Method for evaluating core muscle group recovery after spinal surgery in elderly people using electromyography signals

Abstract

The present application relates to the technical field of electromyography, and particularly relates to an electromyography evaluation method for core muscle group recovery after spinal surgery of the elderly. According to the time-frequency difference characteristics between different effective contraction electromyography segments and pain-free reference segments, and the degree of pain, the core muscle group signal segment, the mild mixed segment or the severe mixed segment is judged; according to the amplitude variation trend of the original electromyography signal in the mild mixed segment and the pain-free reference segment, the output core muscle group signal segment is determined, or the severe mixed segment is re-determined; the candidate pain reflex sub-period, the candidate autonomous driving sub-period and the overlapping mixed sub-period of the severe mixed segment are obtained, and a target function of the fusion sub-period signal characteristics is constructed, and the core muscle group signal segment is obtained through iterative optimization; and the muscle group recovery rate is obtained according to the time-frequency characteristics of the core muscle group signal segment corresponding to all effective contraction electromyography segments. The present application can accurately obtain the core muscle group signal segment stripped of pain interference, and improve the accuracy of muscle group recovery evaluation.

Description

Technical Field

[0001] This invention relates to the field of electromyography signal processing technology, specifically to an electromyography signal assessment method for the recovery of core muscle groups after spinal surgery in elderly patients. Background Technology

[0002] Spinal diseases are common among people over 60 years old. In severe cases, symptoms such as numbness in the lower limbs and bowel and bladder dysfunction may occur. Spinal fusion and decompression surgery are the mainstream treatment methods at present. Due to the decline in the body's metabolic capacity in the elderly, postoperative recovery is a key focus in the field of rehabilitation. In particular, the functional recovery of the core muscle groups after surgery is a core indicator that determines the long-term prognosis of patients and reduces the risk of recurrence. Therefore, the assessment of the functional recovery of the core muscle groups after surgery is a necessary means.

[0003] In existing technologies, reflexive electromyography (EMG) signals are usually ignored, and EMG indices are calculated directly using mixed signals. Because the time-domain and frequency-domain characteristics of the two signals in the mixed EMG signals overlap significantly, existing conventional filtering techniques cannot effectively distinguish between the two types of signals. This results in the calculated EMG indices significantly overestimating or underestimating the patient's true muscle control ability, leading to insufficient accuracy in the rehabilitation assessment of elderly patients after spinal surgery. Summary of the Invention

[0004] To address the technical problem that existing conventional filtering techniques cannot effectively distinguish between two types of signals, leading to significant overestimation or underestimation of the patient's actual muscle control ability in the calculated electromyographic indicators, and resulting in insufficient accuracy in the rehabilitation assessment of elderly patients after spinal surgery, this invention aims to provide a method for assessing electromyographic signals of core muscle group recovery after spinal surgery in the elderly. The specific technical solution adopted is as follows: This invention proposes a method for evaluating electromyographic signals in the recovery of core muscle groups after spinal surgery in elderly patients, the method comprising: Obtain raw electromyographic signals of the postoperative patient during different core contraction movements, as well as the degree of pain before each movement; Based on the amplitude characteristics of the original electromyographic signals and the degree of pain, effective contractile electromyographic segments and pain-free baseline segments are selected. Based on the time-frequency difference characteristics between different effective contractile electromyographic segments and pain-free baseline segments, the signal similarity between each effective contractile electromyographic segment and the pain-free baseline segment is obtained. Based on the signal similarity between each effective contractile electromyographic segment and the pain-free baseline segment, as well as the degree of pain, the degree of pain interference of each effective contractile electromyographic segment is obtained and judged as a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment. Based on the amplitude variation trend of the original electromyographic signals in the mild mixed segment and the painless baseline segment, the core muscle group signal segment is determined, or it is re-identified as the severe mixed segment; candidate pain reflex sub-segments, candidate autonomous driving sub-segments, and overlapping mixed sub-segments of the severe mixed segment are obtained, and an objective function for fusing sub-segment signal features is constructed, and the core muscle group signal segment is obtained through iterative optimization. The muscle recovery rate is obtained based on the time-frequency characteristics of the core muscle group signal segments corresponding to all effective contractile electromyographic segments.

[0005] Furthermore, the method for obtaining the effective contractile electromyographic segment and the pain-free reference segment includes: Sliding window analysis is performed on the raw electromyography (EMG) signals to extract continuous EMG segments with root mean square amplitudes continuously greater than a preset noise threshold and durations meeting a preset duration, which are then used as effective contraction EMG segments. In an effective contraction electromyographic segment, if the pain level before the movement is 0 and the rise time of the electromyographic amplitude is greater than or equal to the preset rise time, the corresponding effective contraction electromyographic segment is taken as the pain-free baseline segment.

[0006] Furthermore, the method for obtaining the signal similarity includes: For effective contraction of electromyographic segments or painless reference segments, obtain the average power frequency, median frequency, root mean square value of electromyography and peak power frequency of the original electromyographic signal. Based on the differences in average power frequency, median frequency, root mean square value of electromyography (EMG) and peak power frequency between the effective contraction EMG segments and the painless reference segment, the signal similarity between each effective contraction EMG segment and the painless reference segment was obtained. The differences were all negatively correlated with the signal similarity.

[0007] Furthermore, the method for obtaining the degree of pain interference includes: Obtain the difference between the constant 1 and the signal similarity; normalize the pain level and obtain the mean of the normalization result and the difference result as the degree of pain interference.

