KR20230070431A - Functional electrical stimulation generate system using electromyogram signal - Google Patents

Abstract

An electrical stimulation generating system, according to an embodiment of the present invention, collects an electromyogram (EMG) signal generated by responding to electrical stimulation from a body to control and generate a functional electrical stimulation signal. The electrical stimulation generating system comprises: a voluntary/involuntary muscular contraction detecting unit which extracts a characteristic vector from a frequency area of the EMG signal, applies an artificial intelligence model, and classifies and detects a voluntary muscular contraction signal and an involuntary muscular contraction signal from the characteristic vector; an involuntary muscular contraction signal removing unit which removes the involuntary muscular contraction signal from the EMG signal according to the detection result; and a functional electrical stimulation control unit which based on the EMG signal having the involuntary muscular contraction signal removed therefrom, generates the functional electrical stimulation signal to be applied to the body.

Description

Functional electrical stimulation generation system using EMG signals

The present invention relates to a functional electrical stimulation generating system, and more particularly, to an electrical stimulation generating system for generating a functional electrical stimulation (FES) signal based on an EMG signal.

Electromyography (EMG), which measures the degree of muscle activity by measuring the potential difference generated in muscle cells when muscles are activated, is widely used not only in the medical field but also in the biomechanics field. EMG technology has been developed according to the configuration of electrodes that measure the potential difference of activated muscles, and a commonly used form is EMG equipment in the form of electrodes attached to the skin surface. In addition, electrical stimulation technology is a technology that artificially induces muscle contraction by applying electrical stimulation to muscles in the form of constant current or constant voltage. Electrical stimulation technology has mainly been developed as a functional electrical stimulation (FES) technology that supplements and replaces weakened or lost muscle functions.

Functional electrical stimulation (FES) has been known as the most effective rehabilitation therapy that can be generally performed in hospitals. For treatment using functional electrical stimulation (FES), rehabilitation experts apply electrical stimulation to the affected area while voluntary muscle contraction occurs. Rehabilitation experts perform the procedure by visually determining whether the patient is maintaining or starting muscle contraction and turning on the power of the FES device. In general FES equipment, when a user applies a certain amount of force, it is driven in such a way that electrical stimulation comes out.

Therefore, for rehabilitation treatment using FES, one rehabilitation specialist has no choice but to handle one patient. Therefore, when a large number of patients are present, even if the FES equipment is used, a shortage of manpower is unavoidable. Even in rehabilitation treatment in a home environment using an FES device, it is difficult to proceed with the most effective FES rehabilitation treatment without a rehabilitation specialist. It is difficult to treat several patients because the rehabilitation specialist must determine whether the patient is maintaining or starting voluntary muscle contraction. In conventional products or technologies, since electrical stimulation is controlled to come out when a user applies a certain force or more, the equipment does not recognize whether or not the muscle contracts after the electrical stimulation is output. Since a device for automatically applying electric stimulation when a patient's voluntary muscle contraction occurs is required, a technique for analyzing and determining an input signal is required. Therefore, there is a demand for FES technology that enables patients to perform effective FES rehabilitation treatment on their own without a rehabilitation expert.

Republic of Korea Patent Publication No. 10-2018-0074597

The present invention is to solve the above-described technical problem, the present invention is to provide an electrical stimulation generating system for generating functional electrical stimulation (FES) based on muscle stimulation signals. When muscles are stimulated by electrical stimulation, involuntary muscle contractions also occur. It is to provide an effective FES generation system using artificial muscle contraction signals.

In the electrical stimulation generating device for controlling and generating a functional electrical stimulation signal by collecting EMG signals (EMG) generated in response to electrical stimulation from the body according to an embodiment of the present invention: a characteristic vector in the frequency domain of the EMG signal a voluntary/involuntary muscle contraction detection unit that extracts and detects a voluntary muscle contraction signal and an involuntary muscle contraction signal separately from the feature vector by applying an artificial intelligence model; an involuntary muscle contraction signal removing unit configured to remove the involuntary muscle contraction signal from the EMG signal according to the detection result; and a functional electrical stimulation control unit generating the functional electrical stimulation signal to be applied to the body based on the EMG signal from which the involuntary muscle contraction signal has been removed.

In this embodiment, the device further includes a muscle activity intensity calculation unit for calculating a Root Mean Square (RMS) of the EMG signal from which the involuntary muscle export signal has been removed, and the functional electrical stimulation control unit comprises the An effective value and a threshold value are compared, and the functional electrical stimulation signal to be applied to the body is generated according to the comparison result.

