Hilbert marginal spectrum analysis for automatic seizure detection in EEG signals

In this paper, we present a new technique for automatic seizure detection in electroencephalogram (EEG) signals by using Hilbert marginal spectrum (HMS) analysis. As the EEG signal is highly nonlinear and nonstationary, the traditional Fourier analysis which expands signals in terms of sinusoids cannot appropriately represent the amplitude contribution from each frequency value. The HMS is derived from the empirical mode decomposition (EMD) which decomposes signal into a collection of intrinsic mode functions (IMFs). Since this decomposition is based on the local characteristic time scale of the signal, it can be well applied to nonlinear and nonstationary processes. In this work, the spectral entropies and energy features of frequency-bands of the rhythms using HMS analysis are extracted and fed into the support vector machine (SVM) for seizure detection of EEG signals. A final comparison between the results obtained with the developed technique and results adopted by Polat and coworkers using Fourier analysis with the same database is given to show the effectiveness of this technique for seizure detection.