A Study on the Effects of Repetitive Transcranial Magnetic Stimulation on EEG Microstate in Patients With Parkinson’s Disease

Repetitive transcranial magnetic stimulation (rTMS) is a non-invasive neuromodulation technology that can modulate cerebral cortical excitability. Electroencephalography (EEG) microstate analysis is an important tool for studying dynamic changes in brain functional activity. This study explores the pathophysiological changes in Parkinson’s disease (PD) patients by analyzing the EEG microstate of PD patients, and analyzes the impact of rTMS on the clinical symptoms of PD patients. In a trial, 25 patients with PD and 18 healthy subjects of the same age were included. The clinical scale (the third part of Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale (UPDRS-III) and Montreal Cognitive Assessment (MoCA)) scores of each patient were evaluated and the microstate characteristic parameters of all subjects were calculated. 10 Hz rTMS was used to stimulate the bilateral primary motor cortex (M1) of PD patients. After two weeks of treatment (10 times), the clinical scale score of each patient was re-evaluated and the microstate characteristic parameters were calculated. At the baseline, the occurrence, duration and coverage of microstate C in PD patients were significantly higher than those in healthy controls (P <0.05),and were significantly negatively correlated with the MoCA score (P <0.05). The duration and coverage of microstate D in PD patients were significantly lower than those in healthy controls (P <0.05), and were significantly negatively correlated with UPDRS-III score (P <0.05). After rTMS treatment in the PD group, the scale score of UPDRS-III was significantly reduced (P <0.05) and the scale score of MoCA was significantly increased. Moreover, the occurrence and coverage of microstate B were significantly increased (p <0.05). The occurrence, duration and coverage of microstate C were significantly reduced (P <0.05). The occurrence, duration and coverage of microstate D were significantly increased (P <0.05). This study shows that abnormal brain functional activity of PD patients can change microstate characteristic parameters, and these changes are significantly related to the decline of motor and cognitive functions. Furthermore, rTMS can improve the motor and cognitive functions and adjust the microstate characteristic parameters of PD patients. EEG microstate analysis can reflect the therapeutic effect of rTMS on PD patients.


I. INTRODUCTION
P ARKINSON'S disease (PD) is a neurodegenerative dis- ease mainly related to the progressive degeneration of dopaminergic neurons in the substantia nigra [1], [2].The clinical manifestations of PD are mainly motor symptoms such as resting tremor, myotonia, bradykinesia, and postural balance disorder [3], [4], [5].Some patients are accompanied by non-motor symptoms such as cognitive impairment, pain, sleep disorders, and depression [6], [7].As the disease progresses, the clinical symptoms of the patients will gradually worsen, seriously affecting their quality of life.PD is mostly treated with dopamine drugs, which can increase the neurotransmitter levels of patients and improve clinical symptoms in a short period of time.However, they cannot prevent the clinical process of neurodegenerative changes.Therefore, it is necessary to find an auxiliary rehabilitation treatment method to cooperate with drug treatment and delay the progression of the disease.
Repetitive transcranial magnetic stimulation (rTMS) is a non-invasive neural intervention technology [8], [9] based on the faraday law of electromagnetic induction.The induced magnetic field of the stimulation coil causes the cerebral cortex to generate induced current, changing the brain cortical excitability, thus causing a series of changes in the physiological and biochemical environment in the brain [10], [11].rTMS is non-invasive, safe, and easy to operate.Relevant clinical studies have researched that rTMS is an effective treatment method for neuropsychiatric diseases [12].
With in-depth research in the field of brain science, various brain imaging technologies have developed rapidly.
Electroencephalography (EEG), as a non-invasive means of detecting neuroelectrophysiological activity in the cerebral cortex, can record potential change signals at different functional locations in the cerebral cortex.It has the characteristics of low cost, simple operation, high time resolution, and rich frequency band information [13], [14].The wide application of EEG technology has derived many processing algorithms.Compared with other EEG processing algorithms, microstate analysis can reflect changes in brain dynamics on the sub-second time scale.Microstate analysis simplifies the EEG signal into a series of scalp voltage topography.The instantaneous topology of the topography always remains in a relatively stable state within 60-120 ms, and then transitions to another state that remains relatively stable for a certain period of time [15], [16].Since EEG topography can reflect changes in the overall activity of the brain, these EEG topographies constitute the basic steps for the brain to process information, called "thinking atoms", which are expressed as functional microstates [17].In most of the current research results, four microstate categories A, B, C, and D have been repeatedly identified, and these four microstate topography maps can explain most of the global changes [18], [19].Relevant studies have shown that the time series of EEG microstates changes with behavioral states, personality types, and neuropsychiatric disorders [20].Therefore, feature parameters of microstates can be used to quantify brain pathophysiological changes in various neurological and psychiatric diseases.Several studies have also found that microstate categories are related to known resting state networks defined by functional magnetic resonance imaging (fMRI).The microstate A is related to the speech processing network, B is related to the visual network, C is related to the salience network, and D is related to the attention network [21].This finding could allow microstates analysis to be used to assess complex brain functional networks that are damaged in neurological and psychiatric diseases.In view of the above research results, microstate analysis can be applied to monitor the impact of rTMS on electrophysiological changes in PD patients.Therefore, this study uses the method of microstate analysis and clinical scales to evaluate changes in brain functional activity in PD patients.Furthermore, the study applies 10 Hz rTMS to the primary motor cortex (M1) of PD patients to research the regulatory effect of 10 Hz rTMS on the neural electrical activity of the brains of PD patients.

