Hand function evaluation system and method based on near infrared and myoelectricity analysis and medium

Abstract

The invention provides a hand function evaluation system and method based on near infrared and myoelectricity analysis and a medium, functional near infrared spectrum equipment of the system is provided with a light source and a detector, light emitted by the light source can penetrate through skull and is absorbed by hemoglobin in cerebral cortex, and the detector is used for measuring near infrared data; the flexible hand rehabilitation robot is used for providing movement assistance; the myoelectricity acquisition equipment is provided with a multi-channel myoelectricity acquisition device and an electrode; the multi-channel myoelectricity acquisition device is used for acquiring a surface myoelectricity signal when a subject grasps through the electrode; the main controller is used for evaluating the hand dysfunction according to the near-infrared data and the surface electromyogram signals. The functional near infrared spectrum imaging technology is adopted for monitoring the nerve activation condition of the brain in the functional movement period, the multi-channel collecting device obtains sEMG grabbing data of a subject, the hand muscle activation condition of the hand rehabilitation robot is researched, and therefore the hand dysfunction is evaluated.

Description

Technical Field

[0001] The present invention relates to the technical field of hand function evaluation systems, and particularly to a hand function evaluation system and method, and a medium based on near-infrared and electromyography analysis. Background Art

[0002] In stroke patients, hand function impairment is a common complication, especially in the late stage of stroke. This impairment usually manifests as weakness of hand muscles, reduced coordination, and problems with movement control.

[0003] Currently, rehabilitation intervention has become the core strategy for treating hand function impairment. However, the rehabilitation medical industry faces problems such as excessive workload of therapists, shortage of medical resources, and difficulty in ensuring rehabilitation effects. In this context, the application of a single rehabilitation evaluation and treatment method has certain limitations for patients with hand function impairment after stroke. A hand rehabilitation robot is a robotic device designed specifically for patients with hand function impairment. By simulating the movement characteristics of the human hand and providing personalized rehabilitation training, it helps patients recover hand functions, including grip strength, flexibility, and fine motor skills, thereby improving the patients' quality of life and rehabilitation effects.

[0004] Another important area in stroke rehabilitation is the monitoring of functional changes in brain regions. Functional near-infrared spectroscopy (fNIRS) has become one of the preferred tools for non-invasive brain function imaging due to its high temporal resolution, excellent portability, and adaptability to the movement environment.

[0005] Therefore, there is an urgent need for a hand function impairment evaluation system and method based on near-infrared brain imaging and electromyography analysis. Summary of the Invention

[0006] To achieve the above objects and other advantages of the present invention, the first object of the present invention is to provide a hand function evaluation system based on near-infrared and electromyography analysis, including a functional near-infrared spectroscopy device, a flexible hand rehabilitation robot, an electromyography acquisition device, and a main controller; wherein,

[0007] The functional near-infrared spectroscopy device is configured with a light source and a detector. The light energy emitted by the light source can penetrate the skull and be absorbed by hemoglobin in the cerebral cortex, and the detector is used to measure near-infrared data;

[0008] The flexible hand rehabilitation robot is used to provide movement assistance;

[0009] The electromyography acquisition device is configured with a multi-channel electromyography acquisition device and electrodes. The multi-channel electromyography acquisition device collects surface electromyography signals of the subject during grasping through the electrodes;

[0010] The main controller is used to evaluate hand function impairment based on near-infrared data and surface electromyography signals.

[0011] Further, the functional near-infrared spectroscopy device is configured with light sources of two wavelengths.

[0012] Further, the flexible hand rehabilitation robot is configured as a wearable flexible cable-driven hand rehabilitation robot based on flexible cable drive and linear actuator drive.

[0013] Further, a task module is configured in the main controller, and the task module provides experiment prompts and configures the grasping activities and rest times of the experiment.

[0014] The second object of the present invention is to provide a hand function evaluation method based on near-infrared and electromyography analysis. Using the above system, it includes the following steps:

[0015] Obtain the near-infrared data and surface electromyography signals of the subject;

[0016] Perform invalid channel cropping, intensity-to-optical density conversion, channel motion artifact marking, wavelet analysis-based motion correction, and periodic filtering on the near-infrared data;

[0017] Convert the optical density data into oxygenated and deoxygenated hemoglobin concentrations;

[0018] Under specific experimental conditions, calculate the block average value of the data within a time period, and adjust the baseline of the average value to zero;

[0019] Use a correlation-based signal improvement method to correct motion artifacts and obtain clear and corrected neural activity signals;

[0020] After preprocessing the surface electromyography signals, extract multiple features for each channel respectively;

[0021] Calculate the importance of the features, and sort the importance scores of all features to determine the influence of the features on the classification effect of the electromyography classification prediction model.