[0008] Furthermore, the method for determining whether a signal segment is a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment includes: If the level of pain interference is less than the preset first interference threshold, the corresponding effective contraction electromyographic segment will be used as the core muscle group signal segment. If the pain interference level is greater than or equal to the preset first interference threshold and less than or equal to the preset second interference threshold, the corresponding effective contraction electromyographic segment will be regarded as a mild mixed segment. If the level of pain interference is greater than the preset second interference threshold, the corresponding effective contraction electromyographic segment will be regarded as a severely mixed segment.

[0009] Furthermore, the method for determining the output core muscle group signal segment, or reclassifying it as a heavily mixed segment, includes: Based on the amplitude variation trend of electromyographic signals in the mild mixed segment and the pain-free baseline segment, the pain reflex period and the undisturbed period were obtained; Time series fitting is performed on the raw electromyography (EMG) signals during undisturbed periods to obtain the fitting function; based on the difference between the EMG amplitude at each sampling time during the pain reflex period and the corresponding function value in the fitting function, the core muscle group signal segment is determined, or it is reclassified as a severely mixed segment. If the core muscle group signal segment is determined, calculate the function value of the pain reflex period on the fitting function, replace the value of the electromyographic signal corresponding to the pain reflex period, and output the core muscle group signal segment.

[0010] Furthermore, the method for obtaining the pain reflex period and the undisturbed period includes: For effective contraction electromyographic segments or painless reference segments, the electromyographic amplitudes at different adjacent time points are subjected to first-order difference. The mean of the absolute values ​​of all first-order differences in the painless baseline segment is obtained as the reference difference. If there is a reference difference whose absolute value of the first-order difference corresponding to consecutive sampling times in the effective contraction electromyographic segment is greater than a preset multiple, the range formed by the corresponding consecutive sampling times will be taken as the pain reflex period; the range formed by the remaining consecutive sampling times will be taken as the undisturbed period.

[0011] Furthermore, the method for determining the output core muscle group signal segment, or reclassifying it as a heavily mixed segment, includes: The mean of the absolute differences between the electromyographic amplitude at each sampling time and the corresponding function value in the fitted function during the pain reflex period is obtained as the average residual of the pain reflex period. If the average residual of the pain reflex period is less than the preset residual threshold, the core muscle group signal segment is determined to be output; if the average residual of the pain reflex period is greater than or equal to the preset residual threshold, it is re-determined as a severely mixed segment.

[0012] Furthermore, the method for obtaining the objective function includes: The candidate autonomous driving sub-segment is fitted with a single-peak Gaussian distribution model based on the pain-free baseline segment to obtain the initial autonomous driving signal; the candidate pain reflex sub-segment is fitted with the average waveform of all candidate pain reflex sub-segments as a template to obtain the initial pain reflex signal. The sum of squared residuals between the original electromyography (EMG) signal and the initial spontaneous drive signal within the candidate spontaneous drive sub-period is obtained as the first feature value; the sum of squared residuals between the original EMG signal and the initial pain reflex signal within the candidate pain reflex sub-period is obtained as the second feature value; and the sum of squared residuals between the original EMG signal and the sum of the initial spontaneous drive signal and the initial pain reflex signal within the overlapping mixed sub-period is obtained as the third feature value. Obtain the first product between the preset first weight and the first eigenvalue, obtain the second product between the preset second weight and the second eigenvalue, obtain the third product between the preset third weight and the third eigenvalue, and obtain the sum of the first product, the second product and the third product as the objective function.

[0013] Furthermore, the method for obtaining the muscle group recovery rate includes: Acquire baseline electromyographic signals of patients during different core contraction movements before surgery; All effective contractile electromyography (EMG) segments corresponding to the core muscle group signal segments were merged, and the average power frequency, median frequency, root mean square value of EMG and peak power frequency of the merged signal segments and the baseline EMG signal were obtained respectively. For any feature among average power frequency, median frequency, root mean square value of electromyography (EMG) or peak power frequency, the ratio of the feature of the merged signal segment to the feature of the baseline EMG signal is obtained as the recovery rate of the corresponding feature.

[0014] The present invention has the following beneficial effects: This invention, based on the amplitude characteristics of the original electromyographic (EMG) signal and the degree of pain, filters out effective contractile EMG segments and pain-free baseline segments, achieving the elimination of invalid noise and resting data in the EMG signal, accurately locating the time window of actual muscle work, and significantly reducing the computational load of subsequent algorithms. Based on the time-frequency difference characteristics between different effective contractile EMG segments and pain-free baseline segments, as well as the degree of pain, it determines whether the segment is a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment, achieving precise quantitative grading of the interference level. Based on the original amplitude characteristics of the mildly mixed segment and the pain-free baseline segment... By analyzing the amplitude variation trend of electromyographic signals, the core muscle group signal segment is identified, or it is reclassified as a severely mixed segment, preventing the omission of complex and hidden continuous interference and ensuring the rigor of the dissection results. Candidate pain reflex sub-segments, candidate autonomously driven sub-segments, and overlapping mixed sub-segments of severely mixed segments are obtained, and a target function for fusing the signal features of these sub-segments is constructed. Iterative optimization is then used to obtain the core muscle group signal segment, achieving blind source separation of the two types of highly aliased signals. Based on the time-frequency characteristics of the core muscle group signal segments corresponding to all effective contraction electromyographic segments, the muscle group recovery rate is obtained. This invention improves the accuracy of muscle group recovery assessment by accurately acquiring the core muscle group signal segment that has been stripped of pain interference. Attached Figure Description

[0015] Figure 1 A flowchart illustrating an electromyographic signal assessment method for core muscle group recovery after spinal surgery in elderly individuals, provided as an embodiment of the present invention. Figure 2 A flowchart of a method for determining the output core muscle group signal segment, or re-identifying it as a heavily mixed segment, according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating a method for obtaining an objective function according to an embodiment of the present invention. Detailed Implementation

[0016] The following description, in conjunction with the accompanying drawings, details the specific scheme of the electromyographic signal assessment method for the recovery of core muscle groups after spinal surgery in the elderly provided by this invention.

