[FieldTrip] Removing artifacts by ICA

[Rodrigo Montefusco]

Dear Erfan-Vajdi
In my brief experience, there are a couple of things you should do, in my
humble opinion:
1.- Study the bases of the ICA, what you can do, what you can't... what is
it good for and what is "not good" for.
2.- Explore a lot before applying it to your pipeline. Even knowing what it
does, you should never trust completely a tool without testing it first to
exhaustion.
3.- Train your eyes to detect the artifacts you are interested in removing.
(pretty much reinforcing what I said in point 2).
4.- Develop a strategy to confront independent components rejection. If you
are interested in removing an eyeblink component for example...do that, and
do not remove components associated with muscle activity, for example. If
you want to do both, it's ok... just stay consistent with your strategy.
5.- You can always perform a ft_timelockanalysis or a ft_freqanalysis on
your component if you need extra information.

That's for starters :)

Good luck!

RMS

On Fri, Jun 9, 2023 at 11:56 AM Erfan Vi via fieldtrip <
fieldtrip at science.ru.nl> wrote:

> Dear FieldTrip users,
> FieldTrip toolbox uses EEGLAB toolbox functions to implement ICA
> decomposition to remove artifacts but it doesn't show spectrum power to
> identify artifacts easily. On the other side, EEGLAB has automatic artifact
> labeling to recognize artifacts with more precision. What can we do to
> choose the correct artifact components to remove? Any document to train?
> Thank you,
> Erfan-Vajdi
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://urldefense.com/v3/__https://doi.org/10.1371/journal.pcbi.1002202__;!!HJOPV4FYYWzcc1jazlU!4Rl0rbEeWBSWygP1FQYVU9lUF6iWA43hZOKUVLcIjoCDZFtB1IBg0ntLv41RAQztMURV1giH0XYqKNEQfK5GLx1a5RvZXac$ 
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20230609/dcda7938/attachment.htm>