A method for processing data of a muscle tension neurofeedback electrical signal

Abstract

The present application relates to the technical field of signal processing, and discloses a muscle tension nerve feedback electric signal data processing method, which comprises preliminary noise reduction, noise suppression, feature extraction, determination of muscle tension baseline and nerve feedback threshold, muscle tension data calculation and correction: a multi-stage noise reduction mode combining sliding average filtering, median filtering and wavelet transform is adopted to effectively remove periodic, pulse and high-frequency noise; the root mean square value and discrete Fourier transform are used to extract features from amplitude and frequency components in multiple dimensions, so as to accurately reflect physiological information such as muscle contraction strength and fatigue state; the muscle tension baseline is determined by collecting individual relaxation state signals, and the nerve feedback threshold is set in combination with subjective feeling, so that the individualization and accuracy of evaluation are enhanced; muscle tension data are calculated in real time, and are dynamically corrected according to energy characteristics of different frequency bands, so that the evaluation result is further optimized, and high-precision and reliable data support is provided for muscle function evaluation and nerve feedback treatment.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and specifically to a method for processing electrical signal data of muscle tone nerve feedback. Background Technology

[0002] In the fields of biomedical engineering and neurorehabilitation, accurate analysis of muscle tone neurofeedback electrical signals is crucial for muscle function assessment, rehabilitation therapy, and exercise training optimization. As an important physiological indicator reflecting muscle tension, muscle tone's neurofeedback electrical signals contain rich information about muscle activity characteristics. Processing and analyzing these signals enables real-time monitoring and precise control of muscle status.

[0003] However, existing neurofeedback techniques for muscle tone have many limitations. In terms of signal denoising, single filtering methods are insufficient to effectively suppress complex noise. For example, impulse noise easily leads to signal distortion, while high-frequency noise such as environmental electromagnetic interference can mask the true characteristics of electromyographic signals, affecting the accuracy of subsequent analysis. In the feature extraction stage, most methods rely solely on single-dimensional indicators such as electromyographic amplitude, lacking in-depth analysis of signal frequency components and failing to comprehensively reflect multidimensional physiological states such as muscle fatigue, contraction intensity, and explosive power. Furthermore, existing technologies often employ universal baselines and threshold standards in muscle tone assessment, failing to fully consider individual physiological differences, leading to deviations between muscle tone data and actual muscle state, thus reducing the effectiveness and adaptability of the neurofeedback system.

[0004] With the rapid development of rehabilitation medicine and smart wearable devices, the demand for accuracy, real-time performance, and personalization in processing neurofeedback electrical signals of muscle tone is increasing. Therefore, there is an urgent need for a data processing method that can effectively suppress noise, extract signal features from multiple dimensions, and achieve personalized muscle tone assessment, in order to improve the accuracy and reliability of muscle tone monitoring and meet the application needs of various scenarios such as clinical diagnosis, rehabilitation training, and sports science research. Summary of the Invention

[0005] The purpose of this invention is to provide a method for processing electrical signal data of muscle tone nerve feedback, which solves the technical problems mentioned in the background art.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for processing electrical signal data from muscle tone neural feedback includes the following steps:

[0008] Step 1, Initial Noise Reduction:

[0009] When processing the electrical signals of muscle tone nerve feedback intensity, the raw bioelectrical signals are first acquired; then, a moving average filter is used to perform preliminary noise reduction on the raw bioelectrical signals.

[0010] Step 2, Noise Suppression:

[0011] A method combining median filtering and wavelet transform was used to suppress noise in the pre-denoised bioelectrical signal.

[0012] Step 3: Feature Extraction

[0013] Feature extraction was performed on the noise-suppressed bioelectrical signal to obtain features corresponding to the electromyographic amplitude and frequency components;

[0014] Step 4: Determine the baseline muscle tone and neural feedback threshold:

[0015] Signals were collected from individuals in a relaxed state. The root mean square (RMS) value of electromyography (EMG) amplitude was calculated using the RMS method, and this RMS value was used as the baseline of muscle tension. Subsequently, neural feedback stimulation of different intensities was applied to the individuals, and EMG signals were collected under different intensities of neural feedback stimulation. Combined with the feature extraction step, features corresponding to the amplitude and frequency components of EMG were extracted. At the same time, the subjective feelings of the individuals under different stimulation intensities were taken into account to determine the neural feedback threshold.

[0016] Step 5: Calculation of muscle tone data:

[0017] Electromyographic signals are acquired in real time, and their real-time root mean square value is calculated. Then, muscle tone data is calculated based on the real-time root mean square value, muscle tone baseline, and neural feedback threshold.

[0018] Step 6: Correction of muscle tone data:

[0019] The frequency component features of the real-time acquired electromyographic signals are extracted according to the frequency component extraction steps, and then the muscle tone data D is corrected in combination with the frequency component features.

[0020] As a further aspect of the present invention, the moving average filtering method is as follows:

[0021] The original bioelectrical signal was labeled as X. t t represents the time series, X t This represents the value of the raw bioelectrical signal collected at timestamp t;

[0022] The average value of the original bioelectric signal is calculated within a pre-specified time window N;

[0023] For timestamp t, the formula for calculating the bioelectrical signal after moving average filtering is:

[0024] In the formula, Y t The signal is the bioelectrical signal after preliminary noise reduction, and 'i' is the index variable for summation, with values ​​ranging from... arrive , Indicates to Round down to the nearest integer.

