Individualized precise brain regulation and control method and device based on dynamic brain network traceability

Abstract

The invention discloses an individualized precise brain regulation and control method and device based on dynamic brain network traceability, and relates to the technical field of electroencephalogram regulation and control. The method comprises the steps that resting state electroencephalogram signals of each lead of two resting state tasks are obtained; performing traceability analysis on the resting-state electroencephalogram signals of each lead of the two resting-state tasks, and calculating weighted phase lag indexes of lead pairs according to the traced resting-state data of the two leads in the pairwise lead pairs; constructing dynamic brain networks according to the weighted phase lag indexes of all lead pairs, performing modeling according to the first dynamic brain network and the second dynamic brain network, calculating an abnormal index of each brain region node, performing target stimulation based on the abnormal index of each brain region node by using a multi-channel stimulation controller, and performing target stimulation based on the abnormal index of each brain region node. According to the invention, the problems of inaccurate target spot positioning and single target spot in the existing cranial nerve regulation and control technology are solved.

Description

Technical Field

[0001] This application relates to the field of brain electrical modulation technology, and in particular to a personalized and precise brain modulation method and device based on dynamic brain network tracing. Background Technology

[0002] In the field of neuromodulation, personalized and precise treatment of mental illnesses (such as depression and anxiety) faces three major technological bottlenecks. Existing technologies rely on static norm-based localization methods, such as those based on resting-state EEG frequency domain features and automated brain network analysis systems. While this method achieves automated analysis of brain network topology, its core flaw lies in using norms from healthy individuals as a reference. Such norms typically need to cover a large sample (e.g., more than 1000 normal subjects); however, confounding factors such as age, gender, and region cause the reference standard to deviate from the individual's true pathological characteristics.

[0003] The limitations of stimulation modalities further restrict therapeutic efficacy. While repetitive transcranial magnetic stimulation (rTMS) can precisely modulate deep brain regions (such as the dorsolateral prefrontal cortex), the range of stimulation focus is limited, making it difficult to cover distributed network lesions associated with depression. While transcranial direct current stimulation (tDCS) can achieve multi-target modulation, its spatial resolution is insufficient, and traditional devices cannot dynamically allocate anode / cathode targets based on individual brain network characteristics.

[0004] Device-level issues are also prominent: current split-type modulation systems require localization before stimulation, and stimulation targets are often based on fixed targets, making personalized device applications impossible. Furthermore, current brain modulation devices cannot distinguish between physiological fluctuations and pathological abnormal signals. Summary of the Invention

[0005] The purpose of this application is to provide a personalized and precise brain modulation method and device based on dynamic brain network tracing, which solves the problems of inaccurate target localization and single target in existing brain neuromodulation technologies, and can provide a personalized and precise modulation solution.

[0006] To achieve the above objectives, this application provides the following solution:

[0007] Firstly, this application provides a personalized and precise brain modulation method based on dynamic brain network tracing, including:

[0008] The first and second resting-state EEG signals of each lead of the subject were acquired; the first resting-state EEG signal was acquired when the subject performed the first resting-state task; the second resting-state EEG signal was acquired when the subject performed the second resting-state task; the second resting-state task involved the application of standardized abnormal emotional stimuli.

[0009] Source tracing analysis was performed on the target frequency band signals of the first resting-state EEG signals and the target frequency band signals of the second resting-state EEG signals in each lead to obtain the first source-tracing resting-state data and the second source-tracing resting-state data for each lead.

[0010] For each pair of leads, calculate the first weighted phase lag index of each lead pair based on the first source-after resting state data of the two leads in the pair; calculate the second weighted phase lag index of each lead pair based on the second source-after resting state data of the two leads in the pair.

[0011] A first dynamic brain network and a second dynamic brain network are constructed based on the first weighted phase lag index of all lead pairs and the second weighted phase lag index of all lead pairs.

[0012] Modeling is performed based on the first and second dynamic brain networks, and the abnormality index of each brain region node is calculated.

[0013] Using a multi-channel stimulation controller, target stimulation is performed based on the abnormal index of all brain region nodes.

[0014] Optionally, the multi-channel stimulation controller includes a control unit, a TMS unit, and a tDCS unit;

[0015] The control unit is used to control the TMS unit and the tDCS unit;

[0016] The TMS unit is used to perform transcranial magnetic stimulation on the target site;

[0017] The tDCS unit is used to perform transcranial direct current stimulation on the target site.

[0018] Optionally, a multi-channel stimulation controller is used to target stimulation based on abnormal indices of all brain region nodes, specifically including:

[0019] The abnormality indices of all brain region nodes were sorted from largest to smallest.

[0020] The brain region with the highest abnormal index value was identified as the optimal target.

[0021] Control the TMS unit to perform transcranial magnetic stimulation on the optimal target point;

[0022] The tDCS unit is controlled to perform transcranial direct current stimulation on the remaining target areas; the remaining target areas are brain regions other than the optimal target area.

[0023] Optionally, modeling is performed based on the first dynamic brain network and the second dynamic brain network, and the abnormality index of each brain region node is calculated, specifically including:

[0024] Information propagation dynamics modeling is performed based on the first dynamic brain network and the second dynamic brain network to obtain the first set of activation nodes and the second set of activation nodes; the first set of activation nodes includes several first activation nodes, the second set of activation nodes includes several second activation nodes, and each activation node corresponds to an abnormal brain region node that identifies an abnormal brain state.

