CN112971786A - Apoplexy rehabilitation evaluation method based on brain electromyographic signal wavelet coherence coefficient - Google Patents
Apoplexy rehabilitation evaluation method based on brain electromyographic signal wavelet coherence coefficient Download PDFInfo
- Publication number
- CN112971786A CN112971786A CN202110168009.3A CN202110168009A CN112971786A CN 112971786 A CN112971786 A CN 112971786A CN 202110168009 A CN202110168009 A CN 202110168009A CN 112971786 A CN112971786 A CN 112971786A
- Authority
- CN
- China
- Prior art keywords
- signal
- brain
- wavelet coherence
- wavelet
- signals
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
- A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
- A61B5/6802—Sensor mounted on worn items
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
- A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
- A61B5/6802—Sensor mounted on worn items
- A61B5/6803—Head-worn items, e.g. helmets, masks, headphones or goggles
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7253—Details of waveform analysis characterised by using transforms
Abstract
The invention discloses a stroke rehabilitation evaluation method based on a brain electromyographic signal wavelet coherence coefficient, which belongs to the technical field of rehabilitation medicine and aims to solve the problems that the traditional rehabilitation treatment of a stroke patient is restricted by the subjective clinical experience of a therapist and the rehabilitation progress of the patient is difficult to reflect in real time at present; by constructing an extraction scheme combining electroencephalogram signals and electromyogram signals, the electroencephalogram signals and the electromyogram signals are processed and analyzed, signal characteristics are extracted to identify motion related information, the rehabilitation stage condition of a patient in a real-time state is obtained, and rehabilitation assessment based on the electroencephalogram signals and the electromyogram signals is realized. The method of the invention provides a reference index for the subsequent formulation of a rehabilitation scheme by analyzing the correlation between the coherence of the patient's deltoid muscle sEMG signal and the corresponding brain area EEG signal and the recovery of the upper limb motor function, classifies the two types of patients in the middle and later stroke recovery stages based on wavelet coherence coefficients, and constructs a two-classification evaluation model, wherein the classification accuracy reaches 88.9% +/-4.9%.
Description
Technical Field
The invention relates to a stroke rehabilitation assessment method based on electroencephalogram and electromyogram signals, in particular to a stroke rehabilitation assessment method based on a brain and electromyogram signal wavelet coherence coefficient, and belongs to the technical field of rehabilitation medicine.
Background
Stroke is a clinical syndrome of focal neurological impairment of the cerebral hemisphere or brain stem caused by acute cerebrovascular circulatory disturbance due to various reasons, and is commonly called stroke. China is one of the countries with the highest incidence rate of cerebral apoplexy worldwide, and the average age of cerebral apoplexy is in the trend of gradual rejuvenation. Cerebral hemorrhage or infarction has long been recognized as a leading cause of death and long-term disability worldwide.
When a patient becomes windy, interruption of blood supply to the brain, an increase in intracranial pressure, and toxic effects of the released blood can all cause severe damage to brain tissue. Depending on the location of the lesion, various physical dysfunctions may result, with consequences including muscle weakness, loss of sensation and cognitive deficits, all of which may have a significant impact on the quality of life of stroke survivors. Although tissue damage is generally irreversible, it has been demonstrated that bodily function can be partially restored by exploiting the remodelling ability of the brain, and that rehabilitation programs following stroke can play an important role in the healing process.
With the development of Brain-Computer Interface (BCI), the application of Brain electricity in the rehabilitation of stroke is more and more emphasized. The electroencephalogram signal and the surface electromyogram signal respectively contain body motion control information and functional response information of muscles to brain control intentions, and the rehabilitation state of a patient can be directly reflected. Therefore, an evaluation model aiming at stroke classification can be constructed by effectively integrating the characteristics of the signals, so that the disadvantages of the traditional evaluation are made up.
Disclosure of Invention
The purpose of the invention is: the stroke rehabilitation assessment method based on the brain electromyogram signal wavelet coherence coefficient is provided, a reference index is provided for subsequent rehabilitation scheme formulation by analyzing the correlation between the coherence of the deltoid muscle sEMG signal of a patient and the corresponding brain area EEG signal and the recovery of the motor function of the upper limb, the two types of patients in the middle stroke recovery period and the later stroke recovery period are classified based on the wavelet coherence coefficient, the optimal characteristic is extracted, an off-line model is trained by the optimal characteristic, and a two-classification assessment model is established by classification and recognition through the characteristic, so that the high classification accuracy of the late stroke recovery period and the middle stroke recovery period is realized.