[0025] As a further aspect of the present invention, the median filtering processing method is as follows:

[0026] Within a pre-specified time window M, the median of the original bioelectrical signal is calculated;

[0027] For timestamp t, the bioelectrical signal after median filtering is calculated as follows:

[0028] With Y t Centered on the target, acquire bioelectrical signal values ​​within a time window M: , ... ... ;

[0029] The bioelectrical signal values ​​within the time window M are then sorted, and the median value of the sorted values ​​is taken as the median-filtered bioelectrical signal Z. t .

[0030] As a further aspect of the present invention, the wavelet transform processing method is as follows:

[0031] The median-filtered bioelectrical signal Z t Perform discrete wavelet transform to decompose it into approximate components and detail components;

[0032] Wherein: the approximation component represents the low-frequency part of the bioelectric signal, which contains the main features of the bioelectric signal; the detail component represents the high-frequency part of the bioelectric signal, and noise exists in the high-frequency detail component;

[0033] For bioelectrical signals Z t Perform m-level wavelet decomposition, and after the j-th level decomposition, obtain the approximate component A. j，t and detail component D j，t j = 1, 2, ..., m;

[0034] In the decomposed detail components, the coefficients of the detail components are thresholded by pre-setting a comparison threshold T, as follows:

[0035] For detail component D j，t The coefficient d in j,k0 The processing method is as follows:

[0036]

[0037] Where k0 represents the coefficient index;

[0038] That is, when the absolute value of the coefficient is greater than or equal to the threshold T, the original value of the coefficient is retained; when the absolute value of the coefficient is less than the threshold T, the coefficient is set to 0, thereby removing the high-frequency coefficients corresponding to noise.

[0039] After processing the detail component coefficients, the processed approximate components and detail components are reconstructed using inverse discrete wavelet transform to obtain the noise-suppressed bioelectric signal S. t ;

[0040] The discrete wavelet inverse transform method is an existing technology, so it will not be elaborated upon.

[0041] As a further aspect of the present invention, the electromyography amplitude extraction method is as follows:

[0042] The noise-suppressed bioelectrical signal S within the time interval [t1,t2] t The root mean square method is used to calculate the electromyographic amplitude.

[0043] Where t1 represents the start time of the time interval, and t2 represents the end time of the time interval;

[0044] The calculation formula is:

[0045] Meanwhile, since the signal is acquired discretely, the above integral formula is converted into a discrete form;

[0046] A total of k discrete points were collected within the time interval [t1, t2], and the time intervals between the discrete points were uniform; then the formula for calculating the root mean square value in discrete form is:

[0047] Where e is the index variable for summation, and its value ranges from 1 to k; S t(e) The signal S after noise suppression t The value at the e-th discrete point;

[0048] The root mean square (RMS) value is used to reflect the average energy of a signal over a period of time.

[0049] As a further aspect of the present invention, the frequency component extraction method is as follows:

[0050] The bioelectrical signal S after noise suppression t Perform a Discrete Fourier Transform (DFT) on the bioelectrical signal S. t The corresponding time-domain signal is converted into a frequency-domain signal, and then the frequency domain is divided into different frequency bands. The energy of each frequency band is then calculated, thus obtaining the frequency component characteristics of the frequency-domain signal.

[0051] The specific method is as follows:

[0052] First, discrete bioelectrical signals S are collected within a pre-specified time interval [t3, t4]. t（r） r = 1, 2, ..., q, where r is the index variable for summation, and its value range is 0, 1, 2, ..., q-1. t（r） Represented as the noise-suppressed bioelectrical signal S t The value at the r-th discrete point;

[0053] The calculation formula using the Discrete Fourier Transform:

[0054] Where S(u) represents the frequency domain signal after obtaining the discrete Fourier transform. is the rotation factor, j is the imaginary unit, and u represents the index of the frequency point, which is 0, 1, ..., q-1;

[0055] After obtaining the frequency domain signal S(u) after discrete Fourier transform, the frequency domain is divided into p frequency bands, and the lower limit frequency of each frequency band is marked as F. L,g The upper limit frequency is F U,g ,

[0056] At the same time, the corresponding frequency point index is marked as H. L,g and H U,g ;

[0057] pass:

[0058] Calculate the energy E of the g-th frequency band. g :

[0059] Where g = 1, 2, ..., p, and f is the index variable for summation, with values ​​ranging from H... L,g To H U,g , |S(f)| represents the amplitude of the frequency domain signal S(f).

[0060] As a further aspect of the present invention: the formula for calculating muscle tone data is as follows:

[0061] In the formula, J RMS K is the baseline for muscle tone. RMS S is the neural feedback threshold. RMS The root mean square value is the real-time value, and D represents muscle tone data; the molecule S RMS- J RMS The denominator K represents the amplitude difference between the real-time electromyographic signal and the signal in the relaxed state. RMS- J RMS D represents the amplitude difference between the neural feedback threshold and the relaxed state signal, and the ratio between the two represents the muscle tone level of the real-time electromyography signal relative to the individual's relaxed state and neural feedback threshold.

