Dynamic brain function network analysis method after repeated transcranial magnetic stimulation

Abstract

The invention discloses a dynamic brain function network analysis method after repeated transcranial magnetic stimulation. According to the method, the dynamic tissue of the large-scale brain function network can be captured in the minimum time scale of 60-120 ms from the electroencephalogram micro-state. A dynamic brain function network is constructed by taking micro-state segments as time windows, so that the arbitrariness and dilution effect of a fixed window length are avoided, and instantaneous network recombination and stability change induced by repeated transcranial magnetic stimulation are represented more accurately. Network parameters such as connectivity stability and a clinical scale are subjected to correlation analysis, a systematic quantitative normal form of repeated transcranial magnetic stimulation-micro state-dynamic brain function network-clinical effect is formed, and a reusable technical path is provided for evaluating brain network dynamics before and after treatment and the relation between brain network dynamics and symptom improvement.

Description

Technical Field

[0001] This invention relates to the field of neuromodulation technology, specifically a method for dynamic brain functional network analysis following repetitive transcranial magnetic stimulation. Background Technology

[0002] Repetitive transcranial magnetic stimulation (rTMS), as a non-invasive neuromodulation technique, has shown promising application prospects in areas such as cognitive enhancement, depression relief, and motor function rehabilitation due to its advantages of being non-invasive and capable of inducing plasticity. RTMS modulates the cerebral cortex through magnetic fields and has been used to enhance cognitive function, alleviate depressive symptoms, and promote motor function recovery. However, despite the significant clinical efficacy of RTMS, its integration with electroencephalographic signals remains insufficient. On the one hand, most research frameworks still focus on static brain functional connectivity, making it difficult to reveal the temporal dynamic evolution and transient network reorganization under RTMS intervention, as exemplified by the literature "Pignat, JM, et al. (2025). 50Hz-Repetitive transcranial magnetic stimulation modulates brain connectivity in Parkinson's disease. Clinical Neurophysiology, 174, 96-104." On the other hand, traditional dynamic brain functional networks mostly use fixed sliding windows, which have recognized challenges in terms of sampling bias, non-neural noise contamination, and state confusion. They are prone to low sensitivity or distortion to fast dynamics, as seen in the literature "Laumann TO, et al. Challenges in the measurement and interpretation of dynamic functional connectivity. Imaging Neurosci (Camb). 2024 Nov 19; 2:imag-2-00366.doi:10.1162 / imag_a_00366.".

[0003] Existing EEG analysis methods, such as spectral analysis, event-related potential analysis, and resting-state brain functional network analysis, while providing some information, cannot comprehensively reflect the changes in the brain's temporal and frequency domain characteristics after repetitive transcranial magnetic stimulation (rTMS). Furthermore, these methods often neglect the spatiotemporal dynamic changes before and after stimulation. Current microstate analysis methods mostly focus on static or short-term EEG signal analysis, failing to reflect the immediate changes and long-term stable characteristics of the brain under dynamic stimulation. Therefore, a systematic approach is lacking to analyze the dynamic changes in brain functional networks after rTMS, especially a detailed analysis of microstate parameters and brain functional networks. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for dynamic brain functional network analysis following repetitive transcranial magnetic stimulation.

[0005] The technical solution of this invention to solve the aforementioned technical problem is to provide a method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation, characterized in that the method includes the following steps:

[0006] Step 1: Perform repetitive transcranial magnetic stimulation on the patient, collect patient data, extract the microstate time series from the patient's electroencephalogram (EEG) signal, and calculate the microstate parameters before and after repetitive transcranial magnetic stimulation.

[0007] Step 2: Construct a dynamic brain functional network using the microstate time series obtained in Step 1; then use the dynamic brain functional network to calculate the brain's steady-state coefficient to analyze the connectivity stability of the dynamic brain functional network.

[0008] Step 3: Test the microstate parameters obtained in Step 1 and the brain homeostasis coefficient obtained in Step 2 against the clinical scale; the analysis is complete when significant changes are detected in both the microstate parameters and the brain homeostasis coefficient, and the changes in the microstate parameters and the brain homeostasis coefficient are significantly correlated with the improvement of the clinical score.

