Myoelectrically driven multi-channel functional electrical stimulation upper limb rehabilitation platform and method thereof

Abstract

The application discloses a myoelectricity-driven multi-channel functional electric stimulation upper limb rehabilitation platform and a method thereof. The platform comprises a myoelectricity collection device, a myoelectricity collection data processing system, a host computer and an electric stimulation system. The input end of the myoelectricity collection data processing system is connected with the myoelectricity collection device. The output end of the myoelectricity collection data processing system is connected with the input end of the host computer in a wireless or wired mode. The output end of the host computer is connected with the myoelectricity collection device and the electric stimulation system in a wireless or wired mode. The myoelectricity collection device collects myoelectricity signals on the surface of a healthy limb of a patient. The myoelectricity collection data processing system filters and processes the collected myoelectricity signals of the healthy limb of the patient to obtain myoelectricity data. The host computer performs gesture recognition processing according to the myoelectricity data to obtain electric stimulation parameter configuration and sends the electric stimulation system. The electric stimulation system outputs matched waveform stimulation signals on the surface of a diseased limb of the patient according to the electric stimulation parameter configuration. The application realizes bilateral fine movement by subjective movement control of the healthy limb of the patient on the diseased limb, and accelerates the rehabilitation process.

Description

TECHNICAL FIELD

[0001] The present application relates to a medical rehabilitation platform, in particular to a myoelectrically driven multi-channel functional electrical stimulation upper limb rehabilitation platform and a method thereof. BACKGROUND

[0002] Stroke is a neurological disease that easily occurs when the blood supply to the central nervous system is abnormal, which interferes with the normal physiological activities of the central nervous system and has the characteristics of high disability rate and high mortality rate. Hemiplegia is a common complication after stroke, which manifests as unilateral limb movement disorder in patients, and seriously affects the daily life activities of patients. Reducing the degree of disability to the greatest extent and restoring self-care are the primary needs of such patients.

[0003] Functional electrical stimulation technology provides a more effective rehabilitation approach than traditional motor rehabilitation for hemiplegic patients. Its working principle is to use the electrical excitability of nerve cells, simulate the nerve impulses transmitted by the central nervous system through externally applied current stimulation pulses, change the membrane potential of nerve cells, form action potentials, and propagate along nerve fibers, causing paralyzed muscles to contract, promoting muscle recovery, and inducing central nervous system plasticity. Multiple stimulation channels can achieve coordinated movement of multiple muscle groups and complete fine and complex functional movements. Studies have shown that combining functional electrical stimulation with intention-based movement that integrates the patient's subjective intention, focusing the patient's attention on the affected limb, guiding the synchronous movement of the affected and unaffected limbs, can promote the balance of excitation and inhibition between the cerebral hemispheres, and significantly improve the rehabilitation effect. Therefore, by collecting the surface electromyogram of the unaffected limb of a hemiplegic patient, extracting active movement information, and performing electrical stimulation on the affected limb according to the movement results to achieve similar movements to the unaffected limb, the rehabilitation efficiency can be effectively improved. However, due to the limitation of the number of channels of the electrical stimulation device, this therapy cannot achieve complex and fine motor movements that require the joint action of multiple muscle groups. Functional electrical stimulation therapy can be applied to the rehabilitation of hemiplegic patients caused by stroke and has produced positive therapeutic effects in clinical practice. In particular, the contralateral control type electrical stimulation therapy, which controls the movement of the affected limb through the electromyogram of the unaffected limb, can further improve the treatment efficiency and promote the balance of excitation and inhibition between the left and right cerebral hemispheres. However, the electromyogram acquisition device of this type of therapy is currently low in portability and inconvenient to wear, and the electrical stimulation device has a small number of channels and low stimulation precision, which cannot achieve relatively fine and complex functional movements.

