Tms individualized stimulation method fusing point cloud deep learning and eeg signal

Abstract

The application discloses a TMS individualized stimulation method fusing point cloud deep learning and EEG signals, and comprises the following steps: obtaining individual point cloud data of a target head to be measured; processing the individual point cloud data by using a trained deep learning model to obtain an individualized target point position; wherein the deep learning model is obtained by training a preset deep learning network by using nuclear magnetic data and point cloud data of different individuals; obtaining EEG signals at the individualized target point position and a plurality of positions around the individualized target point position; processing and predicting the EEG signals to obtain a TMS trigger signal giving time, so as to perform individualized frequency stimulation on the target to be measured according to the TMS trigger signal giving time. According to the method, the individualized target point position can be accurately determined only by using individual point cloud data, the high cost of time and economy caused by collecting nuclear magnetic of a patient's head is reduced, and the method has universality and is suitable for positioning of a TMS individualized stimulation system.

Description

Technical Field

[0001] This invention belongs to the field of transcranial magnetic stimulation technology, specifically relating to a personalized TMS (Transcranial Magnetic Stimulation) stimulation method that integrates point cloud deep learning and EEG (Electroencephalogram) signals. Background Technology

[0002] Transcranial magnetic stimulation (TMS) is a painless, non-invasive method of brain stimulation that has been applied in clinical and research fields for nearly 30 years. The basic principle of TMS is to induce a time-varying induced electric field in the cerebral cortex through a time-varying magnetic field applied extracranially, thereby generating an induced current in the brain tissue. When the induced current exceeds the excitation threshold of the nerve tissue, it produces an effect similar to direct electrical stimulation, thus effectively stimulating the corresponding brain tissue. During TMS, a transcranial magnetic stimulator is connected to a stimulation coil, which is placed on the area of ​​the subject's head that needs stimulation. Based on the principle of electromagnetic induction, the pulsed magnetic field generated by the stimulation coil induces a current in the cerebral cortex, thereby stimulating the cortical nerves and producing a series of physiological and biochemical reactions.

[0003] During TMS administration, the coil needs to be positioned at a specific point on the patient's scalp; this point is called the target point. The precise selection of the target point has a significant impact on the effectiveness of TMS administration.

[0004] Currently, existing TMS systems primarily employ two methods for individualized target selection. One method involves acquiring Magnetic Resonance Imaging (MRI) images of the patient's head, which can accurately determine the target location. However, this method requires prior MRI image acquisition and preprocessing for target localization, leading to high costs. The other method utilizes a 10-10 or 10-20 system EEG positioning cap, directly using the pre-marked target area. The advantage of this method is that the cap is reusable, making it cheaper and simpler to operate compared to using individual MRI images for navigation and localization. However, due to individual head variations, errors may occur, resulting in lower positioning accuracy.

[0005] Therefore, there is an objective need to conduct research on TMS stimulation technology that can accurately determine the target location and reduce the high cost of data collection. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, this invention provides a TMS personalized stimulation method that integrates point cloud deep learning and EEG signals. The technical problem to be solved by this invention is achieved through the following technical solution:

[0007] In a first aspect, this invention proposes a TMS personalized stimulation method that integrates point cloud deep learning and EEG signals, comprising:

[0008] Acquire individual point cloud data of the head of the target object under test;

[0009] The individual point cloud data is processed using a trained deep learning model to obtain individualized target point locations; wherein, the deep learning model is obtained by training a preset deep learning network using NMR data and point cloud data of different individuals;

[0010] Acquire the EEG signals at the individualized target location and several surrounding locations;

[0011] The EEG signal is processed and predicted to obtain the TMS trigger signal time, so as to provide individualized frequency stimulation to the target under test according to the TMS trigger signal time.

[0012] Secondly, this invention proposes a TMS personalized stimulation system that integrates point cloud deep learning and EEG signals, comprising a target localization subsystem and a TMS personalized stimulation subsystem; wherein,

[0013] The target localization subsystem includes a point cloud data acquisition device and a deep learning module;

[0014] The point cloud data acquisition device is used to acquire individual point cloud data of the head of the target to be tested;

[0015] The deep learning module is used to process the individual point cloud data using a trained deep learning model to obtain individualized target locations; wherein, the deep learning model is obtained by training a preset deep learning network using MRI data and point cloud data of different individuals.