[0009] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0010] (1) This invention can capture the rapid dynamic changes of the brain network before and after repetitive transcranial magnetic stimulation on a millisecond time scale, overcoming the limitations of information loss in traditional static EEG analysis and the non-adaptive nature of fixed sliding window method, and providing a systematic and reproducible analysis path for revealing the electrophysiological mechanism of repetitive transcranial magnetic stimulation.

[0011] (2) In the data preprocessing and microstate extraction stages, this invention utilizes multi-band filtering (theta, alpha, beta, gamma) combined with microstate analysis to extract features from neural oscillations at different frequency bands, improving the comprehensiveness of EEG signal analysis. Independent component analysis is used to remove artifacts from eye movements, electromyography, and electrocardiograms, effectively improving data quality and making the analysis results more credible. Microstate clustering combined with cross-validation and global explained variance index ensures the objectivity and stability of template selection, thereby improving the accuracy of subsequent brain functional network analysis. Microstate analysis enables meticulous feature extraction of EEG signals, leading to a deeper understanding of the dynamic changes in brain functional networks. This method can capture the short-term fluctuations and long-term stable characteristics of brain functional networks, providing a new perspective for exploring how brain functional networks respond to external stimuli.

[0012] (3) In the construction of dynamic brain functional network, the present invention uses micro-state window as time segmentation unit to avoid window length selection bias in fixed sliding window analysis and more realistically reflects the instantaneous functional state of the brain; and adopts phase-locked value to quantify the phase synchronization between EEG channels. The phase-locked value obtains the functional connectivity of paired EEG channels in each micro-state window by quantifying the phase synchronization of the signal at a specific frequency. The phase-locked value is suitable for dynamic analysis of instantaneous phase changes, can capture complex phase relationships between signals, and has good robustness to noise.

[0013] (4) In the network stability assessment stage, this invention introduces a brain steady-state coefficient, which characterizes the stability of a dynamic network by the correlation of network connectivity at adjacent time points, thus overcoming the shortcomings of traditional static graph theory indicators that cannot reflect temporal correlation. This coefficient can intuitively reflect the integration and information transmission efficiency of the brain network. Its increase indicates that the connection is more stable and efficient, and it has a stronger explanatory power compared to relying solely on static indicators such as average degree and clustering coefficient.

[0014] (5) In the clinical relevance analysis stage, this invention statistically analyzes the brain functional network parameters before and after stimulation with clinical scales, enabling the establishment of a correlation between brain electrical activity and behavioral performance. This method achieves multi-level integrated analysis from electrophysiological signals to network parameters and then to clinical symptoms, which helps to establish an objective and repeatable evaluation system. This comprehensive analysis helps to understand how changes in brain function affect patient behavior and promotes the development of personalized treatment plans. Attached Figure Description

[0015] Figure 1 This is an overall flowchart of the present invention;

[0016] Figure 2 This is a diagram illustrating the intermittent theta burst magnetic stimulation pulse delivery pattern of Embodiment 1 of the present invention;

[0017] Figure 3 This is a microstate clustering result diagram of a Parkinson's disease patient before and after receiving intermittent theta burst magnetic stimulation in Embodiment 1 of the present invention;

[0018] Figure 4 The diagram shows the microstate parameters of a Parkinson's disease patient before and after receiving intermittent theta burst magnetic stimulation, according to Example 1 of the present invention.

[0019] Figure 5 This is a diagram showing the inter-channel results of the brain homeostasis coefficients of patients before and after intermittent theta burst magnetic stimulation intervention in different channels of different frequency bands in Embodiment 1 of the present invention.

[0020] Figure 6 The graph shows the correlation analysis results between the changes in microstate parameters and the Δ Unified Parkinson's Disease Rating Scale-III in Example 1 of the present invention.