[0004] Surface electromyography has the characteristics of non-stationary, low amplitude, low signal-to-noise ratio, and the amplitude level is positively correlated with the contraction degree and number of the muscle under the electrode, which can effectively reflect the muscle movement state and evaluate muscle activity. In the aspect of surface electromyography collection, it is necessary to provide a collection circuit with high common-mode rejection ratio, and to perform data filtering and other processing at the hardware and software levels to remove other physiological and external environmental interference signals. The current common surface electromyography collection device has low portability, many connection cables, and the wearing method needs to be improved, and cannot efficiently realize motion action recognition. In the aspect of limb motion action recognition, it is necessary to extract various electromyography features in time domain, frequency domain and time-frequency domain, and to perform feature dimension reduction and feature classification to obtain a high-performance machine learning model and apply it to real-time motion action discrimination. SUMMARY

[0005] In view of the above problems, the present application provides an electromyography-driven multi-channel functional electrical stimulation upper limb rehabilitation platform and a method thereof. The platform is composed of eight groups of collection units surrounding the arm to form an arm ring, which is connected to an electromyography collection data processing system. The platform has the characteristics of small size, high portability and convenient wearing. At the same time, the present application supports twelve independent programmable electrical stimulation waveform outputs through a multi-channel functional electrical stimulation system, and can realize relatively fine motor actions according to multi-channel coordinated stimulation. The present application realizes bilateral fine movement by subjective movement control of the patient's healthy side limb to the affected side limb, which is expected to further improve the treatment efficiency of hemiplegic patients and speed up the rehabilitation process.

[0006] In order to solve the problems of the prior art, the present application adopts the following technical solutions:

[0007] An electromyography-driven multi-channel functional electrical stimulation upper limb rehabilitation platform, comprising an electromyography collection device, an electromyography collection data processing system, a host computer and an electrical stimulation system; the input end of the electromyography collection data processing system is connected with the electromyography collection device; the output end thereof is connected with the input end of the host computer in a wired or wireless manner; the output end of the host computer is connected with the electrical stimulation system in a wired or wireless manner; wherein:

[0008] The electromyography collection device collects surface electromyography signals of the patient's healthy side limb.

[0009] The electromyography collection data processing system filters and processes the collected electromyography signals of the patient's healthy side limb to obtain electromyography data.

[0010] The host computer identifies and processes the electromyography data to obtain electrical stimulation parameter configuration and sends the electrical stimulation system.

[0011] The electrical stimulation system outputs matching waveform stimulation signals to the surface of the patient's affected side limb according to the electrical stimulation parameter configuration.

[0012] Further, the myoelectric collection data processing system comprises a power module, an analog signal processing module, a main control module and a communication module; wherein:

[0013] The power module is composed of a lithium battery charging circuit, a 3.3V conversion circuit and a ±2.5V conversion circuit; wherein: the lithium battery charging circuit uses USB to charge the battery, the 3.3V conversion circuit converts the 5V voltage input by the USB or the 3.7V voltage of the battery into a stable 3.3V voltage, which is used for the main control single-chip microcomputer and the USB-to-serial module; the ±2.5V conversion circuit further converts the 3.3V voltage into a stable 2.5V and -2.5V voltage, which is used for the analog front-end chip;

[0014] The analog signal processing module is composed of a pre-processing circuit, an analog amplification chip and a peripheral circuit; wherein: the analog signal processing module amplifies and converts the myoelectric analog signal after low-pass filtering the myoelectric signal of the healthy side of the patient's limb according to the following formula:

[0015]

[0016] Wherein, fc is the filter cutoff frequency, R is the filter resistance value, and C is the filter capacitance value;

[0017] The main control module performs digital filtering processing on the myoelectric analog signal to obtain myoelectric data according to the following formula:

[0018]

[0019] Wherein, y(n) is the filtered sequence, x(n) is the pre-filtered sequence, a i ,b i are filter coefficients respectively.

[0020] Further, the upper computer comprises a data connection module, a myoelectric display module, a gesture recognition module and an electrical stimulation control module, wherein:

[0021] The data connection module is used to control the connection with the myoelectric collection device, the myoelectric collection data processing system and the electrical stimulation system;

[0022] The myoelectric display module supports user interactive control for the waveform display area of the myoelectric data;

[0023] The gesture recognition module extracts the rehabilitation action feature vector from the myoelectric data based on the gesture recognition algorithm of support vector machine, constructs an individualized rehabilitation model and outputs the action classification result;

[0024] The electrical stimulation control module loads the electrical stimulation parameters corresponding to various actions pre-saved and sends them to the electrical stimulation system through a serial port protocol.

