Closed-loop rTMS regulation and control system driven by biomarker

Abstract

The invention provides a closed-loop rTMS regulation and control system driven by a biomarker. Relates to the technical field of closed-loop rTMS diagnosis and treatment. In the previous closed-loop rTMS intervention process, mostly evoked potentials, such as action potentials and local field potentials, which have limited representation degrees on autonomous motion functions are monitored; therefore, a closed-loop rTMS intervention system capable of dynamically monitoring spontaneous rhythmic oscillation PMBS with high representativeness on the upper limb movement function is established, the existing closed-loop rTMS intervention paradigm mostly depends on operation tasks, scales or EMG measurement and other means to evaluate the intervention curative effect, interaction between hemispheres represented by EEG is considered in the assessment of the upper limb movement rehabilitation capacity, and the intervention effect is improved. Therefore, the movement coordination of the upper limbs on the two sides is evaluated.

Description

Technical Field

[0001] The present invention relates to the technical field of closed-loop rTMS diagnosis and treatment, and in particular to a closed-loop rTMS regulation system driven by biomarkers. Background Art

[0002] Biomarkers are disease states characterized by electrophysiological measurement methods monitored in real time by a closed-loop rTMS system. In the past, physiological indicators monitored included evoked potentials such as action potentials and local field potentials that depend on external stimuli (Price et al., 2020). However, evoked potentials are exogenous and have limited representation of autonomous motor functions. To solve this problem, the present invention adopts a method of monitoring spontaneous rhythmic oscillations. Spontaneous rhythmic oscillations are stable and continuous outputs by subjects according to autonomous intentions or movements without external stimulation. They can more accurately reflect the upper limb motor function of stroke patients and are expected to become the core parameters of the new generation of closed-loop brain control technology (Espenhahn et al., 2020). Studies have shown that motor impairment caused by stroke is related to abnormal spontaneous oscillations of the beta rhythm of EEG (Rossiter et al., 2014; Briley et al., 2020; Espenhahn et al., 2020). Movement-Related Beta Desynchronization (MRBD) and Post-Movement Beta Synchronization (PMBS) are two beta rhythm spontaneous oscillations that appear successively during movement. Both MRBD and PMBS are modulated by the neural pathway mediated by gamma-aminobutyric acid (GABA) neurons (Zhang et al., 2024; Rossiter et al., 2014). MRBD and PMBS in the M1 region alternately release excitatory and inhibitory signals, jointly modulating the excitability and inhibition of the spinal nerve fiber bundles, thereby achieving precise control of upper limb movements. Previous studies have shown that MRBD and PMBS can be stably induced by index finger extension movements (Pfurtscheller et al., 1996), and abnormal MRBD and PMBS in stroke patients have a crucial impact on upper limb motor control impairment (Rossiter et al., 2014). Therefore, it is feasible to use beta rhythm spontaneous oscillations as a biomarker to characterize upper limb motor control impairment in stroke patients.

[0003] In addition, the current closed-loop rTMS stroke rehabilitation system still has the problem of inappropriate evaluation of motor rehabilitation function. In previous practices, most of the motor rehabilitation status was evaluated using operational tasks, scales or electromyography (EMG) (Chang et al., 2010; Avenanti et al., 2012; Johnson et al., 2018; Singh et al., 2023), but there was a lack of evaluation of the synergy of brain function after motor rehabilitation. The present invention evaluates the intervention efficacy from two aspects: EEG electrophysiology and clinical evaluation. Because the present invention intervenes in the M1 area on both sides at the same time, the electrophysiological evaluation evaluates the coordination of the bilateral upper limbs based on the interaction between the hemispheres.

[0004] Based on this, the present invention aims to construct an rTMS stimulation system, which uses task-induced and dynamic monitoring of endogenous neural markers during the intervention process, adjusts the stimulation parameters in real time, and achieves precise regulation of upper limb motor rehabilitation after stroke. Interhemispheric interactivity is considered when evaluating the intervention effect, so as to more comprehensively understand the neural reorganization caused by the intervention. Summary of the invention

[0005] The purpose of the present invention is to provide a closed-loop rTMS regulation system driven by biomarkers in order to solve the above-mentioned problems.

[0006] To achieve the above object, the technical solution adopted by the present invention is:

[0007] According to one aspect of the present invention, a closed-loop rTMS regulation system driven by biomarkers is provided, including a pre- and post-test module, which determines the effect of rTMS intervention based on whether there is a significant difference between the pre- and post-test data.

