阅读说明：本技术 基于蝶腭神经节刺激的脑电信号特征提取方法 (Electroencephalogram signal feature extraction method based on sphenopalatine ganglion stimulation ) 是由 俞孝儒 徐文龙 于 2021-01-19 设计创作，主要内容包括：本发明提供了一种基于蝶腭神经节刺激的脑电信号特征提取方法,包括以下步骤：步骤1：利用超声波通过两侧鼻翼实现对蝶腭神经的刺激；步骤2：在头皮上放置脑电采集电极,利用脑电采集装置记录不同蝶腭神经节刺激阶段的脑电信号；步骤3：进行脑电信号预处理操作,对经预处理后的脑电信号进行脑电相对功率特征提取和脑电有效连接特征提取；步骤4：根据所述脑电相对功率特征和脑电有效连接特征评估超声波对蝶腭神经的刺激效果。(The invention provides an electroencephalogram signal feature extraction method based on sphenopalatine ganglion stimulation, which comprises the following steps: step 1: the stimulation of sphenopalatine nerves is realized through the nasal wings at two sides by using ultrasonic waves; step 2: placing an electroencephalogram acquisition electrode on the scalp, and recording electroencephalogram signals of different sphenopalatine ganglion stimulation stages by using an electroencephalogram acquisition device; and step 3: performing electroencephalogram signal preprocessing operation, and performing electroencephalogram relative power feature extraction and electroencephalogram effective connection feature extraction on the preprocessed electroencephalogram signals; and 4, step 4: and evaluating the stimulation effect of the ultrasonic waves on the sphenopalatine nerve according to the electroencephalogram relative power characteristic and the electroencephalogram effective connection characteristic.)

1. A method for extracting electroencephalogram signal features based on sphenopalatine ganglion stimulation is characterized by comprising the following steps: the method comprises the following steps:

step 1: the stimulation of sphenopalatine nerves is realized through the nasal wings at two sides by using ultrasonic waves;

step 2: placing an electroencephalogram acquisition electrode on the scalp, and recording electroencephalogram signals of different sphenopalatine ganglion stimulation stages by using an electroencephalogram acquisition device;

and step 3: performing electroencephalogram signal preprocessing operation, and performing electroencephalogram relative power feature extraction and electroencephalogram effective connection feature extraction on the preprocessed electroencephalogram signals;

and 4, step 4: and evaluating the stimulation effect of the ultrasonic waves on the sphenopalatine nerve according to the electroencephalogram relative power characteristic and the electroencephalogram effective connection characteristic.

2. The sphenopalatine ganglion stimulation-based electroencephalogram signal feature extraction method of claim 1, which is characterized in that: the electroencephalogram signal preprocessing operation comprises the following steps: performing band-pass filtering on the electroencephalogram signals, wherein the pass band frequency is 0.5-80 Hz; performing band elimination filtering on the electroencephalogram signals, wherein the notch frequency is 50 Hz; reconstructing an electroencephalogram signal reference electrode as an average reference; processing the electroencephalogram signals by adopting a pseudo shadow space reconstruction technology, removing bad channels and cutting bad data; processing the electroencephalogram signals by adopting an independent component analysis technology, and removing eye movement artifacts and myoelectricity artifacts; dividing the EEG signal into 5 signal rhythms according to frequency bands, wherein the signal rhythms include delta rhythm, theta rhythm, alpha rhythm, beta rhythm and gamma rhythm.

3. The sphenopalatine ganglion stimulation-based electroencephalogram signal feature extraction method of claim 1, which is characterized in that: the extraction of the electroencephalogram relative power features in the step 4 specifically comprises the following steps:

calculating to obtain a power spectral density value of the electroencephalogram signal by adopting a Welch power spectrum estimation algorithm;

calculating the relative power characteristic of the brain electrical signal by adopting the following formula:

wherein, welch (f)_s，f_e) Representing frequency from f_sTo f_eTo the sum of the power density values.

4. The sphenopalatine ganglion stimulation-based electroencephalogram signal feature extraction method of claim 3, which is characterized in that: the Welch power spectrum estimation algorithm adopts a Hanning window function with 4-second duration, and the overlapping rate is 50%.