In this embodiment, the feature vector includes at least one of a percentile spectral cumulative sum (PoSCS) and a log power spectrum detected in the frequency domain of the EMG signal (EMG).

In this embodiment, the involuntary muscle contraction signal removal unit includes: a window unit selecting a window of the EMG signal (EMG); and a fast Fourier transform unit for processing signals included in the selected window through fast Fourier transform. includes

In this embodiment, the involuntary muscle contraction signal removal unit includes: a magnitude and phase calculation unit for calculating the magnitude and phase of the signal output from the fast Fourier transform unit; more includes

In this embodiment, the involuntary muscle contraction signal removal unit includes: a peak detection unit for detecting a peak in the magnitude of the signal; and a peak remover removing a noise signal corresponding to the detected peak. more includes

According to the electrical stimulation generating system according to an embodiment of the present invention, a functional electrical stimulation (FES) signal can be generated by distinguishing a voluntary muscle contraction signal and an involuntary muscle contraction signal from an EMG signal (EMG) with high accuracy. Therefore, it is possible to implement an electrical stimulation generating system that provides high-accuracy functional electrical stimulation (FES) without a rehabilitation specialist or expensive equipment.

1 is a block diagram exemplarily showing an electrical stimulation generating system according to an embodiment of the present invention.
FIG. 2 is a block diagram showing the configuration of the electrical stimulation generating system of FIG. 1 as an example.
3 is a flowchart showing a processing method in a frequency domain for extracting a percentile spectral cumulative sum (PoSCS) as an example of feature extraction.
4 is a graph showing a method of extracting a percentile spectral cumulative sum (PoSCS).
5 shows probability density functions (PDFs) representing the result of extracting the percentile spectral cumulative sum (PoSCS) for each frequency from the EMG signal (EMG).
6 is a flowchart illustrating a learning method of an artificial intelligence operation unit for separating voluntary muscle contraction signals and involuntary muscle contraction signals according to an embodiment of the present invention.
7 is a diagram briefly showing the structure of the LSTM algorithm for identifying voluntary and involuntary muscle contraction signals through sequential electromyogram (EMG) data in the time domain of the present invention.
8 is a flowchart showing an actual operation and a test operation of the voluntary/involuntary muscle contraction detection unit of the present invention.
FIG. 9 is a block diagram showing the involuntary muscle contraction signal removal unit shown in FIG. 2 as an example.
10 and 11 are graphs showing frequency analysis results of EMG data in the peak detection unit and the peak suppression unit 1126 .
12 is a graph showing a waveform (red) with pre-processing and a waveform (black) without pre-processing in the inverse transform unit.
13 to 14 are diagrams showing results of testing the performance of the electrical stimulation generating system for generating functional electrical stimulation (FES) based on muscle stimulation signals according to the present invention.

Hereinafter, embodiments of the present invention will be described clearly and in detail to the extent that those skilled in the art can easily practice the present invention.

1 is a block diagram exemplarily showing an electrical stimulation generating system according to an embodiment of the present invention. Referring to FIG. 1 , the electrical stimulation generating system 1100 performs preprocessing on EMG signals (EMG) obtained by applying electrical stimulation, and generates functional electrical stimulation (FES) for a patient using the preprocessed data. do.

The electrical stimulation generating system 1100 applies electrical stimulation (ES) to the patient's muscle or skin, and generates functional electrical stimulation (FES) based on an EMG signal (EMG) provided in response to the electrical stimulation (ES). . Unlike general functional electrical stimulation (FES) signals, the electrical stimulation generating system 1100 applies a preprocessing technique for separating and removing involuntary muscle contraction signals from EMG signals. Of course, the EMG signal to be collected includes an electrical stimulation (ES) signal. The electrical stimulation generating system 1100 extracts a voluntary muscle contraction signal by removing an electrical stimulation (ES) signal and an involuntary muscle contraction signal from the electromyogram signal (EMG), and functional electrical stimulation (based on the voluntary muscle contraction signal) FES) signal can be generated. Therefore, in that the functional electrical stimulation (FES) signal of the present invention is generated based on the EMG signal (EMG), it will hereinafter be referred to as EMG-Controlled FES (ECF).

