Method for detecting electroencephalogram signal electromyogram artifacts

Abstract

The present application proposes a method for detecting myoelectric artifacts in EEG signals, which includes: dividing the EEG signal into frames and calculating the myoelectric artifact detection index of each frame; the myoelectric artifact detection index of each frame is the product of the first-order difference zero-crossing rate and relative energy of the frame; the myoelectric artifact detection index of each frame constitutes an EEG artifact detection index curve; the frame corresponding to the portion of the EEG artifact detection index curve that is greater than a predetermined threshold is the frame where the EEG artifact is located. The present application realizes the detection and positioning of the myoelectric artifacts emitted in the EEG signal by combining the first-order difference zero-crossing rate and relative energy, laying the foundation for the overall improvement of the filtering of myoelectric artifacts in long-term EEG signals.

Description

Technical Field

[0001] The present application relates to electroencephalogram (EEG) signal processing technology, and in particular to a method for detecting long-term EEG electromyographic artifacts. Background Art

[0002] Electroencephalogram (EEG) monitoring is widely used in various clinical and research scenarios, including sleep studies, epileptic seizure monitoring, and cognitive function research. In these cases, EEG signals are easily affected by various artifacts, including oculoculographic artifacts, myoelectric artifacts, and cardiac artifacts. However, the appearance of these artifacts is not always continuous, but rather occurs irregularly and occasionally. For example, the myoelectric artifacts observed in sleep studies are mainly caused by occasional movements of the head, limbs, jaw, etc. during sleep, and do not appear continuously. If traditional artifact removal methods such as independent component analysis, canonical correlation analysis, modal decomposition, and wavelet decomposition are used to directly process artifacts on long-term EEG signals during sleep, the amount of data is extremely large, which leads to extremely low efficiency.

[0003] Therefore, for long-term EEG signal monitoring application scenarios similar to sleep research, if there are occasional EMG artifacts in the EEG signals, if a simple and fast method is used to first detect and identify the EMG artifacts, then the subsequent EMG artifact removal will only be performed in the artifact occurrence interval, which can improve the overall efficiency of removing EMG artifacts in the EEG signals. Summary of the Invention

[0004] In view of the above problems, the present application aims to propose a method for detecting myoelectric artifacts of EEG signals, which can simply and quickly detect myoelectric artifacts.

[0005] The method for detecting electroencephalogram (EEG) and myoelectricity (EMG) artifacts of the present application comprises:

[0006] The EEG signal is divided into frames, and the EMG artifact detection index of each frame is calculated; the EMG artifact detection index of each frame is the product of the first-order difference zero-crossing rate and the relative energy of the frame;

[0007] The EMG artifact detection index of each frame constitutes an EMG artifact detection index curve;

[0008] The frame corresponding to the portion of the electromyographic artifact detection index curve that is greater than a predetermined threshold is the frame where the electromyographic artifact is located.

[0009] Preferably, for each frame, the first-order difference zero-crossing rate of the frame is the number of first-order difference zero-crossing times of the frame.

[0010] Preferably, for each frame, the relative energy of the frame is the energy of the mean line of the upper envelope and the lower envelope relative to the frame; the calculation formula is as follows:

[0011] ;

[0012] ;

[0013] in, is the electromyographic artifact detection index, is the mean line, is the upper envelope, is the lower envelope, and N is the frame length.

[0014] Preferably, the EEG signal is a single-channel signal.

[0015] Preferably, the EEG signal includes multiple channel signals; for each channel signal, the electromyographic artifact detection index of each frame of the channel signal is calculated respectively; the product of the electromyographic artifact detection indexes of the corresponding frames of each channel signal is the electromyographic artifact detection index of the corresponding frame of the EEG signal.

[0016] Preferably, the EEG signal is framed using a Hanning window.

[0017] Preferably, the frame length is , the frame shift is , where Fs is the sampling rate of the EEG signal.

[0018] This application detects myoelectric artifacts by calculating the short-term first-order difference zero-crossing rate and short-term relative energy of single-channel or multi-channel EEG signals. The proposed method, with its simplicity and speed, can be applied to identify sporadic myoelectric artifacts in long-term EEG, improving the efficiency of filtering out myoelectric artifacts in long-term EEG. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Schematic diagram of the original EEG signal, upper and lower envelopes, and the mean lines of the upper and lower envelopes;

[0020] FIG2 is a schematic diagram of the original EEG signal and its short-time zero-crossing rate, short-time relative energy, and short-time relative energy*short-time zero-crossing rate;

[0021] Figure 3 For Figure 2 Schematic diagram of the vertical axis stretching of d in the middle;

[0022] Figure 4 ZcrMulEnengy calculated for the three EEG channels. DETAILED DESCRIPTION

[0023] Taking the EEG of polysomnography as an example, the long-term EEG is divided into frames and windowed, and the first-order difference zero-crossing rate and relative energy of each frame are calculated. The window function is the Hanning window, and the frame length is , the frame shift is , where Fs is the sampling rate of the EEG signal, here .

[0024] First-order difference zero-crossing rate Refers to the number of times the first-order difference crosses zero in each frame, that is, the number of positive and negative switching of the first-order difference. The initial value is 0. The specific calculation process is as follows: for the Mth frame signal after framing , find its first-order difference ,when When the M-th frame signal of The value is increased by 1. Traverse the M-th frame signal , that is, the M-th frame signal is obtained of .