[0017] Please see Figure 1 The document illustrates a flowchart of an electromyography (EMG) signal assessment method for core muscle recovery in elderly individuals following spinal surgery, provided by an embodiment of the present invention. The specific method includes: Step S1: Obtain the raw electromyographic signals of the postoperative patient during different core contraction movements, as well as the degree of pain before each movement.

[0018] In the embodiments of the present invention, patients aged 60-85 years who were 1-4 weeks post-spinal surgery were included, while patients with severe cognitive impairment, amyotrophic lateral sclerosis, peripheral neuropathy, or other conditions that prevented them from cooperating with the movements were excluded. Muscle relaxants were discontinued 24 hours before collection, and short-acting analgesics were discontinued 1 hour before collection.

[0019] The SENIAM surface electromyography (EMG) acquisition was performed according to the SENIAM standard. The core muscle groups to be tested were the multifidus, erector spinae, transversus abdominis, and rectus abdominis muscles in the L3-L4 segments bilaterally. Ag / AgCl disposable electrodes were used, with an electrode diameter of 1 cm and an electrode spacing of 2 cm. Before application, the skin was wiped with 75% alcohol until it was slightly red. The electrode impedance was tested with an impedance meter and found to be ≤10 kΩ. The electrode leads were fixed along the direction of the muscle fibers to avoid artifacts caused by the leads shaking during movement. Patients were placed in a supine position with pressure sensors placed below the lumbar spine. The position was adjusted to maintain the lumbar spine pressure in the range of 30-40 kPa to ensure that the lumbar curvature was in a neutral position and to avoid the EMG baseline deviation caused by lumbar lordosis or kyphosis. Both hip and knee joints were flexed at 30 degrees, and both lower limbs were relaxed naturally.

[0020] A portable surface electromyography (EMG) instrument was used. Considering that the highest frequency of EMG signals is 500Hz, the sampling frequency was set to 2000Hz to meet the Nyquist sampling requirements of EMG signals, that is, a sampling time was acquired every 0.5 milliseconds. The analog bandwidth was 20-500Hz to filter out low-frequency motion artifacts and high-frequency noise. The common-mode rejection ratio was set to ≥110dB to effectively suppress power frequency interference. The gain was set to 1000 times to ensure the resolution of weak EMG signals and avoid signal saturation. It was required that the acquired EMG amplitude did not exceed 80% of the ADC range. If the amplitude exceeded the range, the gain was reduced to 500 times.

[0021] During the standardized movement pattern data collection, the patient first relaxes their entire body, breathes steadily, and rests for 5 minutes to collect resting electromyographic signals as the benchmark for subsequent calculations. Then, the patient performs three core contraction movements of varying difficulty for data collection. All movements are performed under the unified guidance of a rehabilitation therapist, with a 2-minute rest between each movement: Abdominal breathing: Keeping the chest still, the abdomen expands during inhalation and contracts during exhalation, holding each position for 3 seconds, relaxing for 5 seconds, and repeating 10 times; Dead bug pose: Extending both arms straight towards the ceiling, bending both legs at 90 degrees, slowly lowering the opposite arm and leg, keeping the lower back pressed against the bed, holding each position for 3 seconds, relaxing for 5 seconds, and repeating 10 times; Bridge pose: Raising the hips until the shoulders, hips, and knees form a straight line, holding for 3 seconds, relaxing for 5 seconds, and repeating 10 times.

[0022] Ten seconds before each movement begins, the therapist uses the visual analog scale (VAS) to record the patient's pain level, with a score ranging from 0 to 10. The higher the score, the greater the pain.

[0023] Considering the presence of noise in the signal due to random system errors of the acquisition equipment or power grid interference, preprocessing of the electromyography (EMG) signal is necessary. This includes: using the mean EMG value of the first 10 seconds of resting state as the baseline, subtracting the baseline value from all signals to eliminate baseline drift; using a 50Hz second-order Butterworth notch filter to remove power grid interference; and using a 5-layer wavelet soft thresholding method to remove motion artifacts, with the wavelet basis set to db4 and the threshold set to 3 times the noise standard deviation. The specific methods are well-known to those skilled in the art and will not be elaborated here.

[0024] Step S2: Based on the amplitude characteristics of the original electromyographic signals and the degree of pain, select effective contractile electromyographic segments and pain-free reference segments; based on the time-frequency difference characteristics between different effective contractile electromyographic segments and pain-free reference segments, obtain the signal similarity between each effective contractile electromyographic segment and the pain-free reference segment; based on the signal similarity between each effective contractile electromyographic segment and the pain-free reference segment, as well as the degree of pain, obtain the degree of pain interference of each effective contractile electromyographic segment, and determine whether it is a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment.

[0025] The amplitude characteristics of the raw electromyographic signal reflect the intensity and activation level of the muscle's active contraction; the degree of pain reflects the probability that painful stimuli will induce unconscious electromyographic activity; based on the amplitude characteristics of the raw electromyographic signal and the degree of pain, effective contractile electromyographic segments and pain-free baseline segments are selected.

[0026] Preferably, in one embodiment of the present invention, the method for obtaining the effective contraction electromyographic segment and the pain-free reference segment includes: The first step is to perform sliding window analysis on the raw electromyography (EMG) signal and extract continuous EMG segments with root mean square amplitude continuously greater than a preset noise threshold and duration meeting a preset duration, as effective contraction EMG segments. Based on this, an effective contractile electromyographic segment refers to the electromyographic signal during the active contraction phase of the core muscle group, which excludes the invalid signals during the resting and relaxation phases and can reflect the recovery status of the core muscle group.