[0021] Figure 7 This is a graph showing the correlation analysis results between the Δ brain homeostasis coefficient of each significant channel in Embodiment 1 of the present invention and the Δ Unified Parkinson's Disease Rating Scale-III. Detailed Implementation

[0022] Specific embodiments of the present invention are given below. These specific embodiments are only used to further illustrate the present invention in detail and do not limit the scope of protection of the present invention.

[0023] This invention provides a method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation (hereinafter referred to as the method), characterized by the following steps:

[0024] Step 1: Perform repetitive transcranial magnetic stimulation on the patient, collect patient data, extract the microstate time series from the patient's electroencephalogram (EEG) signal, and calculate the microstate parameters before and after repetitive transcranial magnetic stimulation.

[0025] Preferably, step 1 specifically includes the following steps:

[0026] S11. Collect raw EEG data and clinical scale scores of patients before and after receiving repetitive transcranial magnetic stimulation as the raw data basis for subsequent analysis.

[0027] S12. Preprocessing of raw EEG data: First, bandpass filtering (0.5-80Hz) and notch filtering (50Hz) are applied to the raw EEG data. Then, artifacts of blinking, horizontal eye movement, electromyography, and electrocardiogram are corrected using independent component analysis. The sampling frequency of the data is reduced and converted to an average reference. Based on the distribution frequency band of neural oscillation activity, the EEG signals are divided into frequencies according to theta (4-7Hz), alpha (8-12Hz), beta (13-30Hz), and gamma (30-80Hz). At the same time, the bandpass-filtered EEG data of 2-20Hz is retained for microstate analysis.

[0028] Preferably, in step S12, the data sampling frequency is reduced to 250Hz.

[0029] S13. Based on the EEG data obtained in step S12, calculate the global field potential of the EEG signal to be analyzed for each subject.

[0030] Preferably, in step S13, when the global field potential reaches a local peak, the corresponding EEG distribution pattern is considered a representative pattern of the microstate:

[0031]

[0032] In equation (1), GFP represents the global field potential; N is the number of electrodes, u iLet be the voltage value of the i-th electrode. This represents the average voltage value across all electrodes.

[0033] S14. Based on the candidate microstate distribution patterns extracted in step S13, the improved K-means clustering algorithm is used to perform global field potential time series clustering.

[0034] Preferably, in step S14, the number of clusters is set to 2 to 8, the clustering process is repeated 100 times to ensure the stability of the results, and the maximum number of iterations for each clustering is 1000.

[0035] S15. Based on the clustering results of step S14, the optimal number of microstate templates is selected by evaluating the microstate fit using the cross-validation criterion and the global explained variance between the original EEG of each group and each microstate category.

[0036] S16. Import the optimal number of microstate templates determined in step S15 into the original EEG data and perform retrospective fitting to identify the specific location of each microstate in the time series and smooth the fitted microstate sequence to obtain the microstate time series.

[0037] S17. Calculate the microstate parameters using the microstate time series obtained in step S16.

[0038] Preferably, in step S17, the micro-state parameters are global explained variance, duration, coverage, and occurrence rate.

[0039] Preferably, in step S17, the global interpretation variance is used to quantify the overall interpretability of each microstate to the topographic map variability, reflecting the accuracy of the microstate in representing topographic features.

[0040] Duration is the average duration of each microstate, describing state stability.

[0041] Coverage is: a quantification of the proportion of the dominant microstates in the total analysis time, reflecting the spatiotemporal dominance of different states;

[0042] The occurrence rate is the number of times each microstate is presented within a unit of time (1 second), which characterizes the dynamic transformation characteristics of the system.

[0043] Step 2: Construct a dynamic brain functional network using the microstate time series obtained in Step 1; then use the dynamic brain functional network to calculate the brain's steady-state coefficient to analyze the connectivity stability of the dynamic brain functional network.

[0044] Preferably, step 2 specifically includes the following steps:

[0045] S21. Using the microstate time series obtained in step 1 as the basis for time division, the EEG signal is divided into n non-overlapping microstate windows, and the EEG data corresponding to each window has a specific microstate class.

