REGRESS_EOG yields the regression coefficients for
correcting EOG artifacts in EEG recordings.
The correction of a single record is obtained like this:
[R,S2] = regress_eog(S1, EL, OL)
[R,S2] = regress_eog(filename, EL, OL)
[R,S2] = regress_eog(filename)
OL = IDENTIFY_EOG_CHANNELS(filename)
EL are all remaining channels
Corrected data is obtained through
[R] = regress_eog(covm(S1,'E'), EL, OL)
S2 = S1 * R.r0; % without offset correction
S2 = [ones(size(S1,1),1),S1] * R.r1; % with offset correction
S1 recorded data
EL list of eeg channels: those channels will be corrected
OL eog/ecg channels.
if OL is a vector, it represents the list of noise channels
if OL is a matrix, OL derives the noise channels through rereferencing.
This is useful if the EOG is recorded monopolar, but the bipolar EOG
should be used for for artefact reduction (because the global EEG should remain),
One can define OL = sparse([23,24,25,26],[1,1,2,2],[1,-1,1,-1])
resulting in two noise channels defined as bipolar channels #23-#24 and #25-#26
R.r1, R.r0 rereferencing matrix for correcting artifacts with and without offset correction
R.b0 coefficients of EOG influencing EEG channels
S2 corrected EEG-signal
see also: IDENTIFY_EOG_CHANNELS, SLOAD
Reference(s):
[1] Schlogl A, Keinrath C, Zimmermann D, Scherer R, Leeb R, Pfurtscheller G.
A fully automated correction method of EOG artifacts in EEG recordings.
Clin Neurophysiol. 2007 Jan;118(1):98-104. Epub 2006 Nov 7.
http://dx.doi.org/10.1016/j.clinph.2006.09.003
http://pub.ist.ac.at/~schloegl/publications/schloegl2007eog.pdf