A high-precision transcranial direct current stimulation device based on a navigation system

Abstract

The application discloses a high-precision transcranial direct current stimulation device based on a navigation system and relates to the technical field of electric digital data processing. In the method, a brain electric data acquisition module of the high-precision transcranial direct current stimulation device first acquires brain electric data of a subject, and then a functional network analysis module performs spectrum analysis and functional connection analysis on the data. After the high-precision transcranial direct current stimulation device compares the two analysis results with standard data of the same age group, abnormal functional sub-networks are identified, and a deviation degree is quantified. A navigation positioning module determines multi-target point stimulation positions based on an individual three-dimensional navigation model and the identified abnormal sub-networks, thereby improving the spatial accuracy of the multi-target point transcranial direct current stimulation positions. A transcranial direct current stimulation scheme generation module takes the age, the deviation degree and the multi-target point positions as input features, generates a personalized collaborative stimulation scheme through a machine learning model, and improves the accuracy of current parameter combinations corresponding to electrode layout schemes for different individuals in the transcranial direct current stimulation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital data processing, and particularly relates to a high-precision transcranial direct current stimulation device based on a navigation system. BACKGROUND

[0002] Transcranial direct current stimulation (tDCS) is a non-invasive neuromodulation technique that applies a weak direct current to the scalp surface to modulate the excitability of neurons in the cerebral cortex, and has broad application prospects in the fields of neural rehabilitation, cognitive enhancement and mental illness treatment.

[0003] The transcranial direct current stimulation technology in the related art mainly determines the transcranial direct current stimulation scheme based on the brain anatomical structure, determines the position of the target brain area through an anatomical positioning method (such as the international 10-20 system or MRI guidance), and places an electrode in the corresponding scalp area for stimulation. This method selects a specific anatomical region related to the target disease as the stimulation target based on the known anatomical function correspondence of the brain area, and adjusts the excitability of neurons in this region to achieve the treatment purpose. This method also optimizes and adjusts the electrode position and stimulation parameters according to the feedback of the stimulation effect.

[0004] However, although the anatomical structure provides physical position information of the brain area, it is difficult to reflect the functional connection state between different brain areas and the abnormal mode of the brain functional network. When the target disease involves connection abnormalities between multiple brain areas or disorder of the entire functional sub-network, the electrode layout scheme and the corresponding current parameter combination obtained based on a single or a small number of anatomical regions are not accurate enough, and it is difficult to effectively regulate the functional state of the entire abnormal network. SUMMARY

[0005] The present application provides a high-precision transcranial direct current stimulation device based on a navigation system to improve the accuracy of the electrode layout scheme and the corresponding current parameter combination.

[0006] In a first aspect, a high-precision transcranial direct current stimulation device based on a navigation system is provided, comprising: a brain electrical data acquisition module configured to acquire brain electrical data of a subject; a functional network analysis module configured to perform digital filtering processing on the brain electrical data and perform frequency spectrum analysis and functional connection analysis on the filtered brain electrical data, wherein the frequency spectrum analysis obtains power ratios of each brain electrical wave at each electrode position, and the functional connection analysis constructs a functional connection network by calculating functional connection strengths between brain regions, and compares the power ratios and the functional connection network with a pre-set standard electroencephalogram spectrum and functional network mode of the same age group to identify abnormal functional sub-networks and calculate deviation degrees of the power ratios and the functional connection network; a navigation positioning module configured to determine a multi-target transcranial direct current stimulation position based on a three-dimensional navigation model constructed based on individual brain structure image data of the subject and the abnormal functional sub-networks; a transcranial direct current stimulation scheme generation module comprising a pre-set machine learning model configured to generate a multi-target collaborative stimulation scheme by taking the age, the deviation degree and the multi-target transcranial direct current stimulation position of the subject as input features; and a transcranial direct current stimulation execution module configured to implement the multi-target collaborative stimulation scheme on the subject.

[0007] By adopting the above technical solutions, the brain electrical data acquisition module first acquires brain electrical data of a subject, and the functional network analysis module then performs frequency spectrum analysis and functional connection analysis on the data, wherein the frequency spectrum analysis calculates power ratios of each brain electrical wave at different electrode positions, and the functional connection analysis constructs a functional network by calculating connection strengths between brain regions. After comparing the two analysis results with standard data of the same age group, abnormal functional sub-networks can be identified and deviation degrees can be quantified. The navigation positioning module determines a multi-target stimulation position based on the individual three-dimensional navigation model and the identified abnormal sub-networks, thereby improving the spatial accuracy of the multi-target transcranial direct current stimulation position. The transcranial direct current stimulation scheme generation module takes the age, the deviation degree and the multi-target position as input features, and generates a personalized collaborative stimulation scheme through a machine learning model, thereby improving the accuracy of current parameter combinations corresponding to electrode layout schemes for different individuals in transcranial direct current stimulation.

