[Eeglablist] using IClabel combinations to flag artifacts

[Cedric Cannard]

Dear Maruti,

I could be wrong, but I believe the thresholds correspond to the classification's confidence, so you never want to use low values, because you would be making a decision based on an unreliable classification (i.e. you are keeping or rejecting something that was not classified with high confidence, so you don't really know what it is basically). 
You could keep only the Brain class with confidence > .9, but this is not recommended because some components may have not been classified well (because of less obvious features) and yet contain valuable brain signals that you would be rejecting. 
The default is to reject ocular and muscular classes with confidence > .9 (i.e. 90% confidence), I would just follow that and look at the components that are rejected to check visually. e.g. pop_selectcomps(EEG, 1:EEG.nbchan);


Cedric Cannard



------- Original Message -------
On Wednesday, November 9th, 2022 at 6:11 AM, Maruti Mishra <maruti.mishra07 at gmail.com> wrote:


> Hi,
> I am using EEGLAB IClabel on a 5 min eye open resting state data.
> I wanted to know if it is ok to use threshold combinations (Brain <0.1 &
> Other >0.75 & muscle >0.8) to reject ICs, or its just good to say remove
> 
> components Brain <0.1 and then eye <0.75 etc
> 
> Thank You,
> Maruti
> _______________________________________________
> 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