I'm new to the topic so forgive the trivial question.
I am trying to understand data preprocessing for EEG signals. As I understand, after recording continuous EEG signals, they should be band-pass filtered and segmented into epochs (so the window of time that is relevant to the experiment). After that, these epochs should be baseline corrected. Can anyone explain what this concept of baseline correction means?
A baseline is generally an unwanted artifact that alter interpretation or computation on signals. The concept is not totally well-defined, and several different names coexist, with similar meaning: drift, continuum, trend, background.
It is often described as a large-scale, slowly-varying reference level, that should be removed to compare, or adjust the signal with respect to a globally flat or zero-reference. It can be seen as a generalized (non-constant) offset or bias correction. When removed, it is generally simpler to further process the signal, quantify parameters, extract features.
In some domains, this smooth reference may also include random noise.
Similar problems are answered here:
