Analysis of P300 containing EEG through three non-linear methods

Along the electroencephalography signal one may identify several frequency intervals as well as negative and positive potentials which are related to conscience states (e.g. awake and asleep) and to neuronal processes such as the response to a variety of stimuli and their interpretation. The P300 is among the latter. The fact that these potentials are the result of the sum of several neurons activation, which are in turn shaped by cognitive processes, makes the EEG signal a good candidate for three types of non-linear analysis: DFA, sample entropy and phase synchronization. All 3 methods showed a higher information production rate and hemisphere synchronization for EEG segments containing P300, when applied to the recordings of 2 subjects from a BCI database. The most significant one was a γ=0.83 for the electrode pair PO3-PO4. Given that the differences between both studied patterns (visual ERP with and without P300), the measures obtained in this paper may be used as viable characteristics for the identification of such patterns.