[0008] In some embodiments combined with the first aspect, in some embodiments, the preset machine learning model training process in the transcranial direct current stimulation scheme generation module comprises: a decision tree ensemble construction unit configured to construct a gradient boosting decision tree model based on the age, the deviation, the multi-target transcranial direct current stimulation position, and the effective historical multi-target synergistic stimulation scheme of the subject; a gradient boosting training unit configured to perform iterative training by using a negative gradient as an approximation of a residual, and in each iteration, train a new decision tree to fit the residual generated by the last round of prediction; a feature importance evaluation unit configured to determine the weight contribution of the age, the deviation, and the multi-target transcranial direct current stimulation position of the subject in the prediction by calculating the number of splits and the information gain of each input feature in all decision trees; and a stimulation parameter output unit configured to weight and sum the prediction results of all decision trees according to a preset learning rate, and output a multi-target synergistic stimulation scheme composed of the current intensity, polarity, and stimulation duration parameters of each electrode.

[0009] By adopting the above technical solution, the gradient boosting training unit trains a new decision tree to fit the residual of the last round in each iteration, while the feature importance evaluation unit calculates the number of splits and the information gain of each input feature in parallel, so that the model training process and the feature weight determination process are performed synchronously, reducing the total training time of model construction. The decision tree ensemble construction unit combines multiple weak learners into a strong learner through the gradient boosting mechanism, which can capture the complex nonlinear relationship between age, deviation, and target position compared to a single prediction model. The stimulation parameter output unit weights and sums the prediction results of all decision trees according to a preset learning rate, directly outputs the accurate values of current intensity, polarity, and stimulation duration, skips the multiple manual debugging and verification steps in traditional methods, reduces the uncertainty in the parameter determination process, and improves the prediction accuracy and generation efficiency of the multi-target synergistic stimulation scheme.

[0010] In some embodiments combined with the first aspect, in some embodiments, the method further comprises an individual difference adaptation module, comprising: a head geometry feature extraction unit configured to measure geometry parameters from the brain structure image data of the subject, the geometry parameters at least including head circumference, skull thickness, cerebrospinal fluid thickness, and gray matter volume; an individualized conductivity calculation unit configured to calculate the conductivity of each layer of the brain of the subject based on the statistical relationship between age and brain tissue conductivity; an electric field conduction efficiency evaluation unit configured to calculate the electric field conduction efficiency coefficient of the electrode to each target brain region based on the conductivity of each layer of the brain of the subject and the head geometry features; and a stimulation parameter individualized adjustment unit configured to individually adjust the multi-target synergistic stimulation scheme output by the preset machine learning model according to the electric field conduction efficiency coefficient, and increase the current intensity by a preset adjustment ratio for individuals with a conduction efficiency less than a preset efficiency threshold.

[0011] By adopting the technical scheme, a head model highly realistic in anatomical morphology and physical properties is constructed for each subject through head geometric feature extraction and individualized conductivity calculation. Based on the model, the electric field conduction efficiency evaluation unit simulates the electric field distribution actually generated in the head of the subject under the preset stimulation scheme by the finite element method and quantifies the conduction efficiency coefficient. Finally, the stimulation parameter individualized adjustment unit compensatively adjusts the current intensity according to the coefficient, thereby improving the accuracy of the transcranial direct current stimulation current intensity for different individuals.

[0012] In some embodiments in combination with the first aspect, in some embodiments, the electric field conduction efficiency evaluation unit comprises: a finite element mesh generation subunit, configured to divide the three-dimensional navigation model into tetrahedral mesh units with a side length not greater than a preset mesh size; a boundary condition setting subunit, configured to map the electrode position generated by the transcranial direct current stimulation scheme module to the tetrahedral mesh units constituting the scalp surface as a voltage boundary condition, and set other regions of the scalp as zero current density boundary conditions, the other regions being regions other than the electrode position; a potential distribution solving subunit, configured to solve the Laplace equation by using the conjugate gradient method, calculate the potential values of each mesh node by the constraint condition that the divergence of the conductivity and the potential gradient is zero; and a conduction efficiency calculation subunit, configured to calculate the conduction efficiency coefficient by the ratio of the abnormal function sub-network electric field gradient amplitude to the electrode current.

[0013] By adopting the technical scheme, the complex three-dimensional navigation model is discretized into high-quality tetrahedral meshes by the finite element mesh generation subunit. Subsequently, the boundary condition setting subunit converts the physical stimulation (voltage or current) of the electrode into mathematical constraints of the model. The potential distribution solving subunit solves the Laplace equation by using the efficient conjugate gradient method, thereby improving the efficiency of calculating the potential distribution of the whole brain in the huge mesh system.

[0014] In some embodiments in combination with the first aspect, in some embodiments, the data preprocessing module further comprises: an artifact identification unit, configured to identify artifact components of the electroencephalogram data by using an independent component analysis method, the artifact components at least including artifact components caused by blinking, electromyography and power supply interference; an artifact removal unit, configured to filter out the artifact components according to a preset artifact removal threshold; a data quality evaluation unit, configured to score the quality of the preprocessed electroencephalogram data based on signal-to-noise ratio, spectral characteristics and signal integrity; and a preprocessing parameter optimization unit, configured to adjust the preset artifact removal threshold according to the data quality score result.

[0015] By adopting the technical scheme, the artifact identification and removal unit separates and removes the noise interference in the electroencephalogram data by using the independent component analysis method. The data quality evaluation unit quantifies the signal-to-noise ratio and integrity of the preprocessed data to form an objective quality score. The score is then used as a feedback signal by the preprocessing parameter optimization unit to adaptively adjust the parameters of filtering and artifact removal, thereby improving the purity of the electroencephalogram data.

[0016] In some embodiments of the first aspect, in some embodiments, the functional network analysis module comprises: an electroencephalogram signal extraction unit configured to extract electroencephalogram signals of preset frequency bands from the electroencephalogram data, the preset frequency bands at least including delta waves, theta waves, alpha waves, beta waves, and gamma waves; a spectrum analysis unit configured to perform frequency domain conversion on the electroencephalogram data using fast Fourier transform, calculate power spectral densities of each electrode position in each frequency band, and normalize to obtain power proportions of each electroencephalogram wave at each electrode position; a functional connectivity calculation unit configured to calculate functional connectivity strengths between brain regions in each frequency band using a coherence analysis method, and construct a multi-band functional connectivity network; a deviation calculation unit configured to statistically compare the power proportions and the multi-band functional connectivity network of the subject with a preset standard electroencephalogram spectrum of the same age group and a preset functional network mode of the same age group, and calculate a deviation; and an abnormal functional sub-network identification unit configured to identify a combination of brain regions with abnormal power distribution and abnormal connection mode based on the deviation, and determine an abnormal functional sub-network.

[0017] By adopting the technical scheme, the deviation calculation unit statistically compares the power proportions and the multi-band functional connectivity network of the subject with the preset standard electroencephalogram spectrum of the same age group and the preset functional network mode of the same age group, and converts the physiological measurement values into a deviation with statistical significance. The abnormal functional sub-network identification unit improves the efficiency of integrating abnormal information of power distribution and connection mode to determine the abnormal functional sub-network based on the deviation.

[0018] In some embodiments of the first aspect, in some embodiments, the deviation calculation unit comprises: a power proportion standard score calculation sub-unit configured to calculate power proportion standard scores of each electrode position in each frequency band and the preset standard electroencephalogram spectrum of the same age group; a functional connectivity standard score calculation sub-unit configured to calculate functional connectivity standard scores of functional connectivity strengths between brain regions in each frequency band and corresponding connections of the preset standard functional network mode of the same age group; and a comprehensive deviation matrix sub-unit configured to combine the power proportion standard scores and the functional connectivity standard scores to form a comprehensive deviation matrix, and use the comprehensive deviation matrix as the deviation.

[0019] By adopting the technical solution, the power ratio standard score calculation subunit and the functional connection standard score calculation subunit convert two brain function indexes (power ratio and connection strength) of different properties and dimensions into a dimensionless common scale of standard scores through statistical methods, so that the abnormal degrees of the two indexes can be directly compared and integrated. The comprehensive deviation matrix subunit combines the two standard scores to form a structured comprehensive deviation matrix. The machine learning model is provided with an input feature with higher information density and richer dimensions, so that the machine learning model can make decisions based on more comprehensive abnormal information, thereby improving the accuracy of the stimulation scheme generation.

[0020] In combination with some embodiments of the first aspect, in some embodiments, the method further comprises a change amount calculation module, including: a post-stimulation data acquisition unit configured to acquire the EEG data of the subject at a preset time point after the stimulation ends; a correspondence analysis processing unit configured to process the post-stimulation EEG data through the same digital filtering processing, spectral analysis and functional connection analysis process as the pre-stimulation EEG data; a wave band power change calculation unit configured to calculate the difference in power ratio before and after the stimulation for each electrode and each frequency band, and quantify the power change degree of each frequency band in each brain region; a network connection change calculation unit configured to calculate the intensity change of the corresponding connection of the functional connection network before and after the stimulation, and identify the connection mode of connection strength enhancement, weakening or recovery; and a comprehensive change amount integration unit configured to integrate the wave band power change and the network connection change into a comprehensive neuromodulation effect evaluation index.

[0021] By adopting the technical solution, the correspondence analysis processing unit analyzes the pre-stimulation and post-stimulation EEG data under the same processing flow, thereby reducing the variables introduced by inconsistent processing methods. The wave band power and network connection change calculation unit quantifies the neurophysiological changes caused by the stimulation from the local and network levels. Finally, the comprehensive change amount integration unit integrates these scattered change amounts into a comprehensive evaluation index, thereby providing a unified evaluation standard for the efficiency comparison between different schemes.