[0016] The TMS personalized stimulation subsystem includes electrodes, a data processing module, and a stimulation device.

[0017] The electrode is used to acquire EEG signals at the individualized target location and several surrounding locations.

[0018] The data processing module is used to process and predict the EEG signal to obtain the TMS trigger signal time.

[0019] The stimulation device is used to provide individualized frequency stimulation to the target under test based on the time given by the TMS trigger signal.

[0020] The beneficial effects of this invention are:

[0021] 1. The TMS personalized stimulation method integrating point cloud deep learning and EEG signals provided by this invention trains a deep learning model by collecting a large amount of individual point cloud data and individual MRI data as training data. This model can describe the relationship between individual point clouds and target points relatively accurately. The accurate determination of individual target point locations can be achieved using only individual point cloud data, reducing the high time and economic costs caused by collecting patient head MRI. Furthermore, because a large amount of different individual data was used in the early training of the deep learning model, the deep learning model has universality and is suitable for the localization of TMS personalized stimulation systems.

[0022] 2. The TMS personalized stimulation method that integrates point cloud deep learning and EEG signals provided by this invention also proposes a method that is compatible with signal analysis of different bands, and obtains accurate TMS trigger signal delivery time, making the analysis of brain signals more comprehensive.

[0023] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating a TMS personalized stimulation method that integrates point cloud deep learning and EEG signals, as provided in an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of the EEG electrode bonding position provided in an embodiment of the present invention;

[0026] Figure 3 This is a structural block diagram of the TMS personalized stimulation system that integrates point cloud deep learning and EEG signals provided in an embodiment of the present invention.

[0027] Figure 4 This is a flowchart of the TMS personalized stimulation system that integrates point cloud deep learning and EEG signals, provided in an embodiment of the present invention. Detailed Implementation

[0028] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0029] Example 1

[0030] Deep learning is a machine learning method that processes and analyzes large-scale data by simulating the workings of the human brain's neural networks. It uses neural network structures to learn and extract features from the data and is commonly used in image recognition, natural language processing, and prediction. The advantage of using deep learning is its ability to learn complex feature representations from raw data.

[0031] Point cloud-based deep learning methods are techniques for processing and analyzing point cloud data, which typically consists of a large number of discrete points, each containing location coordinates to represent objects or scenes in three-dimensional space. These methods are widely used in fields such as computer vision, robotics, and medical imaging.

[0032] In conjunction with the practical application of TMS personalized stimulation, and based on existing point cloud deep learning technology, this embodiment proposes to train a deep learning model using individual point cloud data and individual MRI data to obtain a relatively accurate relationship between individual point clouds and target points, thereby realizing a high-precision and low-cost target point determination method for application in TMS personalized stimulation.

[0033] Please see Figure 1 , Figure 1 This is a flowchart illustrating a TMS personalized stimulation method that integrates point cloud deep learning and EEG signals, as provided in an embodiment of the present invention. The TMS personalized stimulation method integrating point cloud deep learning and EEG signals provided in this embodiment specifically includes the following steps:

[0034] Step 1: Obtain individual point cloud data of the head of the target to be tested.

[0035] Specifically, individual point cloud data of the head of the target object can generally be obtained using devices such as a head positioning bracket, positioning probe, and binocular camera. The specific process is as follows:

[0036] First, a positioning bracket needs to be worn on the head of the target to be tested. The positioning bracket has three infrared reflective positioning balls.

[0037] Secondly, a positioning probe is needed for auxiliary positioning. It is handheld and has a four-branch structure, with infrared reflective beads at each branch. The lower branch of the positioning probe is needle-shaped, and the support is relatively long to indicate the selected point.

[0038] Next, a binocular camera is needed to track the infrared reflective spheres, and then 3D reconstruction is performed to obtain their 3D coordinates in the camera coordinate system. The camera coordinate system is set at the optical center of the left camera of the binocular camera, with the Z-axis coinciding with the optical axis and pointing towards the observed object, the Y-axis pointing vertically upward, and the X-axis perpendicular to both the Z-axis and Y-axis, pointing towards the right camera.