The electrical stimulation generating system 1100 may adjust functional electrical stimulation (FES) based on an EMG signal (EMG) that measures muscle activity according to muscle contraction. The electrical stimulation generating system 1100 may adjust the intensity of the electrical stimulation according to the root mean square (RMS) size of the EMG signal. Through this, the electrical stimulation generating system 1100 can provide a service in which the electrical stimulation is turned on when a certain amount of force is applied and the electrical stimulation is turned off when the power is less than a certain amount. In addition, the electrical stimulation generating system 1100 may provide an assisting service by applying electrical stimulation to supplement insufficient power when power to be applied for a specific motion is not provided.

The electrical stimulation generating system 1100 of the present invention uses EMG-based functional electrical stimulation (ECF). Pad-shaped electrodes may be used for this purpose. The electrical stimulation pad 1111 may include an electrical stimulation pad that applies electrical stimulation (ES) and functional electrical stimulation (ECF) based on an EMG signal. For example, the electrical stimulation pad 1111 may be used once or multiple times in a wet form. Alternatively, the electrical stimulation pad 1111 may be manufactured using a dry, high-adhesive material to transmit a user's biosignal or an electrical stimulation signal of the innervated muscle. For example, the electrical stimulation pad 1111 may be made of a conductive dry adhesive electrode pad using a carbon nanomaterial. The electrical stimulation measurement pad 1112 is used for electromyography (EMG) measurement. The electrical stimulation measurement pad 1112 may include an EMG sensor for sensing an EMG measurement of the thigh. In addition, the reference pad 1113 is provided as an electrode pad to provide a ground level for the electrical stimulation pad 1111 or the electrical stimulation measurement pad 1112.

FIG. 2 is a block diagram showing the configuration of the electrical stimulation generating system of FIG. 1 as an example. Referring to FIG. 2 , the electrical stimulation generating system 1100 includes a voluntary/involuntary muscle contraction detection unit 1110, an involuntary muscle contraction signal removal unit 1120, a muscle activity intensity calculator 1130, and functional electrical stimulation. A controller 1140 may be included.

The voluntary/involuntary muscle contraction detection unit 1110 receives electromyogram signals (EMG) collected in response to electrical stimulation (ES). The voluntary/involuntary muscle contraction detecting unit 1110 may remove the electrical stimulation (ES) included in the input EMG signal (EMG) and distinguish the voluntary muscle contraction signal from the involuntary muscle contraction signal. It is difficult to distinguish a voluntary muscle contraction signal from an involuntary muscle contraction signal only with the amplitude of the signal. Therefore, an artificial intelligence (AI) model is needed to separate voluntary and involuntary muscle contraction signals.

For high-performance signal classification, in addition to a high-performance deep learning model, high-performance feature vectors to improve the performance of the model are also important. Therefore, in the present invention, a sampling rate of 850 Hz, a frame size of 320 samples, a shift size of 20 samples, and an FFT size of 512 may be applied for feature extraction. Since the size of the frame size is 320 samples, 320 samples will be sequentially stored in the buffer. In addition, after a time of 320 samples has passed, signals of 20 samples may be sequentially stored in a buffer. After updating the buffer, feature vectors will be extracted using feature extraction techniques.

The involuntary muscle contraction signal removing unit 1120 removes the detected involuntary muscle contraction signal. The muscle activity intensity calculation unit 1130 calculates the RMS in a noise-removed state so that it is possible to intuitively determine how much force is applied.

The functional electrical stimulation controller 1140 generates EMG-based functional electrical stimulation (ECF). That is, the functional electrical stimulation control unit 1140 may turn ON or OFF the application of the functional electrical stimulation by comparing a specific threshold and RMS. For example, the functional electrical stimulation control unit 1140 may apply the functional electrical stimulation if the RMS is greater than or equal to the threshold value, and may not apply the functional electrical stimulation if the RMS is less than or equal to the threshold value. Alternatively, the functional electrical stimulation controller 1140 may determine the intensity of the functional electrical stimulation according to the RMS. The functional electrical stimulation control unit 1140 can control the electrical stimulation to increase as the RMS increases and to weaken as the RMS decreases.

