Facial paralysis assessment method and system based on myoelectricity detection and treatment system

Abstract

The invention discloses a facial paralysis assessment method and system and a treatment system based on myoelectricity detection, and the method comprises the steps: collecting time-domain myoelectric signals of frontal muscles, eye wheels and orbicular orbicular muscles of a to-be-assessed facial paralysis patient under a certain standard expression, carrying out the filtering, extracting the features of the time-domain myoelectric signals, inputting the features into a trained SVM classifier, and predicting the level of facial paralysis. And targeted facial paralysis treatment can be carried out according to the facial paralysis level. By dynamically analyzing the characteristics of the facial electromyographic signals under different actions, the facial paralysis evaluation is accurately quantified, the accuracy of facial paralysis evaluation is improved, and targeted treatment is performed through the treatment system.

Description

Technical Field

[0001] The present invention relates to the technical field of medical detection, and specifically, to a facial paralysis evaluation method, system and treatment system based on electromyogram detection. Background Art

[0002] Nowadays, in the field of facial paralysis detection and treatment, the House-Brackmann Facial Nerve Paralysis Grading and the Sunnybrook Facial Nerve Assessment System are the two most commonly used scales in the facial paralysis industry, and are also important bases for doctors to grade facial paralysis for patients. However, the current two mainstream assessment systems have the following problems: the grading results highly depend on the clinical experience of doctors, and there is a judgment difference of ±1 level between different assessors, resulting in a large subjectivity of the assessment results; the existing standards only provide overall function scores and lack independent quantitative analysis of local muscle groups such as the frontalis muscle and the orbicularis oris muscle, making it difficult to identify subtle injuries; patients cannot independently judge the recovery status through objective indicators like body temperature monitoring, which severely restricts the compliance of rehabilitation training. During the existing treatment process, it usually takes a period of time to understand the treatment effect through a follow-up visit, and the treatment plan cannot be adjusted in a timely manner according to the actual response of the patient.

[0003] In addition, there are still many deficiencies in the current facial paralysis detection and treatment technologies: when analyzing features, the existing methods mostly rely on single-time domain or frequency domain features and ignore the dynamic deviation characteristics of bilateral electromyogram signals; at the same time, due to the large differences between patients, there is currently a lack of analysis of individual patient differences, and the threshold cannot be adjusted dynamically with the patient, and a universal and personalized assessment result cannot be achieved. Summary of the Invention

[0004] Aiming at the problems existing in the prior art, the present invention provides a facial paralysis evaluation method, system and treatment system based on electromyogram detection, dynamically analyzes the characteristics of facial electromyogram signals under different movements, accurately quantifies the facial paralysis rating, improves the accuracy of facial paralysis evaluation, and conducts targeted treatment through the treatment system.

[0005] To achieve the above technical objectives, the present invention adopts the following technical solutions: A facial paralysis evaluation method based on electromyogram detection, comprising the following steps:

[0006] Step S1, collect the time-domain electromyogram signals of the frontalis muscle, orbicularis oculi muscle and orbicularis oris muscle of different facial paralysis patients under different expressions, and determine the facial paralysis grade corresponding to the expression;

[0007] Step S2, perform frequency-domain and time-domain filtering on the time-domain electromyogram signals, and extract the characteristics of the time-domain electromyogram signals;

[0008] Step S3: Use the extracted features as the input of the SVM classifier, and use the facial paralysis level corresponding to the expression as the label of the SVM classifier to train the SVM classifier;

[0009] Step S4: Repeat Step S3 until the mean squared error loss function of the predicted facial paralysis level and the corresponding label converges, and complete the training of the SVM classifier;

[0010] Step S5: Collect the time-domain electromyography signals of the frontalis muscle, orbicularis oculi muscle, and orbicularis oris muscle of the facial paralysis patient to be evaluated under a certain standard expression, filter them, extract the features of the time-domain electromyography signals, input them into the trained SVM classifier, and predict the facial paralysis level.

[0011] Further, the process of determining the facial paralysis level of the expression in Step S1 is as follows: Determine according to the Sunnybrook Facial Nerve Paralysis Scale or the House-Brackmann Facial Nerve Paralysis Scale.

