Common Cross-Spectral Patterns of Electroencephalography for Reliable Cognitive Task Identification

Haifeng Li,Ruifeng Xu,Zhongliang Xu,Lin Ma,Shang Feng,Tzyy-Ping Jung

doi:10.1109/access.2020.2967814

Abstract

A current popular feature extraction method of classifying cognitive states and task engagements from electroencephalographs (EEG) is common spatial patterns (CSP). However, the classical CSP only focuses on the correlation between the signals and ignores all characteristics of the signals in the time domain and the frequency domain. In this paper we propose Common cross-spectral patterns (CCSP) a novel EEG feature extraction method for combining spectral and spatial patterns based on cross-spectral density (CSD) to overcome the disadvantages of classical CSP. In CCSP method. the cross-power matrices (CPMs) are extracted to measure the spatial correlation of each task in the band of interest. Then, an orthogonal linear transformation is constructed by simultaneously diagonalizing the CPMs of two tasks. Finally, each band's logarithmic power of the transformed signals is extracted for the support vector machine (SVM) classifier. The experiment results on multiple datasets showed that the CCSP algorithm is fully applicable to multi-channel EEG for reliable multi-cognitive-task identification.

Highlights

The human brain is always at a particular cognitive state or in transition from one state into another one, and any changes in brain activities will be reflected in scalp electroencephalograph (EEG)
We propose common cross spectral patterns (CCSP), which is a spatial-spectral transformation method as a feature extraction method for EEG signals
Common cross-spectral patterns (CCSP) PERFORMANCE COMPARED WITH OTHER METHODS To evaluate the performance of CCSP, we used the 1989 Keirn and Aunon data set of the Brain-Computer Interfaces Laboratory in Colorado State University [38]

Summary

Introduction

The human brain is always at a particular cognitive state or in transition from one state into another one, and any changes in brain activities will be reflected in scalp electroencephalograph (EEG). It is of great significance to properly measure and analyze brain activities for the research of human cognition, brain computer interfaces (BCIs) and the clinical diagnosis of various nervous and mental diseases. Motor imagery was a cognitive task research focus on the BCI Competition II, III and IV [9]–[11] from the start of the 21st century. Since the training process is time-consuming and demanding for the subject, the training sets are relatively small compared to the data collected in real applications. It is essential for the feature extraction algorithm to make the most of the limited training data.

Objectives

Methods

Results

Conclusion