The multiscale entropy: Guidelines for use and interpretation in brain signal analysis

Area of Science:

• Neuroscience
• Signal Processing
• Complexity Science

Background:

• Multiscale entropy (MSE) is used to assess brain signal variability, particularly in aging studies.
• The theoretical basis and interpretation of MSE in neuroscience are not fully established.
• Clarifying MSE's relationship with signal properties is crucial for its accurate application.

Purpose of the Study:

• To provide an intuitive explanation of Multiscale Entropy (MSE).
• To investigate the relationship between MSE, signal frequency content, and underlying dynamics (linearity, stochasticity).
• To clarify the interpretation of MSE in the context of brain signal analysis.

Main Methods:

• Utilized both simulated and experimental data for analysis.
• Examined the connection between MSE curves and the power spectrum.
• Explored MSE's ability to capture linear and nonlinear autocorrelations.

Main Results:

• MSE curve features correlate with the power spectrum in linear systems.
• MSE captures nonlinear autocorrelations and their interaction with stochastic processes.
• Observed MSE curve crossings in EEG data are linked to linear stochastic processes and slower time constants in younger adults.

Conclusions:

• MSE features can be explicitly linked to signal power content and autocorrelation measures.
• MSE provides insights into the temporal structure of brain activity fluctuations.
• Combining MSE with other metrics can prevent misinterpretations of underlying stochastic processes.