[0025] Based on the Bayesian source tracing model, reverse reasoning is performed to obtain the first key node probability of each first activated node and the second key node probability of each second activated node.

[0026] The abnormality index of each brain region node is calculated based on the first key node probability of each first activated node and the second key node probability of each second activated node.

[0027] Alternatively, using the Louvain algorithm, community detection can be performed on the first and second dynamic brain networks respectively, and the difference in intra-community connectivity density of each brain region before and after the second resting-state task stimulus can be calculated; the difference in intra-community connectivity density of each brain region can be determined as the abnormality index of each brain region node.

[0028] Optionally, information propagation dynamics modeling is performed based on the first dynamic brain network and the second dynamic brain network to obtain the first set of activated nodes and the second set of activated nodes, specifically including:

[0029] Constructing an information propagation dynamics model of dynamic brain functional networks based on the SIS model;

[0030] Information propagation dynamics modeling is performed based on the information propagation dynamics model of dynamic brain functional network, the first dynamic brain network and the second dynamic brain network, and the state of brain region nodes is updated to obtain the probability of each brain region node being in an activated state.

[0031] The brain region nodes in the first dynamic brain network whose probability of being in an active state is greater than a set probability value are identified as first active nodes, and the brain region nodes in the second dynamic brain network whose probability of being in an active state is greater than a set probability value are identified as second active nodes; all first active nodes constitute the first active node set, and all second active nodes constitute the second active node set.

[0032] Optionally, the formula for calculating the probability of a critical node is as follows:

[0033]

[0034] Where P(i) is the critical node probability of activating node i; I(k) represents the set of nodes in the active state at time step k; βE ij E represents the probability that node j is activated given seed node i; ijβ represents the functional connection strength between nodes i and j; β represents the information propagation rate; V represents the set of key nodes; E represents the... zj The functional connection strength between node z and node j.

[0035] Optionally, the formula for calculating the abnormality index of brain region nodes is as follows:

[0036]

[0037] Among them, BAI i ε is the abnormality index of brain region node i; P1(i) and P2(i) are the probabilities of the first key node in the first resting-state task and the second key node in the second resting-state task, respectively; N is the number of activated nodes; P1(j) and P2(j) are the probabilities of the first key node in the first resting-state task and the second key node in the second resting-state task, respectively; ε is the numerical stability factor.

[0038] Secondly, this application provides a personalized and precise brain modulation device based on dynamic brain network tracing, comprising:

[0039] The target localization module is used to implement the steps of the individualized and precise brain modulation method based on dynamic brain network tracing described in the first aspect, and to determine the abnormal index of each brain region node.

[0040] A multi-channel stimulation controller is used to target stimulation based on the abnormality index of all brain region nodes.

[0041] Optionally, the multi-channel stimulation controller includes:

[0042] Control unit, used to control the TMS unit and tDCS unit;

[0043] The TMS unit is used for transcranial magnetic stimulation of the target site.

[0044] The tDCS unit is used to perform transcranial direct current stimulation on the target site.

[0045] Optionally, in terms of target stimulation based on abnormal indices of all brain region nodes, the multi-channel stimulation controller is used to:

[0046] The control unit controls the TMS unit to perform transcranial magnetic stimulation on the optimal target point; the optimal target point is the brain region with the largest abnormality index value.

[0047] The control unit controls the tDCS unit to perform transcranial direct current stimulation on the remaining target areas; the remaining target areas are brain regions other than the optimal target area.

[0048] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0049] This application provides a personalized and precise brain modulation method and device based on dynamic brain network tracing. It acquires resting-state EEG signals from each lead of two resting-state tasks; performs source tracing analysis on the resting-state EEG signals from each lead of the two resting-state tasks; calculates the weighted phase lag index of the lead pairs based on the source-traced resting-state data of two leads in each pair; constructs a first dynamic brain network and a second dynamic brain network based on the weighted phase lag index of all lead pairs; performs information propagation dynamics modeling based on the first and second dynamic brain networks to obtain the abnormality index of each brain region node; and uses a multi-channel stimulation controller to perform target stimulation based on the abnormality index of each brain region node. This solves the problems of inaccurate target localization and single target in existing brain neuromodulation technologies, and can provide a personalized and precise modulation scheme. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating a personalized and precise brain modulation method based on dynamic brain network tracing, as provided in Embodiment 1 of this application.

[0052] Figure 2 This is a schematic diagram of the structure of the multi-channel stimulation controller provided in Embodiment 1 of this application.

[0053] Figure 3 This is a schematic diagram of the overall connection logic of the multi-channel stimulation controller provided in Embodiment 1 of this application. Detailed Implementation

[0054] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0055] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] Example 1

[0057] In one exemplary embodiment, such as Figure 1As shown, a personalized and precise brain modulation method based on dynamic brain network tracing is provided, which includes the following steps 201 to 206.

[0058] Step 201: Acquire the first and second resting-state EEG signals of each lead of the subject; the first resting-state EEG signal was collected when the subject performed the first resting-state task; the second resting-state EEG signal was collected when the subject performed the second resting-state task; the second resting-state task was to apply standardized abnormal emotional stimuli.

[0059] Step 202: Perform source tracing analysis on the target frequency band signals of the first resting-state EEG signals and the second resting-state EEG signals of each lead to obtain the first source-traced resting-state data and the second source-traced resting-state data of each lead.

[0060] Step 203: For each pair of leads, calculate the first weighted phase lag index of each lead pair based on the first traced resting state data of the two leads in the pair; calculate the second weighted phase lag index of each lead pair based on the second traced resting state data of the two leads in the pair.

[0061] Step 204: Construct a first dynamic brain network and a second dynamic brain network based on the first weighted phase lag index of all lead pairs and the second weighted phase lag index of all lead pairs.

[0062] Step 205: Model the brain based on the first dynamic brain network and the second dynamic brain network, and calculate the abnormality index of each brain region node.

[0063] Step 206: Using a multi-channel stimulation controller, target stimulation is performed based on the abnormality index of all brain region nodes.

[0064] Implementing steps 201 to 206 above solves the problems of inaccurate target localization and single target in existing brain neuromodulation technologies, and can provide individualized and precise modulation solutions.

[0065] The above method is divided into two stages: target localization and stimulus execution.

[0066] (I) Target localization stage

[0067] Execution Entity: Embedded Processor

[0068] ① During the first resting-state task, high-density EEG signals (≥64 leads) were collected from the subjects and recorded as the first resting-state EEG signal EEGR1.

[0069] ② Apply standardized abnormal emotional stimuli (using a negative picture / video library for depression and an uncertainty task for anxiety) as a second resting-state task.

[0070] ③ Collect resting-state EEG signals again, and record them as the second resting-state EEG signal EEGR2.

[0071] ④ The target frequency band signal is the frequency band signal that is closely related to brain regulation. Depression and anxiety are beta or gamma frequency bands, and sleep is delta or theta frequency band signal.

[0072] In a specific example, the β-band signal (13-30Hz) of the two resting-state data is extracted, namely the β-band signal of the first resting-state EEG signal and the β-band signal of the second resting-state EEG signal in each lead. The source analysis of EEGR1 and EEGR2 is performed using the weighted minimum norm estimate (wMNE) to obtain the first and second source-traced resting-state data of each lead.

[0073] The above step 202 specifically includes: using the weighted minimum norm estimation algorithm, performing source tracing analysis on the target frequency band signal of the first resting-state EEG signal EEGR1 and the target frequency band signal of the second resting-state EEG signal EEGR2 in each lead, to obtain the first source-traced resting-state data and the second source-traced resting-state data for each lead.

[0074] ⑤ For the two resting-state data after tracing (the first and second resting-state data after tracing in each lead), the weighted phase lag index (wPLI) is used to calculate the functional connectivity strength: that is, the calculated weighted phase lag index is used as the functional connectivity strength between the two brain region nodes.

[0075] The formula for calculating the weighted phase lag exponent is as follows:

[0076] wPLI=|E{Im(X)}| / E{|Im(X)|} (1);

[0077] Where wPLI is the weighted phase lag exponent of the lead pair; X is the cross spectral density of the two source-tracing resting-state data at a specific frequency; Im(X) is the imaginary part of X; E{Im(X)} is the expected value or average value at all time points; || represents taking the absolute value.

[0078] Where: X = Sxy(f): This is the cross-spectral density of two signals x(t) and y(t) at a specific frequency f. It is a complex number. The real part Re(Sxy(f)) of Sxy(f) contains information about in-phase (0° or 180°) coupling. The imaginary part Im(Sxy(f)) of Sxy(f) contains information about orthogonal (90° or 270°) coupling, which reflects the phase lag (lead / lag relationship) between the signals. Im(X) = Im(Sxy(f)): This is the imaginary part of the cross-spectral density X, which directly reflects the instantaneous phase relationship (lag or lead) of the two signals at frequency f. E{}: Represents the expected value or average value. In actual calculations, this is usually achieved by averaging over a time window or across multiple trials.

[0079] For all pairwise lead pairs, the weighted phase lag index wPLI is calculated to further construct an N*N (N is the number of EEG leads) fully connected brain network. The data from the two resting-state tasks can be used to construct the first dynamic brain network and the second dynamic brain network, respectively.

[0080] Dynamic brain network information propagation dynamics modeling was performed using two sets of resting-state EEG data to obtain the set of activated nodes I. k (i) to identify abnormal brain regions in abnormal brain states (or diseases).

[0081] In step 205 above, modeling is performed based on the first dynamic brain network and the second dynamic brain network, and the abnormality index of each brain region node is calculated. This can be achieved using either of the following two methods:

[0082] (1) The first method in step 205:

[0083] Step 11: Perform information propagation dynamics modeling based on the first dynamic brain network and the second dynamic brain network to obtain the first set of activation nodes and the second set of activation nodes; the first set of activation nodes includes several first activation nodes, the second set of activation nodes includes several second activation nodes, and each activation node corresponds to an abnormal brain region node that identifies an abnormal brain state.

[0084] Step 12: Perform reverse reasoning based on the Bayesian source tracing model to obtain the first key node probability of each first activated node and the second key node probability of each second activated node;

[0085] Step 13: Calculate the abnormality index of each brain region node based on the first key node probability of each first activated node and the second key node probability of each second activated node.