[0022] In combination with some embodiments of the first aspect, in some embodiments, the method further comprises a feedback optimization module configured to update the machine learning model with the neuromodulation effect evaluation index as feedback data.

[0023] By adopting the technical solution, the neuromodulation effect evaluation index is used as feedback data to update the machine learning model, thereby improving the accuracy of the transcranial direct current stimulation scheme output by the transcranial direct current stimulation scheme generation module.

[0024] In some embodiments of the first aspect, in some embodiments, the stimulation safety assurance module further comprises: a current density monitoring unit configured to monitor the current density of the skin under each electrode; an impedance change detection unit configured to measure the impedance of the contact between the electrode and the skin at a preset detection frequency, and determine that the electrode is off or in poor contact when the impedance change value is greater than a preset change threshold; a temperature monitoring unit configured to monitor the temperature change of the skin at the electrode site, and reduce the current intensity by a preset adjustment ratio when the temperature is greater than a first preset temperature threshold; an emergency stop execution unit configured to cut off the power supply of all electrodes within a preset response time when the current density is greater than a preset safe current density threshold, or the impedance change value is greater than the preset change threshold, or the temperature is greater than a second preset temperature threshold; and a safety event recording unit configured to record the information of each safety event and establish a safety archive.

[0025] By using the above technical solution, the key physical parameters of the electrode-skin interface are monitored by the current density, impedance and temperature monitoring units. When the temperature exceeds the first threshold, the current is reduced in proportion to adjust and avoid unnecessary stimulation interruption. When the current density is greater than the preset safe current density threshold, or the impedance change value is greater than the preset change threshold, or the temperature is greater than the second preset temperature threshold, the emergency stop execution unit directly cuts off all power supplies, thereby improving the safety of transcranial direct current stimulation.

[0026] The one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0027] 1. The high-precision transcranial direct current stimulation device first acquires the brain electrical data of the subject through the brain electrical data acquisition module. The functional network analysis module then performs spectral analysis and functional connectivity analysis on the data. The spectral analysis calculates the power proportion of each brain electrical wave at different electrode positions, and the functional connectivity analysis constructs a functional network by calculating the connection strength between brain regions. After comparing these two analysis results with standard data of the same age group, abnormal functional sub-networks can be identified and the deviation degree can be quantified. The navigation positioning module determines the multi-target stimulation position based on the individual three-dimensional navigation model and the identified abnormal sub-network, thereby improving the spatial accuracy of the multi-target transcranial direct current stimulation position. The transcranial direct current stimulation scheme generation module takes age, deviation degree and multi-target position as input features, generates a personalized collaborative stimulation scheme through a machine learning model, and thereby improves the accuracy of the current parameter combination corresponding to the electrode layout scheme for different individuals in transcranial direct current stimulation.

[0028] 2、Since the high-precision transcranial direct current stimulation device trains new decision trees to fit the residual of the last round in each iteration through the gradient boosting training unit, and the feature importance evaluation unit calculates the split times and information gain of each input feature in parallel, the model training process and the feature weight determination process are synchronized, reducing the total training time of model construction. The decision tree ensemble construction unit combines multiple weak learners into a strong learner through the gradient boosting mechanism. Compared with a single prediction model, the ensemble model can capture the complex nonlinear relationship between age, deviation, and target position. The stimulation parameter output unit weights and sums all decision tree prediction results according to the preset learning rate, directly outputs the accurate values of current intensity, polarity, and stimulation duration, and skips the multiple manual debugging and verification steps in traditional methods, thereby reducing the uncertainty in the parameter determination process and improving the prediction accuracy and generation efficiency of the multi-target point collaborative stimulation scheme.

