Block term decomposition for modelling epileptic seizures

Borbála Hunyadi,Sabine Van Huffel,Wim Van Paesschen,Maarten De Vos,Daan Camps,Lieven De Lathauwer,Laurent Sorber

doi:10.1186/1687-6180-2014-139

Abstract

Recordings of neural activity, such as EEG, are an inherent mixture of different ongoing brain processes as well as artefacts and are typically characterised by low signal-to-noise ratio. Moreover, EEG datasets are often inherently multidimensional, comprising information in time, along different channels, subjects, trials, etc. Additional information may be conveyed by expanding the signal into even more dimensions, e.g. incorporating spectral features applying wavelet transform. The underlying sources might show differences in each of these modes. Therefore, tensor-based blind source separation techniques which can extract the sources of interest from such multiway arrays, simultaneously exploiting the signal characteristics in all dimensions, have gained increasing interest. Canonical polyadic decomposition (CPD) has been successfully used to extract epileptic seizure activity from wavelet-transformed EEG data (Bioinformatics 23(13):i10–i18, 2007; NeuroImage 37:844–854, 2007), where each source is described by a rank-1 tensor, i.e. by the combination of one particular temporal, spectral and spatial signature. However, in certain scenarios, where the seizure pattern is nonstationary, such a trilinear signal model is insufficient. Here, we present the application of a recently introduced technique, called block term decomposition (BTD) to separate EEG tensors into rank- (L r ,L r ,1) terms, allowing to model more variability in the data than what would be possible with CPD. In a simulation study, we investigate the robustness of BTD against noise and different choices of model parameters. Furthermore, we show various real EEG recordings where BTD outperforms CPD in capturing complex seizure characteristics.

Highlights

Epilepsy is one of the most common neurological disorders, affecting 0.5% to 1% of the global population [1]
We investigate the robustness of the tensor decomposition techniques against physiological noise including background EEG activity and muscle artefacts, the impact of the chosen model parameters, and the advantages and differences of each approach
The ictal source is captured already in the first canonical polyadic decomposition (CPD) component for an signal-to-noise ratio (SNR) > 0.4, and the reconstruction does not benefit from extracting additional components

Summary

Introduction

Epilepsy is one of the most common neurological disorders, affecting 0.5% to 1% of the global population [1]. Independent component analysis (ICA) was proven to outperform PCA and to remove a wide variety of artefacts from the multichannel EEG [4]. It was shown that elimination of artefacts by ICA increases the quality and interpretability of ictal EEG recordings [6]. Canonical correlation analysis (CCA) used as a BSS technique [7] outperformed the ICA JADE algorithm in removing muscle artefacts. Taking into account both performance and numerical complexity, CCA and CoM2 [8] were shown to be the best choice of BSS method for removing muscle artefacts from epileptic EEG [9]. It was shown that EEG source localisation is rendered more reliable if eye and muscle artefacts are removed using spatially constrained ICA and BSS-CCA, respectively [10]. An artefact removal scheme from channel × time × frequency EEG tensors using a multiway analysis technique, namely, canonical polyadic decomposition (CPD), was presented in [11]

Objectives

Methods

Results

Discussion

Conclusion