In order to achieve the purpose, the invention adopts the following technical scheme: a stroke rehabilitation assessment method based on a brain electromyographic signal wavelet coherence coefficient comprises the following steps:
preparation work: s0, designing an experimental stimulation program through E-prime software, and presenting software as a brain and muscle electricity synchronous acquisition experimental paradigm, wherein the experiment comprises two design paradigms: 1 represents raising the left arm and 2 represents raising the right arm.
The method comprises the following operation steps: s1, synchronously acquiring the brain electromyographic signals: the examinee selects an experimental paradigm according to the hemiplegic side, wears the experimental equipment by using a dry electrode brain electricity cap and a myoelectricity electrode, executes tasks according to requirements, and synchronously acquires the deltoid sEMG signals and the corresponding brain area EEG signals when the examinee lifts the arm in the experimental process.
The dry electrode electroencephalogram cap is worn correctly according to the standard, the A1 and A2 electrodes are used as reference electrodes, the sampling frequency is 300Hz, and a DC direct current acquisition mode is adopted; the myoelectric electrodes adopt an expansion channel, the diameter of each electrode is 5mm, and the myoelectric electrodes are spaced by 2cm to form a differential mode for experiment; the subject executes the task and collects data according to the requirement, the upper arm of the affected side is lifted and kept for 5 seconds each time, and the rest is carried out for 10 seconds in the middle.
S2, preprocessing a brain myoelectric signal: extracting effective data from the collected brain-muscle electrical data in the S1 by using a moving average method and carrying out corresponding processing; filtering the electroencephalogram and electromyogram signals collected in the step S1, performing baseline correction on the filtered signals, performing denoising on signal data segments after the baseline correction by adopting a Mallat decomposition algorithm in wavelet transform in combination with a Brige-Massart threshold determination method, and removing ocular artifacts in EEG signals by adopting a second-order blind source separation method; the filtering of the brain electromyographic signals comprises filtering EEG signals within a filtering range of 0.5-45Hz, and filtering sEMG signals within a filtering range of 0.5-100 Hz.
The specific steps for extracting the effective data are as follows:
1) filtering the sEMG signal, wherein the filtering range is 5-50 Hz;
2) calculating the energy of the filtered signal, and setting a proper threshold value to detect the initial position of the movement;
3) detecting the moving segment, processing the signal by adopting a moving average method, and judging the starting point and the stopping point of the action through a threshold value;
4) intercepting data of 0-4 s from the collected original EEG signal and sEMG signal as effective data according to the starting position;
5) and performing wavelet denoising on the effective data, removing artifacts of an EEG signal, and improving the signal-to-noise ratio of the signal.
S3, finding optimal time period and frequency band characteristics: based on the processed effective data extracted in the S2, selecting the EEG signal of the C3 channel and the sEMG signal in the middle of the right arm deltoid muscle to carry out the same operationStep analysis, adopting R in brain-computer interface BCI system2And judging the optimal characteristics by a distinguishable degree method between the two classes to distinguish the rehabilitation stage, and analyzing and searching the optimal time period and frequency band as the characteristics for classification.
R2The discrimination formula for discriminating the distinguishable degree is:
in wavelet coherence estimation, if X1 and X2 represent features of two classes, L1 and L2 represent the number of samples of different classes, and the larger the two classes are distinguished, the larger R2The larger the value; mean represents mean and std represents standard deviation.
By comparing R2According to the upper limb scoring condition of the brunstrom scoring scale, dividing the patient into a middle recovery period (grade 3-4) and a later recovery period (grade 5-6), and analyzing and searching the optimal time period and the optimal frequency band which are 1500 ms-1800 ms of an alpha frequency band, 2200 ms-2500 ms of a beta frequency band and 2700 ms-3000 ms of a gamma frequency band.
S4, performing wavelet coherence analysis on the brain electromyographic signals: wavelet coherence analysis is carried out on the brain electromyographic signals preprocessed in the step S2,
wavelet coherence results from fourier coherence, defined as:
in the formula (I), the compound is shown in the specification,
SWXX(t, f) and SWYY(t, f) the same calculation formula is used; in the formula, SWXY(t, f) represents the scalar product between X and Y, WCo (t, f) is defined by time t and frequency f, and SWXY(t, f) are the same; wherein, ncy is [ t-delta/2, t + delta/2 ] in the formula of ncy/f]The number of cycles in time, generally speaking, ncy values are related to the signal length, short sequences have small values, and long sequences have large values; t is time and f is frequency.