[0022] Further, the steps of performing invalid channel cropping, intensity-to-optical density conversion, channel motion artifact marking, wavelet analysis-based motion correction, and periodic filtering on the near-infrared data include:

[0023] Eliminate the channel data with near-infrared signal intensity lower than the lowest threshold or higher than the highest threshold, and the standard deviation greater than the deviation threshold;

[0024] Convert the measured light intensity data into optical density data;

[0025] By comparing the signal amplitude at any time point with the average amplitude over a period of time and setting a threshold to mark abnormal points, the marked data segments are replaced by interpolation methods, and overly contaminated channels or data blocks need to be discarded;

[0026] Perform wavelet transform on the data and correct motion artifacts by analyzing the wavelet coefficient distribution, setting the coefficients outside the preset range to zero;

[0027] Apply a band - pass filter to process the signal to reduce the influence of physiological noise.

[0028] Furthermore, the steps of calculating the importance of features and sorting the importance scores of all features to determine the influence of features on the classification effect of the EMG classification prediction model include:

[0029] Use a random forest model to evaluate each feature in the dataset and determine the importance score of the feature according to its contribution to the model prediction effect;

[0030] Sort the importance scores of all features in descending order to determine the influence of features on the classification effect of the EMG classification prediction model;

[0031] By gradually increasing the number of selected features, calculate the accuracy under different numbers of features and explore the relationship between the number of features and the classification performance.

[0032] Furthermore, the steps of calculating the accuracy under different numbers of features include:

[0033] Evenly divide the dataset into five subsets and conduct five independent training and validation processes;

[0034] In each iteration, four subsets are combined into a training set, and the remaining one subset is used as a validation set;

[0035] The model uses the validation set for performance evaluation after each iteration;

[0036] The five evaluation results are finally averaged, and the standard deviation is calculated to obtain a comprehensive evaluation of the model performance;

[0037] Among them, the calculation formula for model performance evaluation is:

[0038]

[0039] Among them, Accuracy is accuracy, Precision is precision rate, Recall is recall rate, F1 Score is F1 - score, TP is true positive, FN is false negative, TN is true negative, and FP is false positive.

[0040] The third object of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the steps of the above method are implemented.

[0041] The fourth object of the present invention is to provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] The present invention provides a hand function evaluation system, method, and medium based on near-infrared and electromyography analysis. The system includes a functional near-infrared spectroscopy device, a flexible hand rehabilitation robot, an electromyography acquisition device, and a main controller. Among them, the functional near-infrared spectroscopy device is configured with a light source and a detector. The light energy emitted by the light source can penetrate the skull and be absorbed by hemoglobin in the cerebral cortex, and the detector is used to measure near-infrared data. The flexible hand rehabilitation robot is used to provide motion assistance. The electromyography acquisition device is configured with a multi-channel electromyography acquisition device and electrodes, and the multi-channel electromyography acquisition device acquires the surface electromyography signals of the subject during grasping through the electrodes. The main controller is used to evaluate hand function disorders based on near-infrared data and surface electromyography signals. The present invention uses functional near-infrared spectroscopy imaging technology to monitor the neural activation of the brain during functional movement, and the multi-channel acquisition device obtains the subject's grasping sEMG data, and studies the activation of hand muscles by the hand rehabilitation robot, so as to evaluate hand function disorders.

[0044] The above description is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly and implement it according to the content of the specification, the following takes the preferred embodiments of the present invention and combines the drawings to describe in detail as follows. The specific implementation manner of the present invention is given in detail by the following embodiments and their accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and the illustrative embodiments and descriptions of the present invention are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0046] Figure 1 Schematic diagram of a hand function evaluation system based on near-infrared and electromyography analysis Figure 1 ;

[0047] Figure 2 Schematic diagram of a hand function evaluation system based on near-infrared and electromyography analysis Figure 2 ;

[0048] Figure 3 Schematic diagram of experimental setup;

[0049] Figure 4 It is a schematic diagram of near-infrared data processing;

[0050] Figure 5 It is a schematic diagram of electromyogram data analysis;

[0051] Figure 6 It is a flowchart of a hand function evaluation method based on near-infrared and electromyogram analysis;

[0052] Figure 7 It is a schematic diagram of a computer device;

[0053] Figure 8 It is a schematic diagram of a computer-readable storage medium.

