Stroke electrical stimulation rehabilitation evaluation system and method

Abstract

The invention relates to the technical field of cerebral apoplexy rehabilitation assessment, in particular to a cerebral apoplexy electrical stimulation rehabilitation assessment system and method.The cerebral apoplexy electrical stimulation rehabilitation assessment system comprises a data acquisition unit, a data processing unit and a control unit, the data processing unit is used for recognizing a dynamic motion intention in the head electroencephalogram signals and the limb surface electromyogram signals in the resting period; the functional electrical stimulation unit is used for generating functional contraction matched with the dynamic motion intention through electrical stimulation of corresponding muscles; and the stimulation response evaluation unit is used for evaluating the continuous transmission efficiency of the brain movement instruction and determining the electrical stimulation response degree of the stroke patient. According to the method, the human body movement intention of the user is recognized and restored through the collected head electroencephalogram signals and limb surface electromyogram signals, then the functional electrical stimulation system stimulates contraction of corresponding muscles to achieve the human body hand movement intention, and the response degree of the stroke user to continuous brain movement instructions is evaluated in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of stroke rehabilitation assessment, and particularly relates to a stroke electrical stimulation rehabilitation assessment system and method. Background Art

[0002] Stroke is a common clinical disease and one of the most serious diseases endangering human health today, characterized by high incidence, high disability rate, and high mortality rate. Electrical stimulation therapy for stroke motor dysfunction has been widely used in clinical practice; previous studies have shown that cerebral cortex functional reorganization is an important mechanism for functional recovery after stroke; the treatment effect is closely related to the central cortex functional response. Therefore, evaluating the degree of cortical functional response induced by electrical stimulation therapy is an important means to determine the stroke motor rehabilitation effect.

[0003] In the prior art, the assessment of the degree of electrical stimulation response is mostly achieved in the form of a scale, which has certain subjectivity and lag, and it is difficult to grasp the muscle functional response of electrical stimulation in real time. Moreover, during the assessment, a single functional movement response is used to grasp the response of the patient to this functional movement, and it is impossible to evaluate the switching response of the patient between different functional movements, making it difficult to achieve the goal of deeper rehabilitation effect assessment. Summary of the Invention

[0004] The purpose of the present invention is to provide a stroke electrical stimulation rehabilitation assessment system to solve the technical problem in the prior art that it is impossible to evaluate the switching response of a patient between different functional movements and difficult to achieve the goal of deeper rehabilitation effect assessment.

[0005] To solve the above technical problems, the present invention specifically provides the following technical solutions:

[0006] A stroke electrical stimulation rehabilitation assessment system, comprising:

[0007] A data acquisition unit, configured to acquire the electroencephalogram signal EEG of the head of a stroke patient, the surface electromyogram signal sEMG(T1) of the limb during the rest period, and the surface electromyogram signal sEMG(T2) of the limb during the stimulation period;

[0008] A data processing unit, configured to identify the dynamic movement intention of a stroke patient from the electroencephalogram signal EEG of the head and the surface electromyogram signal sEMG(T1) of the limb during the rest period;

[0009] A functional electrical stimulation unit, configured to generate a functional contraction matching the dynamic movement intention by electrically stimulating the corresponding muscles of a stroke patient;

[0010] A stimulus response evaluation unit, which is used to evaluate the continuous transmission efficiency of the brain motor commands of stroke patients according to the surface electromyogram signal sEMG(T2) during the stimulation period, and determine the degree of electrical stimulation response of stroke patients.

[0011] As a preferred embodiment of the present invention, the data acquisition unit includes an EEG device and an sEMG device. The EEG device is used to collect the electroencephalogram signal EEG of the head of stroke patients, and the sEMG device is used to collect the surface electromyogram signal sEMG(T1) during the static period and the surface electromyogram signal sEMG(T2) during the stimulation period.

[0012] As a preferred embodiment of the present invention, a dynamic motion intention recognition model is set in the data processing unit, and the dynamic motion intention recognition model is:

[0013] (actLabel(next), actLabel) = CNN2(EEG, sEMG(T1));

[0014] In the formula, EEG is the electroencephalogram signal of the head, sEMG(T1) is the surface electromyogram signal of the limb during the rest period, actLabel is the motion intention output by the second convolutional neural network, actLabel(next) is the motion intention of the next time output by the second convolutional neural network, and CNN is the second convolutional neural network.