[0062] When D=0, the individual muscles are in a relaxed state; when D=1, the muscle tension state corresponding to the neural feedback threshold is reached.

[0063] As a further aspect of the present invention, the specific method for correcting muscle tone data is as follows:

[0064] The different frequency bands in the frequency domain include: low frequency band, mid frequency band, and high frequency band;

[0065] Simultaneously, the pre-determined correspondence between different frequency bands and muscle tone, as analyzed experimentally, is extracted;

[0066] Specifically as follows:

[0067] Low frequency band (0-10Hz): Increased energy in this frequency band is usually associated with muscle fatigue. When muscles are in a contracted state for a long time, the energy in the low frequency band will gradually increase.

[0068] Mid-frequency band (10-50Hz): Its energy variation is related to the sustained contraction intensity and stability of the muscle. When the muscle contracts stably, the energy in this frequency band will remain at a relatively stable level; when the contraction intensity changes, the energy will also fluctuate accordingly.

[0069] High frequency band (50-100Hz): Energy changes in the high frequency band are often related to the rapid contraction and explosive power of muscles. When muscles perform rapid and powerful contraction movements, the energy in this frequency band will increase significantly.

[0070] Since the effects of energy changes in different frequency bands on muscle tone vary in different ranges, either a segmented adjustment strategy or a threshold adjustment strategy can be used.

[0071] As a further aspect of the present invention, the segmented adjustment strategy is as follows:

[0072] The adjustment strategies for low-frequency, mid-frequency, and high-frequency bands are consistent;

[0073] Taking the low-frequency band (0-10Hz) as an example;

[0074] Its energy range is divided into three intervals: [0,E1], (E1,E2], and (E2,+∞), and different adjustment coefficients β1, β2, and β3 are set for each interval;

[0075] Where β1 < β2 < β3;

[0076] The adjustment factor indicates that the degree of influence of low-frequency energy changes on muscle tone data D varies in different energy ranges;

[0077] Extract the detected low-frequency energy E low The correction method for muscle tone data D is as follows:

[0078]

[0079] Among them, E c This corresponds to a preset reference energy value in the low-frequency band.

[0080] As a further aspect of the present invention, the threshold adjustment strategy is as follows:

[0081] The threshold adjustment strategies are consistent across low-frequency, mid-frequency, and high-frequency bands;

[0082] Taking the mid-frequency band (10-50Hz) as an example;

[0083] Extract the preset adjustment thresholds TZ1 and TZ2 corresponding to the mid-frequency band;

[0084] Where TZ1 < TZ2;

[0085] Extract the detected mid-frequency energy E mid ;

[0086] When the energy E in this frequency band mid The muscle tone data D should be adjusted accordingly if the following conditions are met:

[0087] When E mid When the value is greater than TZ2, it indicates that the muscle contraction intensity is relatively high and the muscle tone is relatively high. At this time, the muscle tone data D is increased by a fixed value ΔD1, i.e., D new =D+ΔD1;

[0088] When E mid When TZ1 < TZ1, it indicates that the muscles are in a relatively relaxed state with low muscle tension. The muscle tension data D is then reduced by a fixed value ΔD2, i.e., D < TZ1. new =D-ΔD2;

[0089] When TZ1≤E mid If the value is ≤TZ2, the muscle condition is considered normal, and the muscle tone data D remains unchanged.

[0090] The beneficial effects of this invention are:

[0091] This invention effectively reduces periodic noise and random fluctuations in the original bioelectrical signals by calculating the average signal value within a specified time window, providing a smoother base signal for subsequent processing and avoiding noise interference with feature extraction; for example, in the process of electromyography signal acquisition, it can filter out high-frequency spikes caused by poor electrode contact or environmental electromagnetic interference.

[0092] This invention replaces abnormal impulse noise points with normal signal values ​​by sorting the signal values ​​within a time window and taking the median value, thus preserving the true characteristics of the signal.

[0093] This invention decomposes a signal into low-frequency characteristic components and high-frequency noise components through multi-level decomposition, and uses threshold processing to accurately remove the high-frequency coefficients corresponding to the noise, thereby further purifying the signal while preserving the main characteristics of muscle activity. Compared with single filtering methods, the combined strategy can simultaneously handle impulse noise, high-frequency noise, and low-frequency drift, making it suitable for complex bioelectrical environments.

[0094] This invention uses the root mean square value to calculate the electromyographic amplitude, and through discretization processing to adapt to the actual acquired digital signal, it reflects the average energy of the signal within the time interval; it can quantify the muscle contraction intensity and provide direct data support for setting the muscle tension baseline and threshold.

[0095] This invention decomposes electromyographic signals into different frequency bands of energy, corresponding to physiological states such as muscle fatigue, contraction stability, and explosive power. A surge in high-frequency energy indicates rapid muscle contraction, while an increase in low-frequency energy reflects muscle fatigue.

[0096] This invention corrects muscle tension data based solely on amplitude by using the correspondence between energy in different frequency bands and muscle tension, making the results more consistent with the actual muscle function state.

