Research on functional connectivity with portable electroencephalograph recordings for application in brain-computer interfaces

Resumen

The performance of a brain-computer interface (BCI) system is influenced by several factors, such as the acquisition systems used and the methodologies applied. In this dissertation, several applicable methods in BCI systems have been examined and proposed, focusing primarily on commercial and low-cost systems which are more accessible to anyone. Specifically, this dissertation presents a filtering method that uses the graph Laplacian quadratic form and the Phase Locking Value (PLV) to generate a new filtered signal and improve the feature classification algorithms commonly used in BCI. In addition, a comparison between several algorithms using a long short-term memory (LSTM) network is performed, and the time window that maximizes the classification accuracy is studied. Finally, both lines of research are applied in a virtual reality environment, which is proposed as a safe environment to test these methodologies and allow users to try them and train their control without any risk.