[0008] The data acquisition module includes determining the resting motor threshold (RMT) before rTMS intervention, recording the EMG activity of FDI using Ag-AgCl surface self-adhesive disk electrodes during rTMS intervention, and synchronously acquiring EEG data from 64 electrode leads;

[0009] The dynamic change module includes obtaining movement speed based on EMG data; using denoising algorithms and real-time processing algorithms to process the collected EEG data online and dynamically monitor biomarkers;

[0010] The rTMS stimulation module includes a magnetic stimulator, which triggers rTMS pulses through the BiStim module and acts on the corresponding brain area through a figure-of-eight coil.

[0011] Furthermore, when the front and back measurement modules perform neuroelectrophysiological examinations, the patients are required to perform isotonic index finger extension exercises of both hands respectively, and EEG data of 64 electrode leads are collected synchronously, with a sampling rate of 1 KHz, impedance reduced to below 10 KΩ, reference electrodes being bilateral mastoids, and notch filtering set to 50 Hz.

[0012] Furthermore, the pre- and post-test module performs inter-hemispheric causal effect analysis on the collected pre- and post-test EEG data, including using a time domain window method to obtain a short-term stationary time series, establishing a multivariate autoregressive model, and calculating directional coherence through a Granger causality model.

[0013] Furthermore, when the data acquisition module acquires EMG data, the electromyographic signal is amplified and filtered using a bio-amplifier, and then digitized using an analog-to-digital converter;

[0014] When determining the RMT, when the motor evoked potential value induced by at least 5 stimulations out of 10 consecutive TMS pulses was not less than 50 µV, the minimum TMS panel stimulation intensity that met the condition was the RMT, and the trials contaminated by myoelectric activity were eliminated;

[0015] When collecting EEG signals, the sampling rate was 1 KHz, the electrode impedance was reduced to below 10 KΩ, the reference electrodes were bilateral mastoids, and the notch filter was set to 50 Hz.

[0016] Furthermore, the dynamic change module requires the patient to perform isometric index finger extension exercises of both hands during the rTMS intervention process. The index finger extension exercises of each hand include four groups, each group contains 20 trials, and each trial requires the patient to perform an index finger extension exercise as quickly as possible.

[0017] Furthermore, in the dynamic change module, for each trial, based on the EMG data, the average result of the integration of the horizontal axial movement acceleration output by the MPU-6050 accelerometer sensor is recorded as the speed of the movement, and the average speed of each trial of the behavioral data is extracted.

[0018] Furthermore, in the dynamic change module, the denoising algorithm includes time domain filtering, online PCA and online ICA, CSP filtering, and the real-time processing algorithm includes using the multi-window spectrum estimation of the Slepian data window to perform online analysis on the denoised and artifact-free EEG signal, and using the following formula to calculate the event-related spectrum disturbance:

[0019] (30)

[0020] Furthermore, in the rTMS stimulation module, the rTMS pulse was triggered 2s after the online monitoring of beta power recovery baseline. A 10Hz rTMS pulse was used, the duration was 5s, and the initial stimulation intensity was 80%RMT. The PMBS size and movement speed induced by the patient's index finger movement in the previous trial were dynamically adjusted. There were 4 groups of interventions during the index finger extension movement of each hand, each group contained 20 trials, and each trial required the patient to perform an index finger extension movement as quickly as possible. A total of 4000 pulses were delivered during the index finger extension movement of both hands. There was an interval of 5 minutes between groups, and the total duration was about 40 minutes.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] The biomarker-driven closed-loop rTMS control system established by the present invention can dynamically monitor spontaneous rhythmic oscillation PMBS and adjust rTMS intervention parameters in a timely manner, thereby achieving precise manipulation of endogenous neural oscillations and precise intervention of upper limb motor rehabilitation in stroke patients. At the same time, the present invention combines EEG technology when evaluating the intervention efficacy, taking into account the inter-hemispheric interactivity, so as to more comprehensively evaluate the intervention effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flow chart of a biomarker-driven closed-loop rTMS regulation system provided by the present invention;

[0024] Figure 2 A task flow chart provided for the present invention;

[0025] Figure 3 The present invention provides a 64-sample electrode diagram arranged in accordance with the 10-20 international standard lead system;

[0026] Figure 4 This is a flow chart of the rTMS intervention provided by the present invention. DETAILED DESCRIPTION

[0027] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0028] The present invention provides a closed-loop rTMS control system driven by biomarkers, which includes a pre- and post-test module, a data acquisition module, a dynamic change module, and an rTMS stimulation module.