[0086] In step 11 above, information propagation dynamics modeling is performed based on the first dynamic brain network and the second dynamic brain network to obtain the first set of activated nodes and the second set of activated nodes, specifically including:

[0087] Constructing an information propagation dynamics model of dynamic brain functional networks based on the SIS model;

[0088] Information propagation dynamics modeling is performed based on the information propagation dynamics model of dynamic brain functional network, the first dynamic brain network and the second dynamic brain network, and the state of brain region nodes is updated to obtain the probability of each brain region node being in an activated state.

[0089] The brain region nodes in the first dynamic brain network whose probability of being in an active state is greater than a set probability value are identified as first active nodes, and the brain region nodes in the second dynamic brain network whose probability of being in an active state is greater than a set probability value are identified as second active nodes; all first active nodes constitute the first active node set, and all second active nodes constitute the second active node set.

[0090] A dynamic model of information propagation in a dynamic brain functional network was constructed based on the SIS model. In this model, it is assumed that each brain region node V... k The state of (i) (i = 1, 2, ..., N, where N is the number of EEG leads) can be either a susceptible state (S) or an activated state (I), where S indicates that the brain region is inactive and I indicates that the brain region is involved in information transmission. In two time steps k (k = 1, 2) (representing two resting-state task EEG data acquisitions respectively), the state of the brain region nodes is calculated recursively, and based on the dynamic brain functional network G... k =(V k E k The connections between nodes in the midbrain region update the state of each node. Adjacent brain regions influence each other through their functional connections, enabling the propagation of information between brain regions. k Let E be the set of brain regions for the k-th resting-state task. k Let be the edge set of the k-th resting state task. Its state calculation formula is shown below:

[0091] S k+1 (i)=S k (i)-βS k (i)∑ j∈N(i) E ij (k)I k (j)+γI k (i) (2);

[0092] I k+1 (i)=I k (i)+βS k (i)∑ j∈N(i) Eij (k)I k (j)-γI k (i) (3);

[0093] Among them, S k+1 (i) represents the probability that brain region node i is in a susceptible state at time step k+1; S k (i) represents the probability that brain region node i is in a susceptible state at time step k; k (i) represents the probability that brain region node i is in an activated state at time step k; k+1 (i) represents the probability that brain region node i is in an activated state at time step k+1; k (j) represents the probability that brain region node j is in an activated state at time step k; E ij (k) represents the functional connectivity strength between brain region nodes i and j during the k-th resting-state task acquisition; β is the information propagation rate, with an empirical value of 0.65±0.15 (calibrated via fMRI); γ is the recovery rate; and N(i) is the adjacency set of brain region node i, determined by the dynamic brain functional network G. k The edge set E in k Decision. Each edge e ij ∈E k Connecting brain regions i and j, when brain region j is activated, information propagates to its connected brain region i, affecting S. k (i) and I k The change of (i).

[0094] When the probability that a brain region node is active at time step k is greater than a set probability value, the brain region node is identified as an active node. The set probability value can be 0.6.

[0095] The above model can be used to trace the information transmission path between brain regions, and then the most critical brain regions and functional connections in the process of abnormal emotion induction can be identified based on the brain network tracing model.

[0096] The set of activated nodes I obtained from the above steps k (i) Using a Bayesian source tracing model, we perform reverse reasoning to obtain the probability P(i) of whether an activated node in the set of activated nodes is a key node. That is, we track the key nodes of the abnormal emotions (or diseases) obtained above in the dynamic brain functional network.

[0097] This study employs a Bayesian source model to perform reverse reasoning, aiming to trace the propagation path and key nodes of induced abnormal emotions within dynamic brain functional networks. Based on a Bayesian inference framework, this model calculates the probability of brain region nodes serving as initial seed nodes and integrates the dynamic characteristics of brain functional networks. Its goal is to identify the initial brain region nodes most likely to trigger information propagation, thereby revealing the key paths of neural information diffusion.

[0098] Specifically, suppose that at a certain time step k, some brain regions in the network are already in an activated state (i.e., exhibiting abnormal emotion-related characteristics). The Bayesian source tracing model, through a Bayesian inference framework, calculates the probability P(i) of each brain region node i becoming an initial seed node. The probability of brain region node i becoming an initial seed node is the key node probability, and the formula for calculating the key node probability is:

[0099]

[0100] Where P(i) is the critical node probability of activating node i, also known as the resting state initial seed probability; I(k) represents the set of nodes in the active state at time step k; βE ij E represents the probability that brain region node j is activated given seed node i; ij E represents the functional connectivity strength between brain region nodes i and j. ij =wPLI; β is the information propagation rate; V is the set of key nodes; E zj The functional connection strength between node z and node j.

[0101] By calculating the probability P(i) of each node as an initial seed node, the higher the probability, the greater the likelihood that it is a key brain region.

[0102] Based on the critical node probabilities P(i) obtained from the two resting-state task data obtained in the above steps, an abnormality index (BAI) for all brain region nodes is defined, the activated nodes are sorted, and subsequent hardware stimulation is implemented according to the abnormality index sorting results.