[0046] S22. Based on the microstate window EEG data obtained in step S21, the phase-locked value algorithm is used to construct a dynamic brain function network between the EEG channels of each subject.

[0047] Preferably, in step S21, the expression for the phase-locked loop (PLL) algorithm is:

[0048]

[0049] In equation (2), t represents the time point, N represents the number of sample points of the signal, and Δφ(t) represents the phase difference between the two electrodes at time t. The phase-locked value coefficient between each pair of leads is calculated in each window, and the phase-locked value coefficient is used as the weight of the edge in the brain functional network between each electrode pair to obtain a dynamic brain functional network symmetrical about the main diagonal.

[0050] S23. Calculate the brain homeostasis coefficient using the dynamic brain functional network obtained in step S22. This coefficient is used to quantify the connectivity stability of the dynamic brain functional network.

[0051] Preferably, step S23 specifically includes the following steps:

[0052] S231. Using the dynamic brain function network obtained in step S22, extract the connection columns corresponding to electrodes at specific locations in the brain for each time window; since the phase-locked value of the same channel is always 1, the autocorrelation on the diagonal of the matrix is ​​removed.

[0053] S232. Using the connection columns extracted in step S231 as input, based on the time series arrangement relationship, it is assumed that the network connectivity at any given time point is only related to the previous time point and is not related to other time points; then the Pearson correlation coefficient between the connection columns of adjacent time points is calculated, and the average of the correlation coefficients of all time points is taken to obtain the brain homeostasis coefficient.

[0054] Preferably, in step S232, the expression for calculating the brain homeostasis coefficient is:

[0055]

[0056] In equation (3), W p (i,:) represents the column of the phase-locked loop (PLL) value matrix after removing autocorrelation from channel i, and corr represents the Pearson correlation between the two variables; the formula for calculating the Pearson correlation coefficient is:

[0057]

[0058] In equation (4), cov represents the covariance of the two variables, δ X δ represents the standard deviation of variable X. Y It represents the standard deviation of variable Y.

[0059] Step 3: The microstate parameters obtained in Step 1 and the brain homeostasis coefficient obtained in Step 2 are tested against the clinical scale. When significant changes are detected in both the microstate parameters and the brain homeostasis coefficient, and the changes in the microstate parameters and the brain homeostasis coefficient are significantly correlated with the improvement of the clinical score, the analysis is completed. It is believed that this method has successfully revealed the electrophysiological mechanism of repetitive transcranial magnetic stimulation and can be used as a potential efficacy evaluation indicator.

[0060] Preferably, in step 3, the Shapiro-Wilk normality test is used to test the normality of the data; parametric tests are used for normally distributed data, and non-parametric tests are used for non-normally distributed data; paired-samples t-tests are used for normally distributed electrophysiological parameters before and after treatment, and Wilcoxon signed-rank tests are used for non-normally distributed electrophysiological parameters before and after treatment; Pearson correlation analysis is used for the differences in clinical scale data before and after treatment and the differences in electrophysiological parameters before and after treatment; considering the problem of repeated testing for multiple frequency bands and multiple parameters, all statistical results are corrected for multiple comparisons using the false discovery rate correction method.

[0061] Preferably, in step 3, a significant change is defined as p < 0.05, and a significant correlation is defined as p < 0.05.

[0062] Example 1:

[0063] Step 1: Perform intermittent theta burst magnetic stimulation on the patient. Each stimulation in the magnetic stimulation sequence is generated in clusters, with three stimulations per cluster. The intra-cluster frequency is 50 Hz, and the inter-cluster frequency is 5 Hz. After 2 seconds of continuous cluster stimulation, rest for 8 seconds, for a total of 600 pulses, lasting 190 seconds. The stimulation pulse delivery pattern is as follows: Figure 2 As shown; each intervention was spaced 15 minutes apart, and a total of three intermittent theta burst magnetic stimulation sequences were performed, with each subject receiving 1800 pulses unilaterally. Intermittent theta burst magnetic stimulation is an optimized repetitive transcranial magnetic stimulation modality.