[0097] This invention establishes a personalized baseline by collecting the root mean square value of electromyography (EMG) signals in an individual's relaxed state, thus avoiding misjudgment of different individuals by general standards. For example, there are differences in the resting EMG levels of athletes and ordinary people, and the personalized baseline can more accurately reflect their true relaxed state.

[0098] This invention collects electromyographic signals under stimulation of different intensities and combines them with individual subjective feedback to dynamically determine the threshold. Advantages: It balances physiological indicators and subjective experience, avoids muscle damage caused by excessive stimulation or ineffective feedback caused by insufficient stimulation, and improves the safety and effectiveness of neurofeedback training.

[0099] This invention sets differentiated adjustment coefficients according to different frequency band energy ranges to finely characterize the nonlinear effect of energy changes on muscle tension; it triggers fixed value correction by preset thresholds to quickly respond to sudden changes in muscle state; it overcomes the single-dimensional deficiency of relying solely on amplitude, enabling muscle tension assessment to simultaneously integrate energy intensity and functional characteristics, thereby improving the system's adaptability to complex muscle activities.

[0100] In summary, the multi-stage filtering process is suitable for noisy environments such as clinical settings and sports training, ensuring signal quality. Combining amplitude and frequency domain characteristics, it can be used simultaneously for muscle function assessment and athletic performance optimization. Individualized baselines and dynamic correction mechanisms enable the system to provide customized neurofeedback for different users, improving training efficiency and treatment effectiveness. Attached Figure Description

[0101] The invention will now be further described with reference to the accompanying drawings.

[0102] Figure 1 This is a flowchart illustrating a method for processing electrical signal data of muscle tone nerve feedback according to the present invention. Detailed Implementation

[0103] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0104] As an embodiment of the present invention:

[0105] Please see Figure 1 As shown, the present invention is a method for processing electrical signal data of muscle tone neural feedback, comprising the following steps:

[0106] Step 1, Initial Noise Reduction:

[0107] When processing the electrical signal of muscle tone nerve feedback intensity, the raw bioelectrical signal X is first acquired. t ;

[0108] Where t represents the time series, X t This represents the value of the raw bioelectrical signal collected at timestamp t;

[0109] Since the original bioelectrical signals are susceptible to noise from motion artifacts and electromagnetic interference, noise reduction processing is performed on the original bioelectrical signals.

[0110] The original bioelectric signal X was filtered using a moving average method. t Perform initial noise reduction;

[0111] The moving average filtering method is as follows:

[0112] The average value of the original bioelectric signal is calculated within a pre-specified time window N;

[0113] For timestamp t, the formula for calculating the bioelectrical signal after moving average filtering is:

[0114] In the formula, i is the index variable for summation, and its value ranges from... arrive , Indicates to Round down;

[0115] Rounding down is used to ensure that the signal is averaged within a time window of length N centered at time t.

[0116] Moving average filtering can effectively reduce some of the noise in the original bioelectric signal, resulting in a preliminarily denoised bioelectric signal Y. t ;

[0117] Step 2, Noise Suppression:

[0118] The bioelectric signal Y after preliminary noise reduction was processed using a combination of median filtering and wavelet transform. t To suppress noise;

[0119] The specific method is as follows:

[0120] Step M1, Median Filtering:

[0121] Within a pre-specified time window M, the median of the original bioelectrical signal is calculated;

[0122] For timestamp t, the bioelectrical signal after median filtering is calculated as follows:

[0123] With Y t Centered on the target, acquire bioelectrical signal values ​​within a time window M: , ... ... ;

[0124] The bioelectrical signal values ​​within the time window M are then sorted, and the median value of the sorted values ​​is taken as the median-filtered bioelectrical signal Z. t ;

[0125] Median filtering can replace the values ​​of impulse noise points with surrounding normal signal values, thereby achieving the purpose of removing impulse noise.

[0126] Step M2, wavelet transform processing:

[0127] The median-filtered bioelectrical signal Z t Perform discrete wavelet transform to decompose it into approximate components and detail components;

[0128] Wherein: the approximation component represents the low-frequency part of the bioelectric signal, which contains the main features of the bioelectric signal; the detail component represents the high-frequency part of the bioelectric signal, and noise exists in the high-frequency detail component;

[0129] For bioelectrical signals Z t Perform m-level wavelet decomposition, and after the j-th level decomposition, obtain the approximate component A. j，t and detail component D j，t j = 1, 2, ..., m;

[0130] In the decomposed detail components, the coefficients of the detail components are thresholded by pre-setting a comparison threshold T, as follows:

[0131] For detail component D j，t The coefficient d in j,k0 The processing method is as follows:

[0132]

[0133] Where k0 represents the coefficient index;

[0134] That is, when the absolute value of the coefficient is greater than or equal to the threshold T, the original value of the coefficient is retained; when the absolute value of the coefficient is less than the threshold T, the coefficient is set to 0, thereby removing the high-frequency coefficients corresponding to noise.

[0135] After processing the detail component coefficients, the processed approximate components and detail components are reconstructed using inverse discrete wavelet transform to obtain the noise-suppressed bioelectric signal S. t ;

[0136] The discrete wavelet inverse transform method is an existing technology, so it will not be elaborated upon here.