[0027] It should be noted that, in the embodiments of the present invention, the root mean square (RMS) value of the electromyography (EMG) signal of resting-state movement is obtained; the standard deviation of the EMG amplitude at all sampling times in the EMG signal of resting-state movement is obtained; and the sum of the RMS value and the standard deviation of the EMG by a preset multiple is obtained as a preset noise threshold. Based on a normal distribution, the RMS value plus 3 times the standard deviation covers 99.73% of the resting noise fluctuation range. Signals exceeding this threshold are highly likely to be genuine muscle contractions, effectively filtering out random noise misjudgments. Therefore, the preset multiple is set to 3. Active contraction of the core muscle groups in the elderly is usually continuous; based on the physiological contraction characteristics of the elderly and the noise filtering requirements, the preset duration is set to 0.5 seconds. In other embodiments of the present invention, the preset multiple and preset duration can be set according to specific circumstances, and are not limited or elaborated here.

[0028] The second step is to use the effective contraction electromyography segment as the pain reference segment if the pain level before the movement is 0 and the rise time of the electromyography amplitude is greater than or equal to the preset rise time.

[0029] It should be noted that if an effective contractile electromyographic segment with a pain level of 0 before the movement is not selected, painless electromyographic data from the patient's historical rehabilitation record or pre-stored painless electromyographic baseline data from healthy individuals of the same age will be retrieved as the painless baseline segment.

[0030] It should be noted that, in the embodiments of the present invention, the rising edge of electromyography for voluntary contraction is 200-500ms, and the rising edge of electromyography for pain reflex is <100ms; the preset rise time is set to 200ms based on the physiologically recognized characteristic differences; in other embodiments of the present invention, the preset rise time can be set according to specific circumstances, and will not be limited or elaborated here.

[0031] The pain-free baseline segment reflects the signal segment without pain interference. In order to reflect the pain interference of the effective contraction electromyographic segment, the time-frequency difference characteristics between the effective contraction electromyographic segment and the pain-free baseline segment are analyzed. The smaller the difference, the closer the signal is to the pain-free baseline segment, and the greater the signal similarity. Based on the time-frequency difference characteristics between different effective contraction electromyographic segments and the pain-free baseline segment, the signal similarity between each effective contraction electromyographic segment and the pain-free baseline segment is obtained.

[0032] Preferably, in one embodiment of the present invention, the method for obtaining signal similarity includes: For effective contraction of electromyographic segments or painless reference segments, obtain the average power frequency, median frequency, root mean square value of electromyography and peak power frequency of the original electromyographic signal. It should be noted that, in the embodiments of the present invention, the power spectral density is obtained by performing a fast Fourier transform on the original electromyographic signal. The sum of the products of each frequency value and its corresponding power is divided by the total power to obtain the average power frequency, which reflects the frequency domain distribution of the electromyographic signal. The center frequency point after dividing the total power of the power spectral density in half, i.e., the median frequency, reflects the fatigue state of the muscle. The square root of the squared signal amplitude of all sampling points in the electromyographic signal is obtained by averaging and then taking the square root, which reflects the contraction strength of the muscle. The frequency point corresponding to the maximum power in the power spectral density curve is obtained, i.e., the peak power frequency, which reflects the dominant frequency band of the electromyographic signal. The specific means are well known to those skilled in the art and will not be described in detail here.

[0033] Based on the differences in average power frequency, median frequency, root mean square value of electromyography (EMG) and peak power frequency between the effective contraction EMG segments and the painless reference segment, the signal similarity between each effective contraction EMG segment and the painless reference segment was obtained. The differences were all negatively correlated with the signal similarity.

[0034] It should be noted that the difference represents the absolute value of the difference. The greater the difference between each feature, the more inconsistent the state of the effective contraction electromyographic segment and the painless reference segment. Therefore, the differences are all negatively correlated with signal similarity. In one embodiment of the present invention, for any feature among the average power frequency, median frequency, root mean square value of electromyography, or peak power frequency, the ratio of the feature difference to the feature of the painless reference segment is obtained, and the result is expressed as an exponential function with the natural constant as the base. A negative correlation mapping is performed to obtain the similarity of the corresponding features; the mean similarity of all features is obtained as the signal similarity; the specific negative correlation mapping is a technique well known to those skilled in the art and will not be elaborated here.

[0035] The formula for signal similarity is expressed as: ;in, Indicates signal similarity; Indicates effective contraction electromyographic segments Features ; Indicates the pain-free baseline segment Features ; This represents the number of features, with a value of 4.

[0036] Signal similarity reflects the degree of interference from pain reflexes mixed into the effective contractile electromyographic segments. The smaller the signal similarity, the more likely it is that unconscious pain reflex components are mixed in. The degree of pain reflects the probability that pain stimulation will induce unconscious electromyographic activity. Therefore, the degree of pain interference for each effective contractile electromyographic segment can be obtained based on the signal similarity between each effective contractile electromyographic segment and the pain-free baseline segment, as well as the degree of pain.

[0037] Preferably, in one embodiment of the present invention, the method for obtaining the degree of pain interference includes: Obtain the difference between the constant 1 and the signal similarity; normalize the pain level and obtain the mean of the normalization result and the difference result as the degree of pain interference.

[0038] It should be noted that, in the embodiments of the present invention, normalizing the pain level includes: calculating the ratio of the pain level to the theoretical maximum pain score, i.e., normalizing.

[0039] The greater the degree of pain interference, the smaller the signal similarity, the greater the degree of pain, and the more the effective contraction electromyographic segment is affected by pain, thus determining whether it is a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment.

