[FieldTrip] Dealing with artifacts in continuous EEG data before Granger Causality analysis
dao4free at gmail.com
Thu Oct 9 12:19:19 CEST 2014
Dear FieldTrip list,
I'm planning to apply Granger Causality (GC) models to my continuous EEG
data, but I'm not sure about the preprocessing steps.
I have read an article
filtering EEG data before GC analysis, but still it is not clear for me how
to deal with removing some data segments with strong artifacts.
Right now I'm planning to clean the data (2-minutes trials) by removing the
artifacts using visual detection (+filtering notch+detrending) followed by
GC analysis using 2 sec. windows (actually the final length of the windows
should be identified by VAR model) .
Using such preprocessing approach I would receive continuous data with
"gaps" therefore some 2-sec segments may consist of the mixed data before
and after the removed part. Such situations seem to me not appropriate for
As a possible solutions I thought about 1) removing these mixed segments or
2) Dividing raw data (before artifacts rejection) on 2-sec segments and to
remove completely such segments that will contain an artifact. Do I miss
I would appreciate your advice on this matter.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the fieldtrip