Gait event detection method and system based on single electromyography sensor

Abstract

The application discloses a gait event detection method based on a single myoelectric sensor, which is applied to the technical field of gait detection, collects myoelectric signals of the rectus femoris, and carries out pretreatment on the myoelectric signals; RMS feature extraction is carried out on the pretreated myoelectric signals to obtain an RMS feature group; the RMS feature group is subjected to smoothing treatment, and all extreme points are calculated; whether the extreme points are maximum points is judged, if yes, trigger condition judgment is entered, and if not, the myoelectric signals are continuously collected and reprocessed; the numerical difference between the maximum value and the minimum value is compared with a judgment threshold to carry out trigger condition judgment, if the trigger condition is met, the signal corresponding to the trigger time is output, and if the trigger condition is not met, the myoelectric signals are continuously collected and reprocessed. According to the application, the mid-stance gait event can be accurately detected, other gait parameters are obtained, the event time is sent to a lower limb exoskeleton robot in real time, and the lower limb exoskeleton robot can make corresponding control adjustment according to the received gait parameters.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of gait detection, and more particularly to a gait event detection method and system based on a single electromyographic sensor. BACKGROUND

[0002] Hemiplegia is a motor dysfunction caused by nerve injury. The clinical treatment of hemiplegia patients is mainly one-on-one rehabilitation therapy under the guidance of a rehabilitation therapist. This way is time-consuming and costly. Therefore, there is an urgent need for a new and efficient rehabilitation therapy to make up for the shortcomings of traditional rehabilitation training. Rehabilitation exoskeleton robots combine robot technology with rehabilitation medicine, which can free the rehabilitation therapist from the main physical labor and develop a better rehabilitation program for the patient to improve the rehabilitation effect.

[0003] Rehabilitation training has active and passive modes. Studies have shown that active rehabilitation mode has more effective rehabilitation effect than passive rehabilitation mode. Therefore, extracting the patient's movement intention and integrating it into the rehabilitation exoskeleton robot is a technology that can effectively improve the patient's subjective initiative and achieve more efficient rehabilitation.

[0004] Gait in the process of rehabilitation training contains a lot of human kinematics information. Gait has periodicity and regularity, and each gait cycle in walking contains a series of foot position transitions. According to the changes of typical foot positions, gait can be divided into a series of phases. The gait cycle in the process of human walking can be described as the process from the landing of one side of the heel to the landing of the heel of the same leg. A gait cycle can be divided into two stages, the support phase and the swing phase. The support phase starts from the landing of the heel and ends with the lifting of the toe. The swing phase starts from the lifting of the toe and ends with the landing of the heel. The support phase accounts for about 60% of the gait cycle, and the swing phase accounts for about 40% of the gait cycle. The support phase can be further divided into four gait phases, namely, the early support phase, the middle support phase, the late support phase, and the pre-swing phase.

[0005] Accurate detection of gait phase and sending to lower limb exoskeleton rehabilitation robot. The exoskeleton rehabilitation robot can make corresponding control decisions according to the gait phase. At present, there are mainly two methods to realize the recognition of lower limb movement intention, one is to use human bioelectric signals to recognize human movement intention, such as electromyography or electroencephalogram; the other is to detect human-machine interaction force to recognize human movement intention. For gait event intention detection, common sensors include electromyography, plantar pressure, IMU, etc.

[0006] Electromyography signal is the comprehensive result of muscle movement and nerve movement. Surface electromyography (sEMG) contains rich human motion information, which includes joint movement mode, action size, etc., and its acquisition process is convenient and harmless to the human body. However, for lower extremity exoskeleton robots, the patient's leg needs to be bound with the exoskeleton. These bindings occupy a lot of space, affecting the pasting of electromyography sensors. More electromyography sensors can bring more accurate intention estimation results, but the pasting of sensors faces more complex problems.