A. Subjects
This study included 25 patients with PD and 18 age-matched healthy subjects.The rules of inclusion criteria for PD patients are as follows: 1) The subjects had no other neurological and mental diseases.2) The subjects can able to maintain a sitting position for more than 30 min, with stable vital signs.3) The subjects had no communication barriers, good compliance and intellectual function is normal.4) The subjects had no changes in drug type or dosage during treatment.The rules of exclusion criteria: 1) The subjects had a history of stroke-like cerebrovascular accidents, craniocerebral injury, or other conditions that affect muscle tone.2) The subjects had a history of epilepsy, malignant tumors, severe heart disease, or cardiopulmonary disease.3) The subjects had a tendency for intracranial hemorrhage, intracranial bleeding, or other conditions that affect muscle tone.Internal high pressure or skull defect.4) There were metal objects implanted in the treatment area or implanted electronic devices in the body of subjects.5)The subjects were accompanied by unstable vital signs such as fever and electrolyte disorders.6) The subjects had poor compliance and inability to complete the experiment.
All subjects were right-handed, and patients and their families gave written informed consent for treatment.This study was approved by the Biomedical Ethics Committee of Hebei University of Technology, and all procedures followed the Declaration of Helsinki.

B. Research Design
Resting-state EEG data were collected from all subjects, and clinicians conducted detailed motor and neuropsychiatric assessments of PD patients.This study used the third part of Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale (UPDRS-III) to evaluate motor function in patients with PD, and used the Montreal Cognitive Assessment (MoCA) to evaluate multidomain cognitive function in patients with PD.After a total of 10 times of 10 Hz rTMS treatment in the bilateral M1 area of PD patients for two weeks, the PD patients were re-collected for resting-state EEG data collection and clinical scale assessment.

C. rTMS Protocol
The Rapid 2 magnetic stimulator (Magstim company, UK) was used to perform rTMS.The stimulation coil is an "8"shaped coil, and the stimulation target area is the bilateral M1 area of each patient.The stimulation target point is determined according to the location where at least 5 of the 10 consecutive single pulse outputs have an amplitude of myoelectric greater than or equal to 50 µV.The minimum value of the single pulse output is used as the motor threshold.The coil was placed closely to the patient's target point and tangential to the sagittal suture at 45 • .The stimulation intensity is 90% of the motor threshold.The pulse output mode is shown in Fig. 1.The pulse frequency is 10 Hz and the stimulation time is 4 s.The interval is 26 s and the stimulation is continuous for 30 cycles.The lasting time is 15 min on one side and 30 min on both sides in total.The patient wears earplugs at all the time, and the room remains quiet and eliminates external distractions.Treatment is once a day, for a total of 10 treatments in two weeks.

D. EEG Data Collection
The EEG signal acquisition equipment used in this study is the NeuSenW series 32-lead wireless EEG acquisition system (Neuracle Company).The electrodes were placed according to the international standard lead 10-20 system.Before the collection of EEG signal, the impedance of each electrode was reduced to below 10 K , and the bilateral mastoids were selected as the reference electrode placement points.The sampling frequency is 1000 Hz.The EEG collection process is carried out in a quiet and comfortable room, and the patients wear earplugs to avoid interference from external environments such as temperature and noise.Before the treatment of rTMS, the patients collect EEG signal in a resting eyesclosed state.Before collecting EEG, the patients are reminded to stay relaxed and minimize unnecessary movements.After receiving 10 treatments, the patients collect EEG signal in the same state again.Each collection time is 6 min.