[0103] The formula for calculating the abnormality index of brain region nodes is as follows:

[0104]

[0105] Among them, BAI i Let P1(i) be the abnormality index of brain region node i; P1(i) and P2(i) are the probabilities of the first key node in the first resting-state task (also known as the initial seed probability of the first resting state) and the second key node in the second resting-state task (also known as the initial seed probability of the second resting state), respectively; N is the number of activated nodes; P1(j) and P2(j) are the probabilities of the first key node in the first resting-state task and the second key node in the second resting-state task, respectively; ε is the numerical stability factor, ε = 0.01 × max(P1); is the baseline probability, which is the mean of the minimum probabilities of all brain nodes; log() represents the natural logarithm transformation; |P2(i)-P1(i)| is the seed probability change.

[0106] (2) The first method of step 205: Using the Louvain algorithm, community detection is performed on the first dynamic brain network and the second dynamic brain network respectively, and the difference in intra-community connectivity density of each brain region before and after the second resting state task stimulation is calculated; the difference in intra-community connectivity density of each brain region is determined as the abnormality index of each brain region node.

[0107] Functional community clustering can also replace BAI ranking (BAI target classification): A multi-channel stimulation controller is used to stimulate targets based on the abnormality index of all brain region nodes. Specifically, this involves dividing the brain network into multiple functional communities (modules) and replacing the BAI index by calculating the rate of change in intra-module connectivity density within each community. The specific process is as follows:

[0108] Using the Louvain algorithm, community detection was performed on the first and second dynamic brain networks (two resting-state wPLI networks) for two resting-state tasks. The difference in intra-community connectivity density before and after the stimulus was calculated, i.e., the difference in intra-community connectivity density before and after the second resting-state task was calculated: ΔDc=|D c,R2 -D c,R1 | / D c,R1 Where ΔDc is the intra-community connectivity density difference, D c,R1 To determine the intracommunicular connectivity density in the forebrain region stimulated by the second resting-state task, D c,R2 The intracommunity connectivity density of the brain region after the second resting-state task stimulation; intracommunity connectivity density is the strength of functional connectivity between brain region nodes; the intracommunity brain region node with the largest intracommunity connectivity density difference ΔDc value is selected as the TMS target point, that is, the TMS unit is used to perform transcranial magnetic stimulation on the target point (first priority), and other community nodes are assigned to the tDCS multi-target stimulator, that is, the tDCS unit is used to perform transcranial direct current stimulation on the remaining target points.

[0109] The advantage of the second approach is that it reduces computational complexity.

[0110] (II) Stimulus Implementation Phase

[0111] The main actuator is a multi-channel stimulation controller, which includes a control unit, a TMS unit, and a tDCS unit. The multi-channel stimulation controller is as follows: Figure 2 As shown, the overall connection logic is as follows: Figure 3 As shown.

[0112] The control unit is used to control the TMS unit and the tDCS unit;

[0113] The TMS unit is used to perform transcranial magnetic stimulation on the target site;

[0114] The tDCS unit is used to perform transcranial direct current stimulation on the target site.

[0115] The control unit components include: 1) Central processing unit: the main control core; 2) Display screen: displaying the operation interface and parameters; 3) Communication interface: connecting to external systems.

[0116] The TMS unit components include: 1) TMS figure-eight coil: generates a focused magnetic field; 2) cooling module: prevents the coil from overheating; 3) robotic arm system: a precise positioning device; 4) universal joint: flexibly adjusts the angle of the robotic arm system.

[0117] The tDCS unit components include: 1) Electrode matrix substrate: carrying the electrode array; 2) Electrode posts: automatically extendable and adjustable to fit the user's scalp area; 3) Conductive electrode head: current stimulation interface; 4) Helical spring mechanism: maintaining constant pressure; 5) Current control chip: precise current regulation; 6) Impedance detection circuit: safety monitoring system.

[0118] Key connections: The robotic arm system controls the TMS figure-eight coil via a universal joint. The central processing unit simultaneously controls the TMS unit and the tDCS unit. The impedance data collected by the impedance detection circuit is fed back to the central processing unit in real time. When the impedance data collected by the impedance detection circuit is infinite, it indicates that the electrode is detached from the user's scalp and the contact is not tight. Therefore, manual adjustment of the electrode position is required to fit the scalp. The TMS coil and tDCS electrode directly act on the scalp target area. The cooling system is directly controlled by the central processing unit.

[0119] Using a multi-channel stimulation controller, target stimulation is performed based on the abnormality index of all brain region nodes. Specifically, this includes: sorting the abnormality index of all brain region nodes from largest to smallest; determining the brain region with the largest abnormality index value as the optimal target; controlling the TMS unit to perform transcranial magnetic stimulation on the optimal target; controlling the tDCS unit to perform transcranial direct current stimulation on the remaining target; the remaining target is the brain region other than the optimal target.

[0120] ① Brain regions are sorted in descending order of Abnormality Index (BAI) value, and the brain region with the highest value is selected as the optimal target. The optimal target is assigned to the TMS stimulator, that is, the TMS unit is used to perform transcranial magnetic stimulation on the optimal target.

[0121] ② The remaining target points are assigned to the tDCS multi-target stimulator, that is, the tDCS unit is used to perform transcranial direct current stimulation on the remaining target points.

[0122] ③ Initiation of sequential alternating stimulation:

[0123] The first method: rTMS stimulation (frequency 1 / 10Hz, intensity 120%MT) for 5 minutes, followed by a 1-minute interval (synchronous EEG feedback acquisition), and then tDCS stimulation (anodic / cathode stimulation, anodic stimulation for activation and cathode stimulation for inhibition) for 10 minutes. Technical effect: to improve the efficiency of neural plasticity induction through electromagnetic synergy.