[0054] In the figure: 1. Functional near-infrared spectroscopy device; 11. Light source; 12. Detector; 2. Flexible hand rehabilitation robot; 3. Electromyogram acquisition device. Detailed implementation manners

[0055] Next, in combination with the accompanying drawings and specific implementation manners, the present invention will be further described. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. It should be noted that, on the premise of no conflict, the following-described embodiments or technical features can be combined arbitrarily to form new embodiments.

[0056] Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0057] In the present application, the attached drawing numbers are only used to distinguish each step in the solution, and are not used to limit the execution order of each step. The specific execution order shall be subject to the description in the specification.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments, and are not intended to limit the present invention.

[0059] Embodiment 1

[0060] A hand function evaluation system based on near-infrared and electromyogram analysis, as Figure 1 、 Figure 2 shown. The system includes a functional near-infrared spectroscopy device 1, a flexible hand rehabilitation robot 2, an electromyogram acquisition device 3, and a main controller; wherein,

[0061] The functional near-infrared spectroscopy device is configured with a light source 11 and a detector 12. The light energy emitted by the light source can penetrate the skull and be absorbed by hemoglobin in the cerebral cortex. The detector is used to measure near-infrared data;

[0062] The flexible hand rehabilitation robot is used to provide motion assistance;

[0063] The electromyogram acquisition device is configured with a multi-channel electromyogram acquisition device and electrodes. The multi-channel electromyogram acquisition device collects surface electromyogram signals of the subject during grasping through the electrodes; for example, an eight-channel electromyogram acquisition device and Ag / AgCl electrode patches are used.

[0064] The main controller is used to evaluate hand function disorders based on near-infrared data and surface electromyogram signals.

[0065] In order to evaluate the rehabilitation effect of hand function movement disorders in detail and evaluate the actual impact of the hand rehabilitation robot on brain function improvement. In this embodiment, functional near-infrared spectroscopy imaging technology is used to monitor the neural activation of the brain during functional movement, with a sampling rate of 500 Hz. At the same time, in this embodiment, an eight-channel acquisition device with a sampling rate of 500 Hz is used to obtain sEMG data of the subject during grasping, and the hand rehabilitation robot is used to evaluate the activation of hand muscles, so as to evaluate hand function disorders.

[0066] A functional near-infrared spectroscopy device is used to measure brain activation. Further, the functional near-infrared spectroscopy device is configured with light sources of two wavelengths, such as light sources of 730 nm and 850 nm, marked as red dots in Figure 2 These light energies can penetrate the skull and be absorbed by hemoglobin in the cerebral cortex. Using the Beer-Lambert law, the concentrations of oxyhemoglobin (HbO, 850 nm) and deoxyhemoglobin (HbR, 730 nm) are measured by the detector marked as blue dots, with a sampling rate of 11 Hz. The device contains a total of 42 channels.

[0067] In some embodiments, the flexible hand rehabilitation robot is configured as a wearable flexible cable-driven hand rehabilitation robot based on flexible cable drive and linear actuator drive.

[0068] In this embodiment, stroke patients with hand function disorders and healthy subjects are tested to collect near-infrared data and electromyogram data of the subjects, so as to construct an evaluation system.

[0069] Such as Figure 3As shown, stroke patients performed tasks of grasping with the affected hand, grasping with the affected hand wearing a flexible hand rehabilitation robot, and grasping simultaneously with the healthy hand and the affected hand wearing a flexible hand rehabilitation robot. Healthy subjects performed tasks of grasping with the affected hand, grasping with the affected hand wearing a flexible hand rehabilitation robot, and grasping simultaneously with the unaffected hand and the affected hand wearing a flexible hand rehabilitation robot on both hands. Under the condition of robot therapy, the affected hand was provided with movement by a robot glove driven by a flexible lasso. During all movements, the subjects wore fNIRS acquisition devices.

[0070] Furthermore, a task module is configured in the main controller, and the task module provides experimental prompts and configures the grasping activities and rest times of the experiment.