E. EEG Data Preprocessing
The EEG signal is a very weak signal with a large background noise and is susceptible to interference from the external environment.The signal-to-noise ratio of the original EEG signal collected in the experiment is low, and all EEG data need to be preprocessed using the EEGLAB toolbox.Independent component analysis (ICA) is used to remove noise interference in the original signal [22], the EEG signal is down-sampled at 500 Hz, band-pass filtered at 2-20 Hz, and average reference.

F. Microstate Analysis of EEG Data
Fig. 2 shows the EEG microstate analysis process.The first step is to calculate the global field power (GFP) of the EEG signal, which is the standard deviation of all electrode voltages at a certain time: where N is the number of electrodes, u i is the voltage value of the i-th electrode, and ū is the average voltage of all electrodes.The EEG topography at the maximum point of the GFP curve was selected for clustering due to the high signalto-noise ratio of such topography [23], [24].In the second step, the selected EEG maps were clustered by modified K-means clustering algorithm [25].The number of clusters was set to 2-8, and 100 replicate runs were performed for each value of the number of clusters to minimize inter-run error.Crossvalidation criteria were used to determine the optimal number of microstate clusters [26].The optimal microstate category was determined before PD treatment, after PD treatment, and within healthy controls.In the third step, the global map dissimilarity (GMD) between each raw EEG topography and each microstate category in each group was calculated [23]: where N is the number of electrodes, u i is the voltage value of the i-th electrode, v i is the voltage value of the i-th electrode, ū is the average voltage of all electrodes of topographic map u, and v is the average voltage of all electrodes of topographic map v.Each raw EEG topography in each group was fitted back to the microstate category with the smallest value of its GMD.Through the above steps, the raw EEG data becomes a sequence of several microstate categories that appear alternately.Through the sequence, this study calculated the four state parameter [27], [28], respectively is: 1) Global explained variance (GEV): represents the accuracies of all types of topographic maps in the microstate to explain all topographic maps.
2) Occurrence: denotes the number of occurrences of various microstates in 1 s.
3) Duration: represents the average duration of each microstate.
4) Coverage: represents the proportion of various microstates in the overall analysis period.

G. Statistical Analysis
The statistical analysis was performed using SPSS [29] 26.0 software.The Shapiro-Wilk normality test was used for normality testing.Parametric testing was used for normally distributed data, and non-parametric testing was used for non-normal parameters.Independent sample T test and Mann-Whitney test were used to compare the microstate characteristic parameters of the PD group before treatment with the healthy control group, and paired sample T test and Wilcoxon signed rank test were used to compare the microstate characteristic parameters of the PD group before and after treatment.The correlation test used Spearman's correlation test.Topological analysis of variance (TANOVA) [30] was used to analyze the microstate topographies.The significance level of the above tests is set to P < 0.05.

III. RESULT A. Demographic Characteristics and Clinical Scale Scores
As shown in Table I, there were no significant differences in age, gender, and education level between PD patients and healthy subjects (P > 0.05).Compared with baseline, UPDRS-III scores and MoCA scores of PD patients improved significantly after treatment (P < 0.05).

B. Microstate Topographies
Fig. 3 shows the microstate topography of the PD group before treatment, the PD group after treatment, and the healthy control group.The microstate clustering results are all four categories: A, B, C, and D (the clustering results ignore the influence of polarity).The spatial configuration of the group microstate category corresponds to the results of healthy subjects in previous studies [31], [32], that is, microstate A shows the direction from the left occipital lobe to the right frontal lobe, and microstate B shows the direction from the right occipital lobe to the left frontal lobe, microstate C represents the relatively symmetric frontal to occipital direction, while microstate D is characterized by the midfrontal maximum.

TABLE I DEMOGRAPHIC DATA AND CLINICAL SCALE SCORE CHANGES OF PD PATIENTS AND HEALTHY CONTROLS (MEAN ± SD).
* (P < 0.05), * * (P < 0.01), * * * (P < 0.001) Pairwise comparisons of the three sets of microstate topographies using TANOVA showed no significant differences in any of the microstate categories (P > 0.05).