5. The sphenopalatine ganglion stimulation-based electroencephalogram signal feature extraction method of claim 1, which is characterized in that: the extraction of the effective connection features of the electroencephalogram in the step 4 specifically comprises the following steps:

establishing a multivariable autoregressive model of a sliding window, wherein the formula is as follows:

wherein X (t) ═ X₁(t)，X₂(t)，…，X_i(t)]^TIs an electroencephalogram signal with the number of electrodes being iTime sequence, A (n) is an i multiplied by i coefficient matrix, E (t) is a white noise residual term of the electroencephalogram signal, and p is a multivariable autoregressive model order;

and converting the multivariable autoregressive model of the sliding window into a frequency domain by adopting a frequency domain conversion algorithm, wherein the formula is as follows:

wherein H (f) is a transfer matrix, and f is frequency;

the effective connection characteristic of the electroencephalogram signal is calculated by adopting a direct directional transfer function, and the formula is as follows:

wherein, S (f) ═ X (f)^*K represents the total number of electrodes, and i and j represent electrode numbers.

6. The sphenopalatine ganglion stimulation-based electroencephalogram signal feature extraction method of claim 5, which is characterized in that: the order of the multivariate autoregressive model is determined by a Hannan Quinula information criterion algorithm.

7. The sphenopalatine ganglion stimulation-based electroencephalogram signal feature extraction method of claim 1, which is characterized in that: the specific implementation method of the step 1 comprises the following steps:

smearing ultrasonic coupling agent on two sides of the nose wing; respectively attaching ultrasonic transducers to the two nasal ala sides; starting an ultrasonic transducer to apply ultrasonic stimulation with set ultrasonic frequency, set pulse repetition frequency, set pulse length, set stimulation duration and set intensity;

setting the ultrasonic frequency to be 150kHz to 700 kHz; the set pulse repetition frequency in the step 3 is 20Hz to 2000 Hz; the set pulse length is 0.1ms to 100 ms; the set stimulation time is 1-15 minutes; the set intensity is 0.1W/cm2 to 15W/cm 2.

8. The sphenopalatine ganglion stimulation-based electroencephalogram signal feature extraction method of claim 1, which is characterized in that: in step 2, the electroencephalogram signals of different sphenopalatine ganglion stimulation stages comprise: electroencephalogram signals recorded 30 minutes before the sphenopalatine ganglion is stimulated; stimulating the sphenopalatine ganglion and simultaneously recording the electroencephalogram signals; and (3) recording the electroencephalogram signals 30 minutes after the sphenopalatine ganglion stimulation is finished.

Technical Field

The invention belongs to the field of biomedical information processing, and particularly relates to an electroencephalogram signal feature extraction method based on sphenopalatine ganglion stimulation.

Background

The sphenopalatine ganglion is the largest extracranial parasympathetic ganglion of the human body, and its branches are widely distributed in the orbital, oral and nasal cavities. Studies have shown that nerve stimulation of the sphenopalatine ganglion can improve the primary headache and rhinitis conditions. The traditional sphenopalatine ganglion stimulation technology is generally used in invasive stimulation methods such as electrical stimulation, acupuncture stimulation and the like, and has the side effects of muscle spasm, infection of invasive parts and the like. The sphenopalatine ganglion stimulation technology has good curative effect on non-allergic rhinitis, migraine and other partial inflammatory diseases; the nerve stimulation technology is researched by using resting state functional magnetic resonance imaging, and the intrinsic functional activity of a brain marginal system and a primary sensory system is obviously improved after stimulation. Brain electrical activity is the spontaneous rhythmic potential changes of the cerebral cortex. The electrical brain activity recorded on the surface of the scalp with an electroencephalograph is called an electroencephalogram. Studies have demonstrated that electroencephalograms located on the scalp can record electrical signals from deep brain regions.

However, the above prior art also has some problems: 1. the sphenopalatine ganglion stimulation technology in the prior art has the side effects of muscle spasm, wound infection and the like; 2. lack of an effective feature extraction approach makes it difficult to assess the reliability of the stimulation.

Disclosure of Invention

In view of the above, in order to overcome the problems of adverse effects and the like caused by nerve stimulation in the prior art, an electroencephalogram signal feature extraction method based on sphenopalatine ganglion stimulation in an ultrasonic stimulation mode is provided.

The technical scheme of the invention is to provide an electroencephalogram signal feature extraction method based on sphenopalatine ganglion stimulation, which comprises the following steps:

step 1: the stimulation of sphenopalatine nerves is realized through the nasal wings at two sides by using ultrasonic waves;

step 2: placing an electroencephalogram acquisition electrode on the scalp, and recording electroencephalogram signals of different sphenopalatine ganglion stimulation stages by using an electroencephalogram acquisition device;

and step 3: performing electroencephalogram signal preprocessing operation, and performing electroencephalogram relative power feature extraction and electroencephalogram effective connection feature extraction on the preprocessed electroencephalogram signals;

and 4, step 4: and evaluating the stimulation effect of the ultrasonic waves on the sphenopalatine nerve according to the electroencephalogram relative power characteristic and the electroencephalogram effective connection characteristic.

