[Eeglablist] Rejecting/interpolating/reconstructing bad channels

[Tarik S Bel-Bahar]

Hello Christopher, notes below.

*Yes, as you mentioned, one can load a second file into eeglab that
has all the channels. Then in the interpolation GUI, click the button
that results in the difference. The option to select is either "use
specific channels of other dataset" or "use all channels from other
dataset". That should then do your interpolation for you correctly.
The codelet below is an example of the same steps.

%%%%%%have file to-be-interpolated in eeglab memory as the first
loaded file, then..

if EEG.nbchan < 128;  %does it have all the channels?
   EEG = pop_loadset('filename','ALL128CHANS.set','filepath', cd);
%load up allchannels file
   [ALLEEG, EEG, CURRENTSET] = eeg_store( ALLEEG, EEG, 0 ); EEG =
eeg_checkset( EEG );
   [ALLEEG EEG CURRENTSET] = pop_newset(ALLEEG, EEG,
CURRENTSET,'retrieve',1,'study',0);
   EEG = pop_interp(EEG, ALLEEG(2).chanlocs, 'spherical');     %get
all chan from second loaded file and interpolate
   [ALLEEG, EEG, CURRENTSET] = eeg_store( ALLEEG, EEG, CURRENTSET);
   currFILENAME= [EEG.filename(1:end-4) '_Interpolated'] ; %save the
interpolated file
   EEG = pop_saveset_tarik( EEG,
'filename',currFILENAME,'filepath',EEG.filepath);
end;
end;
ALLEEG = pop_delset( ALLEEG, [2] ); eeglab redraw


*For dropping channels, consider review and practice with the eeg
select data option in GUI to drop/remove channels, if you are not
using it already.
*STUDY can and will interpolate electrodes, you don't have to per subject.
*You'll likely have to make your own modifications at various parts of
the pipeline to fit your needs, expectations, and the analyses you are
trying to do.
*ADJUST is great, but have you reviewed your data to make sure that
you know if ICs that have some neural data are not thrown out. In my
experience you have to mess with the settings. Consider comparing
results to SASICA and IC-MARC IC classification plugins.



























On Mon, Aug 1, 2016 at 9:26 PM, Christopher Barkley <barkl025 at umn.edu> wrote:
> Hi Makoto and fellow list-mates,
> Pardon any ignorance on my end but I have been dealing with a particularly
> 'messy' data set (EEG was collected while subjects were heavily medicated)
> so have been tinkering with the best way to pre-process to get the data as
> clean as possible before creating STUDY. So... I basically used Makoto's
> pre-processing pipe-line as a guide, down-sampling and filtering first, then
> using 'trim-outlier' to remove (particularly) bad channels, keeping track of
> which channels I was removing for each subject. The next steps are where I
> may have gone astray: I first rejected non-sterotyped artifacts manually,
> fed the resulting (epoched) data into ICA, rejecting components using the
> ADJUST toolbox, fed the data back into ICA, and then manually rejected the
> resulting pruned data. The goal was to clean the data to this extent so that
> when I replaced/interpolated the removed channels the data would be as clean
> as possible before calculating average reference, the output of which would
> then also be as clean as possible. However when I try to interpolate in the
> GUI, the list of 'removed channels' does not include the ones that I
> removed, and choosing 'select from data channels'  in the GUI menu leads me
> to a list that excludes the already removed channels....  Have I gone
> completely awry, or is there perhaps a way to use 'pop_interp' and a
> EEG.chanlocs file from an initial pre-channel-deletion file in order to get
> at the subject-specific channel-deletions? I know the number of channels
> will need to be consistent for STUDY creation..... Anyways, many apologies
> for the ramble, any advice will be much appreciated, all the best.
>
> Chris Barkley
> Experimental and Clinical Pharmacology
> University of Minnesota
>
> _______________________________________________
> 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