[0029] 3、Since the high-precision transcranial direct current stimulation device constructs a highly realistic head model in terms of anatomical morphology and physical properties for each subject through head geometric feature extraction and individualized conductivity calculation. Based on this model, the electric field conduction efficiency evaluation unit simulates the actual electric field distribution generated by the preset stimulation scheme in the individual's head through the finite element method and quantifies it as a conduction efficiency coefficient. Finally, the stimulation parameter individualized adjustment unit compensates the current intensity based on the coefficient, thereby improving the accuracy of the transcranial direct current stimulation current intensity for different individuals. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 A module architecture of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0031] Figure 2 A data preprocessing module schematic diagram of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0032] Figure 3 A functional network analysis module schematic diagram of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0033] Figure 4 A deviation calculation unit schematic diagram of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0034] Figure 5 An individual difference adaptation module schematic diagram of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0035] Figure 6 An electric field conduction efficiency evaluation unit schematic diagram of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0036] Figure 7A schematic diagram of a stimulation safety guarantee module of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0037] Figure 8 A schematic diagram of a change amount calculation module of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0038] In the figure, 101 is a brain electrical data acquisition module, 102 is a data preprocessing module, 103 is a functional network analysis module, 104 is a navigation positioning module, 105 is a transcranial direct current stimulation scheme generation module, 106 is an individual difference adaptation module, 107 is a transcranial direct current stimulation execution module, 108 is a stimulation safety guarantee module, 109 is a change amount calculation module, 110 is a feedback optimization module, 1021 is an artifact identification unit, 1022 is an artifact removal unit, 1023 is a data quality assessment unit, 1024 is a preprocessing parameter optimization unit, 1031 is a brain electrical signal extraction unit, 1032 is a spectrum analysis unit, 1033 is a functional connection calculation unit, 1034 is a deviation degree calculation unit, 1035 is an abnormal functional sub-network identification unit, 10341 is a power proportion standard score calculation sub-unit, 10342 is a functional connection standard score calculation sub-unit, 10343 is a comprehensive deviation degree matrix sub-unit, 1061 is a head geometric feature extraction unit, 1062 is an individualized conductivity calculation unit, 1063 is an electric field conduction efficiency evaluation unit, 1064 is a stimulation parameter individualized adjustment unit, 10631 is a finite element mesh generation sub-unit, 10632 is a boundary condition setting sub-unit, 10633 is a potential distribution solving sub-unit, 10634 is a conduction efficiency calculation sub-unit, 1081 is a current density monitoring unit, 1082 is an impedance change detection unit, 1083 is a temperature monitoring unit, 1084 is an emergency stop execution unit, 1085 is a safety event recording unit, 1091 is a post-stimulation data acquisition unit, 1092 is a correspondence analysis processing unit, 1093 is a waveband power change calculation unit, 1094 is a network connection change calculation unit, and 1095 is a comprehensive change amount integration unit. DETAILED DESCRIPTION

[0039] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to be limiting of the present application. As used in the specification and the appended claims of the present application, the singular forms "a," "an" and "the" are intended to include both singular and plural forms, unless the context clearly indicates otherwise. It will be further understood that the terms "and / or," as used in the present application, refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0040] Hereinafter, the terms "first", "second", "third", etc. are used only for the purpose of description, and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second", etc. can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0041] The present application provides a high-precision transcranial direct current stimulation device based on a navigation system, to improve the accuracy of electrode layout scheme and corresponding current parameter combination.

[0042] Please refer to Figure 1 , a module architecture diagram of the high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0043] The high-precision transcranial direct current stimulation device comprises:

[0044] The electroencephalogram data acquisition module 101 is configured to acquire electroencephalogram data of a subject.

[0045] The electroencephalogram data refers to the spontaneous and rhythmic electrical activity signals of the brain cortex neuron group recorded by the electrodes placed on the scalp surface, which is a direct reflection of the brain function state. The subject refers to an individual who receives evaluation or regulation by the high-precision transcranial direct current stimulation device.

[0046] Specifically, when performing acquisition, the operator will first wear a multi-channel electroencephalogram acquisition cap on the subject according to the international standard electrode positioning system (such as the 10-20 system), and inject conductive paste between the electrode and the scalp to reduce the contact impedance to a preset low level (for example, less than 5 kilo-ohms). Subsequently, the acquisition program is started to record the voltage signals of all channels synchronously at a relatively high sampling rate (for example, 1000 Hz) and sufficient resolution (for example, 24 bits), and continuously for a preset time (for example, 5 minutes), to obtain data of sufficient length that can reflect the stable brain function state of the subject in resting state or task state.

[0047] The data preprocessing module 102 is configured to preprocess the acquired electroencephalogram data.

[0048] The preprocessing refers to a series of purification and correction operations on the original electroencephalogram data, aiming to eliminate the mixed noise and artifacts of non-brain origin, and convert them into a format suitable for subsequent analysis.

[0049] Specifically, first, the collected continuous data is band-pass filtered, for example, set the passband to 0.5Hz to 45Hz, to filter out the DC drift and high-frequency noise. Next, a 50Hz (or 60Hz) notch filter is applied to eliminate the power frequency interference introduced by the power line. Then, blind source separation techniques such as independent component analysis (ICA) are used to decompose the mixed electroencephalogram signals into multiple independent components. By analyzing the waveform, frequency spectrum and scalp topography characteristics of these components, the artifacts representing eye movement (EOG), muscle activity (EMG) or electrocardiogram (ECG) are automatically or semi-automatically identified and removed or corrected from the original data. Finally, the data is segmented, poor quality data segments are removed, and all channels are re-referenced, for example, converted to a whole brain average reference, to eliminate the bias caused by the selection of reference electrodes.

[0050] The functional network analysis module 103 performs spectral analysis and functional connectivity analysis on the electroencephalogram data. The spectral analysis obtains the power proportion of each electroencephalogram wave at each electrode position. The functional connectivity analysis calculates the functional connectivity strength between brain regions, constructs a functional connectivity network, and compares the power proportion and functional connectivity network with the pre-set standard electroencephalogram spectrum and functional network mode of the same age group to identify abnormal functional sub-networks and calculate the deviation degree of the power proportion and functional connectivity network.

