A transcranial magnetic stimulation electroencephalogram positioning method based on multi-feature fusion

Abstract

The application belongs to the technical field of transcranial magnetic stimulation, and particularly relates to a transcranial magnetic stimulation electroencephalogram positioning method based on multi-feature fusion; the method comprises the following steps: acquiring electroencephalogram data of a subject in real time during a magnetic stimulation process; intercepting and preprocessing the acquired electroencephalogram data, calculating power spectrum density, activity index and permutation entropy of each lead in each frequency band; fusing the above features, calculating comprehensive indexes of each lead; performing feature processing on the comprehensive indexes of each lead, and calculating final features of each lead; performing descending order sorting on the final features of each lead, and selecting three leads with high rankings; solving geometric position centers of the selected three leads, and obtaining position information of a magnetic stimulation point. In the transcranial magnetic stimulation scene, the application can determine the position of transcranial magnetic stimulation without the aid of MRI images, through real-time electroencephalogram feature analysis, so as to reduce positioning cost and improve timeliness.

Description

Technical Field

[0001] This invention belongs to the field of transcranial magnetic stimulation (TMS) technology, specifically relating to a transcranial magnetic stimulation EEG localization method based on multi-feature fusion. Background Technology

[0002] Transcranial magnetic stimulation (TMS) is a technique that uses pulsed magnetic fields applied to the brain to alter the membrane potential of cortical nerve cells, inducing currents that affect brain metabolism and neural electrical activity, thereby triggering a series of physiological and biochemical responses. Currently, TMS is demonstrating its value in neurology. With the development of TMS, the application of repetitive transcranial magnetic stimulation (rTMS), which features continuously adjustable repetitive stimulation, has become possible. This allows for long-term programmed stimulation that simulates the inhibitory and excitatory discharge patterns of brain neurons, enabling neuromodulation of the cerebral cortex. Furthermore, the application of TMS combined with electroencephalography (EEG) for brain network analysis to assess cortical network properties, excitability, and connectivity is emerging as a new direction in the diagnosis and treatment of neurological diseases. Research on TMS and personalized, precise neuromodulation has become an area urgently needing further development in neurology.

[0003] Currently, existing TMS positioning and navigation systems mainly rely on MRI imaging technology to visualize TMS, that is, to use MRI images to determine the brain functional activity areas of the subject. However, each stimulation requires MRI image localization assistance, and the method of using MRI images to determine the stimulation location of the subject has problems such as high cost and poor timeliness. Therefore, there is an urgent need for a method that does not rely on MRI images to determine the magnetic stimulation location. Summary of the Invention

[0004] Based on the problems existing in the current technology, this invention proposes to use TMS combined with EEG to assess the properties, excitability and connectivity of the cortical network. It also aims to evaluate the effect of TMS on cortical activity and the connectivity between motor cortical regions through a single TMS stimulation. By combining TMS and EEG technologies, a quantitative TMS-EEG-brain functional area localization system is constructed, which provides the possibility for real-time and dynamic assessment of the excitability, connectivity and plasticity of the cerebral cortex, and provides an objective basis for the pathogenesis and precise regulation of diseases.

[0005] The present invention provides a transcranial magnetic stimulation EEG localization method based on multi-feature fusion, the method comprising:

[0006] Real-time acquisition of subjects' electroencephalogram (EEG) data during magnetic stimulation;

[0007] The acquired EEG data is then extracted;

[0008] Preprocess the captured EEG data;

[0009] The power spectral density, activity index, and arrangement entropy of each lead were calculated according to the six frequency bands.

[0010] The power spectral density, activity index and arrangement entropy of each lead in each frequency band are fused to calculate the comprehensive index of each lead.

[0011] The comprehensive indicators of each lead are processed to obtain the final characteristics of each lead.

[0012] The final characteristics of each lead are sorted in descending order, and the top three leads are selected.

[0013] The geometric center of the three selected leads is solved to obtain the location information of the transcranial magnetic stimulation points.

[0014] The beneficial effects of this invention are:

[0015] This invention utilizes the combined application of TMS and high-frequency electroencephalography (EEG) technologies to construct a quantitative TMS-EEG-brain functional area localization system. This enables real-time, dynamic assessment of the excitability, connectivity, and plasticity of the cerebral cortex, providing objective evidence for disease pathogenesis and precise regulation. This invention fuses multiple features—power spectral density, activity indices, and permutation entropy—across various frequency bands in each lead, improving the accuracy of the comprehensive indicators for each lead. This facilitates the selection of more accurate lead features, leading to more objective lead location information. This provides clinicians with auxiliary decision-making information, helping them adjust the position and direction of magnetic stimulation in a timely manner, and providing reliable assurance for precise regulation of related diseases. Attached Figure Description

[0016] Figure 1 This is a flowchart of a transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to an embodiment of the present invention;

[0017] Figure 2 This is a flowchart of a transcranial magnetic stimulation EEG localization method based on multi-feature fusion, according to a preferred embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The existing EEG localization methods mentioned in the background section, which use MRI images to determine the stimulation location of the subject, suffer from problems such as high cost and poor timeliness. This embodiment provides a transcranial magnetic stimulation (TMS) EEG localization method based on multi-feature fusion. When stimulating the subject with a TMS device, the proposed solution can locate the position of the magnetic stimulation. The method is described below with reference to specific embodiments.

[0020] It should be noted that the execution subject of the EEG localization method based on multi-feature fusion transcranial magnetic stimulation points in this embodiment can be a device for EEG localization based on multi-feature fusion transcranial magnetic stimulation points. This device can be implemented by software and / or hardware and can be configured in an electronic device, which may include, but is not limited to, a terminal, a server, etc.

[0021] Figure 1 This is a flowchart of a transcranial magnetic stimulation EEG localization method based on multi-feature fusion, as described in this application. Figure 1 As shown, the method includes:

[0022] 101. Real-time acquisition of EEG data from subjects during magnetic stimulation;

[0023] In this embodiment of the invention, the subject needs to wear a transcranial magnetic stimulation (TMS) device. The magnetic stimulation device is placed on the brain region of the subject to be treated. The subject is in a calm state. The magnetic stimulation device is turned on and a single stimulation is given. After the stimulation is completed, the EEG signal acquisition device can acquire the subject's EEG signal in real time. By analyzing and processing the generated EEG signal, the accuracy of the TMS stimulation position can be accurately assessed in real time.

[0024] 102. Extract the acquired EEG data;

[0025] In this embodiment, the acquired EEG data can be processed in the following steps:

[0026] 1) Based on the duration of magnetic stimulation, set the data truncation length T to be consistent with the duration of magnetic stimulation.

[0027] 2) When the EEG acquisition device receives the start stimulation signal from the transcranial magnetic stimulation (rTMS) device, the current time is set as the start time t of the EEG analysis data. start .

[0028] 3) with t start The EEG data is extracted at the starting time and T is the truncation length.

[0029] In this embodiment, EEG data is captured starting at the moment the magnetic stimulation begins, minimizing interference from irrelevant signals and facilitating subsequent location analysis and processing.

[0030] 103. Preprocess the captured EEG data;

[0031] In this embodiment of the invention, the intercepted EEG data needs to be filtered to remove 50Hz power frequency interference, which facilitates subsequent location analysis and processing.

[0032] 104. Calculate the power spectral density, activity index, and arrangement entropy of each lead according to the six frequency bands;

[0033] To better process EEG signals, this invention divides the captured EEG signals into 6 frequency bands. By processing EEG signals in different frequency bands, the characteristics of EEG signals in different frequency bands can be accurately captured.

[0034] In this embodiment, the six frequency bands may sequentially include 0–4Hz, 4–8Hz, 8–12Hz, 12–32Hz, 32–80Hz, and 80–250Hz. The 0–250Hz band contains most of the information in the EEG data; therefore, this embodiment divides the 0–250Hz band into six frequency bands to accurately assess the characteristics of EEG signals in different frequency bands. In this embodiment, the calculation method for the power spectral density of each lead in each frequency band includes:

[0035] A Fourier transform is performed on an EEG signal sequence x(n) of length N to obtain X(f). The power spectral density across the entire frequency band is then calculated using the Fourier-transformed EEG frequency signal.

[0036]

[0037] Where PSD(f) represents the power spectral density of frequency band f, and K is the sampling rate.

[0038] Using the above method, and processing different leads and frequency bands separately, the power spectral density of lead i in six frequency bands—0–4Hz, 4–8Hz, 8–12Hz, 12–32Hz, 32–80Hz, and 80–250Hz—was calculated and denoted as P. i 1 P i 2 P i 3 P i 4 P i 5 P i 6'i' represents the lead number, and in this embodiment, the value of 'i' is [1, 2, ..., N]. t ], N t This indicates the number of leads, which can typically be 16, 32, 64, 128, etc.

[0039] In this embodiment of the invention, the calculation method for the activity index of each lead in each frequency band includes: constructing a data vector from preprocessed EEG data; constructing a time-domain multivariate autoregressive model from the data vector; converting the multivariate autoregressive model to the frequency domain and constructing a frequency-domain multivariate autoregressive model; determining the transfer matrix based on the frequency-domain multivariate autoregressive model; determining the ratio of inflow to outflow between leads based on the transfer matrix; and determining the lead activity index based on the net inflow and outflow values ​​of the leads. Specifically:

[0040] The Directed Transfer Function (DTF) is calculated as follows:

[0041] Multichannel EEG signals can be described as a data vector X of N source signals: X(t)={x1(t),x2(t),......,x N (t)}.

[0042] Constructing a multivariate autoregressive model:

[0043]

[0044] Where E(t) is a multivariate zero-mean uncorrelated white noise vector at time t, and A n This is an N×N matrix of model coefficients, where p is the model order. Then, the multivariate autoregressive model is converted to the frequency domain.

[0045] X(f)=A -1 (f)E(f)=H(f)E(f)

[0046] Where f represents a specific frequency, and the H(f) matrix is ​​the transfer matrix, defined as follows:

[0047]

[0048] Where I is an identity matrix.

[0049] DTF Transfer Matrix H ij The element is defined as follows:

[0050]

[0051] Where, γ ij (f) represents the ratio between the inflow from lead j to lead i and the total inflow to lead i, N t Indicates the number of leads.

[0052] The net inflow value of a specific lead is used as a measure of the activity level of that lead. The inflow value for lead i is the sum of the information flow from all other EEG leads except lead i, and the outflow value for lead i is the sum of the information flow from lead i to all other EEG leads except lead i. This represents the net inflow value of lead i, calculated as the difference between the inflow value and the outflow value of that lead. The net inflow value can be used to determine the direction of information flow, as follows:

[0053]

[0054] in, DTF is the activity index of lead i in frequency band f. ij (f) is the ratio of the inflow from lead j to lead i in frequency band f to the total inflow to lead i, DTF. ji (f) is the ratio of the inflow from lead i to lead j in frequency band f to the total inflow to lead j; N t Indicates the number of leads.

[0055] If the net inflow value of a lead is positive and relatively high, the lead is called an active node. This embodiment uses the Directional Transfer Function (DTF) to calculate the data flow direction and intensity between leads across the entire frequency band. Processing is performed separately for different leads and frequency bands, calculating the activity index of lead i in six frequency bands: 0–4Hz, 4–8Hz, 8–12Hz, 12–32Hz, 32–80Hz, and 80–250Hz, which are expressed as follows: D i 2 D i 3 D i 4 , i represents the lead number, and in this embodiment, the value of i is [1,2,…,N]. t ], N t This indicates the number of leads, which can typically be 16, 32, 64, 128, etc.