[0012] Further, the Sunnybrook Facial Nerve Paralysis Scale needs to score the muscle movement and synkinesis conditions of the forehead, eyes, and mouth under the standard expression: For each standard expression, a score of 1 is given for no muscle movement, a score of 2 is given for mild muscle movement, a score of 3 is given for muscle movement but disorder, a score of 4 is given for muscle movement close to symmetry, and a score of 5 is given for completely symmetrical facial muscle movement; For each standard expression, a score of 0 is given for no synkinesis, a score of 1 is given for mild synkinesis, a score of 2 is given for obvious synkinesis but no disfigurement, and a score of 3 is given for severe synkinesis with disfigurement; The standard expressions include: raising the forehead, gently closing the eyes, smiling with the mouth open, wrinkling the nose, and lip sucking.

[0013] Further, the House-Brackmann Facial Nerve Paralysis Scale needs to evaluate the movement states of the mouth, eyes, and forehead under the standard expression: When the functions of the mouth, eyes, and forehead are all normal, the facial paralysis level is Grade I; When the mouth is slightly asymmetric, the eyes can be completely closed with mild effort, and the forehead function is moderate to good, the facial paralysis level is Grade II; When there is still mild weakness with the maximum effort of the mouth, the eyes can completely close the eyelids with effort, and the forehead has mild to moderate movement, the facial paralysis level is Grade III; When there is still asymmetry with the maximum effort of the mouth, the eyes cannot be completely closed, and the forehead has no movement, the facial paralysis level is Grade IV; When there is only mild movement of the mouth, the eyes cannot be completely closed, and the forehead has no movement, the facial paralysis level is Grade V; When there is no movement of the mouth, eyes, and forehead, the facial paralysis level is Grade VI; The standard expressions include: raising the forehead, gently closing the eyes, smiling with the mouth open, wrinkling the nose, and lip sucking.

[0014] Further, the specific process of filtering the time-domain electromyography signals is as follows:

[0015] Step A: The time-domain electromyography (EMG) signal collected for a standard expression of a muscle part is converted into a frequency-domain EMG signal, filtered using a fourth-order Butterworth low-pass filter, and then the filtered frequency-domain EMG signal is converted back into a time-domain EMG signal.

[0016] Step B: The time-domain EMG signal after frequency-domain filtering is evenly divided, and the root mean square (RMS) value of each divided segment of the time-domain EMG signal is calculated.

[0017] Step C: The average of the three largest and the three smallest RMS values is taken to obtain the RMS threshold of the time-domain EMG signal.

[0018] Step D: The RMS value of each segment of the time-domain EMG signal is compared with the RMS threshold. If the RMS value of the time-domain EMG signal is less than the RMS threshold, it is recorded as 0; otherwise, it is recorded as 1, obtaining an array with the same length as the number of segments into which the time-domain EMG signal is divided, and two 0s are added respectively at the front and end of the array to obtain a sequence.

[0019] Step E: Set the sliding window length to 3, move the sliding window on the sequence. When the array of the sliding window is 001, it is recorded as the start time, and when the array of the sliding window is 100, it is recorded as the end time.

[0020] Step F: If the length of the sequence between the start time and the end time is less than 5, the corresponding time-domain EMG signal is deleted; otherwise, the corresponding time-domain EMG signal is extracted to achieve filtering of the time-domain EMG signal.

[0021] Furthermore, the characteristics of the time-domain EMG signal are all the characteristics under the same standard expression of different muscle parts. For all the characteristics under the same standard expression of the same muscle part, they include: the average amplitude on the left and right sides of the time-domain EMG signal, the deviation rate of the average amplitude on the left and right sides; the RMS value on the left and right sides of the time-domain EMG signal, the deviation rate of the RMS value on the left and right sides; the wavelength on the left and right sides of the time-domain EMG signal, the deviation rate of the wavelength on the left and right sides.

[0022] Furthermore, during the process of training the SVM classifier in Step S3, for the same facial paralysis patient, if the facial paralysis grade predicted by the SVM classifier is different from the facial paralysis grade of the corresponding label, the optimization weight of the parameters in the SVM classifier needs to be adjusted by the ratio of the standard expression score predicted by the SVM classifier to the corresponding standard expression score in the label.

[0023] Furthermore, the present invention also provides a facial paralysis assessment system for the above-mentioned facial paralysis assessment method based on EMG detection, including: an EMG signal acquisition module, an EMG feature extraction module, and a feature analysis module.