[0029] The pre- and post-test modules include pre-test and post-test, and the effect of rTMS intervention is judged based on whether there is a significant difference between the pre- and post-test data.

[0030] Clinical assessment included assessment of upper limb motor function using the Upper Extremity_FMT (UE_FMT) component of the National Institutes of Health Stroke Scale (NIHSS) and the nine-hole pegboard test.

[0031] The neurophysiological examination required the patient to perform isotonic index finger extension exercises on both hands. When the white circle appeared, the index finger extension exercise was performed as quickly as possible. Four sets of trials were performed on each hand, each set of trials included 20 trials. During the task, EEG data of 64 electrode leads were collected synchronously, with a sampling rate of 1KHz, the electrode impedance was reduced to below 10KΩ, the reference electrode was the bilateral mastoid (M1 & M2), and the notch filter was set to 50Hz.

[0032] Inter-hemispheric causal effect analysis was performed on the collected pre- and post-test EEG data, including using the time domain window method to obtain short-term stationary time series, establishing a multivariate autoregressive model, and calculating directional coherence through the Granger causality model.

[0033] Specifically, let 𝑋(𝑡) be two EEG time series signals, then:

[0034] (1)

[0035] Construct a multivariate autoregressive model based on 𝑋(𝑡):

[0036] (2)

[0037] in, The model coefficient matrix is ​​2*2 (the present invention only selects channels C3 and C4), which reflects the time lag The variable dependencies of is the order of the multivariate autoregressive model, which is determined by the minimization information criterion. In the present invention, the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) are used to comprehensively determine is 10, Represents random noise. The multivariate Yule-Walker equation is solved using the Levisnson, Wiggins, Robinson (LWR) algorithm to obtain an estimate of the coefficient matrix.

[0038] The frequency domain representation of EEG data is:

[0039] (3)

[0040] in is the transfer matrix of the system, from which the direction of information flow between the two signals can be obtained, that is, the directional transfer function is defined as , indicating the signal right The causal influence of represents the observed frequency domain EEG data, represents the frequency domain EEG source signal. The direct directional coherence is:

[0041] (4)

[0042] in, (In the present invention , that is, the signals of the two channels C3 and C4) are the signals The power spectrum of .

[0043] The bandpass filter range in the present invention is 8-35Hz, and the slice data length is 0.25s before the execution of the task instruction signal to 2.5s after the task signal, a total of 2.75s. 0.05s is selected as the sliding window (50% overlap), which will be divided into L segments of short-time data.

[0044] The significance level is calculated as follows:

[0045] (5)

[0046] The Z value was calculated according to the 95% confidence interval. When the results of the connectivity coherence analysis and the directional coherence results of the causal effect analysis were greater than the Z value, the connectivity and causal effect were considered significant.

[0047] Perform dynamic functional network feature analysis, including taking each lead as a node and using the time series data of each lead to reflect the changes in each brain region.

[0048] Calculate the functional network change characteristics by phase-locking method:

[0049] (6)

[0050] in, For time series The total number of points, for and The PLV value range is 0 to 1, where 0 means that the two time series are not statistically coupled and 1 means that they are fully coupled.

[0051] In the present invention, nodes represent electrodes (excluding two reference electrodes M1 and M2, N = 62), and edges represent PLV values ​​between electrode pairs. . Considering the different strengths of network edges in brain network connections, dynamic brain network features are obtained by weighted network analysis method. The threshold is set using the ratio threshold method, and the connection matrix is ​​constructed by retaining the ratio P (0 < P < 1) with the strongest weight. In the present invention, the ratio is set to 0.7, 0.8, 0.9 and 1. The clustering coefficient is calculated as follows:

[0052] (7)

[0053] The global efficiency calculation method is as follows:

[0054] (8)

[0055] The characteristic path length is calculated as follows:

[0056] (9)

[0057] in, is the set of network nodes, is the node degree of node 𝑖, is the connection weight between nodes 𝑖 and 𝑗, is the shortest weighted path length between nodes 𝑖 and 𝑗, is the average weighted path length from node 𝑖 to all nodes.

[0058] The data acquisition module is a module that collects physiological data during the rTMS intervention process.

[0059] EMG activity of the First Dorsal Interosseous (FDI) was recorded using Ag-AgCl (10 mm diameter) surface self-adhesive disc electrodes. The EMG signals were amplified and filtered (5-2 kHz) using a bioamplifier (Neurolog System, Digitimer, UK) and then digitized using an analog-to-digital (A / D) converter (CED Micro 1401, Cambridge Electronic Design, UK).

