Localization of Active Brain Sources From EEG Signals Using Empirical Mode Decomposition: A Comparative Study

Abstract

The localization of active brain sources from Electroencephalogram (EEG) is a useful method in clinical applications, such as the study of localized epilepsy, evoked-related-potentials, and attention deficit/hyperactivity disorder. The distributed-source model is a common method to estimate neural activity in the brain. The location and amplitude of each active source are estimated by solving the inverse problem by regularization or using Bayesian methods with spatio-temporal constraints. Frequency and spatio-temporal constraints improve the quality of the reconstructed neural activity. However, separation into frequency bands is beneficial when the relevant information is in specific sub-bands. We improved frequency-band identification and preserved good temporal resolution using EEG pre-processing techniques with good frequency band separation and temporal resolution properties. The identified frequency bands were included as constraints in the solution of the inverse problem by decomposing the EEG signals into frequency bands through various methods that offer good frequency and temporal resolution, such as empirical mode decomposition (EMD) and wavelet transform (WT). We present a comparative analysis of the accuracy of brain-source reconstruction using these techniques. The accuracy of the spatial reconstruction was assessed using the Wasserstein metric for real and simulated signals. We approached the mode-mixing problem, inherent to EMD, by exploring three variants of EMD: masking EMD, Ensemble-EMD (EEMD), and multivariate EMD (MEMD). The results of the spatio-temporal brain source reconstruction using these techniques show that masking EMD and MEMD can largely mitigate the mode-mixing problem and achieve a good spatio-temporal reconstruction of the active sources. Masking EMD and EEMD achieved better reconstruction than standard EMD, Multiple Sparse Priors, or wavelet packet decomposition when EMD was used as a pre-processing tool for the spatial reconstruction (averaged over time) of the brain sources. The spatial resolution obtained using all three EMD variants was substantially better than the use of EMD alone, as the mode-mixing problem was mitigated, particularly with masking EMD and EEMD. These findings encourage further exploration into the use of EMD-based pre-processing, the mode-mixing problem, and its impact on the accuracy of brain source activity reconstruction.

Analysis in the frequency domain of multicomponent oscillatory modes of the human electroencephalogram extracted with multivariate empirical mode decomposition

“…These findings are the evidence of one of the limitations of the empirical mode decomposition: the so called mode-mixing, consisting in the appearance of disparate scales in an IMF, or when a signal with a similar scale appears in different IMF components. This fact has been reported and multiple procedures have been proposed to mitigate its effect (Munoz-Gutierrez et al 2018;Soler et al 2020;Tsai et al 2016;Zheng, Xu 2019).…”

“…The EMD method that was applied to decompose the EEG was the most novel of the multivariate versions, the APIT-MEMD (Hemakom et al 2016) that has been developed as an improvement to the MEMD Ur Rehman et al 2010), with the express purpose to cope with power imbalances and inter-channel correlations of multichannel signals as the EEG signal, and also was used the method proposed by (Xie, Wang 2006) for the calculation of the mean spectral frequencies that mitigate the effect of some frequencies that have associated low values of power spectral density. However, as has been acknowledged by other authors the mode-mixing problem continues to be a big issue for the analysis of multichannel signals like the EEG (Alegre-Cortes et al 2016;Munoz-Gutierrez et al 2018;Soler et al 2020;Zheng, Xu 2019).…”

“…This range corresponds with the expected frequencies of the EEG from the classic delta to the gamma bands. The estimations of the power spectral density expressed in normalized units (%) of the oscillatory components (IMFs) didn't show notable differences between methods (Table 3), and the detected significant differences of the mean weighted frequencies can be ascribed to the fact that the HHT method has a higher instantaneous frequency resolution than the FFT method as noted by (Durstewitz, Deco 2008;Munoz-Gutierrez et al 2018), and the fact that Fourier transform have constant frequencies and weights, while the Hilbert transform allows the frequency as well as the amplitudes to vary over time (Al-Subari et al 2015).…”

Comparison between traditional FFT and marginal spectra using the Hilbert-Huang transform method for the broadband spectral analysis of the EEG in healthy humans

“…The slight but significant differences observed for values of the mean weighted frequency in the different bands may be associated with the ability of the Hilbert-Huang method that enhances the presence of fast low amplitude oscillations, and offers better time-resolution due to its instantaneous frequency property (Munoz-Gutierrez et al 2018), and performs better compared with Fourier analysis (Noshadi et al 2014).…”

Low-Density EEG for Neural Activity Reconstruction Using Multivariate Empirical Mode Decomposition

“…ERP studies. The EMD is a method that has shown the capability to separate signals using a time-frequency decomposition in different contexts, however the EEG signals are challenging due to the frequency proximity of the source activity, this condition makes that EMD solutions generally present mode mixing effects in the IMF decomposition, in which, the MEMD attenuate these effects when sources have a closer frequency as presented in Muñoz-Gutiérrez et al (2018).…”