Standing posture balance muscle control evaluation device and control evaluation method based on partial derivative coherence

Abstract

The invention provides a standing posture balance muscle control evaluation device based on partial derivative coherence. The standing posture balance muscle control evaluation device is characterized by comprising a rotating platform, a VR helmet, an EEG collecting device, an EMG collecting device, a data processing module and a user interface. The invention further provides a standing posture balance muscle control evaluation method based on partial derivative coherence. According to the method, the device is combined, multi-sensory stimulation with consistent and inconsistent sensory is provided, a standing posture balancing task is carried out under different conditions, and data are synchronously collected. Through signal preprocessing, source reconstruction and partial derivative coherence analysis, a directional function connection network between cortex and muscle is constructed. Cortex-muscle coupling is decomposed into the somatosensory information transmission capability of muscles and the control capability of cortex on the muscles, and bidirectional understanding of a neuromuscular interaction mechanism is provided. Not only are scientificity and accuracy of balance control function evaluation improved, but also powerful technical support is provided for clinical diagnosis and rehabilitation of balance disorders.

Description

Technical Field

[0001] The present invention relates to the fields of neuroscience and bioengineering, and in particular to a standing balance muscle control evaluation device and a control evaluation method based on partial derivative coherence. The method combines a rotating platform with virtual reality (VR) technology, comprehensively collects electroencephalogram (EEG) and electromyogram (EMG), and realizes a detailed evaluation of the two-way information flow between the cortex and the muscle, thereby providing a more comprehensive and accurate means of evaluating the balance control function. Background Art

[0002] Balance control is an important function of the human body to maintain posture stability and adapt to changes in the external environment, involving complex neuromuscular coordination mechanisms. Traditional balance assessment methods mainly rely on central pressure (COP) analysis or single analysis of electromyography (EMG) signals. These methods have certain application value in assessing the overall balance state, but have significant limitations in revealing the deep neuromuscular interaction mechanism.

[0003] In recent years, with the development of neuroscience and biological signal processing technology, researchers have begun to try to combine multimodal data, such as electroencephalogram (EEG) and electromyogram (EMG), in order to achieve a more comprehensive evaluation of balance control function. However, existing technologies are still imperfect in analyzing the information flow between the cortex and muscles, especially in the field of directional functional connectivity and coupling decomposition, which is still in its infancy. Specifically, existing methods have difficulty distinguishing between the somatosensory information transmitted by muscles and the control ability of the cortex over muscles, resulting in a lack of in-depth understanding of the balance control mechanism.

[0004] In addition, although virtual reality (VR) technology and rotating platforms have been used to provide multisensory stimulation to simulate different balance task environments, the combination of these technologies has been involved in patents, and has not fully utilized directional functional connectivity analysis to decompose cortical-muscular coupling. Therefore, a new evaluation method is urgently needed that can subdivide cortical-muscular coupling into the muscle's somatosensory information transmission ability and the cortex's control ability over the muscle through directional functional connectivity, thereby improving the accuracy and depth of the balance-related muscle control function evaluation.

[0005] To solve the above problems, the present invention proposes an evaluation method based on directional functional connectivity decomposition. By decomposing the cortex-muscle coupling into the muscle's somatosensory information transmission ability and the cortex's control ability over the muscle, combined with a comprehensive analysis of multimodal data, a more detailed and comprehensive evaluation of the balance-related muscle control function can be achieved, which significantly improves the scientificity and practicality of the balance control function evaluation. Summary of the invention

[0006] The present invention aims to provide a standing balance muscle control evaluation device and control evaluation method based on partial derivative coherence. The method decomposes the cortex-muscle coupling into the somatosensory information transmission ability of the muscle and the control ability of the cortex over the muscle, and combines the comprehensive analysis of multimodal data (such as electroencephalogram (EEG) and electromyogram (EMG) to achieve the evaluation of balance-related muscle control function.

[0007] To achieve the above-mentioned purpose, the technical solution of the present invention is: a standing balance muscle control evaluation device based on partial derivative coherence, including a rotating platform, a VR helmet, an EEG acquisition device, an EMG acquisition device, a data processing module and a user interface; the rotating platform can achieve precise constant speed rotation, the VR helmet has a built-in high-precision inertial measurement unit (IMU), which can adjust the angle of the visual scene in real time, the data processing module includes signal preprocessing, source reconstruction, PDC analysis and coupling decomposition functions, and the user interface provides real-time data monitoring and visualization display functions, which is convenient for operators to observe the experimental process and evaluation results.

[0008] In the above-mentioned standing balance muscle control assessment device based on partial derivative coherence, the data processing module is developed based on MATLAB; the EEG acquisition device adopts the ANT Neuro EEG cap, which has 32 channels and a sampling rate of 1000 Hz, and the electrode arrangement follows the international 10-20 standard; the EMG acquisition device adopts the Delsys Trigno wireless EMG system, which has six channels and is arranged in the calf muscles according to the SENIAM guidelines.

[0009] The present invention also includes a standing posture balance muscle control evaluation method based on partial derivative coherence, including the above-mentioned standing posture balance muscle control evaluation device, comprising the following steps:

[0010] S1. In a standing balance task, a rotating platform and a virtual reality (VR) helmet are used to provide visual and motion stimulation to participants, achieving two conditions: sensory congruent and sensory incongruent. The rotating platform rotates at a constant speed, and the VR helmet adjusts the visual scene in real time according to the platform movement to simulate different balance task environments. Under the sensory congruent condition, the visual information is consistent with the actual movement direction; while under the sensory incongruent condition, the visual information is opposite to the actual movement direction, resulting in sensory conflict.

[0011] S2. Collect the electroencephalogram (EEG) and electromyography (EMG) data of the participants' sensorimotor cortex. The collected data undergoes rigorous preprocessing, including filtering, denoising, and synchronization processing, to ensure the high quality and accuracy of the signals. The specific preprocessing steps include applying the Artifact Subspace Reconstruction (ASR) method to remove large-amplitude artifacts, and further removing interference signals such as blinking, electromyography, and electrocardiogram through independent component analysis (ICA);

[0012] S3, data analysis stage, the partial directed coherence (PDC) method is used to calculate the directional information flow between the cortical area and the muscle in the frequency domain. PDC is a frequency domain causal analysis method based on Granger causality, which can effectively distinguish the directional influence between signals. The calculation formula of PDC is as follows:

[0013]

[0014] Among them, PDC ij (ω) represents the partial derivative coherence from node j to node i at frequency ω, A ij (e jω ) is the transfer function of the system, N is the total number of nodes in the system, and through PDC analysis, the direction and strength of information flow between different cortical areas and muscles can be quantified;

[0015] S4. After obtaining the PDC results, the cortex-muscle coupling is further decomposed into two parts: the muscle's somatosensory information transmission ability and the cortex's control ability over the muscle. The specific decomposition method is as follows:

[0016]

[0017] Among them, Coupling Sensory Represents the overall ability of muscles to transmit somatosensory information to the cortex, Coupling Control Represents the overall ability of the cortex to actively control muscles.

[0018] In the above-mentioned standing balance muscle control assessment method based on partial derivative coherence, the partial derivative coherence (PDC) method calculates the directional information flow between the cortical area and the muscle in the frequency range of 4-48 Hz, and the calculation steps of PDC include:

[0019] S1. Autoregressive (AR) model construction: An autoregressive model is established for each cortical area and muscle signal to describe

[0020] Describing its dynamic characteristics, assuming that the AR model order of the system is p, it can be expressed as:

[0021]

[0022] Among them, X(t) is the signal vector at the current moment, A(k) is the AR coefficient matrix, and ∈(t) is the error term;

[0023] S2. Transfer function calculation: Calculate the transfer function A(e) of the system through the AR model parameter A(k). jω ):

[0024]

[0025] Where I is the identity matrix and ω is the frequency;

[0026] S3, PDC value calculation: According to the transfer function A (e jω ), calculate the PDC value between each node:

[0027]

[0028] Among them, PDC ij (ω) represents the partial derivative coherence from node j to node i at frequency ω, A ij (e jω ) is the element in the transfer function matrix, N is the total number of nodes in the system;

[0029] Through the above steps, the direction and strength of information flow between different cortical areas and muscles are quantified. Subsequently, the PDC analysis results are decomposed into two independent parts: the somatosensory information transmission ability of muscles and the control ability of the cortex over muscles. The specific decomposition formula is as follows:

[0030]

[0031] Among them, Coupling Sensory Represents the overall ability of muscles to transmit somatosensory information to the cortex, Coupling Control Represents the overall ability of the cortex to actively control muscles. Through the above decomposition, the ability of muscles to transmit somatosensory information and the ability of the cortex to control muscles can be evaluated separately, thereby achieving a comprehensive evaluation of balance-related muscle control functions.

[0032] The present invention has the following beneficial effects:

[0033] Compared with the prior art, the present invention has the following significant technical advantages. First, through directional functional connectivity analysis, the somatosensory information transmission ability of muscles and the control ability of the cortex over muscles can be evaluated respectively, realizing a two-way understanding of cortical-muscle coupling. This two-way coupling decomposition method not only reveals the active control mechanism in neuromuscular interaction, but also reflects the influence of passive somatosensory information transmission on balance control. Secondly, the comprehensive use of EEG, EMG and other biological signals improves the comprehensiveness and accuracy of balance control function evaluation, overcoming the limitations of traditional single data source analysis. In addition, advanced signal processing techniques (such as ASR and ICA) are used to effectively remove noise and artifacts, ensuring the reliability and effectiveness of data analysis. Finally, the method and system are not only suitable for the diagnosis of clinical balance disorders, but can also be widely used in sports training and rehabilitation monitoring, as well as basic research on cognition and motor control, with broad application prospects and significant practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 are the method steps of the present invention;

[0035] Figure 2 It is a device diagram of the present invention;

[0036] Figure 3 This is the experimental result of the present invention. DETAILED DESCRIPTION

[0037] like Figure 2As shown, the present invention provides a standing balance muscle control evaluation device based on partial derivative coherence, including a rotating platform 1, a VR helmet 2, an EEG acquisition device 3, an EMG acquisition device 4, a data processing module and a user interface. The rotating platform uses the All Controller (Nanjing, China) model, which has high-precision rotation control capabilities and can achieve accurate constant speed rotation. The VR helmet uses VIVE PRO 2 (HTC Corporation), which has a built-in high-precision inertial measurement unit (IMU) and can adjust the angle of the visual scene in real time. The EEG acquisition device uses the ANT Neuro EEG cap (EE-225, ANT Neuro, the Netherlands), 32 channels, a sampling rate of 1000Hz, and the electrode arrangement follows the international 10-20 standard. The EMG acquisition device uses the Delsys Trigno wireless EMG system (Trigno, Delsys, MA, the United States), six channels, and is arranged in the calf muscles (such as the gastrocnemius, soleus and peroneus longus) according to the SENIAM guidelines. The data processing module is developed based on MATLAB (2021b, MathWorks, MA, USA), which includes signal preprocessing, source reconstruction, PDC analysis and coupling decomposition functions. The user interface provides real-time data monitoring and visualization functions, which facilitates operators to observe the experimental process and evaluate the results.

[0038] The present invention also provides a standing posture balance muscle control evaluation method based on partial derivative coherence. The standing posture balance muscle control evaluation device comprises the following steps:

[0039] S1. In a standing balance task, a rotating platform and a virtual reality (VR) helmet are used to provide visual and motion stimulation to participants, achieving two conditions: sensory congruent and sensory incongruent. The rotating platform rotates at a constant speed, and the VR helmet adjusts the visual scene in real time according to the platform movement to simulate different balance task environments. Under the sensory congruent condition, the visual information is consistent with the actual movement direction; while under the sensory incongruent condition, the visual information is opposite to the actual movement direction, resulting in sensory conflict.

