Figure 1.

Illustration of the trimming procedure. (A) The first stage shows an original signal after filtering through the theta band contaminated with high aberrant values (abnormal and not representative as to be considered to do meaningful comparisons and scaling, and whose effect should be neutralized, reduced, or removed). The second stage shows the effects of inter-channel trimming (with σ_Et = 57), that already accomplishes its purpose of neutralizing these amplitude values from the time series while preserving the rest of the signal. The third stage shows the effect of intra-channel trimming (with σ_It = 11), which is very fine and preserves the signal very well. (B) The same original EEG signal with sequential samples organized per channel (along all the 64 EEG channels) in the horizontal axis, and the amplitude of samples in the vertical axis. High aberrant values are not dispersed or “distributed” throughout the whole signal, but rather clustered or concentrated in isolated channels, a particularity corroborated in our whole dataset after throughout scrutiny. Here, it is revealed that aberrant values going up to 165.8 standard deviations beyond the mean exist, but they are concentrated in Ch 57 only. Amplitudes correspond to voltages in microvolts (μV).