Optionally, the electroencephalogram signal preprocessing operation includes: performing band-pass filtering on the electroencephalogram signals, wherein the pass band frequency is 0.5-80 Hz; performing band elimination filtering on the electroencephalogram signals, wherein the notch frequency is 50 Hz; reconstructing an electroencephalogram signal reference electrode as an average reference; processing the electroencephalogram signals by adopting a pseudo shadow space reconstruction technology, removing bad channels and cutting bad data; processing the electroencephalogram signals by adopting an independent component analysis technology, and removing eye movement artifacts and myoelectricity artifacts; dividing the EEG signal into 5 signal rhythms according to frequency bands, wherein the signal rhythms include delta rhythm, theta rhythm, alpha rhythm, beta rhythm and gamma rhythm.

Optionally, the extracting of the electroencephalogram relative power feature in step 4 specifically includes:

calculating to obtain a power spectral density value of the electroencephalogram signal by adopting a Welch power spectrum estimation algorithm;

calculating the relative power characteristic of the brain electrical signal by adopting the following formula:

wherein, welch (f)_s，f_e) Representing frequency from f_sTo f_eTo the sum of the power density values.

Optionally, the Welch power spectrum estimation algorithm adopts a hanning window function with a duration of 4 seconds, and the overlapping rate is 50%.

Optionally, the extraction of the electroencephalogram effective connection features in step 4 specifically comprises:

establishing a multivariable autoregressive model of a sliding window, wherein the formula is as follows:

wherein X (t) ═ X₁(t)，X₂(t)，…，X_i(t)]^TThe method comprises the steps of A (n) is an i multiplied by i coefficient matrix, E (t) is a white noise residual term of the electroencephalogram signal, and p is a multivariable autoregressive model order;

and converting the multivariable autoregressive model of the sliding window into a frequency domain by adopting a frequency domain conversion algorithm, wherein the formula is as follows:

wherein H (f) is a transfer matrix, and f is frequency;

the effective connection characteristic of the electroencephalogram signal is calculated by adopting a direct directional transfer function, and the formula is as follows:

wherein, S (f) ═ X (f)^*K represents the total number of electrodes, and i and j represent electrode numbers.

Optionally, the order of the multivariate autoregressive model is determined by a hannan quinuclidine information criterion algorithm.

Optionally, the specific implementation method of step 1 is:

smearing ultrasonic coupling agent on two sides of the nose wing; respectively attaching ultrasonic transducers to the two nasal ala sides; starting an ultrasonic transducer to apply ultrasonic stimulation with set ultrasonic frequency, set pulse repetition frequency, set pulse length, set stimulation duration and set intensity;

setting the ultrasonic frequency to be 150kHz to 700 kHz; the set pulse repetition frequency in the step 3 is 20Hz to 2000 Hz; the set pulse length is 0.1ms to 100 ms; the set stimulation time is 1-15 minutes; the set intensity is 0.1W/cm2 to 15W/cm 2.

Optionally, in step 2, the electroencephalogram signals of the different sphenopalatine ganglion stimulation stages include: electroencephalogram signals recorded 30 minutes before the sphenopalatine ganglion is stimulated; stimulating the sphenopalatine ganglion and simultaneously recording the electroencephalogram signals; and (3) recording the electroencephalogram signals 30 minutes after the sphenopalatine ganglion stimulation is finished.

Compared with the prior art, the invention has the following advantages: the invention adopts a non-invasive method, avoids the damage and side effect generated by the traditional electrical stimulation and implanted electrical stimulation; compared with the traditional method, the sphenopalatine nerve stimulation method provided by the invention is simpler and more convenient to operate, has lower cost and does not need strong professional background; the sphenopalatine nerve stimulation method provided by the invention can be compatible with functional magnetic resonance imaging, electroencephalogram detection and electromyogram detection; the sphenopalatine nerve stimulation method provided by the invention is used for carrying out nerve stimulation on the nasal nerve distribution of the sphenopalatine ganglion by using an ultrasonic transduction technology for the first time, and the regulating and controlling effect of the ultrasonic stimulation based on the nose on the nerve is explored by using an electroencephalogram signal recording and analyzing method for the first time through a characteristic extraction method of an electroencephalogram signal.

Drawings

Fig. 1 is a schematic diagram of the principle of the present invention.

Fig. 2 is a diagram of a 10-10 international standard electroencephalogram lead system.