[0071] For example, as Figure 3 shown, each task was designed as a 35-second block, including 20 seconds of grasping activity and 15 seconds of rest, and each condition was repeated five times. During the rest phase, the subjects were required to relax completely. The settings and conditions of the experiment are shown in Figure 3 .

[0072] Experiment 1 (grasping with the affected hand): According to the auditory prompt, the subject used only the affected hand to perform the grasping action, and the glove did not provide movement assistance.

[0073] Experiment 2 (grasping with the affected hand wearing a flexible hand rehabilitation robot): The affected hand of the subject was assisted by the robot glove to perform the grasping action, and the subject was required to avoid actively moving the hand.

[0074] Experiment 3 (simultaneous grasping task of the healthy hand / unaffected hand and the affected hand wearing a flexible hand rehabilitation robot): The subject was required not to actively use the affected hand wearing the robot glove, while the healthy hand or unaffected hand actively performed the grasping according to the auditory feedback.

[0075] Near-infrared data processing was based on Homer3 software and analyzed in the.NIRS file format. The original signals are shown in Figure 4 (a). Considering that stroke patients mainly showed left hand dysfunction in this experiment, the data analysis of the left hand was particularly concerned. The specific processing steps include:

[0076] Invalid channel clipping (hmrR_Prunechannels function): Quality control was performed on the collected original signals, and channels with too weak or too strong signal intensity and too large standard deviation were removed to ensure the reliability of the data.

[0077] Intensity to optical density (OD) (hmrR_intensity2OD function): OD represents the amount of light attenuation and is the negative logarithm of the ratio of the emitted light intensity to the detected light intensity. Convert the measured light intensity data to optical density data for subsequent processing and analysis. The conversion results are as shown in Figure 4 (b).

[0078] Channel motion artifact marking (hmrR_MotionArtifactBychannel function): Identify motion artifacts caused by the relative displacement between the optode probe and the scalp due to the head movement of the subject. Compare the signal amplitude at any time point with the average amplitude over a period of time and set a threshold to mark abnormal points. The marked data segments can be replaced by interpolation methods, and overly contaminated channels or data blocks may need to be discarded.

[0079] Motion correction based on wavelet analysis (hmrR_MotioncorrectWavelet function): Perform wavelet transform on the data and correct motion artifacts by analyzing the wavelet coefficient distribution, setting coefficients outside a certain range to zero.

[0080] Periodic filtering (hmrR_BandpassFilt:Bandpass_Filter_opticalDensity function): Apply a bandpass filter to process fNIRS signals to reduce the influence of physiological noises such as heartbeat (about 1 Hz), respiration (about 0.2 - 0.3 Hz), and random noise, and improve the signal quality. According to the power spectrum diagram, there are noise interferences around frequencies of 0.8 Hz, 1.7 Hz, and 2.5 Hz.

[0081] OD conversion to concentration (hmrR_OD2Conc function): Convert optical density data to oxygenated and deoxygenated hemoglobin concentrations to provide a data basis for further physiological activity analysis. The conversion results are as shown in Figure 4 (c).

[0082] Block average calculation (hmrR_BlockAvg:Block_Average_on_Concentration_Data function): Under specific experimental conditions, calculate the block average of the data within a time period and adjust the baseline of the average value to zero. If the stimulus is too close to the start or end of the data recording, resulting in outliers not being included within the data range, the corresponding experimental data will be excluded.

[0083] Motion correction (hmrR_Motioncorrectcbsi function) based on the Correlation-Based Signal Improvement (CBSI): The motion artifacts are corrected using the correlation-based signal improvement method, which is based on the assumption that oxyhemoglobin (HbO2) and deoxyhemoglobin (HbR) fluctuate in opposite directions under normal conditions, and corrects by identifying the co-directional changes caused by head movement.

[0084] The finally processed data is shown in Figure 4 (b), providing clear and corrected neural activity signals.

[0085] The NIRS_KIT software is used for data analysis, and the input file format is.SNIRF generated by Homer3. After completing the data processing, the analysis is divided into individual level and group level.

[0086] To evaluate the research on the activation of hand muscles by the hand rehabilitation robot, the classification tasks of grasping between the affected hand and the healthy hand and the classification task of whether to wear the flexible hand rehabilitation robot are mainly analyzed. The positions of the data acquisition channels are as shown in Figure 5 (a), and the collected raw data is as shown in Figure 5 (b). First, the collected surface electromyography signals are analyzed to remove any crosstalk. The preprocessed data is as shown in Figure 5 (c), and features are extracted through MATLAB on the main controller.