[0056] In this embodiment of the invention, the calculation method for the permutation entropy of each lead in each frequency band includes: reconstructing the phase space of the preprocessed EEG data to obtain a reconstruction matrix; sorting the reconstruction components corresponding to each row of the reconstruction matrix in ascending order to obtain the column index of each element position and forming a set of symbol sequences; determining the probability of a symbol sequence based on the ratio of the number of times each symbol sequence appears to the total number of times all different symbol sequences appear; and calculating the permutation entropy of the lead based on the probability of the symbol sequence appearing. Specifically:

[0057] Phase space reconstruction is performed on a time series X of length N to obtain the reconstruction matrix Y:

[0058]

[0059] Where m is the embedding dimension, t is the delay time, and K = N - (m-1)t. Each row in matrix Y is a reconstruction component, and there are a total of K reconstruction components.

[0060] Each reconstructed component is rearranged in ascending order to obtain a set of symbol sequences consisting of the column indices of each element in the vector.

[0061] S(l)={j1,j2,...,j m}, l=1,2,...,K, and K≤m!

[0062] Calculate the probability of each symbol sequence by dividing the number of occurrences of each sequence by the total number of occurrences of all m! different symbol sequences (i.e., {P1, P2, ..., P...}). K}).

[0063] The formula for calculating the permutation entropy of time series X is:

[0064]

[0065] Among them, P k Let represent the probability of the k-th symbol sequence appearing, and K represent the number of reconstruction components in the reconstruction matrix.

[0066] By processing each lead separately using the above method, the permutation entropy of lead i is calculated, denoted as H. i 'i' represents the lead number, and in this embodiment, the value of 'i' is [1, 2, ..., N]. t ], N t This indicates the number of leads, which can typically be 16, 32, 64, 128, etc.

[0067] 105. Perform feature fusion on the power spectral density, activity index and arrangement entropy of each lead in each frequency band to calculate the comprehensive index of each lead;

[0068] In this embodiment of the invention, the calculation method for the comprehensive index of each lead is expressed as follows:

[0069] K i =W1*P i 1 +W2*P i 2 +W3*P i 3 +W4*P i4 +W5*P i 5 +W6*P i 6 +W7*D i 1 +W8*D i 2 +W9*D i 3 +W 10 *D i 4 +W 11 *D i 5 +W 12 *D i 6 +W 13 *H i

[0070] Among them, K i P is a comprehensive index of lead i. i 1 ~P i 6 The power spectral densities of lead i in the frequency bands of 0–4 Hz, 4–8 Hz, 8–12 Hz, 12–32 Hz, 32–80 Hz, and 80–250 Hz are shown in order. The activity indicators for lead i in the frequency bands of 0–4 Hz, 4–8 Hz, 8–12 Hz, 12–32 Hz, 32–80 Hz, and 80–250 Hz are as follows: H i Let W1 be the permutation entropy of lead i. 13 These are the corresponding indicator weights, where the specific values ​​of these indicator weights can be specified by those skilled in the art or calculated using certain indicator weighting algorithms.

[0071] 106. Perform feature processing on the comprehensive indicators of each lead to calculate the final characteristics of each lead;

[0072] In this embodiment of the invention, the calculation method for the characteristics of each lead is expressed as follows:

[0073] T i =K i -(Sum i ) / M

[0074] Among them, T i For the characteristics of lead i, Sum i M is the sum of the comprehensive indicators of the adjacent nodes of lead i, and M is the number of neighbors of lead i.

[0075] 107. Sort the final characteristics of each lead in descending order and select the top three leads.

[0076] In this embodiment, by analyzing the characteristic T of each lead i Sort the leads in descending order and select the top 3 leads. These top 3 leads will best reflect the true location information of the magnetic stimulation.

[0077] 108. Solve for the geometric center of the three selected leads to obtain the location information of the transcranial magnetic stimulation points.

[0078] In this embodiment, in order to obtain a more accurate magnetic stimulation location, the final transcranial magnetic stimulation point location information is obtained by calculating the geometric center of these three lead coordinates.

[0079] In a preferred embodiment of the present invention, such as Figure 2 As shown, after determining the magnetic stimulation location, the doctor can compare the final magnetic stimulation location with the brain location where the magnetic stimulation coil is placed to determine whether the magnetic stimulation location is the area that needs treatment. If it is, the magnetic stimulation coil is considered to be placed correctly; otherwise, the magnetic stimulation coil is considered to be placed incorrectly, and the placement of the magnetic stimulation coil needs to be adjusted.