[0024] The myoelectric signal acquisition module collects the time-domain myoelectric signals of the frontalis muscle, orbicularis oculi muscle, and orbicularis oris muscle of the facial paralysis patient by configuring a multi-channel surface electrode array;

[0025] The myoelectric feature extraction module filters the collected time-domain myoelectric signals through frequency-domain filtering technology and adaptive time-domain filtering technology, and extracts the features of the filtered time-domain myoelectric signals;

[0026] The feature analysis module uses an SVM classifier to predict the facial paralysis grade of the facial paralysis patient according to the extracted features of the time-domain myoelectric signals.

[0027] Furthermore, it further includes: a doctor's facial paralysis grading diagnosis database for storing the facial paralysis grades of facial paralysis patients determined by doctors according to the Sunnybrook Facial Nerve Paralysis Rating Scale or the House-Brackmann Facial Nerve Paralysis Rating Scale; the facial paralysis grades of the facial paralysis patients in the doctor's facial paralysis grading diagnosis database are used as the labels for training the feature analysis module to optimize the parameters of the feature analysis module.

[0028] Furthermore, the present invention also provides a facial paralysis treatment system based on myoelectric detection, including: a facial paralysis evaluation system, an electrical stimulation module, and a cloud data management module;

[0029] The facial paralysis evaluation system uses the facial paralysis evaluation system of the myoelectric detection-based facial paralysis evaluation method described above to predict the facial paralysis grade of the facial paralysis patient;

[0030] The electrical stimulation module sends microcurrent stimulation to the surface electrode array corresponding to the muscle with facial paralysis according to the facial paralysis grade;

[0031] The cloud data management module is used to store the facial paralysis grade of the facial paralysis patient and the parameters of the microcurrent stimulation.

[0032] Compared with the prior art, the present invention has the following beneficial effects: The present invention designs a complete set of facial paralysis evaluation methods, systems, and treatment systems. By collecting the time-domain myoelectric signals of the frontalis muscle, orbicularis oculi muscle, and orbicularis oris muscle of the facial paralysis patient, the facial paralysis grade is evaluated, avoiding the evaluation differences caused by the doctor's subjective visual inspection of facial muscle movement; by performing frequency-domain and time-domain filtering on the time-domain myoelectric signals, the noise interference is reduced, ensuring the accuracy of the subsequent extraction of the time-domain myoelectric signal features; by extracting various features of the time-domain myoelectric signals and synthesizing multiple features, the facial muscle state of the facial paralysis patient can be comprehensively evaluated, thereby realizing a more accurate prediction of the facial paralysis grade and reducing the influence of subjective factors. The present invention improves the accuracy of facial paralysis evaluation by precisely quantifying the facial paralysis rating and performs targeted treatment through the treatment system. Description of the Drawings

[0033] Figure 1Flowchart of the facial paralysis assessment method based on electromyogram detection according to the present invention;

[0034] Figure 2 Schematic diagram of the facial paralysis assessment system based on electromyogram detection according to the present invention. Detailed implementation manners

[0035] The technical solution of the present invention will be further explained below with reference to the accompanying drawings.

[0036] As Figure 1 Flowchart of the facial paralysis assessment method based on electromyogram detection according to the present invention. The facial paralysis assessment method includes the following steps:

[0037] Step S1: Collect the time-domain electromyogram signals of the frontalis muscle, orbicularis oculi muscle, and orbicularis oris muscle of different facial paralysis patients under different expressions through a multi-channel surface electrode array, avoiding the evaluation differences caused by the doctor's subjective visual inspection of facial muscle movement, and determining the facial paralysis level corresponding to the expression. Specifically, the facial paralysis level can be determined according to the Sunnybrook Facial Nerve Paralysis Scale or the House-Brackmann Facial Nerve Paralysis Scale.

[0038] In the present invention, the Sunnybrook Facial Nerve Paralysis Scale needs to score the muscle movement and synkinesis of the forehead, eyes, and mouth under standard expressions: for each standard expression, if the muscle has no movement, the score is 1; if the muscle has mild movement, the score is 2; if the muscle has movement but is disordered, the score is 3; if the muscle movement is nearly symmetrical, the score is 4; and if the facial muscle movement is completely symmetrical, the score is 5; for each standard expression, if the muscle has no synkinesis, the score is 0; if the muscle has mild synkinesis, the score is 1; if the muscle has obvious synkinesis but no disfigurement, the score is 2; and if the muscle synkinesis is severely disfiguring, the score is 3; wherein, the standard expressions include: raising the forehead, gently closing the eyes, smiling with the mouth open, wrinkling the nose, and sucking the lips.