[0137] Step 3: Feature Extraction

[0138] Feature extraction was performed on the noise-suppressed bioelectric signal St, and the features corresponding to the electromyography amplitude were obtained.

[0139] The method for extracting electromyographic amplitude is as follows:

[0140] The noise-suppressed bioelectrical signal S within the time interval [t1,t2] t The root mean square method is used to calculate the electromyographic amplitude.

[0141] Where t1 represents the start time of the time interval, and t2 represents the end time of the time interval;

[0142] The calculation formula is:

[0143] In this embodiment, since the signal is discretely acquired in actual calculations, the above integral formula is converted into a discrete form.

[0144] Assume that k discrete points were collected within the time interval [t1, t2], and the time intervals between the discrete points were uniform;

[0145] The formula for calculating the root mean square value in discrete form is:

[0146] Where e is the index variable for summation, and its value ranges from 1 to k; S t(e) The signal S after noise suppressiont The value at the e-th discrete point;

[0147] The root mean square (RMS) value is used to reflect the average energy of a signal over a period of time.

[0148] Step 4: Determine the baseline muscle tone and neural feedback threshold:

[0149] In this embodiment, since there are significant differences in the baseline muscle tone and neural feedback threshold among different individuals, personalized calibration is required to address individual differences.

[0150] First, signals were collected from individuals in a relaxed state. The root mean square (RMS) value of the electromyographic (EMG) signal in the relaxed state was calculated and used as the baseline for muscle tone. RMS ;

[0151] Then, different intensities of neural feedback stimulation were applied to the individual, and electromyographic (EMG) signals were collected under different intensities of neural feedback stimulation. Combined with the feature extraction step, features corresponding to the amplitude and frequency components of the EMG were extracted. At the same time, the individual's subjective feelings under different stimulation intensities were taken into account to determine the neural feedback threshold K. RMS ;

[0152] In this embodiment, for example, when an individual is under a certain stimulus intensity, the S of the electromyographic signal RMS When the value reaches a certain level and the individual's subjective feeling changes significantly, the S value at this point is... RMS The value was determined as the neural feedback threshold RMS. b Individual subjective feelings, such as "comfortable-uncomfortable";

[0153] Step 5: Calculation of muscle tone data:

[0154] After determining the baseline muscle tone J RMS Neural feedback threshold K RMS Subsequently, electromyographic signals were acquired in real time, and their real-time root mean square value S was calculated. RMS ;

[0155] Subsequently, based on the real-time root mean square value S RMS Baseline muscle tone J RMS Neural feedback threshold K RMS To calculate muscle tone data D;

[0156] The calculation formula is:

[0157] In the formula, the molecule S RMS- J RMS The denominator K represents the amplitude difference between the real-time electromyographic signal and the signal in the relaxed state. RMS- J RMSD represents the amplitude difference between the neural feedback threshold and the relaxed state signal, and the ratio between the two represents the muscle tone level of the real-time electromyography signal relative to the individual's relaxed state and neural feedback threshold.

[0158] When D=0, the individual muscles are in a relaxed state; when D=1, the muscle tension state corresponding to the neural feedback threshold is reached.

[0159] Muscle tone data D reflects the level of muscle tone in real-time electromyography relative to an individual's relaxed state and neural feedback threshold.

[0160] Example 1 achieves effective processing of the electrical signals of muscle tone neural feedback intensity through a complete workflow of "preliminary noise reduction - noise suppression - feature extraction - baseline and threshold determination - muscle tone data calculation". A noise reduction method combining moving average filtering, median filtering, and wavelet transform is used to effectively remove noise such as motion artifacts and electromagnetic interference, improving signal quality. The root mean square value is used to calculate the electromyographic amplitude, and the muscle tone baseline and neural feedback threshold are determined based on individual differences, enabling personalized calculation of muscle tone data. This example can accurately reflect an individual's real-time muscle tone level, providing a reliable data processing foundation for muscle tone-related research and clinical applications, and has good versatility and practicality.

[0161] As a second embodiment of the present invention:

[0162] Please see Figure 1 As shown, in specific implementation, compared with Embodiment 1, the technical solution of this embodiment differs from that of Embodiment 1 only in that, in the feature extraction step, feature extraction is performed on the noise-suppressed bioelectric signal St, and features corresponding to frequency components are also obtained.

[0163] The frequency component extraction method is as follows:

[0164] The bioelectrical signal S after noise suppression t Perform a Discrete Fourier Transform (DFT) on the bioelectrical signal S. t The corresponding time-domain signal is converted into a frequency-domain signal, and then the frequency domain is divided into different frequency bands. The energy of each frequency band is then calculated, thus obtaining the frequency component characteristics of the frequency-domain signal.

