The fast Fourier transform (FFT) has been the main tool for electroencephalographic (EEG) Spectral Analysis (SPA). However, as the EEG dynamics show nonlinear and non-stationary behavior, results using the FFT approach may result meaningless. A novel method has been developed for the analysis of nonlinear and non-stationary signals known as the Hilbert–Huang transform method. In this study we analyze the differences for the broadband (SPA) of the EEG using the traditional FFT approach with those calculated with the Hilbert Marginal Spectra (HMS) after decomposition of the EEG with a multivariate empirical mode decomposition algorithm. EEG segments recorded from 19 leads of 47 healthy volunteers were studied. Statistically significant differences between methods were found for almost all leads by variance analyses. The agreement assessment shows that mean weighted frequencies have a good agreement for almost all bands, with the exception of beta-2 and gamma bands where values for the HMS where higher than 3 Hz. Also the HMS method received lower than 5% energy values for alpha activity with an increment in the adjacent bands. The HMS may be considered a good alternative for the SPA of the EEG when nonlinearity or non-stationarity may be present.
Bibliographical notePublisher Copyright:
© 2021 The Authors. Engineering Reports published by John Wiley & Sons, Ltd.
- Hilbert–Huang transform
- multivariate empirical mode decomposition
- non-stationary analysis
- spectral analysis
ASJC Scopus subject areas
- Engineering (all)
- Computer Science (all)