The electrical stimulation generating system 1100 may adjust functional electrical stimulation (FES) based on an EMG signal (EMG) that measures muscle activity according to muscle contraction. The electrical stimulation generating system 1100 may adjust the strength of the electrical stimulation according to the RMS level of the EMG. Through this, the electrical stimulation generating system 1100 can provide a service in which the electrical stimulation is turned on when a certain amount of force is applied and the electrical stimulation is turned off when the power is less than a certain amount. In addition, the electrical stimulation generating system 1100 may provide an assisting service by applying electrical stimulation to supplement insufficient power when power to be applied for a specific motion is not provided.

3 is a flowchart showing a processing method in a frequency domain for extracting a percentile spectral cumulative sum (PoSCS) as an example of feature extraction. Referring to FIG. 3 , a Percentile of Spectral Cumulative Sum (hereinafter referred to as PoSCS) may be extracted through frequency domain processing of an EMG signal.

In step S110, a window in the time domain of the EMG signal to be converted into a frequency spectrum is selected. For example, the window of the EMG signal (EMG) may be selected in sector units or frame units.

In step S120, fast Fourier transform (FFT) and absolute value calculation are performed on the window of the EMG signal (EMG) of the selected section.

In step S130, a spectral cumulative sum (SCS) is extracted in the frequency domain based on the absolute value calculation result.

In step S140, a normalization operation is performed.

In step S150, a percentile spectral cumulative sum (PoSCS) of each of the frequencies is extracted based on the normalized data.

4 is a graph showing a method of extracting a percentile spectral cumulative sum (PoSCS). Referring to FIG. 4 , extraction of the percentile spectral cumulative sum (PoSCS) may be exemplarily performed through the following process.

First, after accumulating the magnitude in the positive x-axis direction in the frequency domain, max-normalization data is utilized. Subsequently, after shifting the horizontal line from 0.05 to 0.30 by 0.05 units based on the y-axis, a frequency bin of a point of contact between the horizontal line and the spectral cumulative sum (SCS) is extracted as a feature. By extracting feature vectors consisting of 6 orders for each frame, the use of the spectral cumulative sum (SCS) can be used to effectively distinguish between involuntary and voluntary muscle contractions. This process is shown in the right graph showing the extracted characteristics of frequency bins.

A noise component in a frequency domain generated by involuntary muscle contraction has an abnormally bounced value, unlike a frequency component of voluntary muscle contraction. Therefore, when involuntary muscle contraction and voluntary muscle contraction exist simultaneously, the characteristics of the spectral cumulative sum (SCS) appear different from when only involuntary muscle contraction exists. In addition, noise components in the frequency domain generated due to involuntary muscle contraction appear differently depending on the frequency parameter of the electrical stimulation (ES). Depending on the electrical stimulation environment, the characteristics that appear prominently in the voluntary muscle contraction section are different. Therefore, in order to construct a high-performance model for all electrical stimulation environments, as mentioned above, a multi-dimensional feature vector should be used. As a result, it can be seen that the percentile spectral cumulative sum (PoSCS) appears prominently in the voluntary muscle contraction section.

5 is probability density functions (PDFs) showing the results of extracting the percentile spectral cumulative sum (PoSCS) for each frequency from the EMG signal (EMG). Referring to FIG. 5 , in the EMG signal (EMG) in the time domain, involuntary muscle contraction and voluntary muscle contraction may be distinguished in each frequency band. However, it can be seen that overlapping parts exist. Therefore, it can be seen that separation operation using an artificial intelligence algorithm is necessary.

After extracting the percentile spectral cumulative sum (PoSCS) for each frequency, the activity density function (PDF) of the characteristics for the involuntary muscle contraction signal is calculated as curves (C11, C12, and C13) at each frequency (10 Hz, 60 Hz, and 90 Hz). appear. In addition, the probability density function of the characteristics of the voluntary muscle contraction signal is represented by curves C21, C22, and C23 at each frequency (10 Hz, 60 Hz, and 90 Hz). According to the extraction result of the percentile spectral cumulative sum (PoSCS), the involuntary muscle contraction signal and the voluntary muscle contraction signal at low frequencies have different averages, so a relatively clear distinction is possible. However, since the extracted features overlap with each other, deep learning or artificial intelligence techniques for the extracted features are required to provide high classification resolution at any frequency. In particular, in the present invention, a Long Short Term Memory (LSTM) algorithm that provides the highest performance for long-term time series data will be used.