[0165] The specific method is as follows:

[0166] First, discrete bioelectrical signals S are collected within a pre-specified time interval [t3, t4]. t（r） r = 1, 2, ..., q, where r is the index variable for summation, and its value range is 0, 1, 2, ..., q-1. t（r） Represented as the noise-suppressed bioelectrical signal S t The value at the r-th discrete point;

[0167] The calculation formula using the Discrete Fourier Transform:

[0168] Where S(u) represents the frequency domain signal after obtaining the discrete Fourier transform. is the rotation factor, j is the imaginary unit, and u represents the index of the frequency point, which is 0, 1, ..., q-1;

[0169] After obtaining the frequency domain signal S(u) after discrete Fourier transform, the frequency domain is divided into p frequency bands, and the lower limit frequency of each frequency band is marked as F. L,g The upper limit frequency is F U,g ,

[0170] At the same time, the corresponding frequency point index is marked as H. L,g and H U,g ;

[0171] pass:

[0172] Calculate the energy E of the g-th frequency band. g :

[0173] Where g = 1, 2, ..., p, and f is the index variable for summation, with values ​​ranging from H... L,g To H U,g , |S(f)| represents the amplitude of the frequency domain signal S(f);

[0174] Simultaneously, after calculating the muscle tone data, the frequency component features of the real-time acquired electromyographic signals are extracted according to the frequency component extraction step, and then the muscle tone data D is corrected in combination with the frequency component features.

[0175] The specific method is as follows:

[0176] The different frequency bands in the frequency domain include: low frequency band, mid frequency band, and high frequency band;

[0177] Simultaneously, the pre-determined correspondence between different frequency bands and muscle tone, based on experimental analysis, is extracted, as follows:

[0178] Low frequency band (0-10Hz): Increased energy in this frequency band is usually associated with muscle fatigue. When muscles are in a contracted state for a long time, the energy in the low frequency band will gradually increase.

[0179] Mid-frequency band (10-50Hz): Its energy variation is related to the sustained contraction intensity and stability of the muscle. When the muscle contracts stably, the energy in this frequency band will remain at a relatively stable level; when the contraction intensity changes, the energy will also fluctuate accordingly.

[0180] High frequency band (50-100Hz): Energy changes in the high frequency band are often related to the rapid contraction and explosive power of muscles. When muscles perform rapid and powerful contraction movements, the energy in this frequency band will increase significantly.

[0181] Subsequently, a segmented adjustment strategy was adopted, as follows:

[0182] The adjustment strategies for low-frequency, mid-frequency, and high-frequency bands are consistent;

[0183] Taking the low-frequency band (0-10Hz) as an example;

[0184] Its energy range is divided into three intervals: [0,E1], (E1,E2], and (E2,+∞), and different adjustment coefficients β1, β2, and β3 are set for each interval;

[0185] Where β1 < β2 < β3;

[0186] The adjustment factor indicates that the degree of influence of low-frequency energy changes on muscle tone data D varies in different energy ranges;

[0187] Extract the detected low-frequency energy E low The correction method for muscle tone data D is as follows:

[0188]

[0189] Among them, E c This corresponds to a preset reference energy value in the low-frequency band;

[0190] For example:

[0191] For example, setting E1=50, E2=100, β1=0.05, β2=0.1, β3=0.15, the low-frequency reference energy value E c =30;

[0192] If low-frequency energy E is detected at present low =60, which is in the interval (E1,E2]. The muscle tone data before correction is D=0.4, then the corrected muscle tone data is:

[0193] .

[0194] Example 2, building upon Example 1, adds frequency component feature extraction to the feature extraction step. It converts the bioelectrical signal into a frequency domain signal using Discrete Fourier Transform, calculates the energy of each frequency band to obtain frequency component features, and, combined with the correspondence between different frequency bands and muscle tone, employs a segmented adjustment strategy to correct the muscle tone data. This allows this example to not only reflect electromyographic amplitude information but also to analyze muscle state in depth from a frequency dimension, such as assessing muscle fatigue, contraction intensity, and explosive power. This further improves the accuracy of muscle tone data and the comprehensiveness of muscle state assessment, providing richer data support for a more accurate understanding of muscle function.

[0195] As an embodiment of the present invention:

[0196] Please see Figure 1 As shown, in specific implementation, compared with Embodiment 1 and Embodiment 2, the technical solution of this embodiment is to combine the solutions of Embodiment 1 and Embodiment 2. The difference between the technical solution of this embodiment and Embodiment 1 and Embodiment 2 is only that when correcting the muscle tone data D in this embodiment, a threshold adjustment strategy can also be adopted. The threshold adjustment strategy is as follows:

[0197] The threshold adjustment strategies are consistent across low-frequency, mid-frequency, and high-frequency bands;

[0198] Taking the mid-frequency band (10-50Hz) as an example;

[0199] Extract the preset adjustment thresholds TZ1 and TZ2 corresponding to the mid-frequency band;

[0200] Where TZ1 < TZ2;

[0201] Extract the detected mid-frequency energy E mid ;

[0202] When the energy E in this frequency band mid The muscle tone data D should be adjusted accordingly if the following conditions are met:

[0203] When E mid When the value is greater than TZ2, it indicates that the muscle contraction intensity is relatively high and the muscle tone is relatively high. At this time, the muscle tone data D is increased by a fixed value ΔD1, i.e., D new =D+ΔD1;

[0204] When E mid When TZ1 < TZ1, it indicates that the muscles are in a relatively relaxed state with low muscle tension. The muscle tension data D is then reduced by a fixed value ΔD2, i.e., D < TZ1. new =D-ΔD2;

[0205] When TZ1≤E mid When the value is ≤TZ2, the muscle condition is considered normal and the muscle tone data D remains unchanged.