[0025] Further, the gesture recognition module extracts the rehabilitation action feature vector from the electromyography data based on the gesture recognition algorithm of the support vector machine to construct an individual rehabilitation model and output the action classification result process; comprising:

[0026] The gesture recognition module is composed of an action label unit, an offline training unit and an online recognition unit; wherein:

[0027] The action label part generates an action instruction sequence according to the action category and action time set by the patient, and the patient makes corresponding actions according to the action prompt label or picture of the action instruction sequence, and generates corresponding action label data after the action is stable to divide the trial;

[0028] The offline training unit first divides the electromyography data according to the action label; secondly, the features in the time domain, frequency domain and time-frequency domain of each trial are extracted by sliding time window, and are spliced into a feature matrix; after obtaining the sample data, the feature matrix is input into the SVM classifier for training and generating an individual rehabilitation model;

[0029] The online recognition unit buffers the electromyography data of a specific time length in real time and extracts the features, inputs the classification model trained in the offline stage for recognition and outputs the classification result.

[0030] Further, the electrical stimulation system is composed of a lithium battery module, a power management module, a main control module, an electrical stimulation module, a stimulation electrode sheet and a communication module

[0031] The lithium battery module includes a lithium battery pack composed of multiple lithium batteries and a power meter unit, and the output voltage range is 11.1-12.6V, connected to the battery port of the power management module;

[0032] The power management module is used for battery charging, power path management and voltage conversion;

[0033] The main control module is used for receiving and analyzing the data packets of each communication module for stimulation parameters and stimulation control commands, parameter legality verification and electrical stimulation parameter setting;

[0034] The electrical stimulation module is composed of an electrical stimulation circuit board, each of which is based on a current source and an H-bridge architecture to realize constant current driving of the waveform, and the built-in controller of each stimulation circuit board can independently adjust each channel waveform.

[0035] The communication module uses a serial-to-USB circuit, an ESP32 wireless module and a CAN communication to realize the setting of electrical stimulation parameters by multiple devices.

[0036] Further, the electrical stimulation module is composed of 3 groups of electrical stimulation circuit boards to realize synchronous independent output of twelve-channel electrical stimulation waveforms.

[0037] Further, the myoelectricity collecting device is an arm ring connected by eight collecting units; each arm ring is composed of a first shell, a second shell and a PCB electrode plate; the second shell is provided with a buckle electrode; each collecting unit is connected with the myoelectricity collecting data processing system through a flexible flat cable.

[0038] The application can also be implemented by adopting the following technical scheme: the platform performs a myoelectricity-driven multi-channel functional electric stimulation upper limb rehabilitation method, including the following steps:

[0039] The myoelectricity collecting device collects surface myoelectricity signals of a patient's limb, and transmits myoelectricity data to an upper computer wirelessly after being processed by a myoelectricity collecting data processing system;

[0040] The upper computer receives myoelectricity data for data processing and waveform display, further extracts time domain, frequency domain and time-frequency domain myoelectricity characteristics, performs offline training and online identification of gestures by a machine learning method, and loads a pre-configured channel electric stimulation parameter configuration file under a specific action according to an identification result;

[0041] The electric stimulation device receives an instruction of the upper computer, realizes electric stimulation channel selection, electric stimulation parameter setting and start-stop control of electric stimulation waveform output, and realizes a target action.