[0040] Preferably, in one embodiment of the present invention, the method for determining whether a signal segment is a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment includes: If the level of pain interference is less than the preset first interference threshold, the corresponding effective contraction electromyographic segment will be used as the core muscle group signal segment. If the pain interference level is greater than or equal to the preset first interference threshold and less than or equal to the preset second interference threshold, the corresponding effective contraction electromyographic segment will be regarded as a mild mixed segment. If the level of pain interference is greater than the preset second interference threshold, the corresponding effective contraction electromyographic segment will be regarded as a severely mixed segment.

[0041] It should be noted that, in one embodiment of the present invention, considering that surface electromyography signals are highly susceptible to small noise and there is no absolute zero interference, the first interference threshold is preset to 0.1 based on relevant clinical experience; the greater the degree of pain interference than the median value, the more severe the pain, and the interference component gradually covers the entire body from local interference; the second interference threshold is preset to 0.5 based on relevant clinical experience; in other embodiments of the present invention, the preset first interference threshold and the preset second interference threshold can be set according to specific circumstances, and are not limited or elaborated here.

[0042] Based on this, there is no pain reflex interference in the core muscle group signal segment, so no further stripping processing is required; the mildly mixed segment is mostly discrete paroxysmal spikes, which do not completely overlap with the effective signal; the severely mixed segment highly overlaps with the effective signal and requires different degrees of subsequent stripping processing.

[0043] Step S3: Based on the amplitude change trend of the original electromyographic signals in the mild mixed segment and the painless baseline segment, determine the output core muscle group signal segment, or re-determine it as a severe mixed segment; obtain the candidate pain reflex sub-segment, candidate autonomous driving sub-segment, and overlapping mixed sub-segment of the severe mixed segment, and construct the objective function of the fusion sub-segment signal features, and iteratively optimize to obtain the core muscle group signal segment.

[0044] The pain reflex in the mild mixed segment is a discrete spike with low overlap with the effective signal. By analyzing the amplitude change trend of the original electromyographic signal in the mild mixed segment and the pain-free baseline segment, the pain reflex period is separated for pain stripping. Based on the amplitude change trend of the original electromyographic signal in the mild mixed segment and the pain-free baseline segment, the output core muscle group signal segment is determined, or it is reclassified as the severe mixed segment.

[0045] Preferably, in one embodiment of the present invention, the method for determining the output core muscle group signal segment, or reclassifying it as a heavily mixed segment, is described in the following reference: Figure 2 It illustrates a flowchart of a method for determining the output core muscle group signal segment, or reclassifying it as a heavily mixed segment, including: Step S201: Based on the amplitude change trend of electromyographic signals in the mild mixed segment and the painless baseline segment, obtain the pain reflex period and the undisturbed period.

[0046] Preferably, in one embodiment of the present invention, the method for obtaining the pain reflex period and the undisturbed period includes: For effective contraction electromyographic segments or painless reference segments, the electromyographic amplitudes at different adjacent time points are subjected to first-order difference. It should be noted that the original electromyographic signal is a high-frequency oscillation, and directly performing first-order difference will not be able to reflect the changes in muscle contraction intensity. In the embodiments of the present invention, the method for obtaining first-order difference is as follows: the original electromyographic signal is rectified by full wave, a smooth electromyographic amplitude envelope is obtained by passing a low-pass filter, and first-order difference is performed on different adjacent moments of the envelope.

[0047] The mean of the absolute values ​​of all first-order differences in the painless baseline segment is obtained as the reference difference. If there is a reference difference whose absolute value of the first-order difference corresponding to consecutive sampling times in the effective contraction electromyographic segment is greater than a preset multiple, the range formed by the corresponding consecutive sampling times will be taken as the pain reflex period; the range formed by the remaining consecutive sampling times will be taken as the undisturbed period.

[0048] It should be noted that in one embodiment of the present invention, the preset multiple is set to 5; in other embodiments of the present invention, the size of the preset multiple can be set according to specific circumstances, and will not be limited or elaborated here.

[0049] Step S202: Perform time series fitting on the original electromyography (EMG) signals of the undisturbed period to obtain the fitting function; determine the output core muscle group signal segment based on the difference between the EMG amplitude at each sampling time in the pain reflex period and the function value at the corresponding time in the fitting function, or re-determine it as a severely mixed segment.

[0050] It should be noted that, in one embodiment of the present invention, the method for determining the output core muscle group signal segment, or reclassifying it as a heavily mixed segment, includes: The mean of the absolute differences between the electromyographic amplitude at each sampling time and the corresponding function value in the fitted function during the pain reflex period is obtained as the average residual of the pain reflex period. It should be noted that the absolute difference represents the absolute value of the difference; in one embodiment of the present invention, the formula for the average residual is expressed as: ;in, This represents the average residual during the pain reflex period; Indicates the period of pain reflex; Indicates the period of pain reflex Number of sampling times within; This indicates the original electromyographic signal at the sampling time. The amplitude of electromyography; This indicates the fitting function at the sampling time. The function value; Represents the absolute value symbol.

[0051] If the average residual of the pain reflex period is less than the preset residual threshold, the core muscle group signal segment is determined to be output; if the average residual of the pain reflex period is greater than or equal to the preset residual threshold, it is re-determined as a severely mixed segment.

[0052] It should be noted that, in the embodiments of the present invention, an autoregressive moving average (ARMA) model is used to perform time series fitting on the original electromyography (EMG) signals during undisturbed periods; the method for obtaining the preset residual threshold is as follows: obtain the standard deviation of EMG amplitude at all sampling times in the EMG signals of resting-state action; to prevent overfitting and under-stripping, obtain the product of the standard deviation of EMG amplitude and the preset tolerance coefficient as the preset residual threshold, and set the tolerance coefficient to 1.5 based on relevant historical experience; the specific autoregressive moving average (ARMA) model is a well-known technical means in the art and will not be described in detail here.