[0206] For example:

[0207] For example, set the mid-frequency thresholds TZ1=80, TZ2=120, ΔD1=0.2, and ΔD2=0.1;

[0208] If the current detected mid-frequency energy E mid =130, the original muscle tone data D=0.5;

[0209] That is, E mid If the value is greater than T2, then the corrected muscle tone data D is... new =0.5+0.2=0.7.

[0210] Example 3 integrates the technical solutions of Examples 1 and 2 and introduces a threshold adjustment strategy to correct muscle tone data. This strategy presets adjustment thresholds for different frequency bands and adjusts the muscle tone data by increasing, decreasing, or keeping it unchanged based on the relationship between the detected frequency band energy and the threshold. Compared to the previous two examples, Example 3 can more intuitively and quickly adjust muscle tone data according to changes in frequency band energy, effectively reflecting changes in muscle contraction intensity. It enhances the sensitivity and adaptability of muscle tone data to real-time changes in muscle state, making the muscle tone assessment results more closely match the actual muscle state and providing a more accurate reference for subsequent decision-making based on muscle tone data.

[0211] As an embodiment of the present invention:

[0212] Please see Figure 1 As shown, in specific implementation, compared with Embodiment 1, Embodiment 2 and Embodiment 3, the technical solution of this embodiment is to combine the solutions of Embodiment 1, Embodiment 2 and Embodiment 3.

[0213] Example 4 comprehensively integrates the technical content of Examples 1, 2, and 3, covering a complete noise reduction process, extraction of electromyographic amplitude and frequency component features, segmented adjustment, threshold adjustment, and other muscle tone data correction strategies. Through multi-dimensional data processing and correction methods, this example maximizes the accuracy and reliability of muscle tone neurofeedback electrical signal data processing, enabling comprehensive and multi-level analysis of muscle state. It adapts to the muscle tone assessment needs of different individuals and complex muscle movement scenarios, providing a more complete and accurate data processing solution for muscle tone-related research, rehabilitation therapy, sports training, and other fields, and has extremely high application value and practical significance.

[0214] It should be stated that all user data collected in this application was collected with the user's consent and authorization, and the use of user data is legal and compliant, and the use and processing of user data comply with the relevant laws, regulations and standards of the relevant regions.

[0215] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0216] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for processing electrical signal data from muscle tone neural feedback, characterized in that, Includes the following steps: Preliminary noise reduction: When processing the electrical signals of muscle tone nerve feedback intensity, the raw bioelectrical signals are first acquired; The original bioelectric signal was initially denoised using a moving average filter. Noise suppression: A combination of median filtering and wavelet transform is used to suppress noise in the bioelectrical signal after preliminary noise reduction. Feature extraction: Feature extraction is performed on the noise-suppressed bioelectrical signal to obtain the features corresponding to the electromyographic amplitude and frequency components; Determining the muscle tone baseline and neural feedback threshold: Signals were collected from individuals in a relaxed state. The root mean square (RMS) value of the electromyography (EMG) signal in the relaxed state was calculated and used as the muscle tone baseline. Subsequently, neural feedback stimuli of different intensities were applied to the individuals, and EMG signals under different intensities of neural feedback stimuli were collected. Combined with the feature extraction step, features corresponding to the amplitude and frequency components of the EMG were extracted. At the same time, the individual's subjective feelings under different stimulus intensities were taken into account to determine the neural feedback threshold. Muscle tone data calculation: Real-time acquisition of electromyographic signals and calculation of their real-time root mean square value; Muscle tone data are then calculated based on real-time root mean square value, baseline muscle tone, and neural feedback threshold. Muscle tone data correction: The frequency component features of the real-time acquired electromyographic signals are extracted according to the frequency component extraction steps, and then the muscle tone data D is corrected in combination with the frequency component features.

2. The method for processing muscle tone neurofeedback electrical signal data according to claim 1, characterized in that, The moving average filtering method is as follows: The original bioelectrical signal was labeled as X. t t represents the time series, X t This represents the value of the raw bioelectrical signal collected at timestamp t; The average value of the original bioelectric signal is calculated within a pre-specified time window N; For timestamp t, the formula for calculating the bioelectrical signal after moving average filtering is: In the formula, Y t The signal is the bioelectrical signal after preliminary noise reduction, and 'i' is the index variable for summation, with values ​​ranging from... arrive , Indicates to Round down to the nearest integer.

3. The method for processing muscle tone neurofeedback electrical signal data according to claim 2, characterized in that, The median filtering process is as follows: Within a pre-specified time window M, the median of the original bioelectrical signal is calculated; For timestamp t, the bioelectrical signal after median filtering is calculated as follows: With Y t Centered on the target, acquire bioelectrical signal values ​​within a time window M: , ... ... ; The bioelectrical signal values ​​within the time window M are then sorted, and the median value of the sorted values ​​is taken as the median-filtered bioelectrical signal Z. t .