[0039] The House-Brackmann Facial Nerve Paralysis Scale needs to evaluate the movement status of the mouth, eyes, and forehead under standard expressions: when the functions of the mouth, eyes, and forehead are all normal, the facial paralysis level is grade I; when the mouth is slightly asymmetrical, the eyes can be completely closed with mild effort, and the forehead function is moderate to good, the facial paralysis level is grade II; when the mouth still has mild weakness with maximum effort, the eyes can completely close the eyelids with effort, and the forehead has mild to moderate movement, the facial paralysis level is grade III; when the mouth is still asymmetrical with maximum effort, the eyes cannot be completely closed, and the forehead has no movement, the facial paralysis level is grade IV; when the mouth has only mild movement, the eyes cannot be completely closed, and the forehead has no movement, the facial paralysis level is grade V; when the mouth, eyes, and forehead have no movement, the facial paralysis level is grade VI; wherein, the standard expressions include: raising the forehead, gently closing the eyes, smiling with the mouth open, wrinkling the nose, and sucking the lips.

[0040] Step S2: Perform frequency-domain and time-domain filtering on the time-domain electromyography (EMG) signal, extract the features of the time-domain EMG signal, reduce noise interference, and ensure the accuracy of subsequent time-domain EMG signal feature extraction. Specifically,

[0041] Step A: Convert the time-domain EMG signal collected for a standard expression of a muscle part into a frequency-domain EMG signal, filter it using a fourth-order Butterworth low-pass filter, set the low-pass cut-off frequency to 1000 Hz, complete the frequency-domain filtering of the EMG signal, and convert the filtered frequency-domain EMG signal back into a time-domain EMG signal.

[0042] Step B: Evenly divide the time-domain EMG signal after frequency-domain filtering and calculate the root mean square (RMS) value of each divided segment of the time-domain EMG signal.

[0043] Step C: Select the three largest and the three smallest RMS values and take their average to obtain the RMS threshold of the time-domain EMG signal.

[0044] Step D: Compare the RMS value of each segment of the time-domain EMG signal with the RMS threshold. If the RMS value of the time-domain EMG signal is less than the RMS threshold, record 0; otherwise, record 1, to obtain an array with the same length as the number of divided segments of the time-domain EMG signal, and supplement two 0s at the front and back ends of the array to obtain a sequence.

[0045] Step E: Set the sliding window length to 3 and move the sliding window on the sequence. As shown in Table 1, there are 8 combinations of arrays. When the array of the sliding window is 001, record it as the start time, and when the array of the sliding window is 100, record it as the end time.

[0046] Table 1: Examples of 8 combinations of arrays

[0047] Array combination t-1 t t+1 1 0 0 0 Do not record the time 2 0 0 1 Record the start time t+1 3 0 1 0 Do not record the time 4 0 1 1 Do not record the time 5 1 0 0 Record the end time t-1 6 1 0 1 Do not record the time 7 1 1 0 Do not record the time 8 1 1 1 Do not record the time

[0048] Step F: If the length of the sequence between the start time and the end time is less than 5, delete the corresponding time-domain EMG signal; otherwise, extract the corresponding time-domain EMG signal to achieve the filtering of the time-domain EMG signal.

[0049] The features of the time-domain electromyography (EMG) signal are all the features under the same standard expression for different muscle parts. For all the features under the same standard expression for the same muscle part, they include: the average amplitude on the left and right sides of the time-domain EMG signal, and the deviation rate of the average amplitude on the left and right sides; the root mean square (RMS) value on the left and right sides of the time-domain EMG signal, and the deviation rate of the RMS value on the left and right sides; the wavelength on the left and right sides of the time-domain EMG signal, and the deviation rate of the wavelength on the left and right sides. The amplitude reflects the intensity of muscle activity, the RMS value reflects the energy of muscle activity, and the combination of the two can sensitively capture the subtle changes in facial muscle activity. The wavelength reflects the periodic characteristics of the signal, can reflect the coordination of muscle contraction, and further reveals the severity of facial paralysis; by calculating the deviation rates of the amplitude, RMS value, and wavelength on the left and right sides, the asymmetry degree of the facial muscles of facial paralysis patients can be quantified, avoiding the error of subjective judgment; at the same time, by comprehensively considering multiple features, the facial muscle state of facial paralysis patients can be comprehensively evaluated, so as to realize a more accurate prediction of the facial paralysis level and reduce the influence of subjective factors. Among them, the average amplitude MAV l 、MAV r are both: The deviation rate POL MAV of the average amplitude on the left and right sides of the time-domain EMG signal is: l -MAV r ) / (MAV l +MAV r )|; The root mean square values RMS l 、RMS r on the left and right sides of the time-domain EMG signal are both: The deviation rate POL RMS of the root mean square values on the left and right sides of the time-domain EMG signal is: l -RMS r ) / (RMS l +RMS r )|; The wavelengths WL l 、WL r on the left and right sides of the time-domain EMG signal are both: The deviation rate POL WL of the wavelengths on the left and right sides of the EMG signal is: l -WL r ) / (WL l +WL r )|; N represents the length of the time-domain EMG signal, i represents the index of N, x i represents the i-th time-domain EMG signal, and μ represents the mean value of the time-domain EMG signal.

