This is a MATLAB pipeline for easy-to-program automated pre-processing of electroencephalogram (EEG) data using independent component analysis (ICA) and statistically-robust detection of artifacts. It is intended to work similar to a plug-in for the EEGLAB (https://sccn.ucsd.edu/eeglab/index.php) software suite, but it additionally contains low-level functions that work on simpler data arrays (e.g., 2- or 3-dimensional matrices) that may be useful for commandline artifact detection and removal. Generally, the EEG pipeline is designed to successively prune EEG artifacts identified in 1) one-second contiguous segments in the continuous data, in 2) grossly bad data channels, in 3) spatiotemporal stereotyped ICA artifacts, and in 4) channel/time-segments (e.g., trial epochs). EEGLAB (version 14.1.1b or above) should be in the user's MATLAB path. Also, noted issues with newer updates of EEGLAB's pop-resample() have prompted the inclusion of a previous pop-resample() from version 9.
Burwell, S. J., Malone, S. M., Bernat, E. M., & Iacono, W. G. (2014). Does electroencephalogram phase variability account for reduced P3 brain potential in externalizing disorders? Clinical Neurophysiology, 125(10), 2007-2015.
- Removal of independent components that match spatial and temporal criterion (e.g., blinks, saccades)
- Robust statistical approaches for identification of artifacts, including
- statistical thresholding using the normalized median absolute deviation (cf. Rousseeuw & Croux, 1993)
- artifact detection through joint use of multiple features (cf. Mognon et al., 2010), vs. a single feature alone
- Identification and interpolation of channel or channel/time-segment artifacts (e.g., Nolan et al., 2010; Junghofer et al., 2000)
- Easy-to-program modular automation via proc_commander(), a tool that takes arguments similar to SPM's spm_jobman() (cf. https://en.wikibooks.org/wiki/SPM/Batch)
- Identification of salt-bridge artifacts based on the electrical distance (Tenke et al., 2001), see get_elecdists()
- Removal of muscle-artifacts (i.e., white-noise signals) based on canonical correlation or independent components (De Clercq et al., 2006), see bsscca_correct_emg() and bssica_correct_emg(), respectively
- Easy-to-implement Current Source Density transformation (http://psychophysiology.cpmc.columbia.edu/Software/CSDtoolbox/), see proc_CSD()
addpath /path/to/eeglab %recommended to use version 14.1.1b or above
eeglab; %keeping EEGLAB GUI open is optional
addpath /path/to/eeg_commander %obtained in bash by "git clone <repo>"
eeg_commander_startup; %add necessary eeg_commander paths
help proc_commander; %e.g., help for overarching proc_commander (pipeline) function
%% Example pipeline usage, see the following scripts:
% example_scripts/mctfr-rseeg/run_import_mctfr_rseeg.m
% example_scripts/mctfr-rseeg/run_export_mctfr_rseeg.m
% example_scripts/alcstress/run_import_alcstress_general.m
% example_scripts/alcstress/run_export_alcstress_npueeg.m
%% Other example functions:
% Identify channel (pairs) where electrical distances <2 uV (i.e., putative salt-bridge)
methods(1).meas = 'Vdist-min';
methods(1).crit = 2 ;
methods(1).stat = 'abs';
salt_bridged = artifact_channels(EEG.data, methods, 0, 1);
% Remove independent component(s) reflecting white noise / muscle artifact. I.e., where the
% time-series of the component has an autocorrelation (where lag = 1) is <.80 for >25% of
% the recording (detrending of the component time-series applied).
EEG = bssica_correct_emg(EEG, 1, .80, 'pctbad',25,'detrend',[1 2 3]);
% Do Current Source Density / Laplacian transform
process.loaddir = 'EEG';
process.EEG = EEG ;
EEG = proc_commander(process,'cfg_CSD');