[0060] Before rTMS intervention, the resting motor threshold (RMT) was determined. The patient's muscles were in a relaxed state, and the TMS panel intensity was adjusted. When at least 5 of the 10 consecutive stimulations induced a motor evoked potential (MEP) value of no less than 50µV, the minimum TMS panel stimulation intensity that met the conditions was the RMT, and trials contaminated by myoelectric activity were eliminated.

[0061] EEG data of 64 electrode leads were collected synchronously. Wet electrodes were used to collect scalp EEG signals, and Neuroscan Synamps2 amplifier was used to amplify the EEG signals. The sampling rate was 1KHz. The electrode impedance during the experiment was reduced to below 10KΩ. The reference electrodes were bilateral mastoids (M1 & M2), and the notch filter was set to 50Hz.

[0062] The dynamic change module is a module that processes and updates the collected data in real time.

[0063] The patients performed isotonic index finger extension exercises for both hands. The index finger extension exercise for each hand consisted of four sets, each set contained 20 trials, and each trial required the patient to perform an index finger extension exercise as quickly as possible.

[0064] For each trial, the movement speed was obtained based on the EMG data, and the average result of the integration of the horizontal axial movement acceleration output by the MPU-6050 accelerometer sensor was recorded as the movement speed, and the average speed of each trial of the behavioral data was extracted.

[0065] For each trial, when the right hand moves, the dynamic change module processes the EEG data of the C3 channel; when the left hand moves, the dynamic change module processes the EEG data of the C4 channel.

[0066] The collected EEG data were denoised online using denoising algorithms, including time domain filtering, online PCA, and online ICA and CSP filtering.

[0067] Time domain filtering included the use of 8-35 Hz bandpass filtering and 50 Hz notch filtering; the EEG data were sliced, starting from 0.25 s before the task instruction signal and ending 2.5 s after the task signal, for a total of 2.75 s; the signal within 80 ms after each rTMS pulse was removed to exclude artifacts such as TMS discharges and spike pulses; and the EEG data were downsampled to 500 Hz.

[0068] Online PCA analysis of synchronously collected EEG data includes the following steps:

[0069] Step 1: When collecting EEG signals, there are n samplings, each sampling has m electrodes (in this invention, the number of electrodes is 62), and the EEG data is represented as a The matrix , the covariance matrix The calculation formula is as follows:

[0070] (10)

[0071] in, is the mean vector of the data.

[0072] Step 2: Use the following formula to perform eigenvalue decomposition on the covariance matrix:

[0073] (11)

[0074] in, is the eigenvector matrix, is a diagonal matrix, and the elements on the diagonal are the eigenvalues.

[0075] Step 3: Select the eigenvectors corresponding to the first k eigenvalues ​​to form the projection matrix , project the data into the principal component space:

[0076] (12)

[0077] in, is the data matrix after dimensionality reduction.

[0078] The basic steps of online ICA are as follows: First, construct an EEG data matrix containing m electrodes , the size is ,in is the number of sampling points. ICA is expressed mathematically as follows:

[0079] (13)

[0080] in, is the observation signal matrix, is the unknown mixing matrix, is the source signal matrix. The purpose of ICA is to find the source signal matrix , specifically including the following steps:

[0081] Step 1: For the data matrix , first use the following formula to decentralize:

[0082] (14)

[0083] in, is the original data matrix The mean of each feature column.

[0084] Step 2: Whiten the data using the following formula:

[0085] (15)

[0086] The covariance matrix is ​​calculated using the following formula Perform eigendecomposition:

[0087] (16)

[0088] in, is the eigenvector matrix, is a diagonal eigenvalue matrix. The whitening transformation can be expressed as follows:

[0089] (17)

[0090] Step 3: Based on maximizing non-Gaussianity, negative entropy is selected as the measure of non-Gaussianity, using a nonlinear function To approximate the maximization of negative entropy and find the projection weight that maximizes negative entropy, its iterative update rule can be expressed by the following formula:

[0091] (18)

[0092] Afterwards, the new weight vector is calculated using the following formula: Normalize:

[0093] (19)

[0094] Step 4: Orthogonalization and convergence, using the Gram-Schmidt process. components, the update rule will take into account the previous Quantity:

[0095] (20)

[0096] right After normalization, check for convergence.