[0015] As a preferred embodiment of the present invention, the electrical stimulation pulse in the functional electrical stimulation unit is synchronously triggered with the dynamic motion intention in the data processing unit and the stimulation period signal acquisition of the sEMG device through a BNC trigger line.

[0016] As a preferred embodiment of the present invention, the continuous transmission efficiency in the stimulus response evaluation unit is the delay time between the electrical stimulation pulse of the motion intention actLabel switching to the electrical stimulation pulse of the next-time motion intention actLabel(next) and the surface electromyogram signal sEMG(T2) generated by actLabel switching to the surface electromyogram signal sEMG(T2) generated by actLabel(next);

[0017] The continuous transmission efficiency is:

[0018] Δt = (t sEMG(T2)(next) ―t sEMG(T2) )―(t actLabel(next) ―t actLabel );

[0019] In the formula, Δt is the delay time, t actLabel is the emission time of the electrical stimulation pulse of the motion intention actLabel, t actLabel(next)The emission time of the electrical stimulation pulse for the movement intention actLabel(next) at the next time, t sEMG(T2) The time of the surface electromyogram signal sEMG(T2) of the limb corresponding to the movement intention actLabel, t sEMG(T2)(next) The generation time of the surface electromyogram signal sEMG(T2) of the limb corresponding to the movement intention actLabel(next);

[0020] When the delay time is less than the preset threshold, it is determined that the electrical stimulation response degree of the stroke patient is high;

[0021] When the delay time is greater than or equal to the preset threshold, it is determined that the electrical stimulation response degree of the stroke patient is low.

[0022] As a preferred embodiment of the present invention, the present invention provides a method for evaluating stroke electrical stimulation rehabilitation, which is applied to a stroke electrical stimulation rehabilitation evaluation system. The method includes the following steps:

[0023] Collect the head electroencephalogram signal EEG and the surface electromyogram signal sEMG(T1) of the limb during the rest period of the stroke patient, and identify the dynamic movement intention of the stroke patient according to the head electroencephalogram signal EEG and the surface electromyogram signal sEMG(T1) of the limb during the rest period through a dynamic movement intention recognition model;

[0024] When the dynamic movement intention is recognized, synchronously trigger the functional electrical stimulation unit to emit an electrical stimulation pulse matching the dynamic movement intention, and synchronously trigger the sEMG device to enter the signal acquisition during the stimulation period to obtain the surface electromyogram signal sEMG(T2) of the limb during the stimulation period;

[0025] Evaluate the continuous transmission efficiency of the brain movement command of the stroke patient according to the surface electromyogram signal sEMG(T2) during the stimulation period, and determine the electrical stimulation response degree of the stroke patient.

[0026] As a preferred embodiment of the present invention, the method for constructing a dynamic movement intention model includes:

[0027] Collect the time series of the head electroencephalogram signal EEG and the time series of the surface electromyogram signal sEMG(T1) of the limb during the rest period of the stroke patient, and mark the movement intention of the stroke patient at each time on the time series to obtain the time series of the movement intention;

[0028] Take the movement intention at the previous time in the time series of the movement intention as the input of the first convolutional neural network, and take the movement intention at the next time in the time series of the movement intention as the output of the first convolutional neural network, and train the first convolutional neural network to obtain a time series prediction model of the movement intention. The time series prediction model of the movement intention is:

[0029] actLabel(next)1 = CNN1(actLabel(list));

[0030] Wherein, actLabel(list) is the motion intention at the previous time in the time series of the motion intention, actLabel(next)1 is the motion intention at the next time output by the first convolutional neural network, and CNN1 is the convolutional neural network;

[0031] Taking the head electroencephalogram signal EEG and the surface electromyogram signal sEMG(T1) of the limb during the rest period at each time as the input of the second convolutional neural network, and taking the motion intention at each time and the motion intention at the next time at each time as the output of the second convolutional neural network, training the second convolutional neural network to obtain a dynamic motion intention model, the dynamic motion intention recognition model is:

[0032] (actLabel(next), actLabel) = CNN2(EEG, sEMG(T1));