[0064] S11. Before and after the stimulation intervention, the EEG signal acquisition equipment used was the NeuSen W series 32-channel wireless EEG acquisition system from Neuracle. The electrodes were distributed according to the international standard 10-20 lead system, and the sampling frequency was 1000Hz. EEG data of the patient with eyes closed were collected in the same environment before and after 10 stimulations. Clinical scale scores of the patients were also collected before and after stimulation. The Unified Parkinson's Disease Rating Scale is the most comprehensive tool for assessing the clinical symptoms of Parkinson's patients. It includes four parts: assessment of mental and behavioral aspects (I), impact on activities of daily living (II), motor assessment (III), and complications (IV). Clinicians and researchers used the Unified Parkinson's Disease Rating Scale-III to clinically assess the motor function of Parkinson's patients. The assessments were conducted before the first stimulation and after the tenth stimulation.

[0065] S12. Preprocess the raw EEG data collected in step S11;

[0066] S13. Based on the preprocessed data obtained in step S12, calculate the global field potential of the EEG signal to be analyzed for each subject;

[0067] S14. Based on the EEG distribution pattern in step S13, the improved K-means clustering algorithm is used to perform global field potential time series clustering; the number of clusters is set to 2 to 8, the clustering process is repeated 100 times to ensure the stability of the results, and the maximum number of iterations for each clustering is set to 1000.

[0068] S15. Based on the clustering results of step S14, the optimal number of microstate templates is selected by evaluating the microstate fit using cross-validation criteria and the global explained variance between the original EEG of each group and each microstate category. In this example 1, the group-level analysis found that the optimal number of categories was 4, and it was assumed that the optimal number of categories for each subject was also 4. The microstate template topography map was obtained, see [link to example]. Figure 3 Microstate A shows the direction from the left occipital lobe to the right frontal lobe, microstate B shows the direction from the right occipital lobe to the left frontal lobe, microstate C represents the relatively symmetrical direction from the frontal lobe to the occipital lobe, and microstate D is characterized by the maximum value in the midfrontal lobe (the clustering results do not consider the influence of polarity). It can be observed that the microstates of types A, B, C, and D in PD patients before and after iTBS treatment show a high degree of consistency with the standard microstate topography.

[0069] S16. Import the optimal number of microstate templates determined in step S15 into the original EEG data and perform retrospective fitting to identify the specific location of each microstate in the time series and smooth the fitted microstate sequence to obtain the microstate time series.

[0070] S17. Based on the microstate time series obtained in step S16, calculate the four types of microstate parameters. The microstate parameter results in Example 1 are shown below. Figure 4 The results showed that the coverage of microstate A after intermittent theta burst magnetic stimulation (IPS) was significantly lower than before treatment in Parkinson's disease patients (p = 0.020); the duration (p = 0.008), coverage (p = 0.002), and incidence (p < 0.001) of microstate B after IPS were significantly lower than before treatment; the duration (p = 0.004), coverage (p < 0.001), and incidence (p = 0.013) of microstate C after IPS were significantly higher than before treatment; and the duration (p = 0.028), coverage (p = 0.010), and incidence (p = 0.007) of microstate D after IPS were significantly higher than before treatment.

[0071] Step 2: Construct a dynamic brain functional network using the microstate time series obtained in Step 1; then use the dynamic brain functional network to calculate the brain's steady-state coefficient to analyze the connectivity stability of the dynamic brain functional network.

[0072] S21. Using the microstate time series obtained in step 1 as the basis for time division, the EEG signal is divided into n non-overlapping microstate windows, and the EEG data corresponding to each window has a specific microstate class.

[0073] S22. Based on the microstate window EEG data obtained in step S21, the phase-locked value algorithm is used to construct a dynamic brain function network between the EEG channels of each subject.