[0042] Beneficial effects

[0043] Compared with the prior art, the application has the following beneficial effects:

[0044] 1. The application solves the technical problems of overcoming the defects of few channels, poor stimulation precision and low portability of the existing functional electric stimulation equipment, collects surface myoelectricity signals of a patient's healthy limb to obtain autonomous motion information by a wearable eight-channel wireless surface myoelectricity collecting device, applies a highest twelve-channel electric stimulation waveform to a diseased limb, realizes synchronous action with the healthy limb, provides a contralateral control functional electric stimulation rehabilitation system based on surface myoelectricity and fusion of autonomous motion control intention for a hemiplegic patient caused by a stroke, and effectively improves the rehabilitation treatment efficiency and accelerates the rehabilitation process of the patient.

[0045] 2. The application is applied to the upper limb rehabilitation scene of a hemiplegic patient, designs a portable eight-channel wireless myoelectricity collecting device, adopts a wearable electrode wearing mode, integrates a patient's subjective motion control intention into a treatment process through surface myoelectricity, and designs a twelve-channel functional electric stimulation device, which can realize coordinated and orderly contraction of target muscles and complete complex and fine motion control by reasonably setting stimulation electrode sites, parameters of each electric stimulation channel and a stimulation time sequence, and has important significance for clinical rehabilitation of a hemiplegic and stroke patient.

[0046] 3, The application is a portable myoelectric driving multi-channel functional electric stimulation upper limb rehabilitation system, which is worn on the target area of the arm through an arm ring connection mode, collects the surface myoelectricity of the healthy side of a stroke hemiplegic patient, obtains the movement intention information of the limb, and synchronously stimulates the muscles of the affected side of the limb through a functional electric stimulation device, realizes similar actions with the healthy side of the limb, makes up for the loss of the function of the affected limb of the patient, promotes the balance of the excitation and inhibition of the two hemispheres of the brain, maximally induces the neural plasticity of the central system, effectively accelerates the rehabilitation process of the patient, and also provides a new idea and method for the rehabilitation of the lower limbs. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 The figure is a structure schematic diagram of the myoelectricity collection device in the application. In the figure, 101 is a first collection electrode unit, 102 is a second collection electrode unit, 103 is a buckle electrode, 104 is a soft wire interface, and 105 is an elastic belt connection port.

[0048] Figure 2 The figure is a structure schematic diagram of an eight-channel wireless myoelectricity collection data system in the application;

[0049] Figure 3 The figure is a structure schematic diagram of a twelve-channel functional electric stimulation system in the application;

[0050] Figure 4 The figure is a block diagram of a host computer in the application; DETAILED DESCRIPTION

[0051] The following will be described in combination with the accompanying Figure 1 ~ the accompanying Figure 4 The application is described as follows:

[0052] The application provides a myoelectricity driving multi-channel functional electric stimulation upper limb rehabilitation platform, which is composed of an eight-channel wireless myoelectricity collection device, a myoelectricity collection data processing system, a twelve-channel functional electric stimulation system and a host computer software.

[0053] For example Figure 1As shown, the eight-channel wireless electromyography acquisition device is an arm ring connected by eight acquisition units; each of the arm rings is composed of a first shell, a second shell and a PCB electrode plate; the second shell is provided with a button electrode; each acquisition unit is connected with the electromyography acquisition data processing system through a flexible flat cable. In order to ensure good signal quality, the electrode of the application adopts a 3.5mm button type disposable gel patch electrode, which is different from the traditional long cable connection, and a new type of electromyography electrode connection mode is proposed. The male buckle of the two patch electrodes of each channel is connected with the female buckle welded on the electrode plate, and the eight-channel electrode plate is cascaded into an arm ring form and worn on the target acquisition area of the arm. The other parts of the device are integrated on a main control circuit board, which is also connected to the electrode arm ring through a flexible flat cable. The main control circuit board is also designed with an electrode female buckle for installing a bias driving electrode, which eliminates the bondage of traditional electrode cables, improves the common mode rejection ratio of the system, greatly reduces the size and weight of the device, and improves the convenience. Among them:

[0054] Figure 1 The middle 101 and 102 constitute a shell of an acquisition channel, containing a PCB electrode plate, and 103 is a 3.5mm button electrode interface. The data processing and communication part of the acquisition device is integrated on a main control circuit board through the flexible flat cable at 104, which is also connected to the electrode arm ring. The main control circuit board is also designed with an electrode female buckle for installing a bias driving electrode, which eliminates the bondage of traditional electrode cables, improves the common mode rejection ratio of the system, greatly reduces the size and weight of the device. The device is worn on the forearm through an elastic band, can transmit the collected electromyography signals in a wireless manner, and has the characteristics of small size, light weight and high portability. The electromyography acquisition device designed in the application applies a new electrode wearing method to the patch gel electrode, improving the portability of the system. The device is not only suitable for motion recognition needs in rehabilitation therapy, but also widely applicable to electromyography acquisition needs in other scenarios.

[0055] As shown in Figure 2 The wireless electromyography acquisition data processing system is composed of a power module, an analog front-end module, a main control module and a communication module. The wireless electromyography acquisition data processing system uses a single lithium battery or USB power supply, wherein:

[0056] The power module is composed of a lithium battery charging circuit, a 3.3V conversion circuit and a ±2.5V conversion circuit; the power module includes a lithium battery charging circuit, a 3.3V conversion circuit and a ±2.5V conversion circuit, the lithium battery charging circuit uses a TP4056 chip, the battery can be charged through USB, and the voltage conversion circuit converts the lithium battery or USB voltage into 3.3V or ±2.5V for the main control single-chip microcomputer, the communication module and the analog front-end chip.

[0057] The analog signal processing module consists of a preprocessing circuit, an analog amplifier chip, and peripheral circuits. The preprocessing circuit receives the electromyographic signals acquired by the electrodes and filters them using a passive RC low-pass filter to remove some high-frequency signal components. Its cutoff frequency f c Calculated by the following formula:

[0058]

[0059] The analog amplifier chip amplifies the pre-processed analog signal, converts it into a digital signal through an ADC with a specific sampling rate, and transmits it to the main control microcontroller for further processing.

[0060] The main control module consists of a microcontroller and peripheral circuits. The microcontroller uses an ARM Cortex-M4F core and has peripherals such as a floating-point unit, serial port, and SPI. The microcontroller reads the electromyography (EMG) signals acquired by the analog front-end chip and uses a bandpass and a 50Hz notch filter (fourth-order IIR filter) to filter out noise and interference signals, retaining the EMG data within the frequency band where the EMG energy is mainly distributed. The calculation formula for the fourth-order IIR filter algorithm is as follows:

[0061]

[0062] Where y(n) is the filtered sequence, x(n) is the unfiltered sequence, and a i ,b i These are the filter coefficients.

[0063] The communication module has two communication methods: USB serial port and BLE wireless. The former uses a wired connection to the PC's serial port, while the latter transmits electromyographic data wirelessly to the host computer software for further processing.

[0064] like Figure 3 As shown, the twelve-channel functional electrical stimulation system consists of a lithium battery module, a power management module, a main control module, an electrical stimulation module, stimulation electrodes, and a communication module. Its key features include support for independent adjustment of twelve-channel electrical stimulation parameters and independent waveform output, and compatibility with various external communication methods.

[0065] The lithium battery module includes a lithium battery pack composed of multiple lithium batteries and a fuel gauge unit. Its output voltage range is 11.1–12.6V, and it is connected to the battery port of the power management module. The fuel gauge module is used to calculate the battery charge and provide multiple safety protections for battery charging and discharging.

[0066] The power management module includes a power path management circuit and a voltage conversion circuit. The core of the power path management circuit is a power management chip, which has two functions: lithium battery charging and system power path management. The voltage conversion circuit converts the adapter's 12V voltage into stable 5V and 3.3V voltages to power other modules.

[0067] The master module adopts an ARM Cortex-M4F core single-chip microcomputer, and has multiple peripherals such as GPIO, timer, serial port, SPI, IIC, CAN, etc. The master module controls the operation of the functional electrical stimulation device, and the main tasks include receiving and analyzing the data packets of the stimulation parameters and stimulation control commands of each communication module, parameter legality verification, and electrical stimulation parameter setting.