[0007] Further, the existing gait event detection method based on electromyography has problems of calculation complexity and insufficient generalization ability. Although more electromyography sensors can detect more intentions, they also bring complexity of sensor installation and complexity of algorithm processing. The method using artificial intelligence can indeed improve the detection accuracy, but has the problem of insufficient generalization ability. The electromyography signals of different individuals are different, and the same AI model cannot be accurately applied to other individuals.

[0008] In view of the problems existing in the prior art, the present application provides a gait event detection method and system based on a single electromyography sensor, which identifies the support mid-phase through a single electromyography sensor, and then obtains gait frequency and step time parameters. Then, the gait frequency and step time are sent to the lower extremity exoskeleton robot. The lower extremity exoskeleton robot can change the motion state of the robot according to the gait frequency and step time, and realize active on-demand rehabilitation. SUMMARY

[0009] Therefore, the present application provides a gait event detection method and system based on a single electromyography sensor to solve the problems in the background art.

[0010] In order to achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0011] On the one hand, the present application discloses a gait event detection method based on a single electromyography sensor, and the specific steps are as follows:

[0012] Collect electromyography signals of the rectus femoris muscle, and pre-process the electromyography signals;

[0013] Extract RMS features from the pre-processed electromyography signals to obtain an RMS feature group;

[0014] Smooth the RMS feature group and calculate all extreme points;

[0015] Determine whether the extreme points are maximum points according to the extreme points, if yes, enter the trigger condition judgment, if not, return to continue collecting electromyography signals and re-processing;

[0016] The numerical difference between the maximum value and the minimum value is compared with a judgment threshold to make a trigger condition judgment. If the trigger condition is met, a signal corresponding to the trigger time is output. If the trigger condition is not met, the EMG signal is continuously collected and reprocessed.

[0017] Preferably, in the above-mentioned gait event detection method based on a single EMG sensor, the method further comprises a termination instruction. When the termination instruction is detected, the program stops running. Otherwise, the method returns to the beginning and continues to collect the EMG signal and process it.

[0018] Preferably, in the above-mentioned gait event detection method based on a single EMG sensor, the trigger time is the mid-stance phase in the corresponding gait phase.

[0019] Preferably, in the above-mentioned gait event detection method based on a single EMG sensor, the EMG signal of the rectus femoris muscle is collected, and the specific steps of preprocessing the EMG signal are as follows: the EMG sensor is pasted at the center of the muscle belly of the rectus femoris muscle to obtain the EMG signal. The preprocessing of the EMG signal includes: using a 4th order Butterworth band-pass filter to perform 20-500 Hz band-pass filtering on the collected EMG signal; and then using a 50 Hz notch filter to process the EMG signal.

[0020] Preferably, in the above-mentioned gait event detection method based on a single EMG sensor, the specific steps of extracting the RMS feature value are as follows: the EMG signal is windowed according to a fixed step size, and the RMS feature value of the EMG signal in each window is extracted. The calculation formula is as follows:

[0021]

[0022] where x i,pre is the i-th preprocessed sample value, N is the total number of windowed EMG data samples, and x t,rms is the RMS feature value corresponding to the t-th window.

[0023] Preferably, in the above-mentioned gait event detection method based on a single EMG sensor, the RMS feature group is smoothed using WMA, and the calculation formula is as follows:

[0024]

[0025] where ω t is the weight value of the t timestamp rms feature value. Each RMS feature value has the same weight value, ω = 1 / γ, and γ is the weight reciprocal.

[0026] Preferably, in the above-mentioned gait event detection method based on a single EMG sensor, when it is detected that the latest detected extreme point is a maximum value point, the trigger condition is judged. The trigger condition is as follows:

[0027]

[0028] wherein, s max[-1] represents the latest detected maximum value point value, s min[-1] represents the latest detected minimum value point value, s max[-2] represents the last detected maximum value point value, and β is a judgment threshold.