[0040] S2. Collect the electroencephalogram (EEG) and electromyography (EMG) data of the participants' sensorimotor cortex. The collected data undergoes rigorous preprocessing, including filtering, denoising, and synchronization processing, to ensure the high quality and accuracy of the signals. The specific preprocessing steps include applying the Artifact Subspace Reconstruction (ASR) method to remove large-amplitude artifacts, and further removing interference signals such as blinking, electromyography, and electrocardiogram through independent component analysis (ICA);

[0041] S3, data analysis stage, the partial directed coherence (PDC) method is used to calculate the directional information flow between the cortical area and the muscle in the frequency domain. PDC is a frequency domain causal analysis method based on Granger causality, which can effectively distinguish the directional influence between signals. The calculation formula of PDC is as follows:

[0042]

[0043] Among them, PDC ij (ω) represents the partial derivative coherence from node j to node i at frequency ω, A ij (e jω ) is the transfer function of the system, N is the total number of nodes in the system, and through PDC analysis, the direction and strength of information flow between different cortical areas and muscles can be quantified;

[0044] S4. After obtaining the PDC results, the cortex-muscle coupling is further decomposed into two parts: the muscle's somatosensory information transmission ability and the cortex's control ability over the muscle. The specific decomposition method is as follows:

[0045]

[0046] Among them, Couplingsens. ry Represents the overall ability of muscles to transmit somatosensory information to the cortex, CouplingControl l Represents the overall ability of the cortex to actively control muscles.

[0047] Furthermore, the partial derivative coherence (PDC) method calculates the directional information flow between the cortical area and the muscle in the frequency range of 4-48 Hz, and the calculation steps of PDC include:

[0048] S1. Autoregressive (AR) model construction: An autoregressive model is established for each cortical area and muscle signal to describe

[0049] Describing its dynamic characteristics, assuming that the AR model order of the system is p, it can be expressed as:

[0050]

[0051] Among them, X(t) is the signal vector at the current moment, A(k) is the AR coefficient matrix, and ∈(t) is the error term;

[0052] S2. Transfer function calculation: Calculate the transfer function A(e) of the system through the AR model parameter A(k). jω ):

[0053]

[0054] Where I is the identity matrix and ω is the frequency;

[0055] S3, PDC value calculation: According to the transfer function A (e jω ), calculate the PDC value between each node:

[0056]

[0057] Among them, PDC ij (ω) represents the partial derivative coherence from node j to node i at frequency ω, A ij (e jω ) is the element in the transfer function matrix, N is the total number of nodes in the system;

[0058] Through the above steps, the direction and strength of information flow between different cortical areas and muscles are quantified. Subsequently, the PDC analysis results are decomposed into two independent parts: the somatosensory information transmission ability of muscles and the control ability of the cortex over muscles. The specific decomposition formula is as follows:

[0059]

[0060] Among them, Coupling Sensory Represents the overall ability of muscles to transmit somatosensory information to the cortex, Coupling Control Represents the overall ability of the cortex to actively control muscles.

[0061] The present invention is further described below by experiments:

[0062] Experimental process and data collection

[0063] This experiment invited 50 college student volunteers to participate, recruited through online advertisements. Participants were required to have no neurological, skeletal or muscular diseases, normal or corrected vision, and agree to the experimental protocol. The experimental design included two sensory consistency conditions: sensory consistency (congruent) and sensory inconsistency (incongruent), which were performed in random order. Participants stood on a rotating platform, held their hands in front of their chests, wore VR helmets, and kept their eyes looking forward (blinking was allowed).

[0064] The platform rotated clockwise at 30 degrees per second. Under the sensory consistency condition, the VR helmet adjusted the visual scene according to the platform movement, so that the participants visually perceived a static environment, while the visual scene rotated counterclockwise at 30 degrees per second relative to themselves. Under the sensory inconsistency condition, the visual scene in the VR helmet rotated clockwise at 60 degrees per second, resulting in the visual information being opposite to the actual direction of movement, and the participants perceived themselves rotating counterclockwise at 30 degrees per second based on the visual information.