[0053] Step S203: If the core muscle group signal segment is determined, calculate the function value of the pain reflex period on the fitting function, replace the value of the electromyographic signal corresponding to the pain reflex period, and output the core muscle group signal segment.

[0054] Based on this, the fitting function reflects the normal force exertion trend and outputs a smooth, continuous core muscle group signal segment that removes pain interference.

[0055] Candidate pain reflex sub-times, candidate autonomously driven sub-times, and overlapping mixed sub-times were obtained for the severely mixed segment; It should be noted that, in the embodiments of the present invention, a sliding window with a window length of 20ms and a step size of 5ms is set to extract all sub-segments. When extracting the main frequency features, the data within the window is padded with zeros at the end to improve the frequency domain calculation resolution. Based on the pain reflex features, since the unconscious electromyographic activity caused by pain bypasses the processing of the cerebral cortex and its nerve conduction speed is extremely fast, consecutive adjacent sliding sub-segments that meet the requirements of 30ms ≤ latency ≤ 80ms and main frequency 70-120Hz are continuously spliced. If the duration of the continuous splicing meets the requirement of 100-300ms, the included sub-segments are marked as candidate pain reflex sub-segments, and the remaining sub-segments are marked as candidate autonomous driving sub-segments. The overlapping part of the two types of sub-segments is marked as an overlapping mixed sub-segment. The latency represents the time difference from the time when the patient receives a pain stimulus or performs an action to the time when the signal reaches its peak.

[0056] Construct an objective function that integrates the signal features of sub-time periods, and iteratively optimize it to obtain the core muscle group signal segment.

[0057] Preferably, in an embodiment of the present invention, the method for obtaining the objective function is described in [reference needed]. Figure 3 It shows a flowchart of a method for obtaining the objective function, including: Step S301: Fit the candidate autonomous driving sub-segment based on the single-peak Gaussian distribution model of the pain-free baseline segment to obtain the initial autonomous driving signal; use the average waveform of all candidate pain reflex sub-segments as a template to fit the candidate pain reflex sub-segments to obtain the initial pain reflex signal.

[0058] It should be noted that, in the embodiments of the present invention, Gaussian fitting is performed on the patient's pain-free baseline segment, and the baseline amplitude, time center, and width are extracted as prior initial values ​​to construct a unimodal Gaussian distribution model. ;in, This represents a single-peak Gaussian distribution model signal; Indicates the amplitude coefficient; Indicates the sampling time within the candidate autonomous driving sub-period; Indicates the center of time; The width is represented by the initial a priori value as the starting point input for the algorithm optimization. The least squares method is used to fit the candidate autonomous driving sub-time period to obtain the initial amplitude, time center and width, and generate the initial autonomous driving signal. The template waveform is generated by aligning the peak values ​​of the electromyographic signals from all candidate pain reflex sub-time periods and averaging them; a parameterized model for the pain reflex signal is then defined. ;in, This represents the initial pain response signal generated by the fitting process; Indicates the amplitude scaling factor; Indicates the template waveform The value at the sampling time; Indicates the independent variable of time; Indicates time offset; amplitude scaling factor Used to adjust the overall amplitude of the template waveform to match the actual signal strength of the current sub-period; time offset. The template waveform is shifted left and right along the time axis to align its phase with the actual trigger time of the candidate pain reflex sub-segment; the candidate pain reflex sub-segment is initially fitted using the least squares method to obtain the initial amplitude scaling factor and time offset, thereby generating the initial pain reflex signal; the specific methods are well known to those skilled in the art and will not be elaborated here.

[0059] Step S302: Obtain the sum of squared residuals between the original electromyography (EMG) signal and the initial voluntary drive signal within the candidate voluntary drive sub-segment as the first feature value; obtain the sum of squared residuals between the original EMG signal and the initial pain reflex signal within the candidate pain reflex sub-segment as the second feature value; obtain the sum of squared residuals between the original EMG signal and the sum of the initial voluntary drive signal and the initial pain reflex signal within the overlapping mixed sub-segment as the third feature value.

[0060] The formula for eigenvalues ​​is expressed as: ; ; ; in, Represents the first eigenvalue; Indicates the second eigenvalue; Indicates the third eigenvalue; This indicates the original electromyographic signal at the sampling time. The amplitude; This indicates the initial autonomous driving signal at the sampling time. The amplitude; This indicates the initial pain reflex signal at the sampling time. The amplitude; Indicates the candidate autonomously driven sub-time period; Indicates candidate pain reflex sub-time periods; This indicates overlapping and mixed sub-time periods.

[0061] Step S303: Obtain the first product between the preset first weight and the first feature value, obtain the second product between the preset second weight and the second feature value, obtain the third product between the preset third weight and the third feature value, and obtain the sum of the first product, the second product and the third product as the objective function.

[0062] It should be noted that, in the embodiments of the present invention, the proportion of the number of sampling moments in the candidate autonomous driving sub-period to the total number of sampling moments in the heavily mixed segment is obtained as a preset first weight; the proportion of the number of sampling moments in the candidate pain reflex sub-period to the total number of sampling moments in the heavily mixed segment is obtained as a preset second weight; and the proportion of the number of sampling moments in the overlapping mixed sub-period to the total number of sampling moments in the heavily mixed segment is obtained as a preset third weight.