[0029] On the other hand, a gait event detection system based on a single electromyographic sensor comprises:

[0030] A collection and preprocessing module is configured to collect electromyographic signals of the rectus femoris muscle and pre-process the electromyographic signals.

[0031] A feature extraction module is configured to extract RMS features from the pre-processed electromyographic signals to obtain an RMS feature group.

[0032] A smoothing processing and calculation module is configured to smooth the RMS feature group and calculate all extreme value points.

[0033] A first judgment module is configured to judge whether an extreme value point is a maximum value point, and if so, trigger condition judgment is entered, and if not, electromyographic signals are continuously collected and reprocessed.

[0034] A second judgment module is configured to compare the numerical difference between the maximum value and the minimum value with a judgment threshold to perform trigger condition judgment, and if the condition is met, the signal corresponding to the trigger time is output, and if the condition is not met, electromyographic signals are continuously collected and reprocessed.

[0035] Compared with the prior art, the gait event detection method and system based on a single electromyographic sensor provided by the present disclosure can detect gait event intentions using only a single electromyographic sensor, accurately detect the mid-stance gait event, and obtain some other gait parameters. The event time is sent to a lower limb exoskeleton robot in real time, and the lower limb exoskeleton robot can make corresponding control adjustments according to the received gait parameters. BRIEF DESCRIPTION OF DRAWINGS

[0036] To more clearly illustrate the technical solutions in the embodiments of the present disclosure or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art without creative effort based on the provided drawings.

[0037] Figure 1 The method flowchart of the present disclosure;

[0038] Figure 2 System diagram of the present application. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the present application.

[0040] It should be understood that the maximum peak point of the rectus femoris muscle is accurately detected by using a single electromyographic sensor, and then the time corresponding to each peak point is obtained. The difference between the time of adjacent maximum peak points is the step time, and then the gait frequency can be obtained according to the step time.

[0041] The embodiment of the present application discloses a gait event detection method based on a single electromyographic sensor, as shown in Figure 1 The specific steps are as follows:

[0042] Collecting the electromyographic signal of the rectus femoris muscle, and pre-processing the electromyographic signal;

[0043] Extracting the RMS feature of the pre-processed electromyographic signal to obtain an RMS feature group;

[0044] Smoothing the RMS feature group, and calculating all extreme points;

[0045] According to the extreme point, it is judged whether it is a maximum value point. If yes, the trigger condition judgment is entered, and if no, the electromyographic signal is collected and reprocessed;

[0046] The numerical difference between the maximum value and the minimum value is compared with the judgment threshold to perform trigger condition judgment. If it is satisfied, the signal corresponding to the trigger time is output, and if it is not satisfied, the electromyographic signal is collected and reprocessed.

[0047] Specifically, the electromyographic signal of the rectus femoris muscle is collected. The electromyographic sensor of the Delsys device is pasted on the center of the muscle belly of the rectus femoris muscle. The pre-processing of the electromyographic signal includes: using a 4th order Butterworth band-pass filter to perform 20-500Hz band-pass filtering on the collected electromyographic signal; using a 50hz notch filter to process the electromyographic signal to remove the 50hz power frequency interference caused by the power supply.

[0048] In the embodiment, the electromyographic signal is windowed at 100ms, and then the RMS feature value of each window electromyographic signal is extracted, and the calculation formula is as follows:

[0049]

[0050] where x i,pre is the i-th pre-processed sample value, N is the total number of windowed EMG data samples, x t,rms is the RMS feature value corresponding to the window. The sampling frequency of the EMG sensor of the Delsys device is 2000 Hz, and N is taken as 200 here. The RMS feature reflects the time-varying characteristics of the amplitude of the EMG signal, and has low latency and high real-time performance. The RMS models the EMG signal as a Gaussian random process and provides the maximum likelihood estimate of constant force and fatigue-free contraction amplitude.

