backprojection any more. Currently,
I'm happy to use Luca's IClabel() to generate probabilistic labels and use
them to pre-select ICs.

I updated my wiki page. If you are interested in, please follow the updated
pipeline.
https://sccn.ucsd.edu/wiki/Makoto%27s_preprocessing_pipeline#Alternative.2C_more_automated_pipeline_using_ASR_.2811.2F13.2F2019_updated.29

Nowadays I almost never perform epoch rejection for the purpose of data
cleaning. In this sense, my 'main pipeline' is something I no longer use.
However, because I use ASR only sparingly, the final data usually still
contains some level of artifacts. To clean up these bad trials, I wrote a
fast and efficient solution for epoch rejection. I liked the result, so I
would like to share it with you. It only targets trials with high > 25 Hz
power as default. The idea is somewhat similar to the pre-emphasis filter
used in ASR.
https://sccn.ucsd.edu/wiki/Makoto%27s_preprocessing_pipeline#An_idea_for_fast_and_efficient_trial_rejection_.2811.2F13.2F2019_Updated.29

I may wrap it up into an EEGLAB plugin if there is request.

Makoto



On Thu, Nov 7, 2019 at 6:45 AM Dillon, Daniel G. <ddillon at mclean.harvard.edu>
wrote:

> Dear Colleagues:
>
> I’m working my way through Makoto’s incredibly helpful preprocessing
> pipeline and am confused about using the STUDY tool for IC rejection (step
> 13 in Makoto’s 5/1/2019 update). I’m using eeglab2019_0, and here’s what’s
> happening:
>
>
>   1.  I choose “File > Create study > Browse for datasets”;
>   2.  In the “Create a new STUDY set” GUI, I then select my three datasets
> (just testing this out on a few subjects to get it working) and make sure
> the “Update dataset info—datasets sorted on disk will be overwritten” box
> is checked;
>   3.  In the GUI I click “Select by r.v.”, and in the next GUI I click OK
> to reject dipoles (a) with > 15% r.v. and (b) that are outside the brain;
>   4.  In MATLAB, this text appears: “Selecting dipoles with less than
> %15.0 residual variance and removing dipoles outside brain volume in
> dataset”;
>   5.  I click OK in the “Create a new STUDY set” GUI and then “Saving
> dataset . . .” appears 3x (once per subject) in the MATLAB Command Window.
> I then see “Rebuilding designs . . .” and “Done.”
>
> This all seems correct, but if I then load any of my subjects’ datasets
> and choose “Tools > Locate dipoles using DIPFIT > Plot component dipoles”,
> I have the same number of dipoles as I did before creating the STUDY. In
> other words, it doesn’t look like any dipoles have been removed, which is
> what I thought was happening in steps 4 and 5 above. Then I thought that
> perhaps the dipoles aren’t actually removed but are instead just marked for
> removal, but I’m not seeing anything consistent with that when I look at
> EEG.dipfit.model or EEG.reject.gcompreject for individual subjects.
>
> Am I missing something here? I was expecting to end up with ~10-20 clean
> ICs per subject, and instead I still have the ~90 ICs that I started with.
> Not sure if it matters, but I am doing all this with continuous data.
> Consistent with MM’s pipeline, my plan was to complete this step and then
> epoch the IC-rejected data before creating a final STUDY.design for the
> group-level analysis.
>
> I’d be grateful for anyone’s insights—using the STUDY tool to reject ICs
> in this way seems like a particularly efficient way to go (if I can get it
> working).
>
> Thanks!
>
> Dan Dillon
> McLean Hospital
>
>
> The information in this e-mail is intended only for the person to whom it
> is
> addressed. If you believe this e-mail was sent to you in error and the
> e-mail
> contains patient information, please contact the Partners Compliance
> HelpLine at
> http://www.partners.org/complianceline . If the e-mail was sent to you in
> error
> but does not contain patient information, please contact the sender and
> properly
> dispose of the e-mail.
> _______________________________________________
> 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