C. Microstate Characteristic Parameter Results
The microstate parameters of the PD group before and after treatment and the healthy control group are shown in Table II and Fig. 4.There was no significant difference in the characteristic parameters of microstate A between the PD group and the healthy control group (P > 0.05), and there was still no significant change after rTMS treatment (P > 0.05).There was no significant difference in the characteristic parameters of microstate B between the PD group and the healthy control group (P > 0.05).After receiving rTMS treatment, the occurrence and coverage of microstate B increased significantly (P < 0.05).The occurrence, duration and coverage rate of microstate C characteristic parameters in the PD group were significantly higher than those in the healthy control group (P < 0.05).All three parameters were significantly reduced after rTMS treatment (P < 0.05).The duration and coverage rate of microstate D characteristic parameters in the PD group were significantly lower than those in the healthy control group (P < 0.05), and all three parameters were significantly increased after rTMS treatment (P < 0.05).but we found that the microstate of PD patients The C characteristic parameters were significantly higher than those in healthy controls (Fig. 4), indicating abnormal cognitive function in PD patients.The duration and coverage rate of microstate D were significantly negatively correlated with UPDRS-III (P < 0.05), indicating that the more normal the motor function of PD patients, the higher the duration and coverage rate of microstate D. However, it can be seen that the microstate D characteristic parameters of PD patients are significantly lower than those of healthy controls (Fig. 4), indicating abnormal motor function in PD patients.There was no significant correlation between other microstate characteristic parameters and UPDRS-III and MoCA scores (P > 0.05).

D. Correlation Between Microstate Characteristic Parameters and Clinical Scales
IV. DISCUSSION As a non-invasive, convenient and effective neuromodulation technology, rTMS has a good auxiliary role in the treatment of neuropsychiatric diseases.This study used EEG microstate analysis to study brain dynamics changes in PD patients before and after rTMS treatment and healthy controls subjects and evaluated the temporal characteristics of brain activity on the subsecondtime scale.In addition, this article Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.also explores the relationship between microstate characteristic parameters and clinical scale scores in PD patients.
Previous fMRI studies in healthy subjects found close connections between resting-state networks (RSNs) and microstates.Microstate A is associated with negative blood oxygen level-dependent (BOLD) activation in the bilateral parietal and middle lobes, two regions responsible for speech processing [33].Microstate B is strongly associated with the functional integrity of the visual network and is related to negative BOLD activation in the bilateral occipital cortex [34].Microstate C is associated with salience networks responsible for emotional communication and is associated with positive BOLD activation in the dorsal anterior cingulate cortex, bilateral inferior frontal cortex, and right insular regions [35].Microstate D is associated with negative BOLD activation in the right dorsal and ventral areas of the frontal and parietal cortex, considered an attentional network [36].
We found no significant difference in microstate B characteristic parameters between PD group and healthy control group before rTMS treatment.PD patients are characterized by motor symptoms in the early stage of the disease, and non-motor symptoms occur with the extension of the course of the disease.Microstate B is closely related to the function of the visual network.In this study, there are individual differences in the course of PD patients, and the impairment of visual network function in some patients is mild or not obvious, which may be the reason why there is no significant difference in microstate B characteristic parameters between PD group and healthy control group before rTMS treatment.After rTMS treatment, the occurrence and coverage of microstate B in PD patients increased significantly, the MoCA scores of PD patients increased significantly.There is a positive correlation trend between microstate B characteristic parameters and MoCA score, indicating that the higher the microstate B characteristic parameters are, the better the cognitive function of PD patients.Visual network function is closely related to the correct acquisition of visual information, and there is a positive correlation between the correct acquisition of visual information and cognitive function [37].Previous studies have shown that the loss of dopamine neurotransmitters in PD patients may cause cognitive fatigue [38].rTMS stimulation can increase dopamine release in normal healthy [39], and occurrence of microstate B is significantly increased in PD patients after taking dopaminergic drugs [40].We infer that dopamine secretion is increased in PD patients after rTMS treatment.Dopamine activates the visual network generator in PD patients, and the cognitive process time is shortened [41], [42].The cognitive functionin patients with impaired visual network improved, and the visual impaired cognitive functionin patients with lighter network get promoted.Therefore, the frequency and coverage of microstate B in PD group were significantly increased after rTMS treatment.More research is needed to explore the correlation between microstate B and clinical symptoms of PD patients.
The results of this study show that occurrence, duration and coverage of microstate C in PD patients are significantly higher than those in healthy controls.The duration and coverage of microstate C are significantly negatively correlated with the MoCA score, that is, the higher the duration and coverage of microstate C in PD patients, the worse the cognitive domain-related functions.Previous research has shown that microstate C characteristic parameters can reflect the cognitive level of PD patients [43].Microstate C was associated with positive BOLD activation in the posterior part of the anterior cingulate, bilateral inferior frontal gyrus, right anterior insula, and left claustrum, brain regions associated with self-representation and executive control in areas of cognitive function functions are closely related [44].Therefore, we infer that the increase in microstate C characteristic parameters reflects BOLD abnormalities in salience network brain areas, which in turn causes the decline of cognitive function in PD patients.After rTMS treatment, the three parameters of microstate C were significantly reduced, and the MoCA score was significantly increased.Therefore, we speculate that rTMS treatment can normalize BOLD activation in microstate C-related brain areas and improve the cognitive function of PD patients.Microstate C characteristic parameters can reflect changes in cognitive function in PD patients.
In this study, the duration and coverage of microstate D in PD patients were significantly lower than those in healthy controls.We also found that the duration and coverage of microstate D in PD patients were significantly negatively correlated with the UPDRS-III score, which means that the more severely impaired the motor function of PD patients, the lower the duration and coverage of microstate D. Several studies have shown that both PD patients and schizophrenia patients have reduced dopamine secretion [45].Occurrence and duration of microstate D are significantly reduced in patients with schizophrenia [46], [47], [48].The duration and coverage of microstate D in PD patients were significantly reduced [49], which is the same as the results of this study.Reduced dopamine secretion is the main cause of motor symptoms such as tremor and bradykinesia in PD patients [50].Combined with the significant negative correlation between the duration and coverage of microstate D and the UPDRS-III score in this study, we speculate that the reduction in microstate D characteristic parameters in PD patients is related to the reduction in dopamine secretion.After rTMS treatment, occurrence, duration and coverage of microstate D in PD patients were significantly increased, and the UPDRS-III score was significantly increased.Studies have shown that occurrence, duration and coverage of microstate D are significantly increased in PD patients treated with dopamine drugs [40].Therefore, we infer that the increase in microstate D parameters in PD patients after rTMS treatment may be related to the increase in dopamine secretion, which in turn improves the motor symptoms of PD patients.Microstate D can be used as a powerful indicator of changes in motor symptoms in PD patients.

