SUPERIOR CLASSIFICATION OF EEG STATES USING RECURRENCE QUANTIFICATION ANALYSIS
DOI:
https://doi.org/10.34185/1991-7848.itmm.2025.01.029Keywords:
EEG classification, recurrence quantification analysis, spectral analysis, brain states, phase space reconstruction.Abstract
This study investigates the classification of brain states using electroencephalography (EEG) data, comparing recurrence quantification analysis (RQA) with traditional spectral analysis. The goal is to distinguish between eyes-open and eyes-closed states using EEG data. Experimental results demonstrate that RQA provides superior classification accuracy, particularly at the O1 electrode, where accuracy improved from 86% to 95%. The study also identifies optimal phase space reconstruction parameters and the most informative recurrence features for classification. RQA captures nonlinear dynamics of brain activity more effectively than frequency-based spectral methods. The findings support the use of RQA for improving classification in portable EEG systems. This enables more accurate analysis in real-time applications such as cognitive training and brain-computer interfaces.
References
Chelidze T., Matcharashvili T. Dynamical Patterns in Seismology. Recurrence Quantification Analysis: Theory and Best Practices. ред. Jr. Webber Charles L., Norbert Marwan. Cham : Springer International Publishing, 2015. С. 291–334. DOI:10.1007/978-3-319-07155-8_10.
EEG Motor Movement/Imagery Dataset. URL: https://archive.physionet.org/pn4/eegmmidb/ (accessed 02/21/2024).
Rawald T., Sips M., Marwan N. PyRQA—Conducting recurrence quantification analysis on very long time series efficiently. Computers & Geosciences. Vol. 104, 01.07.2017. P. 101–108. DOI:10.1016/j.cageo.2016.11.016.
