[Eeglablist] automatic epoch rejection

[Bradley Voytek]

Brion:

I've played around with a few different methods to get this working.

The simplest is to use EEGLAB's artifact rejection GUI, which allows
you to explicitly specify electrodes to use when looking for
artifacts.

What I find to be the most accurate for my purposes (removing trials
with eye movements and blinks, rather than using ICA to remove the
components), is to create surrogate artifact channels--one for blinks
and one for saccades--and run a thresholding algorithm across these
channels. I create a surrogate blink channel by multiplying my
inferior and superior ocular channels. Because these channels are
highly anticorrelated during blinks, this gives me a *huge* negative
deflection at the time of the blink (well above noise levels), which
is easy for an automated algorithm to extract. I do the same for
saccades using my two lateral ocular channels. You can also do an
effective automated removal of muscle artifacts by creating a
surrogate channel that is the sum of the power in high(er) frequency
bands (>20 Hz) across a subset of frontal and central electrodes.

::bradley voytek


On Wed, Oct 22, 2008 at 3:34 PM, Brion Woroch <bworoch2 at illinois.edu> wrote:
> Thanks Manos,
> You are correct, but setting a low threshold also seems to detect things
> that are not eyeblinks, such as alpha noise, fluctuating channels, HF noise,
> etc.   When I set a low threshold for detecting eyeblinks and it seems to
> detect them just fine.  However, it also marks any epoch for rejection in
> which any one of the scalp channels fluctuates above that threshold.  I am
> looking for a way to either restrict the thresholded fluctuation detection
> to the EOG channels or set a low threshold to EOG channels and a less
> conservative threshold to the scalp electrodes.
> Brion
> On Oct 22, 2008, at 4:51 PM, Klados Manousos wrote:
>
> Dear Brion
>
> If you search in literature there exist a certain paper which suggest that
> eye blink's voltage is larger than 50 ΞΌV. So you can put a threshold of
> 50ΞΌV for eye blinks. If i understand you you need an automatic procedure to
> reject contaminated epochs. One very easy algorithm is to find epochs with
> values larger than 50 ΞΌV and exclude them for your study. I can't
> understand the reason you need different threshold for different channels.
> You can indetify contaminated epochs from one signal and in the others the
> epochs have to be the same...correct me if i uderstood wrong.
>
> Greetings
> Manos
>
> 2008/10/22 Brion Woroch <bworoch2 at illinois.edu>
>>
>> Hello,
>> I am looking for a way to set the threshold for  the "detection of
>> extremely large fluctuations" in the automatic epoch rejection
>> differentially for different channels.  I would like to set one
>> threshold for EOG channels and another for scalp electrodes.  Does
>> anybody know how I can accomplish this?  Thanks in advance.
>> Brion Woroch
>>
>> _______________________________________________
>> 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
>
>
>
> --
> Klados A. Manousos
> Graduate Student, Research Assistant
> Lab of Medical Informatics, Medical School
> Aristotle University of Thessaloniki
> Thessaloniki, Greece
> _________________________________________________
> Tel: +30-2310-999332
> Website :
> http://lomiweb.med.auth.gr/gan/index_files/Page1269.htm
>
>
>
> _______________________________________________
> 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
>