6 is a flowchart illustrating a learning method of an artificial intelligence operation unit for separating voluntary muscle contraction signals and involuntary muscle contraction signals according to an embodiment of the present invention. Referring to FIG. 6 , the voluntary/involuntary muscle contraction detection unit 1110 (see FIG. 2 ) may perform learning of LSTM, which is a type of recurrent neural network (RNN), using an input EMG signal. High-resolution identification of voluntary muscle contraction signals and involuntary muscle contraction signals is possible through learning.

In step S210, the voluntary/involuntary muscle contraction detection unit 1110 may collect electromyography (EMG) data. The voluntary/involuntary muscle contraction detection unit 1110 may apply electrical stimulation (ES) to body muscles and measure electromyography (EMG) data.

In step S220, the voluntary/involuntary muscle contraction detection unit 1110 may analyze electromyography (EMG) data and extract a feature vector. The voluntary/involuntary muscle contraction detection unit 1110 may extract a feature vector related to muscle strength or muscular endurance after removing a noise signal included in electromyogram (EMG) data.

In step S230, the voluntary/involuntary muscle contraction detection unit 1110 performs learning of an artificial intelligence (AI) model based on the feature vector. The voluntary/involuntary muscle contraction detection unit 1110 generates learning data for artificial intelligence learning. The voluntary/involuntary muscle contraction detection unit 1110 may create a database (DB) for learning based on the feature vector (S231). The voluntary/involuntary muscle contraction detector 1110 may initialize LSTM weights (S232). The voluntary/involuntary muscle contraction detection unit 1110 shuffles the learning database (DB). That is, the voluntary/involuntary muscle contraction detection unit 1110 may provide learning data to a fully connected neural network (FCNN) and process it as a learning operation (S233). The voluntary/involuntary muscle contraction detector 1110 may calculate a current LSTM model error (S234). The voluntary/involuntary muscle contraction detection unit 1110 determines whether the epoch learned so far is smaller than the total epoch (S235). The voluntary/involuntary muscle contraction detection unit 1110 ends when the epochs learned so far are not smaller than the total epoch (NO). On the other hand, the voluntary/involuntary muscle contraction detection unit 1110 updates the LSTM weights when the epochs learned so far are smaller than the total epoch (YES), and returns to step S233.

7 is a diagram briefly showing the structure of the LSTM algorithm for identifying voluntary and involuntary muscle contraction signals through sequential electromyogram (EMG) data in the time domain of the present invention. Referring to FIG. 7 , weights for electromyogram (EMG) data sequentially provided temporally by an LSTM algorithm are updated, that is, learning is performed.

The structure of the LSTM algorithm is composed of LSTM cells that process sequentially input data (Dt). Each of the LSTM cells determines how much past data to store or discard based on the state at the current time, reflects the current output on the result, and transmits it to the next LSTM cell. For this function, one LSTM cell is composed of a forget gate for processing the current input data (Dt), an input gate, and an output gate.

8 is a flowchart showing an actual operation and a test operation of the voluntary/involuntary muscle contraction detection unit of the present invention. Referring to FIG. 8 , the voluntary/involuntary muscle contraction detection unit 1110 distinguishes between voluntary and involuntary muscle contraction signals using the LSTM learned in FIG. 7 based on the input EMG signal. can

In step S310, the voluntary/involuntary muscle contraction detector 1110 may collect electromyogram (EMG) data. The voluntary/involuntary muscle contraction detection unit 1110 may apply electrical stimulation (ES) to body muscles and measure electromyography (EMG) data.

In step S320, the voluntary/involuntary muscle contraction detection unit 1110 may analyze electromyography (EMG) data and extract a feature vector. The voluntary/involuntary muscle contraction detection unit 1110 may extract a feature vector related to muscle strength or muscular endurance after removing noise electrical signals included in electromyogram (EMG) data.

In step S330, the voluntary/involuntary muscle contraction detection unit 1110 performs an LSTM operation based on the sequentially input feature vectors in time series. In step S340, the parameter Wo of the output layer provided as a result of the LSTM operation is provided. The voluntary/involuntary muscle contraction detector 1110 provides an output value y using the parameter Wo. In step S350, the voluntary/involuntary muscle contraction detection unit 1110 finally performs classification using a threshold value using the output value y, and outputs the result.

FIG. 9 is a block diagram showing the involuntary muscle contraction signal removal unit shown in FIG. 2 as an example. Referring to FIG. 9, the involuntary muscle contraction signal removal unit 1120 includes a window unit 1121, a fast Fourier transform (FFT) unit 1122, and a magnitude and phase calculator 1123. , 1124), a peak detector 1125, a peak remover 1126, and an inverse FFT (IFFT) 1127.