[0074] S23. Calculate the brain homeostasis coefficient using the dynamic brain functional network obtained in step S22;

[0075] Example 1 statistically analyzed channels in Parkinson's disease patients before and after receiving intermittent theta burst stimulation, showing significant differences (p<0.05) in brain homeostasis coefficients at theta, alpha, beta, and gamma frequency bands. The results between channels are as follows: Figure 5As shown in the figure, the results indicated that the brain homeostasis coefficients of the theta band T3 (p = 0.040) and Cp5 (p = 0.005) channels significantly increased after treatment compared to before treatment; the brain homeostasis coefficient of the alpha band Cp5 (p = 0.040) channel significantly increased after treatment; the brain homeostasis coefficients of the beta band F8 (p = 0.047) and T3 (p = 0.037) channels significantly increased after treatment; and the brain homeostasis coefficient of the gamma band T3 (p = 0.020) channel significantly increased after treatment. These results suggest that some nodes in the brain functional network of each frequency band exhibit strong dynamic connectivity over time, resulting in improved stability.

[0076] S3. The microstate parameters obtained in step 1 and the brain homeostasis coefficient obtained in step 2 are tested against the clinical scale. When significant changes are detected in both the microstate parameters and the brain homeostasis coefficient, and the changes in the microstate parameters and the brain homeostasis coefficient are significantly correlated with the improvement of the clinical score, it is considered that this method has successfully revealed the electrophysiological mechanism of repetitive transcranial magnetic stimulation and can be used as a potential efficacy evaluation indicator.

[0077] The Unified Parkinson's Disease Rating Scale-III (UPARDS-III) is an important tool for clinically assessing patients' motor function and has high sensitivity and specificity in the diagnosis of Parkinson's disease. Therefore, in Example 1, the UPARDS-III score was used to quantify the motor ability of Parkinson's disease patients. The UPARDS-III assessment results showed that the UPARDS-III score was significantly lower after intermittent theta-burst magnetic stimulation (IPS) treatment compared to before stimulation (p<0.001), indicating improvement in patients' motor symptoms. A correlation analysis was performed between the ΔUPARDS-III score (before IPS-III treatment - after IPS-III treatment) and changes in microstate parameters before and after IPS-III treatment in Parkinson's disease patients to assess the relationship between changes in microstate parameters and the improvement of motor impairment in Parkinson's disease patients. The results are as follows: Figure 6As shown, Δ represents the result of parameters before and after intermittent theta-burst magnetic stimulation (IPS) treatment, and significant correlations (p<0.05) are marked with dashed boxes. The results indicate that the duration of Δ in microstate B (r=0.479, p=0.028), Δ coverage (r=0.549, p=0.010), and Δ incidence (r=0.567, p=0.007) are positively correlated with Δ in the Unified Parkinson's Disease Rating Scale-III (UPARS-III). That is, the greater the reduction in microstate B characteristic parameters, the greater the reduction in UPARS-III, and the greater the improvement in patients' motor symptoms. Microstate B is closely related to the functional integrity of the visual network. Previous studies have shown that gait freezing in PD patients is often accompanied by visual network dysfunction, and the microstate B network of Parkinson's disease patients exhibits higher spatiotemporal variability than that of healthy controls. Therefore, this invention infers that intermittent theta burst magnetic stimulation may alleviate compensatory abnormalities in the visual network of Parkinson's disease patients, improve their visual-motor integration function, and thus alleviate motor symptoms such as gait freezing. The Δ duration (r = -0.620, p = 0.003), Δ coverage (r = -0.75, p < 0.001), and Δ incidence (r = -0.685, p < 0.001) of microstate D are negatively correlated with the Δ Unified Parkinson's Disease Rating Scale-III, meaning that the greater the increase in microstate D characteristic parameters, the greater the decrease in the Unified Parkinson's Disease Rating Scale-III, and the greater the improvement in the patient's motor symptoms. Previous studies have shown that repetitive transcranial magnetic stimulation of the motor cortex in healthy subjects can induce contralateral striatal dopamine release, and the increase in dopamine levels enhances the activation of the dorsal attention network. Multiple studies have shown that increased activation of the dorsal attention network is related to compensatory mechanisms for impaired motor function: on the one hand, by enhancing attentional control to optimize gait and balance; on the other hand, by improving motor perception to compensate for control deficits. Furthermore, topological remodeling of the dorsal attention network in Parkinson's disease patients is correlated with motor function scores. Activation of the dorsal attention network manifests as an increase in microstate D parameters at the neurophysiological level, and dopaminergic drug therapy can improve motor symptoms of Parkinson's disease by regulating microstate D parameters. Therefore, the increase in microstate D parameters in Parkinson's disease patients after intermittent theta burst magnetic stimulation therapy in this invention may be a manifestation of this compensatory activation of the dorsal attention network. Microstate D abnormalities are also seen in other dopamine system disorders (such as attention deficit hyperactivity disorder and schizophrenia). This invention hypothesizes that intermittent theta burst magnetic stimulation may increase microstate D parameters by regulating dopamine secretion in Parkinson's disease patients, thereby increasing the activation of the dorsal attention network and compensating for motor function impairment in Parkinson's disease patients, thus improving their motor function.