[0097] Step 5: Independent component analysis, find all weight vectors:

[0098] (twenty one)

[0099] The independent components are estimated as follows:

[0100] (twenty two)

[0101] According to the PCA and ICA results, the EMG and other noises are removed.

[0102] CSP filtering includes the following steps:

[0103] Step 1: Create different covariance matrices for different categories of EEG signals. The matrices are as follows:

[0104]

[0105] in, Represents the variance of the factor and is calculated using the following formula:

[0106] (twenty three)

[0107] in, For different types of EEG signals, is the mean vector of the data.

[0108] Represents the covariance of two factors and is calculated using the following formula:

[0109] (twenty four)

[0110] in, For different types of EEG signals, is the mean vector of the data.

[0111] Step 2: Generalized eigenvalue decomposition is used on these covariance matrices. The basic idea is as follows: For a set of matrices , ... and ,in, yes dimensional matrix, yes -dimensional positive definite matrix. The generalized eigenvalue problem is of the form:

[0112] (25)

[0113] in is a non-zero eigenvector, is the generalized eigenvalue. The generalized eigenvalue decomposition in this general form can be expressed as follows:

[0114] (26)

[0115] in, yes dimensional matrix, yes -dimensional matrix whose columns are eigenvectors combination. is a diagonal matrix whose diagonal elements are generalized eigenvalues . Finally, the generalized eigenvalue decomposition will generate a set of eigenvalues ​​and corresponding eigenvectors.

[0116] Step 3: Select the eigenvectors with the largest or smallest eigenvalues ​​to construct filters. Select the eigenvector with the largest eigenvalue to enhance the brain activity of interest, and select the eigenvector with the smallest eigenvalue to suppress other brain activities. Finally, a set of CSP filters is generated, each filter corresponds to a different brain activity pattern, some of which are used for enhancement and some for suppression.

[0117] Biomarkers are monitored using real-time processing algorithms. The denoised and artifact-free EEG signals are analyzed online using the multi-window spectrum estimation of the Slepian data window to obtain the spectrum estimation of the time series signal S(n), which is calculated as follows:

[0118] (27)

[0119] in, is the number of Slepian data windows (also called order), is the kth Slepian data window. The spectrum energy is concentrated between 2W, and the time-bandwidth product is:

[0120] (28)

[0121] reflects the degrees of freedom of the variance of the coordinated spectrum estimate, and the orthogonal window order and degrees of freedom It provides a trade-off between frequency domain resolution and spectral estimation variance. A Slepian data window can reduce spectrum leakage, so it is often used orthogonal windows. If the eigenvalue corresponding to the kth eigenspectrum is , then the minimum sidelobe leakage spectrum estimation result is:

[0122] (29)

[0123] A third-order Slepian orthogonal window is used to control spectrum leakage, and the frequency smoothing window is , with a step length of 25ms. The absolute value of the square of the convolution of all trials in each channel task is averaged to obtain the time-frequency representation (TFR) of the power. The 250ms before the instruction task is selected as the baseline for correction, and the event-related spectrum perturbation is calculated as:

[0124] (30)

[0125] Usually, the beta band (14-30Hz) data in TFR is reduced to the time domain to intuitively display the event-related power changes in the beta band (including the two processes of desynchronization and synchronization). The present invention mainly focuses on the changes near the PMBS peak of the target brain area.

[0126] The time when beta power increased by 10% compared with the baseline was defined as the start of the PMBS burst, and the time when beta power dropped to 110% of the baseline was defined as the time when beta power returned to the baseline.

[0127] The rTMS stimulation module performs rTMS intervention based on the real-time changes of biomarkers monitored by the dynamic change module.

[0128] The present invention uses two Magstim Rapid2 magnetic stimulators produced by the British company Magstim (Magstim, Whitland, Carmarthenshire, UK), which form rTMS mode through the BiStim module and can act on the corresponding brain area through an eight-shaped coil (the outer diameter of the coil is 90 mm).

[0129] The rTMS pulse was triggered 2s after the online monitoring of beta power returned to the baseline. A 10Hz rTMS pulse was used with a duration of 5s, and the initial stimulation intensity was 80%RMT. The intensity was adjusted dynamically according to the size and speed of the PMBS induced by the patient's index finger movement in the previous trial. When the index finger extension movement speed after rTMS intervention was not less than the average speed obtained offline before the intervention and the PMBS did not decrease, there was no need to adjust the stimulation intensity; when the index finger extension movement speed was less than the average speed obtained offline before the intervention for four consecutive trials after rTMS intervention or the PMBS decreased, the rTMS stimulation intensity was increased, and the adjustment range was 80%-110%RMT, in 10% steps.

