Method for automatically removing muscle artifacts in single-channel EEG signal

Abstract

The invention discloses a method for automatically removing muscle artifacts in a single-channel EEG signal. The method includes the steps that the EEG signal is subjected to SSA decomposition to obtain P signal components; the P signal components are spliced into a P-dimension data matrix by row; the P-dimension data matrix is subjected to time delay processing to obtain several data matrixes; the several data matrixes are subjected to blind source separation with MCCA to obtain a source estimation matrix S and a hybrid matrix A; sources related to muscle artifacts in the source estimation matrix are recognized; muscle artifacts in the source estimation matrix are removed, the sources recognized to be muscle artifacts are set to be zero to obtain a source estimation matrix S' with muscleartifacts eliminated, and a multi-channel EEG signal X'=A*S' with muscle artifacts removed is obtained through reconstruction; all rows of the multi-channel EEG signal X' are summated, and a single-channel EEG signal x' with muscle artifacts removed can be finally obtained. By means of the method, EEG information is retained as far as possible while the muscle artifacts are removed, and thus the accuracy of EEG signal analysis is improved.

Description

technical field [0001] The invention relates to the technical field of electroencephalogram signal processing, in particular to a method for automatically removing myoelectric artifacts in single-channel electroencephalogram signals. Background technique [0002] Electroencephalogram (Electroencephalograph, EEG) is the reflection of the electrophysiological activity of brain nerve cells on the scalp. Due to its high temporal resolution, non-invasiveness, low cost, and suitability for long-term monitoring, EEG has been widely used to study brain function and pathological brain mechanisms. But while EEG records brain activity, it also records electrical signals produced by other activities outside the brain, and these recorded electrical signals that are not produced by brain activity are called artifacts. Common artifacts include Electrooculogram (EOG), Electromyography (EMG), Electrocardiography (ECG), and noise caused by circuits or external devices. Compared with other t...

AI Technical Summary

Problems solved by technology

Artifact rejection is to remove the entire EEG period containing artifacts. The main disadvantage of this method is that it will also eliminate important EEG information, resulting in data loss; and artifact correction refers to the use of various methods to remove artifact components while retaining as much EEG information as possible

Artifact correction technology mainly includes: 1) filtering, which removes EMG artifacts in EEG signals by using filters of specific frequency bands; Therefore, while removing myoelectric artifacts, a part of the EEG signal will also be lost.

2) Wavelet transform (wavelet transform, WT), the observed signal is decomposed into a series of basis functions by wavelet transform, and then the coefficients are thresholded to remove myoelectric artifacts, and finally the pure EEG signal is obtained by wavelet reconstruction; but wavelet Transformation methods require extensive experimentation to select appropriate wavelet basis functions and decomposition levels

EMD is a data-driven method for processing non-stationary, nonlinear, and stochastic processes, so it is very suitable for EEG signal analysis and processing, but EMD has high computational complexity and generally takes several minutes, which is not easy. suitable for real-time processing

Images