Low-Rank Linear Dynamical Systems for Motor Imagery EEG

Wenchang Zhang,Fuchun Sun,Shaobo Liu,Chuanqi Tan

doi:10.1155/2016/2637603

Wenchang Zhang, Fuchun Sun + Show 2 more

Open Access

https://doi.org/10.1155/2016/2637603

Copy DOI

Abstract

The common spatial pattern (CSP) and other spatiospectral feature extraction methods have become the most effective and successful approaches to solve the problem of motor imagery electroencephalography (MI-EEG) pattern recognition from multichannel neural activity in recent years. However, these methods need a lot of preprocessing and postprocessing such as filtering, demean, and spatiospectral feature fusion, which influence the classification accuracy easily. In this paper, we utilize linear dynamical systems (LDSs) for EEG signals feature extraction and classification. LDSs model has lots of advantages such as simultaneous spatial and temporal feature matrix generation, free of preprocessing or postprocessing, and low cost. Furthermore, a low-rank matrix decomposition approach is introduced to get rid of noise and resting state component in order to improve the robustness of the system. Then, we propose a low-rank LDSs algorithm to decompose feature subspace of LDSs on finite Grassmannian and obtain a better performance. Extensive experiments are carried out on public dataset from “BCI Competition III Dataset IVa” and “BCI Competition IV Database 2a.” The results show that our proposed three methods yield higher accuracies compared with prevailing approaches such as CSP and CSSP.

Highlights

With the development of the simpler brain rhythm sampling technique and powerful low-cost computer equipment over the past two decades, a noninvasive brain-computer interface (BCI) called electroencephalography (EEG) has attracted more and more attention than other BCIs such as magnetoencephalography (MEG), functional magnetic resonance imaging, and near infrared spectroscopy (NIRS)
We present the features of EEG dynamic textures by linear dynamical systems (LDSs) modeling and apply machine learning (ML) algorithms to capture the essence of dynamic textures for feature extraction and classification
We propose three methods for EEG pattern recognition: LDSs, LR+common spatial pattern (CSP), and low-rank LDSs (LR-LDSs)

Summary

Introduction

With the development of the simpler brain rhythm sampling technique and powerful low-cost computer equipment over the past two decades, a noninvasive brain-computer interface (BCI) called electroencephalography (EEG) has attracted more and more attention than other BCIs such as magnetoencephalography (MEG), functional magnetic resonance imaging (fMRI), and near infrared spectroscopy (NIRS). The MI pattern recognition is still a challenge due to the low signal-to-noise ratio, highly subject-specific data, and low processing speed For these reasons, more and more digital signal processing (DSP) methods and machine learning algorithms are applied to the MI-BCI analysis. Compared with CSP method, LDSs have the following advantages: first, LDS can simultaneously generate spatiospectral dual-feature matrix; second, there is no need to preprocess or postprocess signals, and the raw data can be directly fed into the model; third, it is easy to use and of low cost; last, the extracted features from the LDS are much more effective for classification.

LDSs Modeling

Low-Rank Matrix Decomposition

LR-LDSs on Finite Grassmannian

Classification Algorithm

Experimental Evaluation

Conclusion