[0050] Step S3: Use the extracted features as the input of the SVM classifier, and use the facial paralysis level corresponding to the expression as the label of the SVM classifier to train the SVM classifier:

[0051] Step S4. Repeat step S3 until the mean square error loss function between the predicted facial paralysis level and the corresponding label converges, and complete the training of the SVM classifier:

[0052] The hyperplane equation for determining the data classification as much as possible through the training set is: w T x i +b = 0, where w is the normal vector that determines the direction of the plane; b is the bias that determines the distance between the plane and the origin;

[0053] For linearly separable data, the goal of the SVM classifier is to find w and b such that all sample points satisfy: where, y i is the label of the sample;

[0054] To maximize the classification margin, the SVM classifier needs to minimize the objective function: while satisfying the constraint condition

[0055] In practical applications, the soft margin support vector machine allows a certain degree of misclassification. Therefore, slack variables ξ i ≥0 represents the degree to which the data points violate the margin. Therefore, the optimization problem becomes: Its constraint condition is:

[0056] During the training process of the SVM classifier, for the same facial paralysis patient, if the facial paralysis level predicted by the SVM classifier is different from the facial paralysis level of the corresponding label, it is necessary to adjust the optimization weight of the parameters in the SVM classifier through the ratio of the standard expression score predicted by the SVM classifier to the corresponding standard expression score in the label.

[0057] Step S5. Collect the time-domain electromyographic signals of the frontalis muscle, orbicularis oculi muscle, and orbicularis oris muscle of the facial paralysis patient to be evaluated under a certain standard expression, perform filtering, extract the features of the time-domain electromyographic signals, input them into the trained SVM classifier, and predict the facial paralysis level. The present invention improves the accuracy of facial paralysis evaluation by precisely quantifying the facial paralysis rating.

[0058] Such as Figure 2 , the present invention also provides a facial paralysis evaluation system for a facial paralysis evaluation method based on electromyographic detection, including: an electromyographic signal acquisition module, an electromyographic feature extraction module, a feature analysis module, and a doctor's facial paralysis rating diagnosis database;

[0059] The electromyographic signal acquisition module collects the time-domain electromyographic signals of the frontalis muscle, orbicularis oculi muscle, and orbicularis oris muscle of the facial paralysis patient through a configured multi-channel surface electrode array; at the same time, the electromyographic signal acquisition module regularly detects the channel noise energy of the surface electrode array and performs mean zeroing;

[0060] The electromyogram feature extraction module filters the collected time-domain electromyogram signals through frequency-domain filtering technology and adaptive time-domain filtering technology, and extracts the features of the filtered time-domain electromyogram signals;

[0061] The feature analysis module uses an SVM classifier to predict the facial paralysis grade of facial paralysis patients according to the extracted features of the time-domain electromyogram signals;

[0062] The doctor's facial paralysis grading diagnosis database is used to store the facial paralysis grades of facial paralysis patients determined by doctors according to the Sunnybrook Facial Nerve Paralysis Scale or the House-Brackmann Facial Nerve Paralysis Scale; the facial paralysis grades of facial paralysis patients in the doctor's facial paralysis grading diagnosis database are used as the labels for training the feature analysis module to optimize the parameters of the feature analysis module.

