[Eeglablist] EOG Artifact removal before ICA?

[Makoto Miyakoshi]

Dear Alexander,

Here is Makoto's pipeline for your info
http://sccn.ucsd.edu/wiki/Makoto%27s_preprocessing_pipeline

Don't use EOG artifact remover before ICA. Let ICA take care of it, since
it's the best algorithm to decompose EOG and blinks. I compared 33ch vs.
same data but reduced to 3ch (Fz, Cz, Pz) and run ICA. Interestingly, the
first component in both results showed identical ERPimage!

If you apply ICA, apply high-pass filter at 1Hz with transition bandwidth
1Hz (0.25-0.75Hz transition).

You may not need to epoch before ICA. Make sure that you have enough (i.e.
channel^2 x 30 or larger) datapoints if sampled at 250Hz (you can
downsample to 125Hz if you inspect only up to 60Hz or around... this
actually helps ICA)

I have no experience with ADJUST, sorry.

If you re-reference, do it to average reference. According to Jason Palmer,
it's better that you average reference before ICA (but after bad-channel
removal).

Makoto

On Fri, May 29, 2015 at 5:26 AM, Alexander Quent <alexander.quent at rub.de>
wrote:

> Hello,
> I have a more general question concerning the order of steps in
> pre-processing process.
>
> Aim: I want to look at negative mean amplitudes between 440 and 540 msec.
>
> My set up:
> 32 channels of which 4 are used for recording EOG and 1 is used an
> additional mastoid reference electrode.
>
> My general plan:
> I want to use remove artifacts using ICA. What I am asking myself is
> whether it is legitimate to remove EOG artifacts before running the ICA
> using a different plugin? Is the apparent redundancy in any way
> problematic?
> This is what I want to do:
> 1. Removing slow linear trend (high pass filter circa 0.3 Hz)
> 2. Removing EOG artifacts using BSS with AAR plugin
> (http://www.germangh.com/eeglab_plugin_aar/)
> 3. Removing EOG channels
> 4. Visually inspect data for artifacts
> 5. Remove bad channels
> 6. Low pass filter and epoching
> 7. Run ICA
> 8. Run ADJUST 1.1
> 9. Re-reference to linked mastoid and baseline correction
>
> Since this is my first ERP analysis, I am sure whether this pipeline is
> legitimate: Especially due to the fact that I want use BSS and ICA, but
> also whether it is okay to re-reference to linked mastoid at the end.
>
> Maybe you have some advise for me.
>
> With best regards,
> Alexander Quent
> _______________________________________________
> 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
>



-- 
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20150603/0c40d266/attachment.html>