FIG. 3 is a graph showing the comparison of the mean variance of the relative power of the brain electrical signals before and after stimulation of the sphenopalatine ganglion in different brain regions of different frequency bands.

FIG. 4 is a color block diagram of partial brain leads (F3, F4, C3, C4, T7, T8, O1, O2) for brain electrical connection in different frequency bands.

Detailed Description

Preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, but the present invention is not limited to only these embodiments. The invention is intended to cover alternatives, modifications, equivalents and alternatives which may be included within the spirit and scope of the invention.

In the following description of the preferred embodiments of the present invention, specific details are set forth in order to provide a thorough understanding of the present invention, and it will be apparent to those skilled in the art that the present invention may be practiced without these specific details.

As shown in figure 1, before stimulation of the sphenopalatine ganglion, an ultrasonic couplant is coated on the nasal wing of a tested person, and then an ultrasonic transducer probe is attached to the nasal wing to perform ultrasonic stimulation on the tested person. The invention provides an electroencephalogram signal feature extraction method based on sphenopalatine ganglion stimulation, which comprises the following steps:

step 1: the stimulation of sphenopalatine nerves is realized through the nasal wings at two sides by using ultrasonic waves;

step 2: placing an electroencephalogram acquisition electrode on the scalp, and recording electroencephalogram signals of different sphenopalatine ganglion stimulation stages by using an electroencephalogram acquisition device;

and step 3: performing electroencephalogram signal preprocessing operation, and performing electroencephalogram relative power feature extraction and electroencephalogram effective connection feature extraction on the preprocessed electroencephalogram signals;

and 4, step 4: and evaluating the stimulation effect of the ultrasonic waves on the sphenopalatine nerve according to the electroencephalogram relative power characteristic and the electroencephalogram effective connection characteristic.

The electroencephalogram signal preprocessing operation comprises the following steps: performing band-pass filtering on the electroencephalogram signals, wherein the pass band frequency is 0.5-80 Hz; performing band elimination filtering on the electroencephalogram signals, wherein the notch frequency is 50 Hz; reconstructing an electroencephalogram signal reference electrode as an average reference; processing the electroencephalogram signals by adopting a pseudo shadow space reconstruction technology, removing bad channels and cutting bad data; processing the electroencephalogram signals by adopting an independent component analysis technology, and removing eye movement artifacts and myoelectricity artifacts; dividing the EEG signal into 5 signal rhythms according to frequency bands, wherein the signal rhythms include delta rhythm, theta rhythm, alpha rhythm, beta rhythm and gamma rhythm.

The extraction of the electroencephalogram relative power features in the step 4 specifically comprises the following steps:

calculating to obtain a power spectral density value of the electroencephalogram signal by adopting a Welch power spectrum estimation algorithm;

calculating the relative power characteristic of the brain electrical signal by adopting the following formula:

wherein, welch (f)_s，f_e) Representing frequency from f_sTo f_eTo the sum of the power density values.

The Welch power spectrum estimation algorithm adopts a Hanning window function with 4-second duration, and the overlapping rate is 50%.

The extraction of the effective connection features of the electroencephalogram in the step 4 specifically comprises the following steps:

establishing a multivariable autoregressive model of a sliding window, wherein the formula is as follows:

wherein X (t) ═ X₁(t)，X₂(t)，…，X_i(t)]^TThe method comprises the steps of A (n) is an i multiplied by i coefficient matrix, E (t) is a white noise residual term of the electroencephalogram signal, and p is a multivariable autoregressive model order;

and converting the multivariable autoregressive model of the sliding window into a frequency domain by adopting a frequency domain conversion algorithm, wherein the formula is as follows:

wherein H (f) is a transfer matrix, and f is frequency;

the effective connection characteristic of the electroencephalogram signal is calculated by adopting a direct directional transfer function, and the formula is as follows:

wherein, S (f) ═ X (f)^*K represents the total number of electrodes, and i and j represent electrode numbers.

The order of the multivariate autoregressive model is determined by a Hannan Quinula information criterion algorithm.

The specific implementation method of the step 1 comprises the following steps:

smearing ultrasonic coupling agent on two sides of the nose wing; respectively attaching ultrasonic transducers to the two nasal ala sides; starting an ultrasonic transducer to apply ultrasonic stimulation with set ultrasonic frequency, set pulse repetition frequency, set pulse length, set stimulation duration and set intensity;

setting the ultrasonic frequency to be 150kHz to 700 kHz; the set pulse repetition frequency in the step 3 is 20Hz to 2000 Hz; the set pulse length is 0.1ms to 100 ms; the set stimulation time is 1-15 minutes; the set intensity is 0.1W/cm2 to 15W/cm 2.

