[FieldTrip] Decreased baseline level after using ICA in ERP data

Beom Jun Min mbj0310 at gmail.com
Tue Jun 25 13:04:15 CEST 2013


Dear Diego,

Thank you for your kind answer.
The importance of 'quality' you mentioned and the references that you
attached could help me to understand the ICA algorithm further.
And I have an additional question about preprocessing before ICA. Is
detrending needed before the decomposition if there is a linear trend in
the segmented data?
Because I noticed one component showing linearly and consistently decreased
(or increased) activity during one segment in some trials after ICA, I
wondered why that happened.

Apart from that, I found the possible cause of the past problem. It looks
like the ft_rejectcomponent might remove the 'demean' effect.
After I used the function without any component removing, (cfg.component =
[];) the baseline level decreased again but the shape of the ERP does not
change.
However, I have not found the way to correct this decreased baseline yet.
The ft_preprocessing with demean pre-stimulus does not work.

Thanks.

BJ



2013/6/24 Lozano Soldevilla, D. (Diego) <d.lozanosoldevilla at fcdonders.ru.nl>

> Dear Beom Jun,
>
> I see multiple scenarios why this baseline activity decrease could happen.
> First of all, how the component you're rejecting look like (i.e. "blink
> component")? Do you see this activity decrease after the baseline period?
>
> The "quality" of the ICA decomposition, how well your artifact/component
> of interest has been isolated by algorithm in time (i.e. blink time
> courses) and space (marked frontal topography), will determine the activity
> that later on you'll reject/select. If your decomposition is not well
> suited, the rejection of a particular IC activity might have "extra"
> activity you don't want to reject (effect of interest), might be the
> algorithm is not able to isolate the components of interests (i.e.
> artifacts) or a combination of both.
>
> To evaluate the quality of your ICA decomposition you might have a look
> here (http://www.ncbi.nlm.nih.gov/pubmed/19162199). Basically, the
> authors find that the ICA decomposition improves significantly "increased
> by removing the mean EEG at each channel for each epoch of data rather than
> the mean EEG in a prestimulus baseline". In addition (see here:
> http://sccn.ucsd.edu/pipermail/eeglablist/2012/004925.html), high-pass
> filtering above ~1hz improve the results.
>
> It's very important to feed ICA as much relevant data as you can use. The
> more the data, the better the decomposition. There's a rule of thumb that
> says that for a reliable IC decomposition 20 time points per channel2  is
> needed (see here for a reference
> http://www.ncbi.nlm.nih.gov/pubmed/16904745)
>
> I hope that helps,
>
> Diego
> ------------------------------
>
> *From: *"Beom Jun Min" <mbj0310 at gmail.com>
> *To: *"FieldTrip discussion list" <fieldtrip at science.ru.nl>
> *Sent: *Monday, 24 June, 2013 6:27:47 AM
> *Subject: *[FieldTrip] Decreased baseline level after using ICA in ERP
> data
>
>
> Dear all,
>
> I have ERP data and now I am dealing with ICA to remove muscle and eye
> artifacts.
> However, I found that after ft_rejectcomponent, the baseline level of the
> segmented epoch decreased. (The baselinewindow is [-0.2 0].)
> The baseline level decreased even though I rejected only one component.
>
> My script is shown below.
>
> *%% Removing the Artifacts*
> *cfg = [];
> *
> *cfg.component = [ ]; % to be removed component(s)*
> *post_ICA_temp6 = ft_rejectcomponent(cfg, comp_detrand, data_raw);*
> *
> *
> *%% timelocking*
> *
> *
> *cfg = [];*
> *timelock_temp6 = ft_timelockanalysis(cfg, post_ICA_temp6);*
> *
> *
> *%% Plot*
> *
> *
> *figure;*
> *cfg = [];*
> *cfg.layout = lay;*
> *cfg.interactive = 'yes';*
> *cfg.channel = ['all', {'-EKG', '-EMG'}];*
> *ft_multiplotER(cfg, timelock_temp6)*
>
> Is there something that I missed?
>
> Thanks.
>
> BJ
>
> --
> BeomJun Min, M.D.
>
> Department of Medical System Engineering (DMSE)
> Gwangju Institute of Science and Technology (GIST)
> 261 Cheomdan-gwagiro(Oryong-dong), Buk-gu, Gwangju
> 500-712, Republic of Korea (South)
> Phone: +82-62-715-3266 / Fax: +82-62-715-3244
> E-mail: mbj0310 at gmail.com, http://bmssa.gist.ac.kr
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
>
>
>
> --
> PhD Student
> Neuronal Oscillations Group
> Donders Institute for Brain, Cognition and Behaviour
> Centre for Cognitive Neuroimaging
> Radboud University Nijmegen
> NL-6525 EN Nijmegen
> The Netherlands
> http://www.ru.nl/people/donders/lozano-soldevilla-d/
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>



-- 
BeomJun Min, M.D.

Department of Medical System Engineering (DMSE)
Gwangju Institute of Science and Technology (GIST)
261 Cheomdan-gwagiro(Oryong-dong), Buk-gu, Gwangju
500-712, Republic of Korea (South)
Phone: +82-62-715-3266 / Fax: +82-62-715-3244
E-mail: mbj0310 at gmail.com, http://bmssa.gist.ac.kr
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20130625/f3ac3e63/attachment.html>


More information about the fieldtrip mailing list