[0051] (3) The RMS feature set is smoothed using WMA, and the calculation formula is as follows:

[0052]

[0053] where ω t is the weight value of the t timestamp rms feature value, each RMS feature value has the same weight value, ω = 1 / γ, γ is the weight reciprocal, the larger the value of γ, the smoother the curve smoothed using WMA. Here, γ = 8 is taken.

[0054] In this embodiment, all extreme points are calculated, and the Argrelextrema function in the Scipy library is used to calculate all extreme points. The function is a commonly used function for calculating extreme points. The Argrelextrema function is used to process the smoothed EMG data in real time.

[0055] In this embodiment, the trigger condition is judged: when it is detected that the latest detected extreme point is a maximum value point, the trigger condition judgment link is entered, otherwise it returns to the start and continues to collect and process the EMG signal.

[0056] In this embodiment, when it is detected that the latest detected extreme point is a maximum value point, the trigger condition is judged, and the trigger condition is as follows:

[0057]

[0058] where s max[-1] represents the value of the latest detected maximum value point, s min[-1] represents the value of the latest detected minimum value point, s max[-2] represents the value of the last detected maximum value point, and β is a judgment threshold.

[0059] In this embodiment, if the maximum extreme point value does not satisfy the trigger condition, it returns to the start and continues to collect and process the EMG signal. If the maximum extreme point value satisfies the trigger condition, a trigger time signal is sent to the lower limb rehabilitation exoskeleton robot.

[0060] Different individuals have different β values through experiments. Generally, β = 0.4.

[0061] The method does not use deep learning and the like, does not need training, and has high generalization. When used for different individuals, only the judgment threshold β needs to be adjusted.

[0062] In the embodiment, when the program termination instruction is detected, the program running is stopped, otherwise, it is returned to the start, and the electromyographic signal is continuously collected and processed.

[0063] Further, the triggering moment is the support mid-period in the gait phase. By accurately detecting the support mid-period, the gait parameters can be derived.

[0064] T n = t n -t n-1 ;

[0065] T n is the n th gait stride time, t n is the time point when the n th gait support mid-period occurs, t n-1 is the time point when the n-1 th gait support mid-period occurs, and the time interval between the adjacent two gait support mid-periods is the stride time.

[0066] The gait frequency is the number of steps run in one minute:

[0067] ω n = 60 / T n ;

[0068] ω n is the n th gait frequency.

[0069] Further, these gait parameters reflect the gait intention information of the human body, and the lower extremity exoskeleton robot can adjust the rehabilitation strategy in real time according to the gait intention information.

[0070] For example, when the gait frequency is detected to be accelerated, the lower extremity exoskeleton robot increases the gait frequency of the exoskeleton robot in real time to achieve the effect of synchronous adaptation; when the gait frequency is detected to be slowed down, the lower extremity exoskeleton robot reduces the gait frequency of the exoskeleton robot in real time, thereby ensuring the synchronization of the exoskeleton and the human body and improving the wearing comfort;

[0071] Similarly, the opposite adjustment can also be made, that is, when the gait frequency is detected to be accelerated, the lower extremity exoskeleton robot reduces the gait frequency in real time, increases the rehabilitation movement resistance, and makes the patient feel more difficult,

[0072] and stimulates the movement willingness of the patient.

[0073] In another aspect, another embodiment of the present application discloses a gait event detection system based on single myoelectric sensor, as shown in the accompanying drawings, comprising: Figure 2

[0074] A collection and preprocessing module is configured to collect and preprocess the myoelectric signal of the rectus femoris muscle.

[0075] A feature extraction module is configured to extract RMS features from the preprocessed myoelectric signal to obtain an RMS feature group.

[0076] A smoothing processing and calculation module is configured to smooth the RMS feature group and calculate all extreme points.

[0077] A first judging module is configured to judge whether the extreme point is a maximum value point, if yes, the trigger condition is judged, if not, the myoelectric signal is collected and processed again.