[0051] The spectral analysis is used to represent the calculation of the energy or power size of different electroencephalogram waves in different brain regions. The functional connectivity analysis refers to quantifying the degree of cooperation between different brain regions by calculating the statistical correlation between their signals. Electroencephalogram waves represent brain rhythms within a specific frequency range, such as delta waves, theta waves, alpha waves, etc., which are related to different physiological and cognitive states. The power proportion refers to the relative proportion of the power of a certain electroencephalogram wave in the total power. The functional connectivity network is a network model composed of brain regions (nodes) and their functional connection strengths (edges). The pre-set standard electroencephalogram spectrum and functional network mode refers to the normal power distribution graph and functional connection mode obtained from a large number of healthy people of the same age as a benchmark. The abnormal functional sub-network represents a collection of brain regions and their connections that deviate from the normal standard in the subject's brain. The deviation degree refers to a quantitative indicator of this deviation, such as a standard score.

[0052] Specifically, after data preprocessing, the module first calculates the power spectral density of each electrode position for each frequency band using fast Fourier transform (FFT) or Welch's method, and normalizes it to obtain the power ratio of each frequency band in each brain region. Second, the coherence, phase synchronization index or Granger causality of the signals in each frequency band between different electrode pairs is calculated to obtain a functional connectivity strength matrix, thereby constructing a multi-band functional connectivity network. Then, the power ratio of each frequency band in each brain region and the functional connectivity strength of the subject are compared with the standard atlas and pattern of the same age group stored in the database point by point and edge by edge (such as calculating the standard score), and the set of standard scores constitutes the deviation degree. Finally, the brain regions and connections with deviation degree exceeding the preset threshold (such as |standard score|>1.96) are identified to form a visual abnormal functional subnetwork.

[0053] The navigation positioning module 104 is used to determine the multi-target transcranial direct current stimulation position based on the three-dimensional navigation model constructed based on the individual brain structure image data of the subject and the abnormal functional subnetwork.

[0054] Among them, the individual brain structure image data generally refers to the high-resolution three-dimensional image obtained by magnetic resonance imaging (MRI) scanning, which can show the anatomical structures of the subject's brain gray matter, white matter, cerebrospinal fluid, etc. The three-dimensional navigation model refers to the digital model reconstructed in the computer, which is consistent with the geometric morphology of the subject's real head and brain, serving as the basis for spatial positioning. The multi-target transcranial direct current stimulation position refers to a plurality of specific coordinates determined on the scalp surface for placing stimulation electrodes, aiming to jointly regulate the abnormal functional subnetwork by applying current through these positions.

[0055] Specifically, the module is executed after the functional network analysis module identifies the abnormal functional subnetwork. First, the weighted MRI data of the subject is imported into the navigation software, and a multi-layer three-dimensional navigation model including the scalp, skull, cerebral cortex, etc. is generated through image segmentation and surface reconstruction technology. Next, the abnormal functional subnetwork determined in the previous step is projected or fused onto the individualized three-dimensional cerebral cortex surface through spatial registration algorithm. Finally, considering the spatial distribution of the abnormal subnetwork, the placement positions of the multiple anode and cathode electrodes that can optimally cover these abnormal brain regions are planned on the scalp model automatically or under the guidance of experts, i.e. the multi-target stimulation position is determined.

[0056] The transcranial direct current stimulation scheme generation module 105 includes a preset machine learning model, which is used to generate a multi-target collaborative stimulation scheme with the age, deviation degree and multi-target transcranial direct current stimulation position of the subject as input features.

[0057] The preset machine learning model refers to an algorithm model that can simulate and predict complex relationships through a large amount of historical input and output data, such as gradient boosting decision tree or deep neural network. The input features represent the basis for the model to make decisions, which include the subject's basic information, quantitative indicators of functional state, and stimulation target information. The multi-target point collaborative stimulation scheme refers to a complete instruction set containing specific parameters to be set on all target electrodes, which usually includes current intensity, polarity (anode or cathode) of each electrode, and stimulation duration.

[0058] Specifically, the preset machine learning model uses the age data, deviation data, and multi-target point transcranial direct current stimulation position data of historical subjects as the training set, and learns the specific pattern of deviation data and electrode polarity and current intensity, the distribution of multi-target point transcranial direct current stimulation position and the current collaborative relationship between different electrodes. The age data is used as a tuning variable to fine-tune the intensity and duration of the overall scheme.

[0059] Based on the specific pattern of deviation data and electrode polarity and current intensity, the distribution of multi-target point transcranial direct current stimulation position and the current collaborative relationship between different electrodes of historical subjects, the preset machine learning model predicts a multi-target point collaborative stimulation scheme based on the input age, deviation, and multi-target point transcranial direct current stimulation position of the subject.