[0063] Based on this, in the embodiments of the present invention, the gradient descent method is used to iteratively adjust the amplitude and phase parameters of the initial autonomous driving signal and the initial pain reflex signal until the objective function converges to the minimum value. The obtained initial autonomous driving signal is the core muscle group signal segment stripped of the pain reflex component. The specific gradient descent method is a well-known technique to those skilled in the art and will not be described in detail here.

[0064] It should be noted that for the core muscle group signal segments after stripping, the signal similarity of the stripped signal relative to the pain-free baseline segment is obtained based on the signal similarity acquisition method. If the signal similarity is greater than the preset similarity threshold, it means that the stripping effect meets the requirements; if the signal similarity is less than the preset similarity threshold, it is marked as a suspicious movement and manually reviewed by the rehabilitation therapist to avoid errors caused by special circumstances; if a rejection instruction is received, the suspicious movement is treated as an invalid waste piece and is not included in subsequent calculations; if a forced confirmation or manual selection and boundary adjustment instruction is received, the confirmed movement output by the system is used as the final core muscle group signal segment for feature extraction.

[0065] The greater the signal similarity, the closer the stripped signal is to the painless baseline segment, and the more it meets the requirements. Therefore, based on relevant historical experience, the preset similarity threshold is set to 0.85. In other embodiments of the present invention, the size of the preset similarity threshold can be set according to specific circumstances, and will not be limited or elaborated here.

[0066] Step S4: Obtain the muscle group recovery rate based on the time-frequency characteristics of the core muscle group signal segments corresponding to all effective contraction electromyographic segments.

[0067] Considering that painful stimuli can trigger unconscious spinal reflex electromyographic activity, and that reflex electromyographic signals can mix with signals from the core muscle groups driven by the patient's voluntary contraction, thus affecting the recovery status, analysis based on the acquired core muscle group signal segments can more accurately assess the muscle group recovery status.

[0068] Preferably, in one embodiment of the present invention, the method for obtaining the muscle group recovery rate includes: Acquire baseline electromyographic signals of patients during different core contraction movements before surgery; It should be noted that the electromyography (EMG) signals of preoperative patients were acquired using the same method as those acquired in postoperative patients.

[0069] All effective contractile electromyography (EMG) segments corresponding to the core muscle group signal segments were merged, and the average power frequency, median frequency, root mean square value of EMG and peak power frequency of the merged signal segments and the baseline EMG signal were obtained respectively. For any feature among average power frequency, median frequency, root mean square value of electromyography (EMG) or peak power frequency, the ratio of the feature of the merged signal segment to the feature of the baseline EMG signal is obtained as the recovery rate of the corresponding feature.

[0070] Based on this, and by obtaining the recovery rate of each feature, rehabilitation suggestions can be provided; that is, according to relevant historical experience, if the recovery rate is <30%, it is recommended to maintain low-intensity core activation training; if the recovery rate is 30%-70%, it is recommended to increase the training load and resistance training; and if the recovery rate is >70%, it is recommended to advance to daily activity adaptation training.

[0071] In summary, this invention determines the segment as a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment based on the time-frequency difference characteristics between different effective contractile electromyographic segments and the pain-free baseline segment, as well as the degree of pain. Based on the amplitude variation trend of the original electromyographic signals in the mildly mixed segment and the pain-free baseline segment, it determines the output core muscle group signal segment, or re-identifies it as a severely mixed segment. It obtains candidate pain reflex sub-segments, candidate autonomously driven sub-segments, and overlapping mixed sub-segments for the severely mixed segment, and constructs an objective function to fuse the signal characteristics of these sub-segments, iteratively optimizing to obtain the core muscle group signal segment. Based on the time-frequency characteristics of the core muscle group signal segments corresponding to all effective contractile electromyographic segments, it obtains the muscle group recovery rate. This invention improves the accuracy of muscle group recovery assessment by accurately acquiring the core muscle group signal segment that is free from pain interference.

[0072] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0073] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for evaluating electromyographic signals in the recovery of core muscle groups after spinal surgery in elderly patients, characterized in that, The method includes: Obtain raw electromyographic signals of the postoperative patient during different core contraction movements, as well as the degree of pain before each movement; Based on the amplitude characteristics of the original electromyographic signals and the degree of pain, effective contractile electromyographic segments and pain-free baseline segments are selected. Based on the time-frequency difference characteristics between different effective contractile electromyographic segments and pain-free baseline segments, the signal similarity between each effective contractile electromyographic segment and the pain-free baseline segment is obtained. Based on the signal similarity between each effective contractile electromyographic segment and the pain-free baseline segment, as well as the degree of pain, the degree of pain interference of each effective contractile electromyographic segment is obtained and judged as a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment. Based on the amplitude variation trend of the original electromyographic signals in the mild mixed segment and the painless baseline segment, the core muscle group signal segment is determined, or it is re-identified as the severe mixed segment; candidate pain reflex sub-segments, candidate autonomous driving sub-segments, and overlapping mixed sub-segments of the severe mixed segment are obtained, and an objective function for fusing sub-segment signal features is constructed, and the core muscle group signal segment is obtained through iterative optimization. The muscle recovery rate is obtained based on the time-frequency characteristics of the core muscle group signal segments corresponding to all effective contractile electromyographic segments.

2. The electromyographic signal assessment method for core muscle group recovery after spinal surgery in the elderly according to claim 1, characterized in that, The methods for obtaining the effective contractile electromyographic segments and the pain-free reference segments include: Sliding window analysis is performed on the raw electromyography (EMG) signals to extract continuous EMG segments with root mean square amplitudes continuously greater than a preset noise threshold and durations meeting a preset duration, which are then used as effective contraction EMG segments. In an effective contraction electromyographic segment, if the pain level before the movement is 0 and the rise time of the electromyographic amplitude is greater than or equal to the preset rise time, the corresponding effective contraction electromyographic segment is taken as the pain-free baseline segment.