[0065] In each condition, participants were required to stand and balance for 36 seconds, with 30 seconds of rest and 5 minutes of rest before and after. The platform acceleration phase was completed within 2 seconds. To reduce the impact of starting and stopping on data stability, only the data from 8 to 28 seconds after the platform started were analyzed.

[0066] Data preprocessing and coupling decomposition

[0067] The collected EEG and EMG were preprocessed using a custom MATLAB script. First, a 1-48 Hz bandpass filter was applied, and the 50 Hz power line noise was removed using the EEGLAB Cleanline plugin. The EMG data was further processed by full-wave rectification. Subsequently, the Artifact Subspace Reconstruction (ASR) method was applied to remove large-amplitude artifacts caused by body movement, etc., and the threshold was set to 20 standard deviations to ensure that at least 65% of the data segments were retained. To keep the data synchronized, the same artifact segments in the EMG data were deleted synchronously. Next, independent component analysis (ICA) was applied to the EEG data, and non-brain source signals such as eye blinks, electromyography, and electrocardiogram were separated and removed using the ICLabel plugin.

[0068] The preprocessed EEG data are input into the source reconstruction unit, and the source reconstruction is performed based on the international standard brain model (such as ICBM-152) using the Beamformers algorithm to extract the time series data of the cortical areas of interest (such as SMC, PPC, etc.).

[0069] After the source reconstruction is completed, the partial derivative coherence (PDC) method is used to calculate the directional information flow between the cortical area and the muscle in the frequency range of 4-48Hz. The calculation steps of PDC include:

[0070] 1. Autoregressive (AR) model construction: An autoregressive model is established for each cortical area and muscle signal to describe its dynamic characteristics. Assuming that the order of the AR model of the system is p, it can be expressed as:

[0071]

[0072] Among them, X(t) is the signal vector at the current moment, A(k) is the AR coefficient matrix, and ∈(t) is the error term.

[0073] 2. Transfer function calculation: Calculate the transfer function A(e) of the system through the AR model parameter A(k). jω ):

[0074]

[0075] Where I is the identity matrix and ω is the frequency.

[0076] 3. PDC value calculation: According to the transfer function A(ejω ), calculate the PDC value between each node:

[0077]

[0078] Among them, PDC ij (ω) represents the partial derivative coherence from node j to node i at frequency ω, A ij (e jω ) is the element in the transfer function matrix, and N is the total number of nodes in the system.

[0079] Through the above steps, the direction and strength of information flow between different cortical areas and muscles are quantified. Subsequently, the PDC analysis results are decomposed into two independent parts: the somatosensory information transmission ability of muscles and the control ability of the cortex over muscles. The specific decomposition formula is as follows:

[0080]

[0081] Among them, Coupling Sensory Represents the overall ability of muscles to transmit somatosensory information to the cortex, Coupling Control Represents the overall ability of the cortex to actively control muscles. Through the above decomposition, the ability of muscles to transmit somatosensory information and the ability of the cortex to control muscles can be evaluated separately, thereby achieving a comprehensive evaluation of balance-related muscle control functions.

[0082] Implementation Results

[0083] PDC analysis showed that under sensory inconsistency conditions, the cortex's ability to control muscles was significantly reduced (mean difference = 0.12, 95% confidence interval = 0.010-0.16, p < 0.001), while the muscle's ability to transmit information to the cortex was significantly enhanced (mean difference = 0.0040, 95% confidence interval = 0.0025-0.0060, p < 0.001). Specific connection pair analysis showed that, except for a few connection pairs, most cortical-muscle connections showed significant differences under sensory inconsistency conditions, verifying the effectiveness and significance of the method of the present invention in evaluating balance-related muscle control functions. Figure 3 shown.

Claims

1. A standing balance muscle control evaluation device based on partial derivative coherence, characterized in that: The method comprises a rotating platform (1), a VR helmet (2), an EEG acquisition device (3), an EMG acquisition device (4), a data processing module and a user interface; the rotating platform (1) can realize precise constant speed rotation, the VR helmet (2) has a built-in high-precision inertial measurement unit (IMU) and can adjust the angle of the visual scene in real time, the data processing module includes signal preprocessing, source reconstruction, PDC analysis and coupling decomposition functions, and the user interface provides real-time data monitoring and visual display functions, so as to facilitate operators to observe the experimental process and evaluate the results.