[0087] Matlab is used to extract 36 features for each channel respectively, and 36×8 = 288 features can be obtained for one sample. The self-selected features include 36 features such as MAV, WL, ZC, SSC, RMS, signal energy, and Willison amplitude.

[0088] First, the conventional Random Forest (RF) model is used to calculate the importance of features. Specifically, RF is used to evaluate each feature in the dataset, and the importance score of the feature is determined according to its contribution to the model prediction effect. Subsequently, the importance scores of all features are sorted in descending order to determine which features have the greatest impact on the classification effect of the model. By gradually increasing the number of selected features, the accuracy rates under different numbers of features are calculated, and the relationship between the number of features and the classification performance is explored.

[0089] The calculation of classification accuracy adopts the five-fold cross-validation method. This method evenly divides the dataset into five subsets and conducts five independent training and validation processes. In each iteration, four subsets are combined into a training set, while the remaining one subset is used as a validation set. The model is evaluated for performance using the validation set after each iteration, usually measured by metrics such as accuracy, precision, recall, etc. The evaluation results of the five times are finally averaged, and the standard deviation is calculated to obtain a comprehensive evaluation of the model performance. Five-fold cross-validation effectively reduces the randomness impact brought by the data splitting method through repeated training and validation processes, thus enabling a more accurate estimation of the model's generalization ability and helping to identify overfitting or underfitting problems of the model.

[0090] To comprehensively evaluate the model performance, accuracy, precision, recall, and F1-score are used as evaluation metrics for the classification model. The performance evaluation of the model is based on the comparison between the predicted labels and the true labels, and then the numbers of true positives (TP), false negatives (FN), true negatives (TN), and false positives (FP) are determined. Specifically, true positive TP is the number of positive samples correctly predicted as positive labels, false negative FN is the number of positive samples wrongly predicted as negative labels, false positive FP is the number of negative samples wrongly predicted as positive labels, and true negative TN is the number of negative samples correctly predicted as negative labels. The calculation formulas for these four model evaluation metrics are shown in Equations (1) to (4):

[0091]

[0092] To improve the classification accuracy and select the most effective features for classification, feature selection is performed on the 288 features in the dataset. RF is used to calculate the feature importance and the ExtraTrees model is used to calculate the accuracy.

[0093] Based on the performance metrics of the activation analysis results of near-infrared electroencephalogram and the myoelectric classification prediction model in the above experimental process, a hand function disorder evaluation system based on near-infrared brain imaging and myoelectric analysis is constructed by establishing hand function disorder evaluation metrics through prior system data.

[0094] In this embodiment, stroke patients and healthy individuals are used as research objects, and the brain activation patterns under different sensory stimulation conditions are explored through fNIRS technology. Three control tasks are designed to construct a hand function evaluation system based on near-infrared and myoelectric analysis. The brain activation levels and functional connectivity patterns of the healthy group and the stroke group under different task conditions are compared to provide a theoretical basis and technical support for hand function rehabilitation after stroke.

[0095] Example 2

[0096] A hand function evaluation method based on near infrared and electromyography analysis adopts a hand function evaluation system based on near infrared and electromyography analysis provided in Example 1. For a detailed description of the hand function evaluation system based on near infrared and electromyography analysis provided in Example 1, reference can be made to the corresponding description of Example 1, which will not be repeated here. Figure 6 As shown, the method comprises the following steps:

[0097] S1. Obtain the near infrared data and surface electromyography signal of the subject; the original near infrared data is as follows: Figure 4 (a) shown.

[0098] In some embodiments, the near infrared data processing is based on Homer3 software and is analyzed in .NIRS file format. Considering that stroke patients in this experiment mainly show left hand dysfunction, special attention is paid to the data analysis of the left hand.

[0099] S2, performing invalid channel clipping, intensity conversion to light density, channel motion artifact marking, motion correction based on wavelet analysis, and periodic filtering on the near-infrared data;

[0100] In some embodiments, the steps of performing invalid channel clipping, intensity conversion to light density, channel motion artifact marking, wavelet analysis-based motion correction, and periodic filtering on the near-infrared data include:

[0101] Invalid channel pruning (hmrR_Prunechannels function): Perform quality control on the collected raw signals and remove channels with too weak or too strong signal strength and too large standard deviation to ensure data reliability. Specifically, remove channel data with near-infrared signal strength lower than the minimum threshold or higher than the maximum threshold, and standard deviation greater than the deviation threshold.