[0130] There were 4 interventions during the index finger extension exercise on each hand, each containing 20 trials, and each trial required the patient to perform an index finger extension exercise as quickly as possible. A total of 4,000 pulses were delivered during the index finger extension exercise on both hands. There was a 5-minute interval between groups, and the total duration was about 40 minutes.

[0131] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0132] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A closed-loop rTMS control system driven by biomarkers, including a pre- and post-test module, which determines the effect of rTMS intervention based on whether there is a significant difference between the pre- and post-test data, and is characterized by: The data acquisition module includes determining the resting motor threshold (RMT) before rTMS intervention, recording the EMG activity of FDI using Ag-AgCl surface self-adhesive disk electrodes during rTMS intervention, and synchronously acquiring EEG data from 64 electrode leads; The dynamic change module includes obtaining movement speed based on EMG data; using denoising algorithms and real-time processing algorithms to process the collected EEG data online and dynamically monitor biomarkers; The rTMS stimulation module includes a magnetic stimulator, which triggers rTMS pulses through the BiStim module and acts on the corresponding brain area through a figure-of-eight coil.

2. The biomarker-driven closed-loop rTMS control system according to claim 1, characterized in that: When the front and back test modules perform neuroelectrophysiological examinations, the patients are required to perform isotonic index finger extension exercises of both hands respectively, and EEG data of 64 electrode leads are collected synchronously, with a sampling rate of 1 KHz, impedance reduced to below 10 KΩ, reference electrodes being bilateral mastoids, and notch filtering set to 50 Hz.

3. The biomarker-driven closed-loop rTMS control system according to claim 1, characterized in that: The pre- and post-test module performs inter-hemispheric causal effect analysis on the collected pre- and post-test EEG data, including obtaining a short-term stationary time series using a time domain window method, establishing a multivariate autoregressive model, and calculating directional coherence through a Granger causality model.

4. The biomarker-driven closed-loop rTMS control system according to claim 1, characterized in that: When the data acquisition module collects EMG data, the electromyographic signal is amplified and filtered by a biological amplifier, and then digitized by an analog-to-digital converter; When determining the RMT, when the motor evoked potential value induced by at least 5 stimulations out of 10 consecutive TMS pulses was not less than 50 µV, the minimum TMS panel stimulation intensity that met the condition was the RMT, and the trials contaminated by myoelectric activity were eliminated; When collecting EEG signals, the sampling rate was 1 KHz, the electrode impedance was reduced to below 10 KΩ, the reference electrodes were bilateral mastoids, and the notch filter was set to 50 Hz.

5. The biomarker-driven closed-loop rTMS control system according to claim 1, characterized in that: The dynamic change module requires the patient to perform isometric index finger extension exercises of the left and right hands during the rTMS intervention process. The index finger extension exercise of each hand includes four groups, each group contains 20 trials, and each trial requires the patient to perform an index finger extension exercise as quickly as possible.

6. The biomarker-driven closed-loop rTMS control system of claim 1, characterized in that: In the dynamic change module, for each trial, based on the EMG data, the average result of the integration of the horizontal axial movement acceleration output by the MPU-6050 accelerometer sensor is recorded as the speed of the movement, and the average speed of each trial of the behavioral data is extracted.

7. The biomarker-driven closed-loop rTMS control system of claim 1, characterized in that: In the dynamic change module, the denoising algorithm includes time domain filtering, online PCA and online ICA, CSP filtering, and the real-time processing algorithm includes using the multi-window spectrum estimation of the Slepian data window to perform online analysis on the denoised and artifact-free EEG signal, and using the following formula to calculate the event-related spectrum disturbance: (30) 。 8. The biomarker-driven closed-loop rTMS control system of claim 1, wherein: In the rTMS stimulation module, the rTMS pulse was triggered 2s after the online monitoring of beta power recovered to the baseline. A 10Hz rTMS pulse was used with a duration of 5s. The initial stimulation intensity was 80%RMT. The intensity was dynamically adjusted according to the size and speed of the PMBS induced by the patient's index finger movement in the previous trial. There were 4 groups of interventions during the index finger extension movement of each hand, each containing 20 trials. Each trial required the patient to perform an index finger extension movement as quickly as possible. A total of 4000 pulses were delivered during the index finger extension movement of both hands. There was an interval of 5 minutes between groups, and the total duration was about 40 minutes.

Images