[0060] The predicted multi-target point collaborative stimulation scheme is then compared with the corresponding effective historical multi-target point collaborative stimulation scheme, and a preset loss function is used to quantify the gap or error between the two. By continuously adjusting the parameters within the model (such as weights in neural networks or structure of decision trees), the value of this loss function is minimized.

[0061] Through repeated prediction-comparison-adjustment cycles on all training data, the multi-target point collaborative stimulation scheme output by the model will continuously and systematically approach the historical multi-target point collaborative stimulation schemes that have been verified as best practices.

[0062] The following is an example of a pre-trained gradient boosting decision tree (GBDT) regression model, and the implementation steps are as follows:

[0063] 1. Input data vectorization processing:

[0064] First, this module receives the deviation matrix (for example, a 64x64 floating-point matrix) and performs flattening processing, i.e., converting it into a one-dimensional floating-point vector with a length of 4096 in row-major or column-major order.

[0065] Then, the module receives M three-dimensional position coordinates, and concatenates them into a one-dimensional float number vector with length Mx3, in a pre-defined order (e.g. by electrode number).

[0066] Finally, the subject's age (a single number), the flattened deviation vector, and the concatenated position vector are concatenated into a one-dimensional feature vector X, in the order determined during model training.

[0067] 2. Model loading and inference execution:

[0068] Load a pre-trained and saved model file (e.g. a scikit-learn GBDT model object serialized by Python's joblib or pickle library) from local storage (e.g. hard disk).

[0069] Call the predict method of the model object, and input the feature vector X generated in the input data vectorization process as an input parameter, i.e. execute Y_raw = model.predict(X). The model can learn the potential relationship between different output parameters inside the model. The model will sequentially calculate through hundreds of decision trees, and the outputs of all trees will be weighted and summed according to the pre-defined learning rate, to finally obtain an original numerical output vector Y_raw. The dimension of the vector and the meaning of each element are pre-defined by the model training phase.

[0070] 3. Output vector analysis and post-processing:

[0071] The module parses the original output vector Y_raw according to the pre-defined decoding rules. For example, assume that the model is designed to output 2 parameters (current intensity, polarity) for each of the M electrodes, and additionally output 1 global time parameter, then the length of Y_raw is Mx2+1.

[0072] The module processes the values in Y_raw one by one: for current intensity values, apply boundary constraints to clip them to the clinically safe range (e.g. [0.5, 2.0] mA), and optionally quantize them to 0.1 mA steps; for polarity values (a continuous number), apply threshold judgment, e.g. greater than 0 is set to anode, otherwise to cathode. For time values, also perform range constraints (e.g. seconds) and take the integer part.

[0073] 4. Structured protocol generation:

[0074] Combine the various parameters obtained after post-processing, which have clear physical meaning, with the corresponding electrode ID and position information, to generate a structured data object (e.g. in JSON format). This object is the final multi-target point coordinated stimulation protocol, which is then sent to the transcranial direct current stimulation execution module (107) for execution.

[0075] An individual difference adaptation module 106 is configured to adjust the multi-target synergistic stimulation scheme according to individual physiological characteristics of the subject.

[0076] Here, the individual physiological characteristics mainly refer to individual-specific anatomical and biophysical parameters that affect the conduction of electric current in the head, such as head circumference, skull thickness, cerebrospinal fluid layer thickness, and electrical conductivity of different tissues.

[0077] Specifically, the module is executed after the transcranial direct current stimulation scheme generation module generates the preliminary scheme and before the stimulation is executed. First, the key geometric parameters affecting the electric field distribution are measured by using the individualized three-dimensional head model constructed by the navigation positioning module. At the same time, the electrical conductivity of each layer of tissue (such as the scalp, skull, and gray matter) of the subject is queried or estimated from a priori knowledge base according to the age and other information of the subject. Then, based on these geometric parameters and electrical conductivity parameters, a fast electric field simulation is performed on the preliminary stimulation scheme by using the finite element method (FEM), and the conduction efficiency coefficient of the actual current reaching the target brain region under the current parameters is calculated. Finally, the calculated conduction efficiency is compared with an ideal efficiency threshold value, and if the conduction efficiency is lower than the threshold value, the current intensity in the stimulation scheme is increased by a preset proportion or function relationship to compensate for the conduction loss caused by individual differences.

[0078] A transcranial direct current stimulation execution module 107 is configured to implement the multi-target synergistic stimulation scheme on the subject.

[0079] Here, the implementation refers to the process of converting the digital stimulation scheme finally determined by the transcranial direct current stimulation scheme generation module 105 and the individual difference adaptation module 106, which contains parameters such as current intensity, polarity, and duration, into precise and stable simulated direct current, and applying it to the subject through the electrodes placed at the target positions on the scalp.