[0102] Intensity to optical density (OD) (hmrR_intensity2OD function): OD represents the light attenuation, which is the negative logarithm of the ratio of the emitted light intensity to the detected light intensity. The measured light intensity data is converted into optical density data for subsequent processing and analysis. The conversion result is as follows: Figure 4 (b) as shown.

[0103] Channel motion artifact marking (hmrR_MotionArtifactBychannel function): Identify motion artifacts caused by the relative displacement of the optode probe and the scalp due to the subject's head movement. Abnormal points are marked by comparing the signal amplitude at any time point with the average amplitude over a period of time and setting a threshold. The marked data segments can be replaced by interpolation methods, and channels or data blocks with excessive pollution may need to be discarded.

[0104] Motion correction based on wavelet analysis (hmrR_MotioncorrectWavelet function): Perform wavelet transform on the data, correct motion artifacts by analyzing the wavelet coefficient distribution, and set coefficients outside a certain range to zero.

[0105] Periodic filtering (hmrR_BandpassFilt:Bandpass_Filter_opticalDensity function): Apply a bandpass filter to process the fNIRS signal to reduce the influence of physiological noises such as heartbeat (about 1 Hz), respiration (about 0.2 - 0.3 Hz), and random noise, etc., and improve the signal quality. According to the power spectrum, there are noise interferences around frequencies of 0.8 Hz, 1.7 Hz, and 2.5 Hz.

[0106] S3. OD conversion to concentration (hmrR_OD2Conc function): Convert optical density data into oxygenated and deoxygenated hemoglobin concentrations, providing a data basis for further physiological activity analysis. The conversion results are as shown in Figure 4 (c).

[0107] S4. Block average calculation (hmrR_BlockAvg:Block_Average_on_Concentration_Data function): Under specific experimental conditions, calculate the block average of the data within a time period, and adjust the baseline of the average to zero. If the stimulus is too close to the start or end of the data recording, resulting in outliers that cannot be included within the data range, the corresponding experimental data will be excluded.

[0108] S5. Motion correction based on the Correlation - Based Signal Improvement (CBSI) method (hmrR_Motioncorrectcbsi function): Use the correlation - based signal improvement method to correct motion artifacts. This method is based on the assumption that oxygenated hemoglobin (HbO2) and deoxygenated hemoglobin (HbR) fluctuate in opposite directions under normal conditions, and corrects by identifying the co - directional changes caused by head movement.

[0109] The finally processed data is shown in Figure 4 (d), providing clear and corrected neural activity signals.

[0110] In this embodiment, the NIRS_KIT software is used for data analysis, and the input file format is.SNIRF generated by Homer3. After completing data processing, the analysis is divided into individual level and group level.

[0111] S6. After preprocessing the surface electromyogram signal, extract multiple features for each channel respectively;

[0112] In this embodiment, in order to evaluate the research on hand muscle activation by the hand rehabilitation robot, the classification tasks of grasping between the affected hand and the healthy hand and the classification task of whether to wear the flexible hand rehabilitation robot are mainly analyzed. The positions of the data acquisition channels are as shown in Figure 5 (a), and the original data collected is as shown in Figure 5 (b). First, the collected surface electromyography signals are analyzed to remove any crosstalk. The preprocessed data is as shown in Figure 5 (c), and features are extracted through MATLAB on the main controller.

[0113] Using Matlab, 36 features are extracted for each channel respectively, and 36×8 = 288 features can be obtained for one sample. The self-selected features include 36 features such as MAV, WL, ZC, SSC, RMS, signal energy, and Willison amplitude.

[0114] S7. Calculate the importance of the features, and sort the importance scores of all the features to determine the influence of the features on the classification effect of the electromyography classification prediction model.

[0115] In some embodiments, the step of calculating the importance of the features and sorting the importance scores of all the features to determine the influence of the features on the classification effect of the electromyography classification prediction model includes:

[0116] First, a conventional Random Forest (RF) model is used to calculate the importance of the features. Specifically, RF is used to evaluate each feature in the dataset, and the importance score of the feature is determined according to its contribution degree to the model prediction effect.