[0068] The electrical stimulation module is composed of multiple electrical stimulation circuit boards. Each stimulation circuit board supports four-way arbitrary programmable waveform output. To adapt to the change of the impedance between electrodes, each stimulation circuit board realizes constant current driving of the waveform based on a current source and an H-bridge architecture. Meanwhile, the current amplitude, frequency and other parameters of each waveform can be independently adjusted through the built-in controller of each stimulation circuit board, meeting the needs of upper limb electrical stimulation rehabilitation scenarios.

[0069] The stimulation electrode is a gel electrode patch for physiotherapy, which is attached to the target muscle. The communication module is a serial-to-USB circuit, an ESP32 wireless module and a CAN communication interface, which can realize the setting of electrical stimulation parameters by various devices through wired and wireless ways.

[0070] The system has three electrical stimulation modules, supporting twelve-channel electrical stimulation waveform synchronous independent output. By designing different electrode attachment positions and electrical stimulation waveform parameters of each channel, the multiple muscles of the arm can produce different degrees of contraction in a specific time sequence, which can realize complex and fine rehabilitation movements. In addition, upper limb rehabilitation is one of the application scenarios of the device. By adjusting the amplitude, pulse width, frequency and other parameters of the stimulation waveform, the rehabilitation treatment of lower limb paralysis, swallowing disorders and other scenarios can be met.

[0071] As shown in Figure 4 The host computer software includes a data connection module, an electromyographic display module, a gesture recognition module and an electrical stimulation control module. Among them:

[0072] The data connection module manages all devices connected to the host computer, including an eight-channel wireless electromyographic acquisition device and a twelve-channel functional electrical stimulation device, for controlling the connection and disconnection of devices, connection parameters, etc.

[0073] The electromyographic display module receives and analyzes the electromyographic data packets sent by the eight-channel wireless electromyographic acquisition device in real time and draws them in the waveform display area. The display of the waveform supports user interaction control, including display amplitude adjustment and display time range adjustment. The module also integrates low-pass, high-pass and notch filters, which can be controlled by the user.

[0074] The gesture recognition module is divided into three parts, action label part, offline training part and online recognition part. The action label part generates an action instruction sequence according to the action category and action time set by the user, the patient makes corresponding actions according to the action prompt label or picture of the action instruction sequence, and the corresponding action label data is generated after the action is stable to divide the trial. In the offline training part, first, the eight-channel electromyography data is divided according to the action label; second, the features in the time domain, frequency domain and time-frequency domain of each trial are extracted by sliding time window, and are spliced into a feature matrix; after obtaining sufficient sample data, the feature matrix is input into the SVM classifier for training and generating a model. In the online recognition stage, the eight-channel electromyography data of a specific time length is cached in real time and the features are extracted, which are input into the classification model trained in the offline stage for recognition and output of the classification result.

[0075] The function of the electrical stimulation control module is to receive the electrical stimulation parameter values and commands input by the user, and send them to the twelve-channel functional electrical stimulation device through the serial port protocol to complete the stimulation parameter setting. The upper computer supports pre-defined electrical stimulation treatment plan management, which can save the stimulation parameters as configuration files to avoid manual repeated setting. During the treatment process, the module can receive the action classification results output by the gesture recognition module, and load the pre-saved electrical stimulation parameter files corresponding to various actions according to the results, and immediately apply corresponding electrical stimulation to the affected limb, so as to realize the synchronous action of the healthy limb driving the affected limb.

[0076] (3) Upper computer software

[0077] As shown in Figure 3 , the upper computer software includes a data connection module, an electromyography display module, a gesture recognition module and an electrical stimulation control module.

[0078] The data connection module manages the devices connected to the upper computer, including the eight-channel electromyography acquisition device and the twelve-channel functional electrical stimulation device, and controls the connection and disconnection, connection parameters, etc. of the devices.

[0079] The electromyography display module receives and analyzes the electromyography data packets sent by the eight-channel electromyography acquisition device in real time and draws them on the waveform display area. The display of the waveform supports user interaction control, including display amplitude adjustment and display time range adjustment. The electromyography display module also integrates low-pass, high-pass and notch filters, which can be controlled by the user.