The wavelet coherence analysis calculation shows that the wavelet coherence coefficient mean value of the late stroke recovery period in an alpha frequency band of 8-12 Hz is smaller than that of a patient in the middle stroke recovery period, the wavelet coherence coefficient mean value of the late stroke recovery period in a beta frequency band of 15-30 Hz and a gamma frequency band of 30-45 Hz is larger than that of the patient in the middle stroke recovery period, and the time of wavelet coherence coefficient difference is mainly distributed between 1500ms and 3500 ms; and (3) calculating the average value of the wavelet coherence coefficients in the optimal time-frequency domain found in the S3 to be used as a characteristic, carrying out rank sum test on the characteristic, finding out that the characteristic has obvious difference through analysis, namely the average value of the wavelet coherence coefficients between 2200ms and 2500ms of the beta frequency band, and using the average value of the wavelet coherence coefficients of the beta frequency band as the optimal characteristic.
S5, constructing a classification model by using the wavelet coherence coefficient mean value of the optimal time-frequency domain: taking 80% of sample data as a training set, performing off-line training on the training set data, and constructing and storing a training model; taking 20% of sample data as a test set, sending the sample data into an SVM classifier for classification, taking the wavelet coherence coefficient mean value of the optimal time-frequency domain obtained by analysis in S4 as a feature for classification and identification, and constructing a two-classification evaluation model; and obtaining the classification precision of the second classification through five-fold cross validation.
The invention has the beneficial effects that: based on the problems that the traditional rehabilitation treatment of the current stroke patient is restricted by the subjective clinical experience of a therapist and the rehabilitation progress of the patient is difficult to reflect in real time, a reference index is provided for the subsequent formulation of a rehabilitation scheme by analyzing the correlation between the coherence of the sEMG signal of the deltoid muscle of the patient and the EEG signal corresponding to the brain area and the recovery of the motor function of the upper limb, the two types of patients in the middle stroke recovery stage and the later stroke recovery stage are classified based on the wavelet coherence coefficient, the optimal characteristic is extracted, the off-line model is trained by the optimal characteristic, and the classified identification is carried out through the characteristic to establish a two-classification evaluation model, so that the high classification accuracy of the later stroke recovery stage and the middle stroke recovery.
Drawings
FIG. 1 is a flow chart of electroencephalogram data processing in the present invention;
FIG. 2 is a wavelet coherence time-frequency diagram of a subject in mid-stroke rehabilitation according to the present invention;
FIG. 3 is a wavelet coherence time-frequency diagram of a subject in late stroke rehabilitation according to the present invention;
FIG. 4 is a graph showing R of wavelet coherence coefficients of patients at different recovery periods in the present invention2A drawing;
fig. 5 is a table of wavelet coherence coefficient rank sum test calculated values for each frequency band.
Detailed Description
The invention is further explained below with reference to the figures and the embodiments.
Example (b): as shown in fig. 1 to 5, the stroke rehabilitation assessment method based on the brain electromyogram signal wavelet coherence coefficient according to the present invention includes the following steps:
preparation work: s0, designing an experimental stimulation program through E-prime software, and presenting software as a brain and muscle electricity synchronous acquisition experimental paradigm, wherein the experiment comprises two design paradigms: 1 represents raising the left arm and 2 represents raising the right arm.
The examinee sits on a comfortable chair, the eyes of the examinee stares at the front display, corresponding actions are completed according to screen prompts, in the experimental process, the examinee is required to avoid eye movement, swallowing and unnecessary limb actions as much as possible, meanwhile, the upper limb of the affected side is lifted according to the experimental paradigm, and the arm is lifted to the position parallel to the ground as much as possible.
The method comprises the following operation steps: s1, synchronously acquiring the brain electromyographic signals: the examinee selects an experimental paradigm according to the hemiplegic side, wears the experimental equipment by using a dry electrode brain electricity cap and a myoelectricity electrode, executes tasks according to requirements, and synchronously acquires the deltoid sEMG signals and the corresponding brain area EEG signals when the examinee lifts the arm in the experimental process.
Multi-channel electroencephalogram signal acquisition: the testee is provided with a DSI-24 electroencephalogram cap of the bearable sensing company, wherein 19 leads collect electroencephalogram signals, the electrode position adopts the international 10/20 system standard, an expansion channel collects surface Electromyography (EMG) signals at the middle part of the right arm deltoid, a dry electrode electroencephalogram cap is correctly worn on the head of the testee, the electrode is fully contacted with the scalp, the electroencephalogram collection system is checked at the same time, whether the impedance is reduced to a proper range is observed, and the A1 and A2 electrodes are used as reference electrodes; the sampling frequency is 300Hz, and the DC direct current acquisition mode is adopted.
Collecting electromyographic signals: collecting surface electromyographic signals of the middle part of the right arm deltoid by adopting an expansion channel, wiping the arm deltoid part by using a scrub test subject, and then attaching electrodes, wherein the diameter of the electrodes of the electromyographic electrodes is 5mm, and the electrodes are spaced by 2cm to form a differential mode for carrying out experiments; the examinee performs the task and collects the data according to the requirements after having a full rest, the upper arm on the affected side is lifted up and kept for 5 seconds each time, and the rest is carried out for 10 seconds in the middle, and after having a full rest.