[0063] In a technical solution of the present invention, a facial paralysis treatment system based on electromyogram detection is further provided, including: a facial paralysis evaluation system, an electrical stimulation module, and a cloud data management module;

[0064] The facial paralysis evaluation system uses a facial paralysis evaluation system based on the facial paralysis evaluation method of electromyogram detection to predict the facial paralysis grade of facial paralysis patients;

[0065] The electrical stimulation module sends microcurrent stimulation to the surface electrode array corresponding to the muscles with facial paralysis according to the facial paralysis grade, and different electrical stimulation parameters such as low frequency and medium frequency can be used to be applicable to the treatment of different grades of facial paralysis, and the output current range is 0 mA to 60 mA;

[0066] The cloud data management module is used to store the facial paralysis grades of facial paralysis patients and the parameters of microcurrent stimulation in the cloud, which is convenient for doctors to remotely view and manage the treatment records of patients.

[0067] The above is only the preferred implementation manner of the present invention, and the protection scope of the present invention is not limited to the above implementation manner. All technical solutions falling within the idea of the present invention belong to the protection scope of the present invention. It should be pointed out that for those of ordinary skill in the art, several improvements and modifications made without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A facial paralysis evaluation method based on electromyogram detection, characterized in that, It includes the following steps: Step S1: Collect the time-domain electromyography (EMG) signals of the frontalis muscle, orbicularis oculi muscle, and orbicularis oris muscle of different facial paralysis patients under different expressions, and determine the facial paralysis grade corresponding to the expression; Step S2: Perform frequency-domain and time-domain filtering on the time-domain EMG signals to extract the characteristics of the time-domain EMG signals; Step S3: Use the extracted characteristics as the input of the SVM classifier, and use the facial paralysis grade corresponding to the expression as the label of the SVM classifier to train the SVM classifier; Step S4: Repeat Step S3 until the mean squared error loss function of the predicted facial paralysis grade and the corresponding label converges, and complete the training of the SVM classifier; Step S5: Collect the time-domain EMG signals of the frontalis muscle, orbicularis oculi, and orbicularis oris muscles of the facial paralysis patient to be evaluated under a certain standard expression, perform filtering, extract the characteristics of the time-domain EMG signals, input them into the trained SVM classifier, and predict the facial paralysis grade.

2. The facial paralysis evaluation method based on electromyogram detection according to claim 1, wherein The process of determining the facial paralysis grade of the expression in Step S1 is as follows: It is determined according to the Sunnybrook Facial Nerve Paralysis Rating Scale or the House-Brackmann Facial Nerve Paralysis Rating Scale.

3. The facial paralysis assessment method based on electromyogram detection according to claim 2, wherein The Sunnybrook Facial Nerve Paralysis Rating Scale needs to score the muscle movement and synkinesis of the forehead, eyes, and mouth under the standard expression: For each standard expression, a score of 1 is given if there is no muscle movement, a score of 2 if there is mild muscle movement, a score of 3 if there is muscle movement but disorder, a score of 4 if the muscle movement is nearly symmetrical, and a score of 5 if the facial muscle movement is completely symmetrical; For each standard expression, a score of 0 is given if there is no synkinesis, a score of 1 if there is mild synkinesis, a score of 2 if there is obvious synkinesis but no disfigurement, and a score of 3 if there is severe synkinesis with disfigurement; The standard expressions include: raising the forehead, gently closing the eyes, smiling with the mouth open, wrinkling the nose, and lip sucking.

4. The facial paralysis assessment method based on electromyogram detection according to claim 2, characterized in that The House-Brackmann Facial Nerve Paralysis Rating Scale needs to evaluate the movement status of the mouth, eyes, and forehead under the standard expression: When the functions of the mouth, eyes, and forehead are all normal, the facial paralysis grade is Grade I; When there is mild asymmetry of the mouth, the eyes can be completely closed with mild effort, and the function of the forehead is moderate to good, the facial paralysis grade is Grade II; When there is still mild weakness when the mouth uses the maximum force, the eyes can completely close the eyelids with effort, and the forehead has mild to moderate movement, the facial paralysis grade is Grade III; When there is still asymmetry when the mouth uses the maximum force, the eyes cannot be completely closed, and the forehead has no movement, the facial paralysis grade is Grade IV; When there is only mild movement of the mouth, the eyes cannot be completely closed, and the forehead has no movement, the facial paralysis grade is Grade V; When there is no movement of the mouth, eyes, and forehead, the facial paralysis grade is Grade VI; The standard expressions include: raising the forehead, gently closing the eyes, smiling with the mouth open, wrinkling the nose, and lip sucking.