[0025] Relative energy The relative energy of each frame is calculated. The relative energy here is to first calculate the mean line of the upper and lower envelopes of the signal, and then use this mean line as a reference to calculate the signal energy. Specifically, Figure 1 As shown, the blue line is the original EEG signal, and the upper and lower envelopes are the red lines. and green lines , the mean line is the black line:

[0026] (Formula 1)

[0027] Relative energy is also the same as first dividing the original EEG into frames and adding windows to obtain the Mth frame data , then calculate the M-th frame data :

[0028] (Equation 2)

[0029] Where N is the frame length.

[0030] The method of the present application combines the short-time zero-crossing rate and the short-time relative energy to realize the myoelectric artifact recognition in the EEG signal, namely the myoelectric artifact detection index

[0031]

[0032] like Figure 2 As shown, a is the original EEG signal, b is calculated by 1 per frame The waveform curve is formed by connecting the values, c is calculated 1 per frame The waveform curve is formed by connecting the values, d is calculated for each frame The waveform curve is formed by connecting the values. Figure 2 It can be seen from d that the location of the electromyographic artifact is The value of becomes significantly larger, and the threshold is set to 0.5*10 5 After that, the location of the EMG artifact can be detected.

[0033] Although the combination of short-time zero-crossing rate and short-time relative energy, combined with the threshold (1*10 5 ) can locate the position of the electromyographic artifact, but for the accurate determination of the electromyographic artifact range, Figure 2 For Figure 2 The identification of the green part of the electromyographic artifact in a is not sensitive enough, such as Figure 3 As shown, it is Figure 2 The longitudinal stretch of d in the middle shows that although 10 seconds to 11 seconds is also the range of myoelectric artifacts, The value is not prominent compared to the non-electromyographic artifact part, which means that the 10-11 second electromyographic artifact part cannot be identified by the threshold. If the threshold is too large (such as 0.5*10 5 ), can only detect the electromyographic artifact interval of 9-10 seconds, and the electromyographic artifact of 10-11 seconds is missed; the threshold is too small (such as 2*10 4 ), although the EMG artifact interval of 10-11 meters can be detected, the EEG signal of 5-6 seconds is mistakenly detected as EMG artifact.

[0034] Therefore, we try to take the same treatment on EEG signals of multiple channels, and then multiply the results of multiple channels. The cumulative effect will make the identification of EMG artifacts more accurate. Figure 4 As shown, this is the EEG signal of 3 channels It can be seen that the three channels calculated Value, relative to the clean EEG area, within the range of EMG artifact occurrence The value is very prominent, by setting the threshold to 2*10 14 , the position where the EMG artifact occurs can be detected at approximately 9-11 seconds. This shows that the cumulative calculation of the three channels is more sensitive and accurate in identifying EMG artifacts.

[0035] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, and documents, cited in this application is hereby incorporated by reference in its entirety. This includes any application history documents that are inconsistent with or conflicting with the present application, including any documents (currently or subsequently attached to this application) that limit the broadest scope of the claims of this application. It should be noted that if the descriptions, definitions, and / or terminology used in the accompanying materials are inconsistent or conflicting with the present application, the descriptions, definitions, and / or terminology used in this application will control.

[0036] Finally, it should be understood that the embodiments described in this application are merely illustrative of the principles of the embodiments of this application. Other variations may also fall within the scope of this application. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this application may be considered consistent with the teachings of this application. Accordingly, the embodiments of this application are not limited to the embodiments explicitly introduced and described in this application.

Claims

1. A method for detecting electroencephalogram (EEG) and myoelectricity (EMG) artifacts, comprising: Divide the EEG signal into frames and calculate the EMG artifact detection index of each frame; The EMG artifact detection index of each frame is the product of the first-order difference zero-crossing rate and the relative energy of the frame; The EMG artifact detection index of each frame constitutes an EMG artifact detection index curve; The frame corresponding to the portion of the electromyographic artifact detection index curve that is greater than a predetermined threshold is the frame where the electromyographic artifact is located; For each frame, the first-order difference zero-crossing rate of the frame is the number of first-order difference zero-crossings of the frame. The number of first-order difference zero-crossings is the number of positive and negative switching of the first-order difference. The initial value of the number of first-order difference zero-crossings is 0. For the M-th frame signal X(n) after framing, find its first-order difference diff_X(n)=diff(X(n)). When diff_X(n)*diff_X(n-1)<0, add 1 to the ShortZcr value of the M-th frame signal X(n). After traversing the M-th frame signal X(n), the first-order difference zero-crossing rate of the M-th frame signal X(n) is obtained. For each frame, the relative energy of the frame is the energy of the mean line of the upper envelope and lower envelope relative to the frame; the calculation formula is as follows: ； ； in, For relative energy, the original EEG is first framed and windowed to obtain the M-th frame data X(n), and then the ShortEnergy of the M-th frame data X(n) is calculated; is the mean line, is the upper envelope, is the lower envelope, and N is the frame length.

2. The method for detecting electroencephalogram (EEG) and myoelectricity (EMG) artifacts according to claim 1, wherein: The EEG signal is a single-channel signal.

3. The method for detecting EEG and myoelectric artifacts according to claim 1, wherein: The EEG signal includes multiple channel signals; for each channel signal, the EMG artifact detection index of each frame of the channel signal is calculated respectively; the product of the EMG artifact detection indexes of the corresponding frames of each channel signal is the EMG artifact detection index of the corresponding frame of the EEG signal.

4. The method for detecting EEG and myoelectric artifacts according to claim 1, wherein: The EEG signal is framed using the Hanning window.

5. The method for detecting EEG and myoelectric artifacts according to claim 4, wherein: The frame length is , the frame shift is , where Fs is the sampling rate of the EEG signal.

Images