According to the upper limb scoring condition of a brunnstrom scoring scale, patients are divided into three grades, namely a recovery early stage (grade 1-2), a recovery middle stage (grade 3-4) and a recovery later stage (grade 5-6), and synchronous analysis of the brain and muscle electricity is carried out aiming at the recovery middle stage and the recovery later stage of stroke.
S2, preprocessing a brain myoelectric signal: extracting effective data from the collected brain-muscle electrical data in the S1 by using a moving average method and carrying out corresponding processing; because the original electroencephalogram signals contain more noise and artifacts, the directly acquired signals cannot be directly used for research, data must be preprocessed, the signal-to-noise ratio is improved, the electroencephalogram signals acquired in S1 are filtered, the filtered signals are subjected to baseline correction, the noise removal is completed on signal data segments after the baseline correction by adopting a Mallat decomposition algorithm in wavelet transform in combination with a Brige-Massart threshold determination method, and the ocular artifacts in EEG signals are removed by adopting a second-order blind source separation method.
The specific steps for extracting the effective data are as follows:
1) filtering the sEMG signal, wherein the filtering range is 5-50 Hz;
2) calculating the energy of the filtered signal, and setting a proper threshold value to detect the initial position of the movement;
3) detecting the moving segment, processing the signal by adopting a moving average method, and judging the starting point and the stopping point of the action through a threshold value;
4) intercepting data of 0-4 s from the collected original EEG signal and sEMG signal as effective data according to the starting position;
5) and performing wavelet denoising on the effective data, removing artifacts of an EEG signal, and improving the signal-to-noise ratio of the signal.
Band-pass filtering: the original EEG signal is filtered using the EEGLAB toolbox with a filtering range of 0.5-45Hz, the original sEMG signal is filtered to 0.5-100Hz, and the filtered signal is baseline corrected.
Wavelet denoising: performing 8-layer wavelet decomposition on the signals by adopting a Mallat decomposition algorithm on the data fragments, determining the threshold value of the wavelet coefficient obtained by wavelet decomposition of each layer by using a Brige-Massart strategy, adjusting the wavelet coefficient obtained by wavelet decomposition of each layer according to the threshold value, reconstructing clutter signals by the adjusted wavelet coefficient of each layer, and subtracting the reconstructed clutter signals from the original signals to obtain pure electroencephalogram signals.
Removing the artifacts: manually removing waveforms with large interference, and removing ocular artifacts in the EEG signals by adopting a second-order blind source separation method.
S3, finding optimal time period and frequency band characteristics: based on the effective data extracted and processed in the S2, an EEG signal of a C3 channel and an sEMG signal in the middle of the right arm deltoid are selected for synchronous analysis, and an R is adopted in a brain-computer interface BCI system2And judging the optimal characteristics by a distinguishable degree method between the two classes to distinguish the rehabilitation stage, and analyzing and searching the optimal time period and frequency band as the characteristics for classification.
R2The discrimination formula for discriminating the distinguishable degree is:
in wavelet coherence estimation, if X1 and X2 represent features of two classes, L1 and L2 represent the number of samples of different classes, and the larger the two classes are distinguished, the larger R2The larger the value; mean represents the mean, std represents the standard deviation.
By comparing R2According to the upper limb scoring condition of the brunstrom scoring scale, dividing the patient into a middle recovery period (grade 3-4) and a later recovery period (grade 5-6), and analyzing and searching the optimal time period and the optimal frequency band which are 1500 ms-1800 ms of an alpha frequency band, 2200 ms-2500 ms of a beta frequency band and 2700 ms-3000 ms of a gamma frequency band.
S4, performing wavelet coherence analysis on the brain electromyographic signals: performing wavelet coherence analysis on the brain-electromyographic signals preprocessed in the step S2, wherein the traditional coherence analysis is completed based on Fourier transform, the premise is that the signals are stable, but EEG signals and sEMG signals are non-stable signals, the wavelet coherence can locate an instantaneous frequency domain, the coherence between the two signals can be calculated at any frequency,
wavelet coherence results from fourier coherence, defined as:
in the formula (I), the compound is shown in the specification,
the calculation formulas of SWXX (t, f) and SWYY (t, f) are the same; in the formula, SWXY(t, f) represents the scalar product between X and Y, WCo (t, f) is defined by time t and frequency f, and SWXY(t, f) are the same; wherein, ncy is [ t-delta/2, t + delta/2 ] in the formula of ncy/f]The number of cycles in time, ncy, is related to the signal length, short sequences are small, long sequences are large, t is time, and f is frequency. The idea of wavelet coherence is that a narrow integration window is used for calculating a high-frequency signal, the wavelet coherence can embody the coherent time domain information between two signals, and the instantaneous characteristics between the two signals can be well represented.