The window unit 1121 performs windowing from a time domain input signal (eg, an EMG signal) to a frequency domain signal. The window unit 1121 may shift in units of 20 samples in real time, configure a frame in units of 512 samples, and operate with an FFT size of 512. The fast Fourier transform unit 1122 performs Fourier transform, and the calculators 1123 and 1124 calculate a magnitude and a phase. The peak detector 1125 and the peak remover 1126 detect noise by detecting the peak of the waveform and perform peak suppression through permutation. The involuntary muscle contraction component appears as a peak-type magnitude like an impulse. Accordingly, the peak detector 1125 detects an involuntary muscle contraction component whose magnitude appears like an impulse. The peak remover 1126 may replace the detected peak with an eps (= 2.2204e-16) value, and then perform IFFT to obtain a signal from which involuntary muscle contraction components are removed. The inverse transformation unit 1127 performs inverse transformation using the magnitude of the waveform that has passed through the peak detection unit 1125 and the peak removal unit 1126 and the phase of the previously calculated waveform, and generates an output signal. .

As a result, the involuntary muscle contraction signal canceling unit 1120 uses an adaptive noise suppression algorithm that detects and then removes a peak signal related to the involuntary muscle contraction signal in the frequency domain. As the frequency of the electrical stimulation (ES) changes, the frequency component of the involuntary muscle contraction signal also changes. Involuntary muscle contraction signals of varying frequencies can be effectively removed using this adaptive noise suppression algorithm. When a method with a fixed filter band is used, since performance deviation may occur depending on situations or users, the involuntary muscle contraction signal removal unit 1120 using an adaptive noise suppression algorithm can provide stable performance. there is.

10 and 11 are graphs showing frequency analysis results of EMG data in the peak detector 1125 and the peak suppressor 1126. 12 is a graph showing a waveform (red) with pre-processing in the inverse transform unit 1127 and a waveform (black) without pre-processing.

The involuntary muscle contraction signal removal unit 1120 may be implemented as an adaptive noise suppression algorithm in the form of finding and removing peak signals related to involuntary muscle contraction (ie, noise) in the frequency domain. there is. As the frequency of electrical stimulation changes, the frequency component of involuntary muscle contraction changes. The involuntary muscle contraction signal remover 1120 may adaptively remove the involuntary muscle contraction component in order to effectively remove it.

If a fixed filter is used, performance deviation may occur depending on the situation or person. The involuntary muscle contraction signal removal unit 1120 is intended to solve this problem, and can reduce performance deviations according to situations or people.

The involuntary muscle contraction signal removal unit 1120 may operate in the following manner. For example, the involuntary muscle contraction signal removal unit 1120 shifts in units of 20 samples in real time, constructs a frame in units of 512 samples, and sets the FFT size to 512 to drive the algorithm. can do. The involuntary muscle contraction signal remover 1120 performs FFT on a predefined frame, calculates the magnitude and phase, and removes the involuntary muscle contraction component that appears like an impulse from the magnitude. You can detect peaks to find them.

The involuntary muscle contraction signal removal unit 1120 replaces the peak with an eps (= 2.2204e-16) value to remove the involuntary muscle contraction component, and then performs IFFT to obtain a signal from which the involuntary muscle contraction component has been removed. can be obtained. The involuntary muscle contraction signal removal unit 1120 is finally classified as an involuntary muscle contraction through an involuntary muscle contraction signal removal algorithm and can obtain a signal by removing the magnitude by 6 dB when it is not a contraction section.

Referring to FIG. 12 , the electrical stimulation generating system 1100 using the EMG-based functional electrical stimulation (ECF) of the present invention can provide high removal efficiency of involuntary muscle contraction signals. The involuntary muscle contraction signal included in the EMG signal in the time domain can be effectively removed by applying an adaptive noise suppression algorithm. Accordingly, the electrical stimulation generating system 1100 (see FIG. 1 ) of the present invention can generate functional electrical stimulation (ECF) based on the low-noise voluntary muscle contraction signal. Therefore, it is possible to generate functional electric stimulation with high reliability without relying on experts or expensive devices.

