[FieldTrip] Preprocessing before connectivity analysis and model order

Carlo Huber carlo.huber at uk-koeln.de
Wed Feb 27 10:21:27 CET 2013

Dear Subbu,

PDC seems to be influenced by filtering, see http://www.ncbi.nlm.nih.gov/pubmed/20026279 by Florin et al., 2010, Neuroimage.
I would not recommend therefore applying PDC to source data.
Other views are much welcome.

Von: fieldtrip-bounces at science.ru.nl [fieldtrip-bounces at science.ru.nl]" im Auftrag von "Subramaniam Iyer [eeguser at hotmail.com]
Gesendet: Mittwoch, 27. Februar 2013 00:07
An: FieldTrip
Betreff: Re: [FieldTrip] Preprocessing before connectivity analysis and model order

Dear Jan-Mathjis

Thanks for your reply. So if I understand correctly, performing PDC on EEG data can still be affected due to volume conduction, hence I will have to do source localization first.
I have never done source localization before, but I will try it by going through the extensive FT documentation.

However, is performing ICA before computing inverse solution recommended  or is it better to simply reject artifactual sections as you mentioned even in this case ?


From: jan.schoffelen at donders.ru.nl
Date: Tue, 26 Feb 2013 21:52:37 +0100
To: fieldtrip at science.ru.nl
Subject: Re: [FieldTrip] Preprocessing before connectivity analysis and model order

Dear Subbu,

In general I would discourage people to perform connectivity analysis on (non-invasively obtained) sensor-level data, with or without ICA cleaning. If, however, you are bold enough to try, I'd recommend to discard epochs that are contaminated by eye-blinks/EMG activity etc., rather than using ICA for cleaning. The main reason is that removing topographies of artifactual components may actually inject correlations onto the recordings. I don't know the wikipedia article myself, but I would say that DTF and PDC are not immune for volume conduction. Also, it is important to keep in mind that your choice of reference is important when interpreting results.



On Feb 25, 2013, at 10:52 PM, Subramaniam Iyer wrote:

Dear FieldTrip users

I  am trying to compute PDC from a set of EEG data from 10 subjects. I have three questions regarding connectivity analysis.

1) Since the data has artifacts (EOG,EMG) , would it be alright to preprocess the data using for ex. ICA ? This wikipedia article says, its not advisable :http://en.wikipedia.org/wiki/Brain_connectivity_estimators#DTF_and_PDC_are_not_influenced_by_volume_conduction. What would be your opinion ? Is there any article that would shed some light on this ?
2) How do I compute the model order in FieldTrip ? Before that, how can I check if the data is stable and stationary before computing the MVAR model in FieldTrip ?
3) How do I compute the threshold for PDC above which I can safely assume, a link exists between two regions ?

Thanks in advance.
fieldtrip mailing list
fieldtrip at donders.ru.nl<mailto:fieldtrip at donders.ru.nl>

Jan-Mathijs Schoffelen, MD PhD

Donders Institute for Brain, Cognition and Behaviour,
Centre for Cognitive Neuroimaging,
Radboud University Nijmegen, The Netherlands

Max Planck Institute for Psycholinguistics,
Nijmegen, The Netherlands

J.Schoffelen at donders.ru.nl<mailto:J.Schoffelen at donders.ru.nl>
Telephone: +31-24-3614793

_______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20130227/14369bc4/attachment-0002.html>

More information about the fieldtrip mailing list