[0117] Then, the importance scores of all the features are sorted in descending order to determine which features have the greatest influence on the classification effect of the model.

[0118] By gradually increasing the number of selected features, the accuracy rates under different numbers of features are calculated, and the relationship between the number of features and the classification performance is explored.

[0119] The calculation of classification accuracy adopts the five-fold cross-validation method. This method evenly divides the dataset into five subsets and conducts five independent training and validation processes. In each iteration, four subsets are combined into a training set, while the remaining one subset is used as a validation set. The model is evaluated for performance using the validation set after each iteration, usually measured by metrics such as accuracy, precision, recall, etc. The evaluation results of the five times are finally averaged, and the standard deviation is calculated to obtain a comprehensive evaluation of the model performance. Five-fold cross-validation effectively reduces the randomness impact brought by the data splitting method through repeated training and validation processes, thereby enabling a more accurate estimation of the model's generalization ability and helping to identify overfitting or underfitting problems of the model.

[0120] To comprehensively evaluate the model performance, accuracy, precision, recall, and F1-score are used as evaluation metrics for the classification model. The performance evaluation of the model is based on the comparison between the predicted labels and the true labels, and then the numbers of true positives (TP), false negatives (FN), true negatives (TN), and false positives (FP) are determined. Specifically, true positive TP is the number of positive samples correctly predicted as positive labels, false negative FN is the number of positive samples wrongly predicted as negative labels, false positive FP is the number of negative samples wrongly predicted as positive labels, and true negative TN is the number of negative samples correctly predicted as negative labels. The calculation formulas for these four model evaluation metrics are shown in Equations (1) to (4):

[0121]

[0122] To improve the classification accuracy and screen out the most effective features for classification, feature selection is performed on the 288 features in the dataset. The RF is used to calculate the feature importance and the ExtraTrees model is used to calculate the accuracy.

[0123] Based on the performance metrics of the activation analysis results of near-infrared electroencephalogram and the myoelectric classification prediction model in the above experimental process, an evaluation index for hand dysfunction is established through prior system data, thereby constructing a hand dysfunction evaluation system based on near-infrared brain imaging and myoelectric analysis.

[0124] Example 3

[0125] A computer device 800, as Figure 7 shown, includes a memory 810, a processor 820, and a computer program 830 stored on the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a hand function evaluation method based on near-infrared and myoelectric analysis. For a detailed description of the method, reference can be made to the corresponding description in the above method embodiments, which will not be elaborated here.

[0126] Example 4

[0127] A computer-readable storage medium, as Figure 8 shown, stores a computer program thereon, and when the computer program is executed by a processor, it implements the steps of a hand function evaluation method based on near-infrared and electromyography analysis. For a detailed description of the method, reference can be made to the corresponding description in the above method embodiments, which will not be elaborated here.

[0128] The number of devices and the scale of processing described here are used to simplify the description of the present invention. Applications, modifications, and variations of the present invention will be obvious to those skilled in the art.

[0129] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to specific details and the examples shown and described here.

[0130] The device, computer device, non-volatile computer storage medium, and method provided in the embodiments of this specification are corresponding. Therefore, the device, computer device, and non-volatile computer storage medium also have beneficial technical effects similar to the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding device, computer device, and non-volatile computer storage medium will not be elaborated here.

[0131] Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program code, the method steps can be logically programmed to enable the controller to be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. to achieve the same function. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be regarded as the structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software units for implementing the method and the structures within the hardware component.

[0132] The systems, devices, or units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. For the convenience of description, when describing the above devices, they are divided into various units according to functions and described separately. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0133] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, the embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0134] This specification is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of this specification. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more of the flows Figure 1 or a plurality of flows and / or blocks

[0135] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one or more of the flows Figure 1 or a plurality of flows and / or blocks

[0136] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one or more of the flows Figure 1 or a plurality of flows and / or blocks

[0137] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising said element.

[0138] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program units. Generally, program units include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The specification may also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program units may be located in local and remote computer storage media including storage devices.

[0139] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and reference can be made to the corresponding parts of the method embodiments for the relevant content.