In step 2, the electroencephalogram signals of different sphenopalatine ganglion stimulation stages comprise: electroencephalogram signals recorded 30 minutes before the sphenopalatine ganglion is stimulated; stimulating the sphenopalatine ganglion and simultaneously recording the electroencephalogram signals; and (3) recording the electroencephalogram signals 30 minutes after the sphenopalatine ganglion stimulation is finished.

The distribution of the electroencephalogram signal acquisition electrodes follows a 10-10 international standard lead system, the sampling rate is set to be 500 Hz, and electroencephalogram signal recording is carried out on four brain areas of the frontal lobe, the parietal lobe, the temporal lobe and the occipital lobe of the brain, and the specific electrode positions are shown in figure 2. Where Cz is set as the reference electrode and Fpz is set as the ground electrode.

The electroencephalogram signals are recorded in a closed sound insulation chamber, and the indoor temperature is controlled at 24 ℃. The tested person needs to clean hair, and the conductive paste is injected at the position of the electrode to ensure that the impedance of the electrode is lower than 5k ohm in the recording process.

The electroencephalogram signal recording testee needs to avoid the activities of the face and the limbs as much as possible and keep clear and relaxed. The recording is divided into 2 phases: (1) 0.5 hours before the start of stimulation, record 5 minutes; (2) after 0.5 hour from the end of the stimulation, 5 minutes were recorded.

And preprocessing the electroencephalogram signal sample. The method comprises the following steps: (1) band-pass filtering is carried out for 0.5-45 Hz, and baseline drift and 50Hz power frequency interference are removed; (2) averaging reference; (3) removing bad channels and data segments by using pseudo shadow space reconstruction (ASR); (4) noise components such as eye movement and myoelectricity are decomposed and removed by Independent Component Analysis (ICA). (5) Dividing the electroencephalogram signals into 4 frequency bands: delta (1-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), beta (13-30 Hz).

The power spectral density of the electroencephalogram signal is obtained by a Welch method, a Hanning window with the time length of 4 seconds is selected, and the overlapping is 50%.

The relative power calculation formula of each frequency band is

Welch (f) in the formula_s，f_e) Indicating the frequency band from f_sTo f_eThe sum of the power density values, the calculation result is shown in fig. 3.

The method comprises the steps of calculating effective connection values of electroencephalogram signals in different frequency bands by adopting a direct directional transfer function (dDTF) based on a Glange causal relationship (GC), and firstly establishing a multivariable Autoregressive Model (AMVAR) of a sliding window.

AMVAR is defined as:

wherein X (t) ═ X₁(t)，X₂(t)，…，X_i(t)]^TIs an electroencephalogram signal time sequence with the number of electrodes being i,

a (n) is an i × i coefficient matrix, E (t) is a white noise residue, and p is the order of a multivariate autoregressive model determined by Hannan Quinula information criterion (HQ).

The multivariate autoregressive model is transformed into the frequency domain using fast fourier transforms,

X(f)＝H(f)E(f)

whereinF is the frequency, which is the transfer matrix of the system.

dDTF is defined as:

wherein the content of the first and second substances,

where k represents the number of channels.

P_ij(f) Is defined as:

wherein the content of the first and second substances,

S(f)＝X(f)X(f)^*。

finally, the dtf matrix of each frequency band is obtained as shown in fig. 4.

The result shows that the relative power and effective connection of the alpha frequency range electroencephalogram and the relative power of the beta frequency range are obviously enhanced after the sphenopalatine ganglion is stimulated. Relevant FMRI studies have demonstrated that nasal application of vibratory stimuli enhances the intrinsic functional activity of the limbic system and primary sensory system, that limbic system activity is positively correlated with power and effective connectivity enhancement in the alpha band of EEG signals, and that power enhancement in the beta band results from activation of the primary somatosensory and motor cortex. The invention applies the sphenopalatine ganglion stimulation and the EEG signal acquisition and analysis method to prove the internal functional activity promotion effect of the sphenopalatine ganglion stimulation on the marginal system and the primary sensory network from the angle of electroencephalogram. The application of the invention can provide a simpler, lower-cost and effective scheme for evaluating the stimulation effect of the sphenopalatine ganglion and evaluating the curative effect of the sphenopalatine ganglion on treating related diseases by stimulation. The above-described embodiments do not limit the scope of the present invention. Any modification, equivalent replacement, and improvement made within the spirit and principle of the above-described embodiments should be included in the protection scope of the technical solution.