V. CONCLUSION
This study still has certain limitations.First, we recruited 23 PD patients and 18 healthy subjects as the control group.
The number of subjects is still small, in the future, the sample size still needs to be increased to further explore the gender, age, and complications of the subjects.Whether demographic characteristics will cause changes in microstate characteristic parameters.Secondly, there are differences in the course of disease, initial symptoms, and drug types and dosages among different patients.We did not conduct classification research on PD patients in this study.Finally, we did not follow up PD patients after rTMS treatment and only studied the short-term effects of rTMS.
Resting-state EEG microstate analysis is based on high-temporal resolution EEG technology.Since it has been confirmed that there is a correlation between microstate categories and RSNs, microstate analysis can evaluate the dynamic changes in functional network activity in brains of each PD patient in a sub-second time scale.The results of this study show that the characteristic parameters of microstate C and microstate D in PD patients will change significantly compared with healthy subjects of the same age.Among them, microstate C is significantly related to the cognitive function of PD patients, and microstate D is significantly related to the motor function of PD patients.After rTMS treatment, the motor and cognitive functions of PD patients are significantly improved, and the characteristic parameters of microstate B, microstate C, and microstate D will change significantly.

Fig. 2 .
Fig. 2. Schematic diagram of the microstate analysis process.(A) GFP peak topography results of EEG data, (B) clustering results of four microstates using the modified K-means clustering algorithm, (C) results of fitting four microstate categories back into the original EEG data.

Fig. 5 and
Fig.5and Fig.6show the correlation (Spearman correlation) between UPDRS-III scores and MoCA scores in PD patients and different microstate parameters before rTMS Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.

Fig. 3 .
Fig. 3. Microstate clustering results of the PD group before and after treatment and the healthy control group.Pre-rTMS represents the PD group before rTMS treatment, Post-rTMS represents the PD group after rTMS treatment, and HC represents the healthy control group.

Fig. 4 .
Fig. 4. Differences in microstate characteristic parameters between the PD group before and after treatment and the healthy control group.Pre-rTMS represents the PD group before rTMS treatment, Post-rTMS represents the PD group after rTMS treatment, HC represents the healthy control group, GEV represents global explained variance, Occ represents occurrence, Dur represents duration, and Cov represents coverage.* (P < 0.05), * * (P < 0.01), * * * (P < 0.001).