[0033] Wherein, EEG is the head electroencephalogram signal, sEMG(T1) is the surface electromyogram signal of the limb during the rest period, actLabel is the motion intention output by the second convolutional neural network, actLabel(next) is the motion intention at the next time output by the second convolutional neural network, and CNN2 is the second convolutional neural network;

[0034] Among them, the loss function of the dynamic motion intention recognition model is:

[0035] Loss = MSE(actLabel, actLabel(list)) + MSE(actLabel(next), actLabel(next)1);

[0036] Wherein, Loss is the total loss, MSE(actLabel, actLabel(list)) is the mean square error between actLabel and actLabel(list), and MSE(actLabel(next), actLabel(next)1) is the mean square error between actLabel(next) and actLabel(next)1.

[0037] As a preferred solution of the present invention, the acquisition method of the surface electromyogram signal sEMG(T2) of the limb during the stimulation period includes:

[0038] Trigger the functional electrical stimulation unit to emit electrical stimulation pulses matching the motion intention actLabel, collect the surface electromyogram signals of the limb during the stimulation period through the sEMG device, and obtain the surface electromyogram signal sEMG(T2) of the limb corresponding to the motion intention actLabel;

[0039] When the next time of the motion intention actLabel arrives, trigger the functional electrical stimulation unit to switch from emitting the electrical stimulation pulses matching the motion intention actLabel to emitting the electrical stimulation pulses matching the motion intention actLabel(next), collect the surface electromyogram signals of the limb during the stimulation period through the sEMG device, and obtain the surface electromyogram signal sEMG(T2) of the limb corresponding to the motion intention actLabel(next).

[0040] As a preferred embodiment of the present invention, the calculation method of the continuous transfer efficiency includes:

[0041] Record the emission time of the electrical stimulation pulses of the motion intention actLabel, and record the emission time of the electrical stimulation pulses of the motion intention actLabel(next);

[0042] Record the generation time of the surface electromyogram signal sEMG(T2) of the limb corresponding to the motion intention actLabel, and the generation time of the surface electromyogram signal sEMG(T2) of the limb corresponding to actLabel(next);

[0043] According to the calculation formula of the continuous transfer efficiency, obtain the continuous transfer efficiency.

[0044] As a preferred embodiment of the present invention, the determination method of the electrical stimulation response degree includes:

[0045] When the delay time is less than the preset threshold, it is determined that the electrical stimulation response degree of the stroke patient is high;

[0046] When the delay time is greater than or equal to the preset threshold, it is determined that the electrical stimulation response degree of the stroke patient is low.

[0047] The present invention has the following beneficial effects compared with the prior art:

[0048] The present invention collects the electroencephalogram signals of the head and the surface electromyogram signals of the limb, so as to identify and restore the human motion intention of the user, and then stimulates the contraction of the corresponding muscles through the functional electrical stimulation system to realize the human motion intention, and evaluates the response degree of the stroke user to the continuous brain motion commands in real time, so as to achieve a deeper rehabilitation effect evaluation goal. Description of the Drawings

[0049] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only exemplary. For those of ordinary skill in the art, without creative efforts, other implementation drawings can also be obtained according to the provided drawings.

[0050] Figure 1 Block diagram of the stroke electrical stimulation rehabilitation evaluation system provided by the embodiment of the present invention;

[0051] Figure 2 Flowchart of the stroke electrical stimulation rehabilitation evaluation method provided by the embodiment of the present invention. Specific embodiments

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0053] As Figure 1 shown, the present invention provides a stroke electrical stimulation rehabilitation evaluation system, including:

[0054] A data acquisition unit for acquiring the electroencephalogram signal EEG of the head of a stroke patient, the surface electromyogram signal sEMG(T1) of the limb during the rest period, and the surface electromyogram signal sEMG(T2) of the limb during the stimulation period;

[0055] A data processing unit for identifying the dynamic movement intention of a stroke patient in the electroencephalogram signal EEG of the head and the surface electromyogram signal sEMG(T1) of the limb during the rest period;

[0056] A functional electrical stimulation unit for generating a functional contraction matching the dynamic movement intention by electrically stimulating the corresponding muscles of a stroke patient;

[0057] A stimulation response evaluation unit for evaluating the continuous transmission efficiency of the brain movement command of a stroke patient according to the surface electromyogram signal sEMG(T2) during the stimulation period and determining the electrostimulation response degree of the stroke patient.

