Steady-state visual evoked potential (SSEVP) from EEG signal modeling based upon recurrence plots

Dealing with electroencephalography (EEG) signals is often not simple. Steady-state visual evoked potentials (SSVEP) are signals even more difficult to determine or detect accurately. Given their non-stationary, lack of predictability, quality of recorded signal or a considerable amount of noise embedded in the signal to be many of the factors that make an analysis of SSVEP signals a hard and time-consuming task. Since EEG signals are non-stationary signals, using nonlinear features such as recurrence quantification analysis (RQA) may be more descriptive than other traditional methods for investigating these signals. The demonstrated ability of recurrence quantification analysis to detect very subtle patterns in time series and extract signals buried in large amounts of noise may be useful for this type of signal. The proposed technique, recurrence quantification analysis, demonstrates the ability to extract signals up to a very low signal-to-noise ratio and to allow an immediate appreciation of their degree of periodicity for SSVEP signals. In this contribution, many experiments on 30 subjects and three separate visual tests for each subject using a commercial apparatus have been performed. RQA was carried out for all valid tests, and patterns were found. Additional tests were carried out to detect false-positive trends, thus demonstrating that this technique is feasible to detect SSVEP events with moderate accuracy. Lastly, the results were compared with results from other methods, and the advantages of using RQA tools above other traditional methods were highlighted.