3. The electromyographic signal assessment method for core muscle group recovery after spinal surgery in the elderly according to claim 1, characterized in that, The method for obtaining signal similarity includes: For effective contraction of electromyographic segments or painless reference segments, obtain the average power frequency, median frequency, root mean square value of electromyography and peak power frequency of the original electromyographic signal. Based on the differences in average power frequency, median frequency, root mean square value of electromyography (EMG) and peak power frequency between the effective contraction EMG segments and the painless reference segment, the signal similarity between each effective contraction EMG segment and the painless reference segment was obtained. The differences were all negatively correlated with the signal similarity.

4. The electromyographic signal assessment method for core muscle group recovery after spinal surgery in the elderly according to claim 1, characterized in that, The methods for obtaining the degree of pain interference include: Obtain the difference between the constant 1 and the signal similarity; normalize the pain level and obtain the mean of the normalization result and the difference result as the degree of pain interference.

5. The electromyographic signal assessment method for core muscle group recovery after spinal surgery in elderly patients according to claim 1, characterized in that, The methods for determining whether a signal segment is a core muscle group signal segment, a mildly mixed segment, or a severely mixed segment include: If the level of pain interference is less than the preset first interference threshold, the corresponding effective contraction electromyographic segment will be used as the core muscle group signal segment. If the pain interference level is greater than or equal to the preset first interference threshold and less than or equal to the preset second interference threshold, the corresponding effective contraction electromyographic segment will be regarded as a mild mixed segment. If the level of pain interference is greater than the preset second interference threshold, the corresponding effective contraction electromyographic segment will be regarded as a severely mixed segment.

6. The electromyographic signal assessment method for core muscle group recovery after spinal surgery in the elderly according to claim 1, characterized in that, The method for determining the output core muscle group signal segment, or reclassifying it as a severely mixed segment, includes: Based on the amplitude variation trend of electromyographic signals in the mild mixed segment and the pain-free baseline segment, the pain reflex period and the undisturbed period were obtained; Time series fitting is performed on the raw electromyography (EMG) signals during undisturbed periods to obtain the fitting function; based on the difference between the EMG amplitude at each sampling time during the pain reflex period and the corresponding function value in the fitting function, the core muscle group signal segment is determined, or it is reclassified as a severely mixed segment. If the core muscle group signal segment is determined, calculate the function value of the pain reflex period on the fitting function, replace the value of the electromyographic signal corresponding to the pain reflex period, and output the core muscle group signal segment.

7. The electromyographic signal assessment method for core muscle group recovery after spinal surgery in elderly patients according to claim 6, characterized in that, The methods for obtaining the pain reflex period and the undisturbed period include: For effective contraction electromyographic segments or painless reference segments, the electromyographic amplitudes at different adjacent time points are subjected to first-order difference. The mean of the absolute values ​​of all first-order differences in the painless baseline segment is obtained as the reference difference. If there is a reference difference whose absolute value of the first-order difference corresponding to consecutive sampling times in the effective contraction electromyographic segment is greater than a preset multiple, the range formed by the corresponding consecutive sampling times will be taken as the pain reflex period; the range formed by the remaining consecutive sampling times will be taken as the undisturbed period.

8. The electromyographic signal assessment method for core muscle group recovery after spinal surgery in the elderly according to claim 6, characterized in that, The method for determining the output core muscle group signal segment, or reclassifying it as a severely mixed segment, includes: The mean of the absolute differences between the electromyographic amplitude at each sampling time and the corresponding function value in the fitted function during the pain reflex period is obtained as the average residual of the pain reflex period. If the average residual of the pain reflex period is less than the preset residual threshold, the core muscle group signal segment is determined to be output; if the average residual of the pain reflex period is greater than or equal to the preset residual threshold, it is re-determined as a severely mixed segment.

9. The electromyographic signal assessment method for core muscle group recovery after spinal surgery in elderly patients according to claim 1, characterized in that, The method for obtaining the objective function includes: The candidate autonomous driving sub-segment is fitted with a single-peak Gaussian distribution model based on the pain-free baseline segment to obtain the initial autonomous driving signal; the candidate pain reflex sub-segment is fitted with the average waveform of all candidate pain reflex sub-segments as a template to obtain the initial pain reflex signal. The sum of squared residuals between the original electromyography (EMG) signal and the initial spontaneous drive signal within the candidate spontaneous drive sub-period is obtained as the first feature value; the sum of squared residuals between the original EMG signal and the initial pain reflex signal within the candidate pain reflex sub-period is obtained as the second feature value; and the sum of squared residuals between the original EMG signal and the sum of the initial spontaneous drive signal and the initial pain reflex signal within the overlapping mixed sub-period is obtained as the third feature value. Obtain the first product between the preset first weight and the first eigenvalue, obtain the second product between the preset second weight and the second eigenvalue, obtain the third product between the preset third weight and the third eigenvalue, and obtain the sum of the first product, the second product and the third product as the objective function.

10. The method for evaluating electromyographic signals of core muscle group recovery after spinal surgery in elderly patients according to claim 1, characterized in that, The method for obtaining the muscle group recovery rate includes: Acquire baseline electromyographic signals of patients during different core contraction movements before surgery; All effective contractile electromyography (EMG) segments corresponding to the core muscle group signal segments were merged, and the average power frequency, median frequency, root mean square value of EMG and peak power frequency of the merged signal segments and the baseline EMG signal were obtained respectively. For any feature among average power frequency, median frequency, root mean square value of electromyography (EMG) or peak power frequency, the ratio of the feature of the merged signal segment to the feature of the baseline EMG signal is obtained as the recovery rate of the corresponding feature.

Images