[0058] The head electroencephalogram signal EEG and the surface electromyogram signal sEMG(T1) during the rest period contain information for identifying the movement intention of stroke patients. Therefore, in order to identify the movement intention of stroke patients, the present invention uses a neural network to construct an automated model for identifying the movement intention, that is, a real-time movement intention recognition model actLabel = CNN(EEG, sEMG(T1)); where EEG is the head electroencephalogram signal, sEMG(T1) is the surface electromyogram signal of the limb during the rest period, actLabel is the movement intention output by the convolutional neural network, and CNN is the convolutional neural network. This model can achieve intelligent and real-time recognition of the movement intention.

[0059] Among them, when identifying the movement intention, the surface electromyogram signal sEMG(T1) of the limb during the rest period is added to the basis of the head electroencephalogram signal EEG, which can exclude the involuntary interference of the patient's muscles during the movement intention recognition and obtain a more accurate intention recognition result.

[0060] On the basis of being able to identify the real-time movement intention, the present invention further introduces temporal dynamic prediction. By training on the movement intention data sequence of an individual patient, a temporal prediction model of the movement intention is obtained: actLabel(next)1 = CNN1(actLabel(list)), which can predict the next movement intention of the patient according to the real-time movement intention. Since the training data (the movement intention data sequence of the individual patient) is a digital manifestation of the movement habits of the individual patient, the prediction result obtained by the temporal prediction model of the movement intention trained by this data conforms to the movement habits of the individual patient.

[0061] The real-time movement intention recognition model introducing temporal dynamic prediction forms a dynamic movement intention recognition model. While identifying the movement intention of stroke patients based on the head electroencephalogram signal EEG and the surface electromyogram signal sEMG(T1) of the limb during the rest period, it integrates the temporal prediction performance of the movement intention, so that the dynamic movement intention recognition model can predict the current movement intention and also the movement intention of the patient at the next moment (or stage).

[0062] After the dynamic motion intention recognition model can predict the current motion intention and also the motion intention of the patient at the next moment (or stage), during the stroke rehabilitation assessment, the followability of the muscle reflex to the motion intention switch will be monitored during the process of the patient making different functional motion switches under different intentions, and the conduction efficiency of the continuous motion switch will be evaluated. The higher the efficiency, the higher the continuous response of the patient's different functional motions, and the better the rehabilitation effect. On the contrary, the lower the efficiency, the lower the continuous response of the patient's different functional motions, and the worse the rehabilitation effect. This makes the stroke electrical stimulation response assessment become a dynamic assessment of functional motion switches, serving as a deeper rehabilitation effect assessment goal, thereby deeper exploring the rehabilitation situation of stroke.

[0063] Moreover, during the process of the patient making different functional motion switches under different intentions, all the intention motion switches conform to the patient's personalized motion habits, so as to realize the capture of individual motion as the basis for the patient's rehabilitation assessment, which is more in line with the individual, and thus make the stroke rehabilitation assessment result tend to be individualized and precise, facilitating the subsequent adjustment of the rehabilitation plan for stroke patients to fit the individual, and providing the possibility for the realization of personalized diagnosis and treatment of stroke.

[0064] The data acquisition unit includes an EEG device and an sEMG device. The EEG device is used to collect the electroencephalogram signal EEG of the head of the stroke patient, and the sEMG device is used to collect the surface electromyogram signal sEMG(T1) of the limb during the static period, and the surface electromyogram signal sEMG(T2) of the limb during the stimulation period.

[0065] A dynamic motion intention recognition model is set in the data processing unit. The dynamic motion intention recognition model is:

[0066] (actLabel(next), actLabel) = CNN2(EEG, sEMG(T1));

[0067] In the formula, EEG is the electroencephalogram signal of the head, sEMG(T1) is the surface electromyogram signal of the limb during the resting period, actLabel is the motion intention output by the second convolutional neural network, actLabel(next) is the motion intention at the next time output by the second convolutional neural network, and CNN2 is the second convolutional neural network.

[0068] The electrical stimulation pulse in the functional electrical stimulation unit is synchronously triggered with the dynamic motion intention in the data processing unit and the signal acquisition during the stimulation period of the sEMG device through the BNC trigger line.