Through data screening, 9 groups of patients in the middle of stroke recovery and 9 groups of patients in the later period of stroke recovery are finally selected, and because muscles of the patients in the initial period can not be controlled completely, the synchronous data of the brain and muscle electricity of the patients in the initial period of stroke recovery are not collected, and the method mainly analyzes the difference between the middle period and the later period.
Fig. 2 and 3 are time-frequency graphs of wavelet coherence means of all patients, fig. 2 is a patient in the middle of stroke rehabilitation, and fig. 3 is a patient in the late stage of stroke rehabilitation. As can be seen from the graphs in FIGS. 2 and 3, in the alpha band of 8-12 Hz, the mean value of the wavelet coherence coefficient in the late stage of stroke recovery is smaller than that of the patient in the middle stage of stroke recovery, while in the beta band of 15-30 Hz and the gamma band of 30-45 Hz, the mean value of the wavelet coherence coefficient in the late stage of stroke recovery is larger than that of the patient in the middle stage of stroke recovery.
FIG. 4 isR of two recovery phase data2Value R2Values are expressed in gradient colors, with the more red the color, the greater the difference. It can be seen from the figure that the time distribution of the wavelet coherence coefficient difference is mainly between 1500ms and 3500ms, and the wavelet coherence coefficient difference is the largest between 1500ms and 1800ms of the alpha band, 2200ms to 2500ms of the beta band, and 2700ms to 3000ms of the gamma band.
Fig. 5 shows that the average value of the wavelet coherence coefficients in the specific time-frequency domain is used as a feature, the average value of the wavelet coherence coefficients in the optimal time-frequency domain found in S3 is used as a feature, the rank and the test are performed on the feature, the analysis shows that the wavelet coherence coefficients with significant difference between the features are the average value of the wavelet coherence coefficients between 2200ms and 2500ms of the beta frequency band, the beta frequency band is used as the optimal frequency band, and the feature is used for classification and identification in the following process.
S5, constructing a classification model by using the wavelet coherence coefficient mean value of the optimal time-frequency domain: taking 80% of sample data as a training set, performing off-line training on the training set data, and constructing and storing a training model; taking 20% of sample data as a test set, sending the sample data into an SVM classifier for classification, taking the wavelet coherence coefficient mean value of the optimal frequency band and time period obtained by analysis in S4 as a feature for classification and identification, and constructing a two-classification evaluation model; and obtaining the classification precision of the second classification through five-fold cross validation.
The wavelet coherence can embody coherent time domain information between two signals and can well represent transient characteristics between the two signals.
An offline classification model is constructed based on wavelet coherence characteristics, and rehabilitation assessment classification of patients in the middle and later stroke recovery periods is achieved.
Dividing the electroencephalogram and electromyography synchronous data of patients in the middle and later stroke recovery periods according to the proportion of 1:1, wherein 9 patients in the middle stroke recovery period and 9 patients in the later stroke recovery period are selected, wavelet coherence coefficient mean values of corresponding frequency bands and time periods are selected as classification features, 80% of sample data are used as a training set, 20% of sample data are used as a test set, the training set and the test set are sent to an SVM classifier for classification, 5-fold cross validation is carried out, and finally the classification accuracy is 88.9% +/-4.9%.
Based on the problems that the traditional rehabilitation treatment of the current stroke patient is restricted by the subjective clinical experience of a therapist and the rehabilitation progress of the patient is difficult to reflect in real time, a reference index is provided for the subsequent formulation of a rehabilitation scheme by analyzing the correlation between the coherence of the sEMG signal of the deltoid muscle of the patient and the EEG signal corresponding to the brain area and the recovery of the motor function of the upper limb, the two types of patients in the middle stroke recovery stage and the later stroke recovery stage are classified based on the wavelet coherence coefficient, the optimal characteristic is extracted, the off-line model is trained by the optimal characteristic, and the classified identification is carried out through the characteristic to establish a two-classification evaluation model, so that the high classification accuracy of the later stroke recovery stage and the middle stroke recovery.