[0039] Next, two coordinate systems need to be constructed to describe the coordinate transformation relationship: one is called the head tracking coordinate system, which is set on the head positioning bracket and can move with the head; the other is called the probe coordinate system, which is set on the probe, with the X, Y, and Z directions the same as the head tracking coordinate system, and the origin at the center of the probe. Then, the three-dimensional coordinates of the probe tip in the probe coordinate system can be obtained. Then, through the stereo camera target tracking and 3D reconstruction process, the transformation relationship between the two coordinate systems is obtained, and the probe tip coordinates are transformed from the probe coordinate system to the head tracking coordinate system.

[0040] Finally, the probe can be moved to different positions on the head to obtain the coordinates of different positions on the head in the head tracking system, forming a point cloud dataset of the head.

[0041] For a detailed explanation of the process of obtaining individual point cloud data, please refer to existing related technologies. This embodiment will not go into detail here.

[0042] Step 2: Process individual point cloud data using the trained deep learning model to obtain individualized target locations; the deep learning model is trained on a pre-defined deep learning network using NMR data and point cloud data from different individuals.

[0043] First, the pre-defined deep learning network needs to be trained using MRI data from different individuals to obtain the deep learning model, specifically including:

[0044] 1) Obtain the MRI images of the heads of different individual objects and the corresponding first point cloud data.

[0045] It should be noted that a large number of MRI images and point cloud data of different individuals' heads need to be collected here for network training. The MRI images can be acquired using an MRI scanner, and the point cloud data can be acquired using the method described in step 1.

[0046] 2) Perform medical image processing and target localization on the MRI images to obtain target coordinates, and convert the target coordinates into second point cloud data to form label data.

[0047] Specifically, the medical image processing here mainly facilitates target localization. The target is selected from commonly used stimulation target coordinates and activation regions. After obtaining the target coordinates, they are converted into point cloud data, thus forming labeled data.

[0048] For a detailed explanation of the implementation process of medical image processing and coordinate transformation, please refer to existing related technologies.

[0049] 3) Using the first point cloud data and label data as the training dataset, train the pre-built deep learning network to obtain the trained deep learning model.

[0050] Optionally, as one implementation method, this embodiment can use an existing Unet network or an improved Unet-based network for training. In practical applications, other network structures can also be used, or a network structure can be designed according to actual needs.

[0051] For details regarding the specific training process of the network and the setting of related loss functions, please refer to the training process of the existing Unet network. This embodiment will not provide a detailed description here.

[0052] By training the network with the first point cloud data and the label data, a relatively accurate relationship between individual point clouds and target points can be trained.

[0053] Then, the individual point cloud data of the target head obtained in step 1 is input into the trained deep learning model to predict the target location.

[0054] It should be noted that by using a trained deep learning model to predict target points from the point cloud data of the individual to be tested, a target point location coordinate can be obtained. In this embodiment, this target point is referred to as an individualized target point.

[0055] As can be seen from the above process, the target prediction provided in this embodiment only requires MRI data during network training. In practical applications, it is not necessary to collect MRI scans of the patient's head; only individual point cloud data is needed to accurately determine the location of individualized target points, saving time and economic costs. Furthermore, since the deep learning model is built on a large amount of individual data, the constructed model has universality, avoiding errors caused by differences in the heads of different individuals.

[0056] Once the individualized target location is obtained, it can be used for the localization of the TMS individualized stimulation system.

[0057] Step 3: Obtain EEG signals at the individualized target location and several surrounding locations.

[0058] Optionally, as one implementation method, this embodiment can use EEG acquisition electrodes to acquire EEG signals at individualized target locations; at the same time, it can use several surrounding electrodes to acquire EEG signals at several locations on a ring with the individualized target location as the center and a preset distance as the radius.

[0059] Preferably, the sampling positions of the surrounding electrodes are evenly distributed along the circular ring.

[0060] It should be noted that the specific number and placement of the electrodes can be set according to the actual situation.

[0061] For example, see Figure 2 , Figure 2This is a schematic diagram of the EEG electrode attachment position provided in an embodiment of the present invention. First, an EEG acquisition electrode is attached at the location of the individualized target point, and then four electrodes, referred to as the surrounding electrodes, are attached around it.