13 to 14 are diagrams showing results of testing the performance of the electrical stimulation generating system for generating functional electrical stimulation (FES) based on muscle stimulation signals according to the present invention. For the test of the electrical stimulation generating system 1100 of the present invention, electrical stimulation (ES) was applied while increasing the frequency by 5 Hz from 10 Hz to 90 Hz. Then, while collecting the electromyographic signal (EMG) while electrical stimulation (ES) was provided, the measurement subject repeated the pattern of resting for 10 seconds, proceeding with voluntary muscle contraction for 20 seconds, and resting for 10 seconds again. . A total of 40 seconds of data was collected per person, and a database was built using the data collected in the same way for a total of 6 people. In addition, the period in which voluntary muscle contraction was maintained was leveled as '1', and the period in which only involuntary muscle contraction existed was leveled as '0'.

The artificial intelligence model uses all extracted features as inputs, and in addition, the random initialization method is applied to the initialization of the artificial intelligence model, and the fine-tuning method is backpropagation. , the number of fully connected layers is set to 1, and the number of units is also set to 1. In addition, the adaptive moment estimation method (Adam: Adaptive Momentum Estimation) was used as an optimization algorithm for determining the weight update method. In addition, Binary cross entropy is used as the cost function, and hyperbolic tangent is used as the active function. The number of cells is 3, and the hidden units per cell are 128. , 64 and 32 were used.

Referring to FIG. 13, the removal performance of involuntary muscle contraction signals in the case of using the artificial intelligence model of the present invention, LSTM, and the case of applying general artificial intelligence models (SVM, ANN, DNN), is shown in Table is shown as According to the test results for a total of two groups (Set1, Set2), the total accuracy (TA) in the case of using the LSTM model was the best at 90.01% and 82.82%, respectively.

Referring to FIG. 14, a graph shows results of area under the curve (AUC) evaluation of artificial intelligence models for two groups (Set1 and Set2). In the case of using the LSTM model for each of the two groups (Set1 and Set2), it was observed that the reliability (AUV) was the highest across all experimental frequencies.

The foregoing are specific examples for carrying out the present invention. In addition to the above-described embodiments, the present invention will also include embodiments that can be simply or easily changed in design. In addition, the present invention will also include techniques that can be easily modified and practiced using the embodiments. Therefore, the scope of the present invention should not be limited to the above-described embodiments and should not be defined, and should be defined by those equivalent to the claims of this invention as well as the claims to be described later.

Claims (6)

In the electrical stimulation generating device for controlling and generating a functional electrical stimulation signal by collecting electromyographic signals (EMG) generated in response to electrical stimulation from the body:
a voluntary/involuntary muscle contraction detection unit that extracts a feature vector from the frequency domain of the EMG signal, applies an artificial intelligence model, and classifies and detects a voluntary muscle contraction signal and an involuntary muscle contraction signal from the feature vector;
an involuntary muscle contraction signal removing unit configured to remove the involuntary muscle contraction signal from the EMG signal according to the detection result;
Including a functional electrical stimulation control unit for generating the functional electrical stimulation signal to be applied to the body based on the EMG signal from which the involuntary muscle contraction signal has been removed.
Functional electrical stimulation signal generation device. The method of claim 1, wherein the device,
a muscle activity intensity calculation unit that calculates a Root Mean Square (RMS) of the EMG signal from which the involuntary muscle export signal has been removed;
Including more,
The functional electrical stimulation controller compares the effective value with the threshold value and generates the functional electrical stimulation signal to be applied to the body according to the comparison result.
Functional electrical stimulation signal generation device. The method of claim 1, wherein the feature vector includes at least one of a percentile spectral cumulative sum (PoSCS) and a log power spectrum detected in a frequency domain of the EMG signal (EMG).
Functional electrical stimulation signal generation device. According to claim 1,
The involuntary muscle contraction signal removal unit:
a window unit that selects a window of the electromyogram signal (EMG); and
a fast Fourier transform unit for processing signals included in the selected window through fast Fourier transform;
including,
Functional electrical stimulation signal generation device. According to claim 4,
The involuntary muscle contraction signal removal unit:
Magnitude and phase calculation unit to calculate the magnitude and phase of the signal output from the fast Fourier transform unit, respectively;
Including more,
Functional electrical stimulation signal generation device. According to claim 5,
The involuntary muscle contraction signal removal unit:
a peak detector for detecting a peak in the magnitude of the signal; and
a peak remover removing a noise signal corresponding to the detected peak;
Including more,
Functional electrical stimulation signal generation device.

Images