[0069] The delay time between the electrical stimulation pulses with a continuous transfer efficiency of the motion intention actLabel switching to the electrical stimulation pulses of the next-time motion intention actLabel(next) and the limb surface electromyogram signal sEMG(T2) generated by actLabel switching to the limb surface electromyogram signal sEMG(T2) generated by actLabel(next) in the stimulation response evaluation unit;

[0070] The continuous transfer efficiency is:

[0071] Δt = (t sEMG(T2)(next) ―t sEMG(T2) )―(t actLabel(next) ―t actLabel );

[0072] In the formula, Δt is the delay time, t actLabel is the emission time of the electrical stimulation pulse of the motion intention actLabel, t actLabel(next) is the emission time of the electrical stimulation pulse of the next-time motion intention actLabel(next), t sEMG(T2) is the time of the limb surface electromyogram signal sEMG(T2) corresponding to the motion intention actLabel, t sEMG(T2)(next) is the generation time of the limb surface electromyogram signal sEMG(T2) corresponding to the motion intention actLabel(next);

[0073] Evaluating the response degree of stroke patients to brain commands is usually determined by the delay time between the limb surface electromyogram signal sEMG(T2) in the stimulation period generated by the motion intention and the motion intention. A long delay time indicates a low response degree to the brain command. This response degree is an evaluation of a single motion intention, a static motion intention evaluation.

[0074] In the present invention, the continuous response degree of the brain commands of stroke patients is evaluated through the continuous transfer efficiency. The continuous transfer efficiency refers to the dynamic followability of the limb surface electromyogram signal sEMG(T2) in the stimulation period to the motion intention, that is, when the adjacent motion intentions change, the delay generated by the limb surface electromyogram signal sEMG(T2) in the stimulation period following the change of the motion intention. A high delay indicates a low continuous response degree to the brain command. Since the electrical stimulation pulses in the functional electrical stimulation unit are synchronously triggered with the dynamic motion intention in the data processing unit through the BNC trigger line, when quantifying the continuous transfer efficiency in the present invention, the dynamic followability of the limb surface electromyogram signal sEMG(T2) in the stimulation period to the electrical stimulation pulses is adopted, that is, when the adjacent motion intentions change, the delay generated by the limb surface electromyogram signal sEMG(T2) in the stimulation period following the change of the electrical stimulation pulses. When the delay time is less than the preset threshold, it is determined that the electrical stimulation response degree of the stroke patient is high;

[0075] When the delay time is greater than or equal to a preset threshold, it is determined that the electrostimulation response degree of the stroke patient is low.

[0076] As Figure 2 As shown, the present invention provides a stroke electrostimulation rehabilitation evaluation method, which is applied to a stroke electrostimulation rehabilitation evaluation system. The method includes the following steps:

[0077] Collect the electroencephalogram signal EEG of the head of the stroke patient and the surface electromyogram signal sEMG(T1) of the limb during the rest period, and identify the dynamic movement intention of the stroke patient according to the electroencephalogram signal EEG of the head and the surface electromyogram signal sEMG(T1) of the limb during the rest period through a dynamic movement intention recognition model;

[0078] When the dynamic movement intention is recognized, synchronously trigger the functional electrostimulation unit to emit an electrostimulation pulse matching the dynamic movement intention, and synchronously trigger the sEMG device to enter the signal acquisition during the stimulation period to obtain the surface electromyogram signal sEMG(T2) of the limb during the stimulation period;

[0079] Evaluate the continuous transmission efficiency of the brain movement command of the stroke patient according to the surface electromyogram signal sEMG(T2) during the stimulation period, and determine the electrostimulation response degree of the stroke patient.

[0080] The dynamic movement intention model construction method includes:

[0081] Collect the time series of the electroencephalogram signal EEG of the head of the stroke patient and the time series of the surface electromyogram signal sEMG(T1) of the limb during the rest period, and mark the movement intention of the stroke patient at each time on the time series to obtain the time series of the movement intention;

[0082] Take the movement intention at the previous time in the time series of the movement intention as the input of the first convolutional neural network, and take the movement intention at the next time in the time series of the movement intention as the output of the first convolutional neural network, and train the first convolutional neural network to obtain the time series prediction model of the movement intention. The time series prediction model of the movement intention is:

[0083] actLabel(next)1 = CNN1(actLabel(list));

