Spatial Time Series Segmentation Method Based on Traveling Waves

Abstract

This paper describes a method for spatial time series segmentation based on the analysis of traveling waves and demonstrates its efficiency using real data.Phase-locking value and coherence measure are used to evaluate the intensity of the traveling waves. The calculation of these indicators has been based on the cross-spectral analysis of the signals corresponding to the spatially adjacent points of observation.The described method can be applied to analysis of spatially organized physical, chemical, and biological processes. In particular, the method can be used to neurophysiological, geothermal and ocean researches and to studies of the distributed solar dynamics.Experimental researches of the method have been carried out on the real data of human brain electroencephalography (EEG). It is shown that the proposed method outperforms the classical methods of EEG segmentation. The results achieved have a simple interpretation.The first section provides a mathematical formulation of the multivariate time series segmentation problem. The second section gives a formal description of the traveling wave and indicators of traveling wave intensity between two spatial points. The third section considers the method of the features extraction based on the characteristics of traveling waves for multivariate time series segmentation. The fourth section presents the simulation results of the proposed segmentation algorithm on real EEG data. In conclusion the main results and future lines of research are discussed.

Keywords

• spatial time series
• travelling-wave
• segmentation
• phase-locking value
• coherence
• eeg microstates
• electroencephalogram