The above description is only for the purpose of illustrating the technical solutions of the present invention and not for the purpose of limiting the same, and other modifications or equivalent substitutions made by those skilled in the art to the technical solutions of the present invention should be covered within the scope of the claims of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims (7)
1. A stroke rehabilitation assessment method based on a brain electromyographic signal wavelet coherence coefficient is characterized by comprising the following steps: the method comprises the following steps:
preparation work: s0, designing an experimental stimulation program through E-prime software, and presenting software as a brain and muscle electricity synchronous acquisition experimental paradigm, wherein the experiment comprises two design paradigms: 1 represents lifting the left arm, and 2 represents lifting the right arm;
the method comprises the following operation steps: s1, synchronously acquiring the brain electromyographic signals: the method comprises the following steps that a testee selects an experimental paradigm according to the hemiplegic side, wears a dry electrode brain electricity cap and a myoelectricity electrode as experimental test equipment, executes tasks according to requirements, and synchronously collects a deltoid sEMG signal and a corresponding brain area EEG signal when the testee lifts the arm in the experimental process;
s2, preprocessing a brain myoelectric signal: extracting effective data from the collected brain-muscle electrical data in the S1 by using a moving average method and carrying out corresponding processing; filtering the electroencephalogram and electromyogram signals collected in the step S1, performing baseline correction on the filtered signals, performing denoising on signal data segments after the baseline correction by adopting a Mallat decomposition algorithm in wavelet transform in combination with a Brige-Massart threshold determination method, and removing ocular artifacts in EEG signals by adopting a second-order blind source separation method;
s3, finding optimal time period and frequency band characteristics: based on the effective data extracted and processed in the S2, an EEG signal of a C3 channel and an sEMG signal in the middle of the right arm deltoid are selected for synchronous analysis, and an R is adopted in a brain-computer interface BCI system2Distinguishing the rehabilitation stage by distinguishing the optimal characteristics by distinguishing the degree method between the two types, and analyzing and searching the optimal time period and frequency band as the characteristics for classification;
by comparing R2According to the upper limb scoring condition of a brunstrom scoring scale, dividing the patient into a middle recovery period (grade 3-4) and a later recovery period (grade 5-6), and analyzing and searching the optimal time period and the optimal frequency band which are 1500 ms-1800 ms of an alpha frequency band, 2200 ms-2500 ms of a beta frequency band and 2700 ms-3000 ms of a gamma frequency band;
s4, performing wavelet coherence analysis on the brain electromyographic signals: wavelet coherence analysis is carried out on the brain electromyographic signals preprocessed in the step S2,
wavelet coherence results from fourier coherence, defined as:
in the formula (I), the compound is shown in the specification,
SWXX(t, f) and SWYYThe formula for calculating (t, f) is the same, wherein SWXY(t, f) represents the scalar product between X and Y, WCo (t, f) is defined by time t and frequency f, and SWXY(t, f) are the same; wherein, ncy is [ t-delta/2, t + delta/2 ] in the formula of ncy/f]The number of cycles in time, generally speaking, ncy values are related to the signal length, short sequences have small values, and long sequences have large values; t is time and f is frequency;
calculating the average value of the wavelet coherence coefficients in the optimal time-frequency domain found in S3 as a feature, carrying out rank sum test on the feature, and selecting the average value of the wavelet coherence coefficients with most significant difference as the optimal feature;
s5, constructing a classification model by using the wavelet coherence coefficient mean value in the optimal time-frequency domain: taking 80% of sample data as a training set, performing off-line training on the training set data, and constructing and storing a training model; and (4) taking 20% of sample data as a test set, sending the sample data into an SVM classifier for classification, performing classification and identification by taking the optimal wavelet coherence coefficient mean value obtained by analysis in S4 as a feature, and constructing a two-classification evaluation model.
2. The stroke rehabilitation assessment method based on the brain electromyogram signal wavelet coherence coefficient according to claim 1, characterized in that: the dry electrode electroencephalogram cap in the operation step S1 is worn correctly according to the standard, and the A1 and A2 electrodes are used as reference electrodes; the myoelectric electrodes adopt an expansion channel, the diameter of each electrode is 5mm, and the myoelectric electrodes are spaced by 2cm to form a differential mode for experiment; the subject executes the task and collects data according to the requirement, the upper arm of the affected side is lifted and kept for 5 seconds each time, and the rest is carried out for 10 seconds in the middle.
3. The stroke rehabilitation assessment method based on the brain electromyogram signal wavelet coherence coefficient according to claim 1, characterized in that: the specific steps of extracting valid data in operation S2 are as follows:
1) filtering the sEMG signal, wherein the filtering range is 5-50 Hz;
2) calculating the energy of the filtered signal, and setting a proper threshold value to detect the initial position of the movement;
3) detecting the moving segment, processing the signal by adopting a moving average method, and judging the starting point and the stopping point of the action through a threshold value;
4) intercepting data of 0-4 s from the collected original EEG signal and sEMG signal as effective data according to the starting position;
5) and performing wavelet denoising on the effective data, removing artifacts of an EEG signal, and improving the signal-to-noise ratio of the signal.