2. The standing balance muscle control evaluation device based on partial derivative coherence according to claim 1, characterized in that: The EEG acquisition device (3) uses an ANT Neuro EEG cap with 32 channels and a sampling rate of 1000 Hz, and the electrode arrangement complies with the international 10-20 standard; the EMG acquisition device (4) uses a Delsys Trigno wireless EMG system with six channels, which is arranged on the calf muscles according to the SENIAM guidelines.

3. A standing balance muscle control evaluation method based on partial derivative coherence, characterized by: The standing balance muscle control evaluation device according to claims 1 to 2 comprises the following steps: S1. In a standing balance task, a rotating platform and a virtual reality (VR) helmet are used to provide visual and motion stimulation to participants, achieving two conditions: sensory congruent and sensory incongruent. The rotating platform rotates at a constant speed, and the VR helmet adjusts the visual scene in real time according to the platform movement to simulate different balance task environments. Under the sensory congruent condition, the visual information is consistent with the actual movement direction; while under the sensory incongruent condition, the visual information is opposite to the actual movement direction, resulting in sensory conflict. S2. Collect the electroencephalogram (EEG) and electromyography (EMG) data of the participants' sensorimotor cortex. The collected data undergoes rigorous preprocessing, including filtering, denoising, and synchronization processing, to ensure the high quality and accuracy of the signals. The specific preprocessing steps include applying the Artifact Subspace Reconstruction (ASR) method to remove large-amplitude artifacts, and further removing interference signals such as blinking, electromyography, and electrocardiogram through independent component analysis (ICA); S3, data analysis stage, the partial directed coherence (PDC) method is used to calculate the directional information flow between the cortical area and the muscle in the frequency domain. PDC is a frequency domain causal analysis method based on Granger causality, which can effectively distinguish the directional influence between signals. The calculation formula of PDC is as follows: Among them, PDC ij (ω) represents the partial derivative coherence from node j to node i at frequency ω, A ij (e jω ) is the transfer function of the system, N is the total number of nodes in the system, and through PDC analysis, the direction and strength of information flow between different cortical areas and muscles can be quantified; S4. After obtaining the PDC results, the cortex-muscle coupling is further decomposed into two parts: the muscle's somatosensory information transmission ability and the cortex's control ability over the muscle. The specific decomposition method is as follows: Among them, Coupling Sensory Represents the overall ability of muscles to transmit somatosensory information to the cortex, Coupling Control Represents the overall ability of the cortex to actively control muscles.

4. The method for evaluating standing balance muscle control based on partial derivative coherence according to claim 3, characterized in that: The partial derivative coherence (PDC) method calculates the directional information flow between the cortical area and the muscle in the frequency range of 4-48 Hz. The calculation steps of PDC include: S1. Autoregressive (AR) model construction: An autoregressive model is established for each cortical area and muscle signal to describe its dynamic characteristics. Assuming that the order of the AR model of the system is p, it is expressed as: Among them, X(t) is the signal vector at the current moment, A(k) is the AR coefficient matrix, and ∈(t) is the error term; S2. Transfer function calculation: Calculate the transfer function A(e) of the system through the AR model parameter A(k). jω ): Where I is the identity matrix and ω is the frequency; S3, PDC value calculation: According to the transfer function A (e jω ), calculate the PDC value between each node: Among them, PDC ij (ω) represents the partial derivative coherence from node j to node i at frequency ω, A ij (e jω ) is the element in the transfer function matrix, N is the total number of nodes in the system; Through the above steps, the direction and strength of information flow between different cortical areas and muscles are quantified. Subsequently, the PDC analysis results are decomposed into two independent parts: the somatosensory information transmission ability of muscles and the control ability of the cortex over muscles. The specific decomposition formula is as follows: Among them, Coupling Sensory Represents the overall ability of muscles to transmit somatosensory information to the cortex, Coupling Control Represents the overall ability of the cortex to actively control muscles.

Images