[0084] In the formula, actLabel(list) is the movement intention at the previous time in the time series of the movement intention, actLabel(next)1 is the movement intention at the next time output by the first convolutional neural network, and CNN1 is the convolutional neural network;

[0085] Use the EEG of the head electroencephalogram signal at each time and the sEMG (T1) of the limb surface electromyogram signal during the rest period as the input of the second convolutional neural network, and use the movement intention at each time and the movement intention at the next time at each time as the output of the second convolutional neural network. Train the second convolutional neural network to obtain a dynamic movement intention model. The dynamic movement intention recognition model is as follows:

[0086] (actLabel(next), actLabel) = CNN2(EEG, sEMG(T1));

[0087] In the formula, EEG is the head electroencephalogram signal, sEMG(T1) is the limb surface electromyogram signal during the rest period, actLabel is the movement intention output by the second convolutional neural network, actLabel(next) is the movement intention at the next time output by the second convolutional neural network, and CNN2 is the second convolutional neural network;

[0088] Among them, the loss function of the dynamic movement intention recognition model is as follows:

[0089] Loss = MSE(actLabel, actLabel(list)) + MSE(actLabel(next), actLabel(next)1);

[0090] In the formula, Loss is the total loss, MSE(actLabel, actLabel(list)) is the mean square error between actLabel and actLabel(list), and MSE(actLabel(next), actLabel(next)1) is the mean square error between actLabel(next) and actLabel(next)1.

[0091] In order to further introduce temporal dynamic prediction on the basis of recognizing real-time motion intentions, the present invention uses the temporal prediction model of motion intentions as the teacher model to guide the training of the dynamic motion intention recognition model. Therefore, the loss function is set to two parts. The first part, MSE(actLabel, actLabel(list)), is to ensure the minimum error between actLabel and actLabel(list), so that the real-time motion intention predicted by the model is closest to the true intention, ensuring the accuracy of real-time motion intention recognition. The second part, MSE(actLabel(next), actLabel(next)1), is to ensure the minimum error between actLabel(next) and actLabel(next)1, so that the motion intention predicted by the model at the next time is closest to the output result of the temporal prediction model of motion intentions, enabling the model to have temporal dynamic prediction performance on the basis of recognizing real-time motion intentions, thereby ensuring that the model obtains a motion habit that conforms to the patient's individual between actLabel(next) and actLabel.

[0092] The acquisition method of the surface electromyogram signal sEMG(T2) of the limb during the stimulation period includes:

[0093] Trigger the functional electrical stimulation unit to emit electrical stimulation pulses matching the motion intention actLabel, and collect the surface electromyogram signal of the limb during the stimulation period through the sEMG device to obtain the surface electromyogram signal sEMG(T2) corresponding to the motion intention actLabel;

[0094] When the next time of the motion intention actLabel arrives, trigger the functional electrical stimulation unit to switch from emitting electrical stimulation pulses matching the motion intention actLabel to emitting electrical stimulation pulses matching the motion intention actLabel(next), and collect the surface electromyogram signal of the limb during the stimulation period through the sEMG device to obtain the surface electromyogram signal sEMG(T2) corresponding to the motion intention actLabel(next).

[0095] The calculation method of the continuous transfer efficiency includes:

[0096] Record the emission time of the electrical stimulation pulses of the motion intention actLabel, and record the emission time of the electrical stimulation pulses of the motion intention actLabel(next);

[0097] Record the generation time of the surface electromyogram signal sEMG(T2) corresponding to the motion intention actLabel, and the generation time of the surface electromyogram signal sEMG(T2) corresponding to actLabel(next);

[0098] Calculate the continuous transfer efficiency according to the calculation formula of the continuous transfer efficiency.

[0099] The method for determining the degree of electrostimulation response includes:

[0100] If the delay time is less than the preset threshold, it is determined that the degree of electrostimulation response of the stroke patient is high;

[0101] If the delay time is greater than or equal to the preset threshold, it is determined that the degree of electrostimulation response of the stroke patient is low.

[0102] The present invention collects the electroencephalogram signals of the head and the surface electromyogram signals of the limbs, so as to identify and restore the human motion intention of the user, and then stimulates the contraction of the corresponding muscles through the functional electrical stimulation system to realize the human motion intention, and evaluates the response degree of the stroke user to continuous brain motion commands in real time, so as to achieve a deeper rehabilitation effect evaluation goal.