[0140] The above is only for the embodiments of this specification and is not intended to limit one or more embodiments of this specification. For those skilled in the art, various changes and modifications can be made to one or more embodiments of this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A hand function assessment system based on near infrared and electromyography analysis, characterized in that: Including functional near-infrared spectroscopy equipment, flexible hand rehabilitation robot, electromyography acquisition equipment, and main controller; among them, The functional near-infrared spectroscopy device is configured with a light source and a detector. The light emitted by the light source can penetrate the skull and be absorbed by the hemoglobin in the cerebral cortex. The detector is used to measure near-infrared data. The flexible hand rehabilitation robot is used to provide movement assistance; The electromyographic acquisition device is equipped with a multi-channel electromyographic acquisition device and electrodes, and the multi-channel electromyographic acquisition device acquires surface electromyographic signals of the subject when grasping through the electrodes; The main controller is used to evaluate hand dysfunction based on near-infrared data and surface electromyography signals.

2. A hand function assessment system based on near infrared and electromyography analysis as claimed in claim 1, characterized in that: The functional near-infrared spectroscopy device is equipped with a light source with two wavelengths.

3. A hand function assessment system based on near infrared and electromyography analysis as claimed in claim 1, characterized in that: The flexible hand rehabilitation robot is configured as a wearable flexible rope-driven hand rehabilitation robot based on flexible rope transmission and linear actuator drive.

4. The hand function evaluation system based on near infrared and electromyography analysis as claimed in claim 1, characterized in that: The main controller is provided with a task module, which provides experimental prompts and configures the grasping activities and rest time of the experiment.

5. A hand function assessment method based on near infrared and electromyography analysis, using the system according to any one of claims 1 to 4, characterized in that: The following steps are involved: Acquire the near infrared data and surface electromyography signals of the subjects; The near-infrared data is subjected to invalid channel clipping, intensity conversion to light density, channel motion artifact marking, wavelet analysis-based motion correction, and periodic filtering processing; Convert optical density data to oxygenated and deoxygenated hemoglobin concentrations; Under a particular experimental condition, data within a time period were block averaged, and the mean values ​​were baseline adjusted to zero; A correlation-based signal improvement method is used to correct motion artifacts and obtain clear and corrected neural activity signals; After preprocessing the surface electromyography signal, a plurality of features are extracted from each channel respectively; The importance of features was calculated and the importance scores of all features were ranked to determine the impact of features on the classification effect of the EMG classification prediction model.

6. A hand function assessment method based on near infrared and electromyography analysis as claimed in claim 5, characterized in that: The steps of performing invalid channel clipping, intensity conversion to light density, channel motion artifact marking, motion correction based on wavelet analysis, and periodic filtering on the near-infrared data include: Eliminate channel data whose near-infrared signal intensity is lower than the minimum threshold or higher than the maximum threshold, and whose standard deviation is greater than the deviation threshold; Convert the measured light intensity data into optical density data; By comparing the signal amplitude at any time point with the average amplitude over a period of time and setting a threshold to mark abnormal points, the marked data segments are replaced by interpolation methods, and channels or data blocks with excessive pollution need to be discarded; Perform wavelet transform on the data and correct motion artifacts by analyzing the distribution of wavelet coefficients, setting coefficients that exceed a preset range to zero; The signal was processed using a bandpass filter to reduce the influence of physiological noise.

7. The hand function assessment method based on near infrared and electromyography analysis as claimed in claim 5, characterized in that: The steps of calculating the importance of features and ranking the importance scores of all features to determine the influence of features on the classification effect of the electromyography classification prediction model include: Use the random forest model to evaluate each feature in the dataset and determine the importance score of the feature based on its contribution to the model's prediction effect; The importance scores of all features are sorted from high to low to determine the impact of the features on the classification effect of the electromyography classification prediction model; By gradually increasing the number of selected features, the accuracy under different numbers of features is calculated, and the relationship between the number of features and classification performance is explored.

8. The hand function assessment method based on near infrared and electromyography analysis as claimed in claim 7, characterized in that: The step of calculating the accuracy under different feature quantities includes: The dataset was evenly divided into five subsets, and five independent training and validation processes were performed; In each iteration, four subsets were combined into the training set, and the remaining one subset was used as the validation set; The model uses the validation set for performance evaluation after each iteration; The five evaluation results were finally averaged and their standard deviations were calculated to obtain a comprehensive evaluation of the model performance; Among them, the calculation formula for model performance evaluation is: Among them, Accuracy is accuracy, Precision is precision, Recall is recall, F1 Score is F1 score, TP is true positive, FN is false negative, TN is true negative, and FP is false positive.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 5 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 5 to 8 are implemented.

Images