[0124] The second method:

[0125] 1) tDCS lead stimulation protocol:

[0126] The timing was adjusted based on clinical research: tDCS pre-stimulation (anodic targeting for 10 minutes) → 1-minute interval → rTMS enhanced stimulation (5 minutes). Mechanism of action: tDCS increases cortical excitability and enhances the subsequent rTMS neural remodeling effect.

[0127] 2) Alternating frequency band stimulation: rTMS focuses on low-frequency (≤1Hz) inhibitory stimulation, while tDCS focuses on high-frequency (≥10Hz) excitatory stimulation. The two devices work synchronously, but the TMS unit uses burst mode (Burst mode: 5 pulses / second, lasting 10 seconds, with a 50-second interval), which has the advantage of shortening the total treatment time.

[0128] This application also provides an application scenario in which the aforementioned personalized and precise brain modulation method based on dynamic brain network tracing is applied. Specifically, the personalized and precise brain modulation method based on dynamic brain network tracing provided in this embodiment can be applied in a personalized and precise brain modulation scenario. The personalized and precise brain modulation scenario includes a content production stage, a content processing link, and a target stimulation stage; the first resting-state EEG signal and the second resting-state EEG signal enter the content processing link from the content production stage, and through human-machine collaboration, obtain the abnormality index of brain region nodes, and then enter the downstream target stimulation stage. The personalized and precise brain modulation method based on dynamic brain network tracing provided in this embodiment belongs to the content processing link. Specifically, in the content processing chain of resting-state EEG signals, source analysis can be performed on the resting-state EEG signals of each lead in two resting-state tasks. The weighted phase lag index of the lead pair can be calculated based on the source-traced resting-state data of two leads in each pair. A dynamic brain network can be constructed based on the weighted phase lag index of all lead pairs. Information propagation dynamics modeling can be performed based on the dynamic brain network to obtain the set of activated nodes. Backward reasoning based on the Bayesian source model can be performed to obtain the probability of key nodes. The abnormality index of brain region nodes can be calculated based on the probability of key nodes of each activated node in the two resting-state tasks. Finally, target stimulation can be performed using a multi-channel stimulation controller in the target stimulation stage.

[0129] Example 2

[0130] Based on the same inventive concept, this application also provides a device for implementing the aforementioned personalized precise brain modulation method based on dynamic brain network tracing. The solution provided by this device is similar to the implementation described in the above method. Therefore, the specific limitations of one or more embodiments of the device for personalized precise brain modulation based on dynamic brain network tracing provided below can be found in the limitations of the personalized precise brain modulation method based on dynamic brain network tracing above, and will not be repeated here.

[0131] In one exemplary embodiment, a personalized precision brain modulation device based on dynamic brain network tracing is provided, comprising:

[0132] The target localization module is used in the steps of the individualized and precise brain modulation method based on dynamic brain network tracing described in Example 1 to determine the abnormal index of each brain region node.

[0133] A multi-channel stimulation controller is used to target stimulation based on the abnormality index of all brain region nodes.

[0134] The multi-channel stimulation controller includes: a control unit for controlling the TMS unit and the tDCS unit; the TMS unit for performing transcranial magnetic stimulation on the target point; and the tDCS unit for performing transcranial direct current stimulation on the target point.

[0135] In terms of target stimulation based on abnormal indices of all brain region nodes, the multi-channel stimulation controller is used for:

[0136] The control unit controls the TMS unit to perform transcranial magnetic stimulation on the optimal target point; the optimal target point is the brain region with the largest abnormality index value.

[0137] The control unit controls the tDCS unit to perform transcranial direct current stimulation on the remaining target areas; the remaining target areas are brain regions other than the optimal target area.

[0138] This application aims to address the core problems in existing brain neuromodulation technologies, such as inaccurate target localization, single target, low efficiency of single-modal stimulation, and lack of closed-loop feedback.

[0139] This study pioneered a dynamic brain network analysis mechanism based on "emotion-induced-resting state comparison": High-density EEG (≥64 channels) was used to collect dual resting-state signals (EEG_R1 / EEG_R2) before and after emotional stimulation. Weighted minimum norm estimation (wMNE) was employed in the β band (13-30Hz) to locate brain region activity. A dynamic functional network was constructed based on the weighted phase lag index (wPLI), utilizing the imaginary part to eliminate volumetric conduction artifacts. The study innovatively integrated the SI propagation model with a Bayesian source tracing algorithm, simulating the "susceptible state-activated state" transition to track information propagation paths and calculating the brain region abnormality index (BAI) to quantify the stimulus response specificity. During the stimulation phase, the optimal target points were assigned to repetitive transcranial magnetic stimulation (rTMS) according to BAI ranking, while the remaining target points were assigned to transcranial direct current stimulation (tDCS). A temporal alternation protocol of "rTMS 5 minutes → 1 minute interval (synchronous EEG feedback) → tDCS 10 minutes" was adopted to achieve electromagnetic synergistic enhancement.