[0103] The above embodiments are only exemplary embodiments of the present application and are not used to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art can make various modifications or equivalent replacements within the essence and protection scope of the present application, and such modifications or equivalent replacements should also be regarded as falling within the protection scope of the present application.

Claims

1. A stroke electrical stimulation rehabilitation assessment system, characterized in that: include: A data acquisition unit, used for acquiring EEG signals of the head of stroke patients, sEMG signals of the surface electromyography of the limbs in the resting period (T1), and sEMG signals of the surface electromyography of the limbs in the stimulation period (T2); A data processing unit, used for identifying the dynamic movement intention of the stroke patient from the head EEG signal and the limb surface electromyography signal sEMG (T1) in the resting period; A functional electrical stimulation unit, used for generating a functional contraction matching the dynamic movement intention by electrically stimulating the corresponding muscles of the stroke patient; The stimulation response evaluation unit is used to evaluate the continuous transmission efficiency of the brain movement commands of the stroke patient according to the limb surface electromyography signal sEMG (T2) during the stimulation period, and determine the degree of electrical stimulation response of the stroke patient.

2. A stroke electrical stimulation rehabilitation assessment system according to claim 1, characterized in that: The data acquisition unit includes an EEG device and a sEMG device. The EEG device is used to collect EEG signals of the head of a stroke patient. The sEMG device is used to collect sEMG (T1) surface electromyography signals of the limbs during a static period and to collect sEMG (T2) surface electromyography signals of the limbs during a stimulation period.

3. A stroke electrical stimulation rehabilitation assessment system according to claim 1, characterized in that: The data processing unit is provided with a dynamic motion intention recognition model, and the dynamic motion intention recognition model is: (actLabel(next),actLabel)=CNN2(EEG,sEMG(T1)); Where EEG is the head electroencephalogram signal, sEMG(T1) is the limb surface electromyography signal in the resting period, actLabel is the movement intention output by the second convolutional neural network, actLabel(next) is the movement intention at the next time output by the second convolutional neural network, and CNN is the second convolutional neural network.

4. A stroke electrical stimulation rehabilitation assessment system according to claim 1, characterized in that: The electrical stimulation pulses in the functional electrical stimulation unit are triggered synchronously with the dynamic movement intention in the data processing unit and the stimulation period signal acquisition of the sEMG device through the BNC trigger line.

5. The stroke electrical stimulation rehabilitation assessment system according to claim 1, characterized in that: The continuous transfer efficiency in the stimulus response evaluation unit is the delay time between the switching of the electrical stimulation pulse of the movement intention actLabel to the electrical stimulation pulse of the movement intention actLabel(next) at the next time and the switching of the limb surface electromyography signal sEMG(T2) generated by actLabel to the limb surface electromyography signal sEMG(T2) generated by actLabel(next); The continuous transfer efficiency is: Δt=(t sEMG(T2)(next) ―t sEMG(T2) )―(t actLabel(next) ―t actLabel ); Where Δt is the delay time, t actLabel is the emission time of the electrical stimulation pulse of the movement intention actLabel, t actLabel(next) is the emission time of the electrical stimulation pulse of the next movement intention actLabel(next), t sEMG(T2) is the time of the limb surface electromyographic signal sEMG (T2) corresponding to the movement intention actLabel, t sEMG(T2)(next) The generation time of the limb surface electromyography signal sEMG (T2) corresponding to the movement intention actLabel (next); When the delay time is less than a preset threshold, it is determined that the stroke patient has a high degree of electrical stimulation response; When the delay time is greater than or equal to a preset threshold, it is determined that the stroke patient has a low electrical stimulation response level.

6. A method for evaluating stroke electrical stimulation rehabilitation, characterized in that: A stroke electrical stimulation rehabilitation assessment system according to any one of claims 1 to 5, the method comprising the following steps: Collecting EEG signals of the head and sEMG signals of the limbs at rest (T1) of the stroke patient, and identifying the dynamic movement intention of the stroke patient based on the EEG signals of the head and sEMG signals of the limbs at rest (T1) by a dynamic movement intention recognition model; When the dynamic movement intention is recognized, the functional electrical stimulation unit is synchronously triggered to emit electrical stimulation pulses matching the dynamic movement intention, and the sEMG device is synchronously triggered to collect signals during the stimulation period, so as to obtain the limb surface electromyography signal sEMG (T2) during the stimulation period; The continuous transmission efficiency of brain motor commands in stroke patients was evaluated based on the limb surface electromyographic signal sEMG (T2) during the stimulation period, and the degree of electrical stimulation response of stroke patients was determined.