[0080] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include ROM, RAM, disk, or optical disk, etc.

[0081] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A transcranial magnetic stimulation EEG localization method based on multi-feature fusion, characterized in that, The method includes: Real-time acquisition of EEG data from subjects during magnetic stimulation; The acquired EEG data is then extracted; Preprocess the captured EEG data; The power spectral density, activity index, and arrangement entropy of each lead were calculated according to the six frequency bands. The power spectral density, activity index and arrangement entropy of each lead in each frequency band are fused to calculate the comprehensive index of each lead. The comprehensive indicators of each lead are processed to obtain the final characteristics of each lead. The final characteristics of each lead are sorted in descending order, and the top three leads are selected. The geometric center of the three selected leads is solved to obtain the location information of the transcranial magnetic stimulation points.

2. The transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to claim 1, characterized in that, The preprocessing of the intercepted EEG data includes filtering the intercepted EEG data to remove 50Hz power frequency interference.

3. The transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to claim 1, characterized in that, The six frequency bands include 0~4Hz, 4~8Hz, 8~12Hz, 12~32Hz, 32~80Hz, and 80~250Hz.

4. The transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to claim 1, characterized in that, The calculation method for the activity index of each lead in each frequency band includes: constructing a data vector from the preprocessed EEG data; constructing a time-domain multivariate autoregressive model from the data vector; converting the multivariate autoregressive model to the frequency domain and constructing a frequency-domain multivariate autoregressive model; determining the transfer matrix based on the frequency-domain multivariate autoregressive model; and calculating the difference between the information outflow and inflow of each lead based on the transfer matrix, which is taken as the net outflow value of that lead, and the net outflow value is the activity index of the lead.

5. The transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to claim 4, characterized in that, The formula for calculating the activity index of each lead in each frequency band is expressed as follows: in, This is an indicator of the activity level of lead i in frequency band f. Let be the ratio between the inflow from lead j to lead i in frequency band f and the total inflow to lead i. The ratio of the inflow from lead i to lead j in frequency band f to the total inflow to lead j; Indicates the number of leads.

6. The transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to claim 1, characterized in that, The calculation method for the permutation entropy of each lead in each frequency band includes reconstructing the phase space of the preprocessed EEG data to obtain a reconstruction matrix; sorting the reconstruction components corresponding to each row in the reconstruction matrix in ascending order to obtain the column index of each element position and forming a set of symbol sequences; and determining the probability of the occurrence of a symbol sequence based on the ratio of the number of occurrences of each symbol sequence to the total number of occurrences of all different symbol sequences. The permutation entropy of the leads is calculated based on the probability of the occurrence of the symbol sequence.

7. The transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to claim 6, characterized in that, The formula for calculating the permutation entropy of each lead is expressed as follows: in, Let i be the permutation entropy of lead i. This represents the probability of the k-th symbol sequence appearing. This indicates the number of reconstructed components in the reconstruction matrix.

8. The transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to claim 1, characterized in that, The calculation method for the comprehensive index of each lead is expressed as follows: in, For the comprehensive index of lead i, ~ The power spectral densities of lead i in the frequency bands of 0~4Hz, 4~8Hz, 8~12Hz, 12~32Hz, 32~80Hz, and 80~250Hz are shown in order. ~ The activity indicators for lead i in the frequency bands of 0~4Hz, 4~8Hz, 8~12Hz, 12~32Hz, 32~80Hz, and 80~250Hz are shown in order. Let i be the permutation entropy of lead i. ~ These are the corresponding indicator weights.

9. The transcranial magnetic stimulation EEG localization method based on multi-feature fusion according to claim 1, characterized in that, The final characteristics of each lead are calculated as follows: in, For the characteristics of lead i, This is the sum of the comprehensive indices of the adjacent nodes of lead i. Let be the number of neighbors of lead i.

Images