[0062] The electrodes are then connected to an EEG signal acquisition device to acquire the EEG signals from these electrodes.

[0063] It should be noted that electrodes also need to be attached to the corner of the eye to collect eye point signals for subsequent removal of electrooculography artifacts.

[0064] Understandably, after acquiring the EEG signal, it is necessary to perform bandpass filtering to remove interference signals from the EEG signal for further processing.

[0065] Optionally, as one implementation method, in this embodiment, the frequency of the bandpass filter is generally taken as 0.1 to 40 Hz.

[0066] Step 4: Process and predict the EEG signal to obtain the TMS trigger signal time, so as to provide individualized frequency stimulation to the target based on the TMS trigger signal time.

[0067] Specifically, the EEG signal is processed and predicted to obtain the TMS trigger signal timing, including:

[0068] 41) Perform independent component analysis on the EEG signal to separate the signals of different channels and obtain independent EEG signals.

[0069] Understandably, due to mutual interference between the channels, it is necessary to perform independent component analysis on the EEG signals acquired by all electrodes in order to separate the signals of different channels and obtain independent EEG signals.

[0070] The specific implementation process of independent component analysis can refer to existing tubular technology. Through independent component analysis, the target EEG signal corresponding to the individualized target location, several surrounding EEG signals corresponding to several surrounding locations, and electrooculography signals can be obtained.

[0071] It should be noted that after obtaining the independent EEG signal, the following is also included:

[0072] Electrooculography artifact removal was performed on the independent EEG signals.

[0073] In this embodiment, removal of electrooculography (EOG) artifacts can be achieved by deleting EOG signals from independent EEG signals. EOG signals are obtained by attaching electrodes to the corner of the eye and using an EEG signal acquisition device.

[0074] 42) The difference between the target EEG signal and the average of several surrounding EEG signals is used to obtain the target EEG signal.

[0075] First, calculate the average value of several surrounding EEG signals; then, subtract the calculated average value from the target EEG signal to remove interference from surrounding signals and obtain the target EEG signal.

[0076] 43) Separate the desired band signal from the target EEG signal and construct an autoregressive Ulwalk model based on the band signal obtained over a period of time.

[0077] Specifically, firstly, the target EEG signal obtained above is bandpass filtered to separate the signal of the desired band from the EEG signal. This band can be a series of common EEG signal bands such as alpha wave, beta wave, theta wave, or delta wave; only one band is used here.

[0078] Then, after collecting the separated band signals for a period of time, the cluttered data before and after the acquisition is removed to reduce edge effects and obtain a valid data segment.

[0079] It should be noted that, in the embodiments, the time period for data collection can be set as needed. For example, a band signal within a 1000ms time period can be collected, and then the data within the 50ms time period before and after it can be removed to obtain a valid data segment.

[0080] Finally, an autoregressive Ulwalk model is constructed using this valid data, a process that can be achieved through relevant simulation software.

[0081] 44) Use the autoregressive Ulwalk model to predict data within a preset time period in the future to obtain predicted data.

[0082] It should be noted that this step can only be performed if the autoregressive Ulwalk model constructed in the previous step is effective. If the model is invalid, the target point needs to be redefined and the EEG signal needs to be collected again.

[0083] Specifically, the effectiveness of the model can be verified by calculating the mean square error of the results; the detailed process will not be described in detail here.

[0084] 45) Find the position of the band signal value corresponding to the instantaneous phase zero value from the prediction data, and analyze the data at that position to obtain the time given by the TMS trigger signal.

[0085] First, perform a Hilbert transform on the predicted data to calculate the instantaneous phase of the signal and find the location of the zero instantaneous phase value.

[0086] Experiments have shown that when the instantaneous phase of a signal is near zero, the separated band signal is at its peak. Therefore, this implementation first performs a Hilbert transform on the predicted data to obtain the instantaneous phase of the signal, thereby obtaining the position where the instantaneous phase is zero.

[0087] Then, the maximum value is taken for all band signals near the instantaneous phase zero position to obtain the peak position.

[0088] Specifically, the position of the band signal value near the instantaneous phase zero value is saved. This position is multiple and segmented continuously. The maximum value of each segment of the signal is taken to obtain the position of all peaks.