[0078] A second judging module is configured to compare the numerical difference between the maximum value and the minimum value with a judging threshold to judge the trigger condition, if satisfied, the signal corresponding to the trigger time is output, if not satisfied, the myoelectric signal is collected and processed again.

[0079] In the present specification, each embodiment is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the same or similar parts between each embodiment can be referred to each other. For the device disclosed by the embodiments, since it corresponds to the method disclosed by the embodiments, the description is relatively simple, and the related parts can be referred to the method part.

[0080] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.​

Claims

1. A single electromyographic sensor based gait event detection method, characterized in that, The specific steps are as follows: The electromyographic signal of the rectus femoris muscle is collected, and the electromyographic signal is pretreated; The RMS characteristic value of the pretreated electromyographic signal is extracted to obtain an RMS characteristic group; The RMS characteristic group is smoothed, and all extreme points are calculated; Whether the extreme point is a maximum value point is judged, if yes, the trigger condition is judged, if not, the electromyographic signal is collected and reprocessed; The numerical difference between the maximum value and the minimum value is compared with the judgment threshold to judge the trigger condition, if satisfied, the signal corresponding to the trigger time is output, if not satisfied, the electromyographic signal is collected and reprocessed; The trigger time is the middle stage of support in the gait phase; When the latest detected extreme point is a maximum value point, the trigger condition is judged, and the trigger condition is as follows: ； wherein, represents the value of the latest detected maximum point, represents the value of the latest detected minimum point, represents the value of the last detected maximum point, is a decision threshold.

2. The gait event detection method based on single electromyography sensor according to claim 1, characterized in that, Also includes: The termination instruction, when the termination instruction is detected, the program is stopped, otherwise, it returns to the start, continues to collect the electromyographic signal and processes.

3. The single electromyographic sensor based gait event detection method of claim 1, wherein, The specific steps of collecting the electromyographic signal of the rectus femoris muscle and pretreating the electromyographic signal are as follows: the electromyographic sensor is pasted at the center of the muscle belly of the rectus femoris muscle to obtain the electromyographic signal, and the pretreatment of the electromyographic signal includes: using a 4th order Butterworth band-pass filter to perform 20-500Hz band-pass filtering on the collected electromyographic signal; then using a 50Hz notch filter to process the electromyographic signal.

4. The single electromyographic sensor based gait event detection method of claim 1, wherein, The specific steps of RMS characteristic value extraction are as follows: the electromyographic signal is windowed according to a fixed step length, the RMS characteristic value of the electromyographic signal of each window is extracted, and the calculation formula is as follows: ； wherein is the first i pre-processed sample value, N is the total number of windowed EMG data samples, is the first t RMS feature value corresponding to the window.

5. The single electromyographic sensor based gait event detection method of claim 4, wherein, The RMS characteristic group is smoothed using WMA, and the calculation formula is as follows: ； wherein, is a weight value for the t time stamp rms eigenvalue, each RMS eigenvalue having the same weight value, , is a weight reciprocal.

6. A single electromyographic sensor-based gait event detection system employing the single electromyographic sensor-based gait event detection method according to any one of claims 1 to 5, characterized in that, It includes: The acquisition and pretreatment module is used for collecting the electromyographic signal of the rectus femoris muscle, and pretreating the electromyographic signal; The feature extraction module extracts the RMS characteristic value of the pretreated electromyographic signal to obtain an RMS characteristic group; The smoothing processing and calculation module smoothes the RMS characteristic group and calculates all extreme points; The first judgment module judges whether the extreme point is a maximum value point, if yes, the trigger condition is judged, if not, the electromyographic signal is collected and reprocessed; The second judgment module compares the numerical difference between the maximum value and the minimum value with the judgment threshold to judge the trigger condition, if satisfied, the signal corresponding to the trigger time is output, if not satisfied, the electromyographic signal is collected and reprocessed.

Images