[0078] Subsequently, a correlation analysis was performed on the ΔUnified Parkinson's Disease Rating Scale-III (before-after intermittent theta-burst magnetic stimulation therapy) and the ΔBrain Homeostasis Coefficient (before-after intermittent theta-burst magnetic stimulation therapy) of the significantly changed channels in each frequency band of Parkinson's disease patients before and after intermittent theta-burst magnetic stimulation therapy. This was to assess the relationship between changes in brain functional networks in each frequency band and the improvement of motor disorders in Parkinson's disease patients. The results are as follows: Figure 7 As shown, significant correlations (p<0.05) are marked with dashed boxes. The results indicate that the Δ brain homeostasis coefficient of the F8 channel in the beta band is negatively correlated with the Δ Unified Parkinson's Disease Rating Scale-III (r=-0.513, p=0.017), meaning that the greater the increase in connectivity stability between the F8 channel and other channels in the beta band, the greater the decrease in the Unified Parkinson's Disease Rating Scale-III, and the greater the improvement in the patient's motor symptoms. Previous studies have shown that Parkinson's disease patients exhibit abnormal beta band oscillations and abnormal brain functional network connectivity. The frontal lobe plays a crucial role in motor function, particularly in voluntary motor control and preparation, and the prefrontal cortex can guide the output of multiple motor centers. In Parkinson's disease patients, the prefrontal cortex exhibits abnormal activation patterns during motor tasks. Combined with evidence that intermittent theta burst magnetic stimulation can induce neural plasticity by modulating cortical excitability, this paper hypothesizes that intermittent theta burst magnetic stimulation enhances the higher-level regulatory role of the prefrontal cortex in motor-related networks by strengthening the connectivity stability between the beta-band frontal lobe and different brain regions, thereby improving motor function in Parkinson's disease patients.

[0079] Any aspects not covered in this invention are applicable to existing technologies.

Claims

1. A method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation, characterized in that, The method includes the following steps: Step 1: Perform repetitive transcranial magnetic stimulation on the patient, collect patient data, extract the microstate time series from the patient's electroencephalogram (EEG) signal, and calculate the microstate parameters before and after repetitive transcranial magnetic stimulation. Step 2: Construct a dynamic brain functional network using the microstate time series obtained in Step 1; then use the dynamic brain functional network to calculate the brain's steady-state coefficient to analyze the connectivity stability of the dynamic brain functional network. Step 3: Test the microstate parameters obtained in Step 1 and the brain homeostasis coefficient obtained in Step 2 against the clinical scale; the analysis is complete when significant changes are detected in both the microstate parameters and the brain homeostasis coefficient, and the changes in the microstate parameters and the brain homeostasis coefficient are significantly correlated with the improvement of the clinical score.

2. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 1, characterized in that, Step 1 specifically includes the following steps: S11. Collect raw EEG data and clinical scale scores of patients before and after receiving repetitive transcranial magnetic stimulation as the raw data basis for subsequent analysis. S12. Preprocessing of raw EEG data: First, bandpass filtering and notch filtering are performed on the raw EEG data. Then, independent component analysis is used to correct artifacts in blinking, horizontal eye movement, electromyography, and electrocardiogram. The sampling frequency of the data is reduced and converted to an average reference. According to the distribution frequency band of neural oscillation activity, the EEG signal is divided into theta, alpha, beta, and gamma. At the same time, the bandpass-filtered EEG data of 2-20Hz is retained for microstate analysis. S13. Based on the EEG data obtained in step S12, calculate the global field potential of the EEG signal to be analyzed for each subject. S14. Based on the candidate microstate distribution patterns extracted in step S13, the improved K-means clustering algorithm is used to perform global field potential time series clustering. S15. Based on the clustering results of step S14, the optimal number of microstate templates is selected by evaluating the microstate fit using the cross-validation criterion and the global explained variance between the original EEG of each group and each microstate category. S16. Import the optimal number of microstate templates determined in step S15 into the original EEG data and perform retrospective fitting to identify the specific location of each microstate in the time series and smooth the fitted microstate sequence to obtain the microstate time series. S17. Calculate the microstate parameters using the microstate time series obtained in step S16.

3. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 2, characterized in that, In step S12, the data sampling frequency is reduced to 250Hz.

4. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 2, characterized in that, In step S13, when the global field potential reaches a local peak, the corresponding EEG distribution pattern is considered a representative pattern of the microstate: In equation (1), GFP represents the global field potential; N is the number of electrodes, u i Let be the voltage value of the i-th electrode. This represents the average voltage value across all electrodes.

5. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 2, characterized in that, In step S17, the micro-state parameters are global explained variance, duration, coverage, and occurrence.

6. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 1, characterized in that, Step 2 specifically includes the following steps: S21. Using the microstate time series obtained in step 1 as the basis for time division, the EEG signal is divided into n non-overlapping microstate windows, and the EEG data corresponding to each window has a specific microstate class. S22. Based on the microstate window EEG data obtained in step S21, the phase-locked value algorithm is used to construct a dynamic brain function network between the EEG channels of each subject. S23. Calculate the brain homeostasis coefficient using the dynamic brain function network obtained in step S22.

7. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 6, characterized in that, In step S21, the expression for the phase-locked loop (PLL) algorithm is: Δφ(t)=φ x (t)-φ y (t)(2) In equation (2), t represents the time point, N represents the number of sample points of the signal, and Δφ(t) represents the phase difference between the two electrodes at time t. The phase-locked value coefficient between each pair of leads is calculated in each window, and the phase-locked value coefficient is used as the weight of the edge in the brain functional network between each electrode pair to obtain a dynamic brain functional network symmetrical about the main diagonal.

8. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 6, characterized in that, Step S23 specifically includes the following steps: S231. Using the dynamic brain function network obtained in step S22, extract the connection columns corresponding to electrodes at specific locations in the brain for each time window; since the phase-locked value of the same channel is always 1, the autocorrelation on the diagonal of the matrix is ​​removed. S232. Using the connection columns extracted in step S231 as input, based on the time series arrangement relationship, it is assumed that the network connectivity at any given time point is only related to the previous time point and is not related to other time points; then the Pearson correlation coefficient between the connection columns of adjacent time points is calculated, and the average of the correlation coefficients of all time points is taken to obtain the brain homeostasis coefficient.

9. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 8, characterized in that, In step S232, the expression for calculating the brain's homeostasis coefficient is: In equation (3), W p (i,:) represents the column of the phase-locked loop (PLL) value matrix after removing autocorrelation from channel i, and corr represents the Pearson correlation between the two variables; the formula for calculating the Pearson correlation coefficient is: In equation (4), cov represents the covariance of the two variables, δ X δ represents the standard deviation of variable X. Y It represents the standard deviation of variable Y.

10. The method for dynamic brain functional network analysis after repetitive transcranial magnetic stimulation according to claim 1, characterized in that, In step 3, a significant change is defined as p < 0.05, and a significant correlation is defined as p < 0.05.

Images