[0089] Finally, the data corresponding to the peak position is divided by the acquisition frequency of the EEG signal to convert it into time information, and the difference is calculated with the inherent delay when the TMS device is triggered to obtain the time when the TMS trigger signal is given.

[0090] By providing a trigger signal at this time, TMS stimulation can be performed at the signal peak, thus completing the process of individualized frequency stimulation at individualized locations.

[0091] The EEG signal analysis method proposed in this embodiment is compatible with different bands and obtains accurate TMS trigger signal delivery time, making the analysis of brain signals more comprehensive.

[0092] Through the above process, individualized frequency stimulation, i.e. closed-loop TMS stimulation, is achieved based on individualized target selection. Furthermore, the point cloud deep learning method reduces the step of acquiring MRI images before stimulation, while ensuring positioning accuracy and reducing costs.

[0093] The TMS personalized stimulation method provided by this invention, which integrates point cloud deep learning and EEG signals, trains a deep learning model by collecting a large amount of individual point cloud data and individual MRI data as training data. This model can describe the relationship between individual point clouds and target points with relatively accurate results. The accurate determination of individual target point locations can be achieved using only individual point cloud data, reducing the high time and economic costs associated with collecting patient head MRI data. Furthermore, because a large amount of different individual data was used in the early training of the deep learning model, the deep learning model has universality and is suitable for the localization of TMS personalized stimulation systems.

[0094] Example 2

[0095] Based on the first embodiment described above, and using the same inventive concept, this embodiment also provides a TMS personalized stimulation system that integrates point cloud deep learning and EEG signals. Please refer to... Figure 3 , Figure 3This is a structural block diagram of the TMS personalized stimulation system that integrates point cloud deep learning and EEG signals, provided in an embodiment of the present invention. It includes a target localization subsystem and a TMS personalized stimulation subsystem; wherein,

[0096] The target localization subsystem includes a point cloud data acquisition device and a deep learning module;

[0097] The point cloud data acquisition device is used to acquire individual point cloud data of the head of the target to be tested;

[0098] The deep learning module is used to process individual point cloud data using a trained deep learning model to obtain individualized target locations; the deep learning model is obtained by training a pre-set deep learning network using NMR data and point cloud data of different individuals.

[0099] The TMS personalized stimulation subsystem includes electrodes, a data processing module, and a stimulation device;

[0100] The electrodes are used to acquire EEG signals at individualized target locations and several surrounding locations.

[0101] The data processing module is used to process and predict EEG signals to obtain the timing of the TMS trigger signal.

[0102] The stimulation device is used to provide individualized frequency stimulation to the target based on the time given by the TMS trigger signal.

[0103] Specifically, the point cloud data acquisition device can consist of a positioning bracket, a positioning probe, a binocular camera, and related software modules for data processing. For detailed structure and usage methods, please refer to existing related technologies. The working principle of the deep learning module is explained in Embodiment 1 above and related existing technologies.

[0104] Conventional EEG electrodes can be used for the electrodes; the working principle of the data processing module can be found in the above embodiment 1; the specific structure and usage of the stimulation device can also be found in existing related technologies.

[0105] Please see Figure 4 , Figure 4 The following is a flowchart of the TMS personalized stimulation system that integrates point cloud deep learning and EEG signals, provided in an embodiment of the present invention:

[0106] First, a positioning bracket is fitted to the head of the target under test and a positioning probe is fixed. Then, a binocular camera is used to track the infrared reflective ball on the positioning bracket and the positioning probe in real time. The point cloud dataset of the target head is obtained through coordinate transformation, which is also known as individual point cloud data.

[0107] Then, the individual point cloud data is processed using a deep learning model to obtain the individualized target point location coordinates.

[0108] Next, EEG acquisition electrodes are attached to the individualized target site to acquire EEG signals. After the processor predicts the signal, the TMS trigger signal output time is obtained.

[0109] Finally, the obtained TMS trigger signal is used to give time for individualized frequency stimulation of the target under test.

[0110] The device provided in this embodiment can implement the method provided in Embodiment 1 above. Therefore, this device can also accurately determine the location of individualized target points using only individual point cloud data, reducing the high time and economic costs associated with collecting MRI scans of patients' heads, and has universal applicability.