5. The facial paralysis assessment method based on electromyogram detection according to claim 2, characterized in that The specific process of filtering the time-domain EMG signals is as follows: Step A: Convert the time-domain EMG signals collected for a standard expression of a muscle part into frequency-domain EMG signals, filter them using a fourth-order Butterworth low-pass filter, and convert the filtered frequency-domain EMG signals back into time-domain EMG signals; Step B: Uniformly divide the time-domain EMG signals after frequency-domain filtering, and calculate the root mean square value of each divided segment of the time-domain EMG signals; Step C: Select the three with the largest root mean square values and the three with the smallest root mean square values and calculate their average to obtain the root mean square threshold of the time-domain electromyography signal; Step D: Compare the root mean square value of each segment of the time-domain electromyography signal with the root mean square threshold. If the root mean square value of the time-domain electromyography signal is less than the root mean square threshold, record 0; otherwise, record 1 to obtain an array with the same length as the number of segments of the time-domain electromyography signal, and supplement two 0s at the front and back ends of the array to obtain a sequence; Step E: Set the sliding window length to 3 and move the sliding window on the sequence. When the array of the sliding window is 001, record it as the start time, and when the array of the sliding window is 100, record it as the end time; Step F: If the length of the sequence between the start time and the end time is less than 5, delete the corresponding time-domain electromyography signal; otherwise, extract the corresponding time-domain electromyography signal to implement filtering of the time-domain electromyography signal.

6. The facial paralysis evaluation method based on electromyogram detection according to claim 5, wherein, The characteristics of the time-domain electromyography signal are all the characteristics under the same standard expression of different muscle parts. For all the characteristics under the same standard expression of the same muscle part, they include: the average amplitude on the left and right sides of the time-domain electromyography signal, the deviation rate of the average amplitude on the left and right sides; the root mean square value on the left and right sides of the time-domain electromyography signal, the deviation rate of the root mean square value on the left and right sides; the wavelength on the left and right sides of the time-domain electromyography signal, the deviation rate of the wavelength on the left and right sides.

7. The facial paralysis assessment method based on electromyogram detection according to claim 6, wherein During the process of training the SVM classifier in Step S3, for the same facial paralysis patient, if the facial paralysis grade predicted by the SVM classifier is different from the facial paralysis grade of the corresponding label, it is necessary to adjust the optimization weight of the parameters in the SVM classifier through the ratio of the standard expression score predicted by the SVM classifier to the corresponding standard expression score in the label.

8. A facial paralysis assessment system for the facial paralysis assessment method based on electromyogram detection according to any one of claims 1-7, characterized in that Including: An electromyography signal acquisition module, an electromyography feature extraction module, and a feature analysis module; The electromyography signal acquisition module collects the time-domain electromyography signals of the frontalis muscle, orbicularis oculi muscle, and orbicularis oris muscle of the facial paralysis patient respectively by configuring a multi-channel surface electrode array; The electromyography feature extraction module filters the collected time-domain electromyography signals through frequency-domain filtering technology and adaptive time-domain filtering technology, and extracts the features of the filtered time-domain electromyography signals; The feature analysis module uses an SVM classifier to predict the facial paralysis grade of the facial paralysis patient according to the extracted features of the time-domain electromyography signal.

9. The facial paralysis evaluation system of the facial paralysis evaluation method based on electromyogram detection according to claim 8, characterized in that, It also includes: A doctor's facial paralysis grading diagnosis database for storing the facial paralysis grade of the facial paralysis patient determined by the doctor according to the Sunnybrook Facial Paralysis Rating Scale or the House-Brackmann Facial Paralysis Rating Scale; The facial paralysis grade of the facial paralysis patient in the doctor's facial paralysis grading diagnosis database is used as the label for training the parameters of the feature analysis module to optimize the parameters of the feature analysis module.

10. A facial paralysis treatment system based on electromyogram detection, characterized in that, Including: A facial paralysis evaluation system, an electrical stimulation module, and a cloud data management module; The facial paralysis evaluation system uses the facial paralysis evaluation system of the electromyography detection-based facial paralysis evaluation method described in Claim 9 to predict the facial paralysis grade of the facial paralysis patient; The electrical stimulation module sends microcurrent stimulation to the surface electrode array corresponding to the muscle with facial paralysis according to the facial paralysis grade; The cloud data management module is used to store the facial paralysis grade of the facial paralysis patient and the parameters of the microcurrent stimulation.

Images