4. The stroke rehabilitation assessment method based on the brain electromyogram signal wavelet coherence coefficient according to claim 1, characterized in that: the filtering of the electroencephalogram and electromyogram signals in the operation step S2 includes filtering the EEG signals with a filtering range of 0.5 to 45Hz, and filtering the sEMG signals with a filtering range of 0.5 to 100 Hz.
5. The stroke rehabilitation assessment method based on the brain electromyogram signal wavelet coherence coefficient according to claim 1, characterized in that: r in the operation S32The discrimination formula for discriminating the degree of discrimination between the two classes is:
in wavelet coherence estimation, if X1 and X2 represent features of two classes, L1 and L2 represent the number of samples of different classes, and the larger the two classes are distinguished, the larger R2The larger the value; mean represents mean and std represents standard deviation.
6. The stroke rehabilitation assessment method based on the brain electromyogram signal wavelet coherence coefficient according to claim 1, characterized in that: the wavelet coherence analysis calculation in the operation step S4 is carried out, the wavelet coherence coefficient mean value of the late stroke recovery period in an alpha frequency band of 8-12 Hz is smaller than that of the patient in the middle stroke recovery period, the wavelet coherence coefficient mean value of the late stroke recovery period in a beta frequency band of 15-30 Hz and a gamma frequency band of 30-45 Hz is larger than that of the patient in the middle stroke recovery period, and the time of the wavelet coherence coefficient difference is mainly distributed between 1500ms and 3500 ms; and (4) calculating the average value of the wavelet coherence coefficients in the optimal time-frequency domain found in the S3 to be used as a feature, carrying out rank sum test on the feature, and finding out that the wavelet coherence coefficient average value between 2200ms and 2500ms of the beta frequency band has significant difference through analysis.
7. The stroke rehabilitation assessment method based on the brain electromyogram signal wavelet coherence coefficient according to claim 1, characterized in that: in the operation step S5, classification and identification are performed by using the wavelet coherence coefficient mean value in the optimal time-frequency domain analyzed in S4 as a feature, an evaluation model of the second classification is constructed, and the classification accuracy of the second classification is obtained through five-fold cross validation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110168009.3A CN112971786A (en) | 2021-02-05 | 2021-02-05 | Apoplexy rehabilitation evaluation method based on brain electromyographic signal wavelet coherence coefficient |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110168009.3A CN112971786A (en) | 2021-02-05 | 2021-02-05 | Apoplexy rehabilitation evaluation method based on brain electromyographic signal wavelet coherence coefficient |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112971786A true CN112971786A (en) | 2021-06-18 |
Family
ID=76348826
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110168009.3A Withdrawn CN112971786A (en) | 2021-02-05 | 2021-02-05 | Apoplexy rehabilitation evaluation method based on brain electromyographic signal wavelet coherence coefficient |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112971786A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114983419A (en) * | 2022-06-14 | 2022-09-02 | 天津大学 | Rehabilitation device for improving motor skills based on electroencephalogram signals |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070038382A1 (en) * | 2005-08-09 | 2007-02-15 | Barry Keenan | Method and system for limiting interference in electroencephalographic signals |
| WO2014089899A1 (en) * | 2012-12-10 | 2014-06-19 | 国家电网公司 | Distribution network phase-to-earth fault location method and location device based on transient signal wavelet transformation |
| CN109620223A (en) * | 2018-12-07 | 2019-04-16 | 北京工业大学 | A kind of rehabilitation of stroke patients system brain-computer interface key technology method |
| CN109669555A (en) * | 2017-10-16 | 2019-04-23 | 山东格橹特康复器具有限公司 | Wearable wireless mouse and its control method based on SEMG control |
| US20190231230A1 (en) * | 2018-01-30 | 2019-08-01 | Soochow University | Cerebral function state evaluation device based on brain hemoglobin information |
| CN110238863A (en) * | 2019-06-17 | 2019-09-17 | 北京国润健康医学投资有限公司 | Based on brain electricity-electromyography signal lower limb rehabilitation robot control method and system |
| CN110472595A (en) * | 2019-08-20 | 2019-11-19 | 郑州大学 | Identification model construction method, device and the recognition methods of EEG signals, device |
| CN111563581A (en) * | 2020-05-27 | 2020-08-21 | 杭州电子科技大学 | Method for constructing brain muscle function network based on wavelet coherence |