4. The method for processing muscle tone neurofeedback electrical signal data according to claim 3, characterized in that, The wavelet transform processing method is as follows: The median-filtered bioelectrical signal Z t Perform discrete wavelet transform to decompose it into approximate components and detail components; For bioelectrical signals Z t Perform m-level wavelet decomposition, and after the j-th level decomposition, obtain the approximate component A. j，t and detail component D j，t j = 1, 2, ..., m; In the decomposed detail components, the coefficients in the detail components are thresholded by pre-setting a contrast threshold T: When the absolute value of the coefficient is greater than or equal to the threshold T, the original value of the coefficient is retained; when the absolute value of the coefficient is less than the threshold T, the coefficient is set to 0. After processing the detail component coefficients, the processed approximate components and detail components are reconstructed using inverse discrete wavelet transform to obtain the noise-suppressed bioelectric signal S. t .

5. The method for processing muscle tone neurofeedback electrical signal data according to claim 4, characterized in that, The method for extracting electromyographic amplitude is as follows: Discrete bioelectrical signals S were collected within a pre-specified time interval. t The root mean square method is used to calculate the electromyographic amplitude. Formula for calculating the root mean square value in discrete form: Among them, S RMS S represents the electromyographic amplitude, e is the index variable for summation, and its value ranges from 1 to k; t(e) The signal S after noise suppression t The value at the e-th discrete point.

6. The method for processing muscle tone neurofeedback electrical signal data according to claim 5, characterized in that, The frequency component extraction method is as follows: The bioelectrical signal S after noise suppression t Perform a Discrete Fourier Transform (DFT) on the bioelectrical signal S. t The corresponding time-domain signal is converted into a frequency-domain signal, and then the frequency domain is divided into different frequency bands. The energy of each frequency band is then calculated, thus obtaining the frequency component characteristics of the frequency-domain signal. The specific method is as follows: First, discrete bioelectrical signals S are collected within a pre-specified time interval. t（r） r = 1, 2, ..., q, where r is the index variable for summation, and its value range is 0, 1, 2, ..., q-1. t（r） Represented as the noise-suppressed bioelectrical signal S t The value at the r-th discrete point; The calculation formula using the Discrete Fourier Transform: Where S(u) represents the frequency domain signal after obtaining the discrete Fourier transform. is the rotation factor, j is the imaginary unit, and u represents the index of the frequency point, which is 0, 1, ..., q-1; After obtaining the frequency domain signal S(u) after discrete Fourier transform, the frequency domain is divided into p frequency bands, and the lower limit frequency of each frequency band is marked as F. L,g The upper limit frequency is F U,g , At the same time, the corresponding frequency point index is marked as H. L,g and H U,g ; pass: Calculate the energy E of the g-th frequency band. g : Where g = 1, 2, ..., p, and f is the index variable for summation, with values ​​ranging from H... L,g To H U,g , |S(f)| represents the amplitude of the frequency domain signal S(f).

7. The method for processing muscle tone neurofeedback electrical signal data according to claim 6, characterized in that, The formula for calculating muscle tone data is: In the formula, J RMS K is the baseline for muscle tone. RMS S is the neural feedback threshold. RMS The root mean square value is the real-time value, and D represents muscle tone data; the molecule S RMS- J RMS The denominator K represents the amplitude difference between the real-time electromyographic signal and the signal in the relaxed state. RMS- J RMS D represents the amplitude difference between the neural feedback threshold and the relaxed state signal, and the ratio between the two represents the muscle tone level of the real-time electromyography signal relative to the individual's relaxed state and neural feedback threshold.

8. The method for processing muscle tone neurofeedback electrical signal data according to claim 7, characterized in that, in, When D=0, the individual's muscles are in a relaxed state; when D=1, the muscle tension state corresponding to the neural feedback threshold is reached.

9. The method for processing muscle tone neurofeedback electrical signal data according to claim 7, characterized in that, The specific methods for correcting muscle tone data are as follows: The different frequency bands in the frequency domain include: low frequency band, mid frequency band and high frequency band, and the adjustment strategies for the low frequency band, mid frequency band and high frequency band are the same; The low-frequency band was selected and its energy range was divided into three intervals: [0,E1], (E1,E2], and (E2,+∞). Different adjustment coefficients β1, β2, and β3 were preset for each interval. Where β1 < β2 < β3; Extract the detected low-frequency energy E low The correction method for muscle tone data D is as follows: Among them, E c This corresponds to a preset reference energy value in the low-frequency band.

10. A method for processing muscle tone neurofeedback electrical signal data according to claim 7, characterized in that, The specific methods for correcting muscle tone data are as follows: The different frequency bands in the frequency domain include: low frequency band, mid frequency band and high frequency band, and the adjustment strategies for the low frequency band, mid frequency band and high frequency band are the same; Select the mid-frequency band and extract the preset adjustment thresholds TZ1 and TZ2 corresponding to the mid-frequency band; Where TZ1 < TZ2; Extract the detected mid-frequency energy E mid ; When the energy E in this frequency band mid The muscle tone data D should be adjusted accordingly if the following conditions are met: When E mid When the muscle tone data is greater than TZ2, the muscle tone data D will be increased by a fixed value ΔD1, i.e., D new =D+ΔD1; When E mid When <TZ1, the muscle tone data D is reduced by a fixed value ΔD2, i.e., D new =D-ΔD2; When TZ1≤E mid When TZ2 is less than or equal to 2, the muscle tone data D remains unchanged.

Images