[0111] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A TMS personalized stimulation system integrating point cloud deep learning and EEG signals, characterized in that, It includes a target localization subsystem and a TMS personalized stimulation subsystem; among which, The target localization subsystem includes a point cloud data acquisition device and a deep learning module; The point cloud data acquisition device is used to acquire individual point cloud data of the head of the target to be tested; The deep learning module is used to process the individual point cloud data using a trained deep learning model to obtain individualized target locations; wherein, the deep learning model is obtained by training a preset deep learning network using MRI data and point cloud data of different individuals. The TMS personalized stimulation subsystem includes electrodes, a data processing module, and a stimulation device. The electrode is used to acquire EEG signals at the individualized target location and several surrounding locations. The data processing module is used to process and predict the EEG signal to obtain the TMS trigger signal time. The stimulation device is used to provide individualized frequency stimulation to the target under test based on the time given by the TMS trigger signal. In the deep learning module, the deep learning model is trained in the following manner: Obtain MRI images of the heads of different individual objects and the corresponding first point cloud data; The MRI image is subjected to medical image processing and target localization to obtain target coordinates, and the target coordinates are converted into second point cloud data to form tag data; Using the first point cloud data and the label data as training datasets, a pre-constructed deep learning network is trained to obtain a trained deep learning model. The data processing module processes and predicts the EEG signal to obtain the TMS trigger signal timing; specifically, it includes: Independent component analysis is performed on the EEG signal to separate the signals of different channels and obtain independent EEG signals; wherein, the independent EEG signal includes the target EEG signal corresponding to the individualized target position and several surrounding EEG signals corresponding to several surrounding positions. The target EEG signal is obtained by subtracting the average value of the target EEG signal from the average value of the surrounding EEG signals. The desired band signals are separated from the target EEG signal, and an autoregressive Ulwalk model is constructed based on the band signals obtained over a period of time; wherein, the desired band signals include alpha waves, beta waves, theta waves, or delta waves; The autoregressive Ulwalk model is used to predict data within a preset time period in the future, and the predicted data is obtained. The location of the band signal value corresponding to the instantaneous phase zero value is found from the predicted data, and the data at that location is analyzed to obtain the TMS trigger signal output time.

2. The TMS personalized stimulation system integrating point cloud deep learning and EEG signals according to claim 1, characterized in that, The deep learning network includes the Unet network.

3. The TMS personalized stimulation system integrating point cloud deep learning and EEG signals according to claim 1, characterized in that, Acquiring the EEG signals at the individualized target location and several surrounding locations includes: The EEG signal at the individualized target location is acquired using EEG acquisition electrodes; simultaneously, EEG signals at several locations on a ring centered at the individualized target location and with a preset distance as the radius are acquired using several surrounding electrodes. The sampling positions of the surrounding electrodes are evenly distributed along the circular ring.

4. The TMS personalized stimulation system integrating point cloud deep learning and EEG signals according to claim 1, characterized in that, After obtaining the EEG signal, and before performing independent component analysis on the EEG signal, the process also includes: The EEG signal is bandpass filtered to remove interference signals.

5. The TMS personalized stimulation system integrating point cloud deep learning and EEG signals according to claim 1, characterized in that, After obtaining the independent EEG signal, it also includes: The independent EEG signals were subjected to electrooculography artifact removal processing.

6. The TMS personalized stimulation system integrating point cloud deep learning and EEG signals according to claim 1, characterized in that, After constructing the autoregressive Ulwalk model, the following is also included: The effectiveness of the autoregressive Ulwalk model was verified.

7. The TMS personalized stimulation system integrating point cloud deep learning and EEG signals according to claim 1, characterized in that, The location of the band signal value corresponding to the instantaneous phase zero value is found from the predicted data, and the data at that location is analyzed to obtain the TMS trigger signal output time, including: Perform a Hilbert transform on the predicted data to calculate the instantaneous phase of the signal and find the location of the zero instantaneous phase value; The peak position is obtained by taking the maximum value of all band signals near the instantaneous phase zero position; The data corresponding to the peak position is divided by the acquisition frequency of the EEG signal to convert it into time information, and the difference is calculated with the inherent delay when the TMS device is triggered to obtain the time when the TMS trigger signal is given.

Images