[0140] At the device level, the integrated protection system incorporates a multi-channel stimulation controller: the control unit is equipped with a central processing unit, display screen, and communication interface to achieve target coordinate analysis and stimulation protocol execution; the TMS unit includes a figure-eight coil, a robotic arm positioning system, and a composite cooling module, which maintains the stability of the coil-target position through a universal joint; the tDCS unit is composed of a multi-channel (≥16 channels) matrix substrate, with each electrode integrating an adaptive telescopic column and a helical spring mechanism, which, together with the conical conductive electrode head, achieves adaptive fit to the scalp curvature; the innovative closed-loop safety system includes an impedance detection circuit (real-time current adjustment to prevent burns) and a pressure sensor (automatic cut-off of stimulation when exceeding the limit), and all components establish bidirectional data interaction through optical fiber / cable, forming a dynamic optimization loop of "positioning-stimulation-feedback".

[0141] This application overcomes the technical bottlenecks of traditional static norm localization bias and incomplete coverage of single-modal stimulation by organically combining dynamic brain network targeted analysis with bimodal spatiotemporal synergistic stimulation, providing an individualized and precise regulation solution for abnormal emotions such as depression and anxiety.

[0142] This application has the following advantages:

[0143] 1. Improved target localization accuracy

[0144] Existing brain modulation techniques rely on static norm networks for localization, which are susceptible to target bias due to factors such as age and gender. This proposed solution pioneers a "dual resting-state dynamic self-comparison analysis" mechanism: it constructs a β-band dynamic network using EEG signals (EEG_R1 / EEG_R2) before and after emotional stimulation, and utilizes the imaginary part of wPLI to eliminate volume conduction artifacts; furthermore, it integrates the SI propagation model to simulate the "susceptible state-activated state" transition process, and combines Bayesian source tracing to reverse the propagation path of abnormal emotions, significantly improving the accuracy of key target localization.

[0145] 2. A significant increase in the efficiency of neural regulation.

[0146] Traditional single-modal stimulation suffers from insufficient depth of action (tDCS) or limited coverage (rTMS). This application proposes a "target hierarchical-spatiotemporal synergy" strategy: the optimal target is assigned to rTMS according to the BAI index, and secondary targets are assigned to multi-channel tDCS; through a temporal alternation protocol of "5 minutes rTMS → 1 minute EEG feedback → 10 minutes tDCS", the electromagnetic field complementarity effect is used to induce the superposition of synaptic plasticity.

[0147] 3. Security is achieved through a multi-dimensional closed-loop protection system.

[0148] Existing equipment often results in scalp burns due to a lack of real-time monitoring. This application proposes a triple protection system:

[0149] 1) The adaptive electrode posts of the tDCS unit maintain a constant contact pressure with the helical spring mechanism to prevent mechanical compression damage;

[0150] 2) The impedance detection circuit automatically cuts off the current when there is poor electrode contact;

[0151] 3) The TMS cooling module ensures that the coil temperature is below the threshold.

[0152] 4. Significant breakthrough in personalized adaptation capabilities:

[0153] It overcomes the inherent bias problem of existing technology group norms through end-to-end individualized modeling:

[0154] Self-comparative analysis was conducted based on high-density data before and after the subjects' own emotional responses.

[0155] Constructing an individual brain anatomical connectivity framework based on wMNE tracing;

[0156] Combined with SIS model dynamic calibration to capture individual neural information propagation characteristics;

[0157] Bayesian source tracing models quantify the origins of abnormal emotions in individuals.

[0158] The aforementioned advantages stem from the synergy of three core technologies: a dynamic brain network targeted analysis method that overcomes the limitations of static localization; a dual-modal spatiotemporal alternation protocol that achieves complementary electromagnetic effects; and a closed-loop safety system (device point) for the hardware device that eliminates clinical risks. This provides a novel technological paradigm for addressing the challenges of precision, safety, and universality in the field of neuromodulation of mental illnesses.

[0159] This application proposes an individualized brain modulation scheme based on the contrast between emotion-induced and resting states: by constructing a dynamic information propagation model to replace the static norm reference, and by using the dual-modal temporal alternation mechanism to overcome the electromagnetic interference barrier, it provides the first closed-loop solution of "dynamic modeling-precise positioning-adaptive stimulation" for mental illness.

[0160] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0161] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A personalized and precise brain modulation method based on dynamic brain network tracing, characterized in that, The personalized and precise brain modulation method based on dynamic brain network tracing includes: The first and second resting-state EEG signals of each lead of the subject were acquired; the first resting-state EEG signal was acquired when the subject performed the first resting-state task; the second resting-state EEG signal was acquired when the subject performed the second resting-state task; the second resting-state task involved the application of standardized abnormal emotional stimuli. Source tracing analysis was performed on the target frequency band signals of the first resting-state EEG signals and the target frequency band signals of the second resting-state EEG signals in each lead to obtain the first source-tracing resting-state data and the second source-tracing resting-state data for each lead. For each pair of leads, calculate the first weighted phase lag index of each lead pair based on the first source-after resting state data of the two leads in the pair; calculate the second weighted phase lag index of each lead pair based on the second source-after resting state data of the two leads in the pair. A first dynamic brain network and a second dynamic brain network are constructed based on the first weighted phase lag index of all lead pairs and the second weighted phase lag index of all lead pairs. Modeling is performed based on the first and second dynamic brain networks, and the abnormality index of each brain region node is calculated. Using a multi-channel stimulation controller, target stimulation is performed based on the abnormal index of all brain region nodes.