| CN111938991A (en) * | 2020-07-21 | 2020-11-17 | 燕山大学 | Hand rehabilitation training device and training method in double active control modes |
-
2021
- 2021-02-05 CN CN202110168009.3A patent/CN112971786A/en not_active Withdrawn
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070038382A1 (en) * | 2005-08-09 | 2007-02-15 | Barry Keenan | Method and system for limiting interference in electroencephalographic signals |
| WO2014089899A1 (en) * | 2012-12-10 | 2014-06-19 | 国家电网公司 | Distribution network phase-to-earth fault location method and location device based on transient signal wavelet transformation |
| CN109669555A (en) * | 2017-10-16 | 2019-04-23 | 山东格橹特康复器具有限公司 | Wearable wireless mouse and its control method based on SEMG control |
| US20190231230A1 (en) * | 2018-01-30 | 2019-08-01 | Soochow University | Cerebral function state evaluation device based on brain hemoglobin information |
| CN109620223A (en) * | 2018-12-07 | 2019-04-16 | 北京工业大学 | A kind of rehabilitation of stroke patients system brain-computer interface key technology method |
| CN110238863A (en) * | 2019-06-17 | 2019-09-17 | 北京国润健康医学投资有限公司 | Based on brain electricity-electromyography signal lower limb rehabilitation robot control method and system |
| CN110472595A (en) * | 2019-08-20 | 2019-11-19 | 郑州大学 | Identification model construction method, device and the recognition methods of EEG signals, device |
| CN111563581A (en) * | 2020-05-27 | 2020-08-21 | 杭州电子科技大学 | Method for constructing brain muscle function network based on wavelet coherence |
| CN111938991A (en) * | 2020-07-21 | 2020-11-17 | 燕山大学 | Hand rehabilitation training device and training method in double active control modes |
Non-Patent Citations (1)
| Title |
|---|
| 刘博: ""基于脑电、肌电信号的上肢康复训练与评估系统"", 《万方数据库》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114983419A (en) * | 2022-06-14 | 2022-09-02 | 天津大学 | Rehabilitation device for improving motor skills based on electroencephalogram signals |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110327036B (en) | Method for extracting respiratory signal and respiratory frequency from wearable electrocardiogram | |
| CN109907752B (en) | Electrocardiogram diagnosis and monitoring system for removing motion artifact interference and electrocardio characteristic detection | |
| CN108577834B (en) | A method of it is detected automatically for epilepsy interphase spike | |
| CN105147248A (en) | Physiological information-based depressive disorder evaluation system and evaluation method thereof | |
| WO2007021645A2 (en) | Method and system for limiting interference in electroencephalographic signals | |
| CN103892829B (en) | A kind of eye based on common space pattern moves signal recognition system and recognition methods thereof | |
| WO2014029764A1 (en) | Method for detection of an abnormal sleep pattern in a person | |
| CN104323773A (en) | EOG (Electrooculography)-based ERG (Electroretinography) signal acquisition and processing system and method | |
| CN103405225B (en) | A kind of pain that obtains feels the method for evaluation metrics, device and equipment | |
| CN106485208A (en) | The automatic removal method of eye electrical interference in single channel EEG signals | |
| CN112914587A (en) | Apoplexy rehabilitation assessment model construction method and assessment method based on resting state electroencephalogram signal coherence brain function network | |
| CN106175754A (en) | During sleep state is analyzed, waking state detects device | |
| CN114886388B (en) | Evaluation method and device for quality of electroencephalogram signal in anesthesia depth monitoring process | |
| Tran et al. | Detection of eeg-based eye-blinks using a thresholding algorithm | |
| CN112971786A (en) | Apoplexy rehabilitation evaluation method based on brain electromyographic signal wavelet coherence coefficient | |
| CN106037729A (en) | Lumbar intervertebral disc protrusion determination method and system | |
| CN106618486B (en) | Sleep state identification method and system in intelligent sleep assistance | |
| CN108464829B (en) | It is a kind of for assessing the method and device of traumatic brain injury | |
| CN106175698B (en) | Sleep cycle detection device in sleep state analysis | |
| CN106983510A (en) | A kind of muscle performance measuring method and device | |
| Fruet et al. | Assessment of Driving Stress Through SVM And KNN Classifiers on Multi-Domain Physiological Data | |
| Colamarino et al. | Adaptive learning in the detection of Movement Related Cortical Potentials improves usability of associative Brain-Computer Interfaces | |
| CN110680315A (en) | Electroencephalogram and electromyogram signal monitoring method based on asymmetric multi-fractal detrending correlation analysis | |
| CN112022151A (en) | Method for processing and identifying brain wave spike slow wave | |
| Waser et al. | A blind source-based method for automated artifact-correction in standard sleep EEG |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210618 |
|
| WW01 | Invention patent application withdrawn after publication |