[0080] The gesture recognition module is divided into three parts, action label part, offline training part and online recognition part. The action label part generates action instruction sequence according to the action category and action time set by the user, the patient makes corresponding action according to the action prompt label or picture of the action instruction sequence, and the corresponding action label data is generated after the action is stable to divide the trial. In the offline training part, first, the trial data is divided according to the action label and eight-channel electromyography data; second, the feature extraction in time domain, frequency domain and time-frequency domain is carried out on the electromyography data of each trial through the sliding time window method, and the feature matrix is spliced; after obtaining sufficient sample data, the classifier based on SVM is input for training and the model is generated. In the online recognition stage, the eight-channel electromyography data of a specific time length is cached in real time and the feature matrix is calculated, then the trained machine learning model is input for recognition and the classification result is output.

[0081] The electric stimulation control module is used for receiving the electric stimulation parameter value and command input by the user, and sending to the eight-channel electromyography acquisition device or twelve-channel functional electric stimulation system through serial port protocol to complete the stimulation parameter setting. The upper computer supports the management of pre-defined electric stimulation treatment scheme, and the stimulation parameters which have achieved good effect can be saved as electric stimulation parameter configuration file to avoid manual repeated setting. During the treatment process, the module can receive the action classification result output by the gesture recognition module, load the electric stimulation parameter file corresponding to various actions according to the result, and immediately start the electric stimulation waveform output, so as to realize the synchronous action of the healthy limb driving the affected limb.

[0082] Although the present application is described above, the present application is not limited to the above-mentioned specific embodiments, the above-mentioned specific embodiments are only illustrative but not restrictive, and those skilled in the art can make many modifications under the inspiration of the present application without departing from the purpose of the present application, and these all belong to the protection of the present application.

Claims

1. A multi-channel functional electrical stimulation upper limb rehabilitation platform driven by electromyography, the platform comprising an electromyography acquisition device, an electromyography acquisition data processing system, an upper computer and an electrical stimulation system; characterized in that The input end of the myoelectric collection data processing system is connected with the myoelectric collection device; The output end of the upper computer is connected with the electric stimulation system through wired or wireless mode; wherein: The myoelectric collection device collects myoelectric signals on the surface of the healthy limb of the patient; The myoelectric collection data processing system filters the collected myoelectric signals to obtain myoelectric data; The upper computer identifies the myoelectric data to obtain electric stimulation parameters, and sends the electric stimulation system; the upper computer comprises a data connection module, a myoelectric display module, a gesture recognition module and an electric stimulation control module, wherein: The data connection module is used for controlling the connection with the myoelectric collection device, the myoelectric collection data processing system and the electric stimulation system; The myoelectric display module draws a waveform display area of the myoelectric data to support user interaction control; The gesture recognition module extracts a feature vector from the myoelectric data based on a support vector machine gesture recognition algorithm, constructs an individualized rehabilitation model and outputs a motion classification result; The electric stimulation control module loads pre-stored electric stimulation parameters corresponding to various motions and sends them to the electric stimulation system through a serial port protocol; The gesture recognition module extracts a rehabilitation motion feature vector from the myoelectric data based on a support vector machine gesture recognition algorithm, constructs an individualized rehabilitation model and outputs a motion classification result; the process comprises: The gesture recognition module comprises a motion label unit, an offline training unit and an online recognition unit; wherein: The motion label unit generates a motion instruction sequence according to the motion categories and motion time set by the patient, the patient makes corresponding motions according to the motion prompt labels or pictures of the motion instruction sequence, and corresponding motion label data is generated after the motion is stable to divide the test; The offline training unit first divides the myoelectric data according to the motion labels; secondly, it extracts features in the time domain, frequency domain and time-frequency domain of each test through a sliding time window, and splices them into a feature matrix; after obtaining sample data, the feature matrix is input into an SVM classifier for training and an individualized rehabilitation model is generated; The online recognition unit buffers myoelectric data of a specific time length in real time and extracts features, inputs the classification model trained in the offline stage for recognition and outputs a classification result; The electric stimulation system outputs matching waveform stimulation signals on the surface of the diseased limb of the patient according to the electric stimulation parameter configuration.

