Nonlinear multivariate analysis of neurophysiological signals.

Area of Science:

• Neurophysiology
• Information Theory
• Nonlinear Dynamical Systems

Background:

• Multivariate time series analysis is crucial in neurophysiology for studying relationships between simultaneous signals.
• Recent advances enable studying synchronization from time series using information theory and nonlinear dynamics.

Purpose of the Study:

• To extend multivariate linear methods for assessing nonlinear interdependence in neurophysiological signals.
• To review and describe nonlinear methods for analyzing synchronization (phase, generalized, event) in neurophysiological data.

Main Methods:

• Description of commonly used multivariate linear methods in neurophysiology.
• Review of entropy and mutual information concepts.
• Detailed explanation of nonlinear methods: phase, generalized, and event synchronization.
• Application of multivariate surrogate data tests to assess interdependence.

Main Results:

• Demonstration of extending linear methods to detect nonlinear interdependence.
• Illustrations of applying nonlinear synchronization methods to neurophysiological data.
• Assessment of interdependence strength (strong/weak) and type (linear/nonlinear) using surrogate data tests.

Conclusions:

• The study provides a framework for analyzing linear and nonlinear interdependence in neurophysiological signals.
• Nonlinear methods offer powerful tools for understanding complex brain dynamics.
• Surrogate data testing is essential for robustly characterizing signal interdependencies.