Tags: faq preprocessing artifact

How can I consistently represent artifacts in my data?

FieldTrip includes multiple functions for automatic artifact rejection, as explained in this tutorial. These functions detect time segments in the data in which an artifact is present by the begin and end sample of that artifact. If there are N artifacts, that results in a Nx2 matrix, with the first column representing the begin samples, and the second column representing the end samples.

If you manually identify time segments with an artifact, you can represent them like

eog.artifact    = N x 2
muscle.artifact = M x 2
jump.artifact   = K x 2

and pass them onto the ft_rejectartifact function in the cfg.artfctdef field like

cfg.artfctdef.eog.artifact    = N x 2
cfg.artfctdef.muscle.artifact = M x 2
cfg.artfctdef.jump.artifact   = K x 2

followed by

cfg = ft_rejectartifact(cfg);

Bad channels

To specify lists of bad channels, you can use a consistent representation, e.g.

linenoise.badchannel = {'C3', 'Fp1'};
emg.badchannel       = {'T3', 'T4'};

and use the following code to merge the bad channels into a single list

hdr = ft_read_header(filename);
sel = false(size(hdr.label));

for i=1:length(linenoise.badchannel)
sel(strmatch(hdr.label, linenoise.badchannel{i})) = true;
end

for i=1:length(emg.badchannel)
sel(strmatch(hdr.label, emg.badchannel{i}) = true;
end

% combine them into one list
badchannel  = hdr.label(sel);
goodchannel = setdiff(hdr.label, badchannel);

This list with good channels can be passed to ft_preprocessing in the cfg.channel option.