2. The personalized and precise brain modulation method based on dynamic brain network tracing according to claim 1, characterized in that, The multi-channel stimulation controller includes a control unit, a TMS unit, and a tDCS unit; The control unit is used to control the TMS unit and the tDCS unit; The TMS unit is used to perform transcranial magnetic stimulation on the target site; The tDCS unit is used to perform transcranial direct current stimulation on the target site.

3. The personalized and precise brain modulation method based on dynamic brain network tracing according to claim 2, characterized in that, Using a multi-channel stimulation controller, target stimulation is performed based on abnormal indices of all brain region nodes, specifically including: The abnormality indices of all brain region nodes were sorted from largest to smallest. The brain region with the highest abnormal index value was identified as the optimal target. Control the TMS unit to perform transcranial magnetic stimulation on the optimal target point; The tDCS unit is controlled to perform transcranial direct current stimulation on the remaining target areas; the remaining target areas are brain regions other than the optimal target area.

4. The personalized and precise brain modulation method based on dynamic brain network tracing according to claim 1, characterized in that, Modeling is performed based on the first and second dynamic brain networks, and the abnormality index of each brain region node is calculated, specifically including: Information propagation dynamics modeling is performed based on the first dynamic brain network and the second dynamic brain network to obtain the first set of activation nodes and the second set of activation nodes; the first set of activation nodes includes several first activation nodes, the second set of activation nodes includes several second activation nodes, and each activation node corresponds to an abnormal brain region node that identifies an abnormal brain state. Based on the Bayesian source tracing model, reverse reasoning is performed to obtain the first key node probability of each first activated node and the second key node probability of each second activated node. The abnormality index of each brain region node is calculated based on the first key node probability of each first activated node and the second key node probability of each second activated node. Alternatively, using the Louvain algorithm, community detection can be performed on the first and second dynamic brain networks respectively, and the difference in intra-community connectivity density of each brain region before and after the second resting-state task stimulus can be calculated; the difference in intra-community connectivity density of each brain region can be determined as the abnormality index of each brain region node.

5. The personalized and precise brain modulation method based on dynamic brain network tracing according to claim 4, characterized in that, Based on the first and second dynamic brain networks, information propagation dynamics modeling is performed to obtain the first set of activated nodes and the second set of activated nodes, specifically including: Constructing an information propagation dynamics model of dynamic brain functional networks based on the SIS model; Information propagation dynamics modeling is performed based on the information propagation dynamics model of dynamic brain functional network, the first dynamic brain network and the second dynamic brain network, and the state of brain region nodes is updated to obtain the probability of each brain region node being in an activated state. The brain region nodes in the first dynamic brain network whose probability of being in an active state is greater than a set probability value are identified as first active nodes, and the brain region nodes in the second dynamic brain network whose probability of being in an active state is greater than a set probability value are identified as second active nodes; all first active nodes constitute the first active node set, and all second active nodes constitute the second active node set.

6. The personalized and precise brain modulation method based on dynamic brain network tracing according to claim 4, characterized in that, The formula for calculating the probability of a critical node is as follows: Where P(i) is the critical node probability of activating node i; I(k) represents the set of nodes in the active state at time step k; βE ij E represents the probability that node j is activated given seed node i; ij β represents the functional connection strength between nodes i and j; β represents the information propagation rate; V represents the set of key nodes; E represents the... zj The functional connection strength between node z and node j.

7. The personalized and precise brain modulation method based on dynamic brain network tracing according to claim 4, characterized in that, The formula for calculating the abnormality index of brain region nodes is as follows: Among them, BAI i ε is the abnormality index of brain region node i; P1(i) and P2(i) are the probabilities of the first key node in the first resting-state task and the second key node in the second resting-state task, respectively; N is the number of activated nodes; P1(j) and P2(j) are the probabilities of the first key node in the first resting-state task and the second key node in the second resting-state task, respectively; ε is the numerical stability factor.

8. A personalized precision brain modulation device based on dynamic brain network tracing, characterized in that, The personalized precision brain modulation device based on dynamic brain network tracing includes: The target localization module is used to implement the steps of the individualized and precise brain modulation method based on dynamic brain network tracing as described in any one of claims 1-7, and to determine the abnormal index of each brain region node. A multi-channel stimulation controller is used to target stimulation based on the abnormality index of all brain region nodes.

9. The personalized precision brain modulation device based on dynamic brain network tracing according to claim 8, characterized in that, The multi-channel stimulation controller includes: Control unit, used to control the TMS unit and tDCS unit; The TMS unit is used for transcranial magnetic stimulation of the target site. The tDCS unit is used to perform transcranial direct current stimulation on the target site.

10. The personalized precision brain modulation device based on dynamic brain network tracing according to claim 9, characterized in that, In terms of target stimulation based on abnormal indices of all brain region nodes, the multi-channel stimulation controller is used for: The control unit controls the TMS unit to perform transcranial magnetic stimulation on the optimal target point; the optimal target point is the brain region with the largest abnormality index value. The control unit controls the tDCS unit to perform transcranial direct current stimulation on the remaining target areas; the remaining target areas are brain regions other than the optimal target area.

Images