2. The myoelectrically driven multi-channel functional electrical stimulation upper limb rehabilitation platform according to claim 1, characterized in that: The myoelectric collection data processing system comprises a power module, an analog signal processing module, a main control module and a communication module; wherein: The power module comprises a lithium battery charging circuit, a 3.3V conversion circuit and a ±2.5V conversion circuit; wherein: the lithium battery charging circuit uses USB to charge the battery, the 3.3V conversion circuit converts the 5V voltage input by the USB or the 3.7V voltage of the battery into a stable 3.3V voltage for use by the main control microcontroller and the USB-to-serial module; the ±2.5V conversion circuit further converts the 3.3V voltage into a stable 2.5V and -2.5V voltage for use by the analog front-end chip; The analog signal processing module is composed of a pre-processing circuit, an analog amplification chip and a peripheral circuit; wherein: the analog signal processing module performs low-pass filtering on the surface electromyography signal of the healthy side of the patient's limb and then amplifies and converts the electromyography analog signal according to the following formula: ； Wherein, fc is the filter cutoff frequency, R is the filter resistance value, and C is the filter capacitance value The main control module performs digital filtering on the electromyography analog signal to obtain electromyography data according to the following formula: ； wherein is the filtered sequence, is the pre-filtered sequence, are filter coefficients, respectively. 3.The myoelectrically driven multi-channel functional electrical stimulation upper limb rehabilitation platform of claim 1, wherein: The electric stimulation system is composed of a lithium battery module, a power management module, a main control module, an electric stimulation module, a stimulation electrode sheet and a communication module The lithium battery module includes a lithium battery pack composed of multiple lithium batteries and a power meter unit, and has an output voltage range of 11.1-12.6V, and is connected to the battery port of the power management module; The power management module is used for battery charging, power path management and voltage conversion; The main control module is used for receiving and analyzing data packets of each communication module for stimulation parameters and stimulation control commands, parameter legality verification and electric stimulation parameter setting; The electric stimulation module is composed of an electric stimulation circuit board, each of which is based on a current source and an H-bridge architecture to realize constant current driving of the waveform, and the built-in controller of each stimulation circuit board can independently adjust each channel waveform. The communication module uses a serial-to-USB circuit, an ESP32 wireless module and CAN communication to realize the setting of electric stimulation parameters by multiple devices.

4. The myoelectrically driven multi-channel functional electrical stimulation upper limb rehabilitation platform according to claim 3, characterized in that: The electric stimulation module is composed of three groups of electric stimulation circuit boards, realizing synchronous independent output of twelve channels of electric stimulation waveforms.

5. The myoelectrically driven multi-channel functional electric stimulation upper limb rehabilitation platform of claim 1, wherein the myoelectricity collection device is composed of eight collection units connected to form an arm ring; each arm ring is composed of a first shell, a second shell and a PCB electrode plate; the second shell is provided with a buckle electrode; and each collection unit is connected to the myoelectricity collection data processing system through a flexible flat cable.

6. The method of claim 1-5, wherein the platform is used for the rehabilitation of the upper limbs by means of multi-channel functional electrical stimulation driven by myoelectric signals. , comprising the following steps: The myoelectricity collection device collects the surface electromyography signal of the patient's limb, and transmits the electromyography data wirelessly to the upper computer after being processed by the myoelectricity collection data processing system; The upper computer receives the electromyography data for data processing and waveform display, further extracts time domain, frequency domain and time-frequency domain electromyography features, performs offline training and online recognition of gestures through a machine learning method, and loads a pre-configured channel electric stimulation parameter configuration file under a specific action according to the recognition result; The electric stimulation system receives the instructions of the upper computer, realizes the selection of electric stimulation channels, the setting of electric stimulation parameters and the start-stop control of electric stimulation waveform output, and realizes the target action.

Images