7. A stroke electrical stimulation rehabilitation assessment method according to claim 6, characterized in that: The dynamic motion intention model construction method includes: The time series of EEG of the head of the stroke patient and the time series of sEMG (T1) of the surface electromyography of the limbs in the resting period are collected, and the movement intention of the stroke patient is marked at each time in the time series to obtain the time series of the movement intention; The motion intention at the previous time in the time series of the motion intention is used as the input of the first convolutional neural network, and the motion intention at the next time in the time series of the motion intention is used as the output of the first convolutional neural network. The first convolutional neural network is trained to obtain a time series prediction model of the motion intention. The time series prediction model of the motion intention is: actLabel(next)1=CNN1(actLabel(list)); In the formula, actLabel(list) is the motion intention of the previous time in the time series of motion intention, actLabel(next)1 is the motion intention of the next time output by the first convolutional neural network, and CNN1 is a convolutional neural network; The head EEG signal at each time and the limb surface electromyography signal sEMG (T1) at rest are used as the input of the second convolutional neural network, and the movement intention at each time and the movement intention at the next time of each time are used as the output of the second convolutional neural network. The second convolutional neural network is trained to obtain a dynamic movement intention model. The dynamic movement intention recognition model is: (actLabel(next),actLabel)=CNN2(EEG,sEMG(T1)); Where, EEG is the head electroencephalogram signal, sEMG(T1) is the limb surface electromyography signal in the resting period, actLabel is the movement intention output by the second convolutional neural network, actLabel(next) is the movement intention at the next time output by the second convolutional neural network, and CNN2 is the second convolutional neural network; Among them, the loss function of the dynamic motion intention recognition model is: Loss=MSE(actLabel,actLabel(list))+MSE(actLabel(next),actLabel(next)1); Where Loss is the total loss, MSE(actLabel,actLabel(list)) is the mean square error between actLabel and actLabel(list), and MSE(actLabel(next),actLabel(next)1) is the mean square error between actLabel(next) and actLabel(next)1.

8. A stroke electrical stimulation rehabilitation assessment method according to claim 7, characterized in that: The method for collecting the surface electromyographic signal sEMG (T2) of the limbs during the stimulation period includes: The functional electrical stimulation unit is triggered to emit an electrical stimulation pulse matching the movement intention actLabel, and the surface electromyographic signal of the limbs during the stimulation period is collected through the sEMG device to obtain the surface electromyographic signal of the limbs corresponding to the movement intention actLabel (T2); When the next time of the movement intention actLabel is reached, the functional electrical stimulation unit is triggered to switch from the electrical stimulation pulse that matches the movement intention actLabel to the electrical stimulation pulse that matches the movement intention actLabel(next), and the surface electromyography signal of the limb during the stimulation period is collected through the sEMG device to obtain the surface electromyography signal sEMG(T2) of the limb corresponding to the movement intention actLabel(next).

9. A stroke electrical stimulation rehabilitation assessment method according to claim 8, characterized in that: The calculation method of the continuous transfer efficiency includes: Record the emission time of the electrical stimulation pulse of the movement intention actLabel, and record the emission time of the electrical stimulation pulse of the movement intention actLabel(next); Record the generation time of the limb surface electromyography signal sEMG(T2) corresponding to the movement intention actLabel, and the generation time of the limb surface electromyography signal sEMG(T2) corresponding to actLabel(next); The continuous transfer efficiency is obtained according to the calculation formula of the continuous transfer efficiency.

10. A stroke electrical stimulation rehabilitation assessment method according to claim 9, characterized in that: The method for determining the degree of electrical stimulation response includes: When the delay time is less than a preset threshold, it is determined that the stroke patient has a high degree of electrical stimulation response; When the delay time is greater than or equal to a preset threshold, it is determined that the stroke patient has a low electrical stimulation response level.

Images