[0080] Specifically, the module is executed after the transcranial direct current stimulation scheme generation module 105 and the individual difference adaptation module 106 are completed. First, the operator places and fixes the stimulation electrodes on the scalp of the subject according to the positions given by the navigation positioning module. Then, the finally determined multi-target synergistic stimulation scheme is loaded into the main control unit of the stimulator. When the stimulation is started, multiple independent constant current sources inside the stimulator will accurately output direct current of specified intensity and polarity for each electrode channel according to the parameters set in the scheme. During the entire stimulation process, the main control unit will strictly control the total duration of the stimulation and automatically stop the current output gently when the preset time is reached to complete a complete stimulation.

[0081] A stimulation safety guarantee module 108 is configured to monitor and guarantee the safety of the subject during the stimulation execution process.

[0082] The change amount calculation module 109 is configured to quantitatively evaluate the neuromodulation effect after the stimulation is completed.

[0083] Specifically, the module is triggered after the transcranial direct current stimulation execution module completes a complete stimulation. The workflow is as follows: first, at a preset time point (for example, 5 minutes) after the stimulation is completed, the electroencephalogram data acquisition module 101 is called again to acquire a period of resting-state electroencephalogram data which is completely identical to the pre-stimulation condition. Then, the electroencephalogram data after the stimulation is sent to the data preprocessing module 102 and the functional network analysis module 103 which are completely identical to the pre-stimulation to obtain the power ratio and the functional connection network after the stimulation. Finally, the analysis results before and after the stimulation are compared in pairs: the power ratio difference is calculated electrode by electrode and frequency band by frequency band to obtain the wave band power change amount. The functional connection strength difference is calculated connection by connection to obtain the network connection change amount. These change amounts can be further integrated into one or more comprehensive neuromodulation effect evaluation indexes.

[0084] The feedback optimization module 110 is configured to update the machine learning model by taking the neuromodulation effect evaluation index as feedback data.

[0085] The feedback data refers to a set of data records containing the complete information of a stimulation (including the pre-stimulation state and the stimulation scheme used) and the exact effect (i.e., the neuromodulation effect evaluation index) generated by the stimulation. Updating the machine learning model refers to the process of retraining or fine-tuning the preset machine learning model using new feedback data to continuously improve its prediction ability and decision level, which is the key to realizing system adaptation and intelligence.

[0086] Specifically, the module is triggered after the change amount calculation module produces the quantitative evaluation index. First, the complete data of this regulation is packaged into a new training sample, which includes the age of the subject, the deviation before the stimulation, the stimulation position and the stimulation parameters of the multi-target stimulation used as input features, and the neuromodulation effect evaluation index calculated by the change amount calculation module as the output label (ground truth). Then, the new training sample is added to the historical training data set of the model. Finally, according to the preset strategy (for example, every 10 new data collected, or during the idle time every day), the updated data set is called to perform incremental training or complete retraining on the machine learning model in the transcranial direct current stimulation scheme generation module 105. Through this process, the model continuously learns from new successful or failed cases, so that it can generate more accurate and effective stimulation schemes in the future when facing similar situations.

[0087] The above describes the module architecture of the high-precision transcranial direct current stimulation device in the embodiment of the application, and each module is described below.

[0088] Reference will now be made to Figure 2 , the data preprocessing module of the high-precision transcranial direct current stimulation device in the embodiment of the present application.

[0089] The data preprocessing module 102 comprises:

[0090] The artifact identification unit 1021 is configured to identify artifact components of the electroencephalogram data by using an independent component analysis method, wherein the artifact components at least include artifact components caused by blinking, electromyography and power supply interference.

[0091] The independent component analysis method (ICA) is a blind source separation signal processing technology, and its core idea is to assume that the mixed signals recorded by the scalp electrodes are linearly superimposed by multiple statistically independent source signals (including brain neuron activity signals and artifact signals not from the brain), and the method aims to find the inverse process of the mixing process, so as to separate the independent source signals. The artifact component represents an independent component separated by the ICA method, and the time domain waveform, frequency domain feature or spatial topographic map of the independent component conforms to the typical characteristics of non-brain activities such as blinking, muscle activity, electrocardiogram or power line interference. The blinking artifact usually appears as a high-amplitude, low-frequency pulse-shaped signal occurring in the forehead region. The electromyography artifact appears as a high-frequency, spike-shaped noise, which often occurs on electrodes close to the muscles (such as the temple and forehead). The power interference artifact appears as a sinusoidal waveform signal with an extremely high peak value at 50 Hz or 60 Hz in the frequency spectrum.

[0092] Specifically, the unit is executed after the electroencephalogram data is preliminarily band-pass filtered. The working process is as follows: first, input the multi-channel electroencephalogram data matrix after preliminary filtering into an ICA algorithm model (such as the commonly used Infomax or FastICA algorithm). The algorithm finds a best demixing matrix through iterative calculation, and when the matrix is multiplied by the original signal matrix, the statistical independence of the output signal matrix (i.e. each independent component) is maximized. After separation, the unit extracts features of each separated independent component (IC), calculates the kurtosis of the time sequence (for identifying spike-shaped electromyography artifacts), the power spectral density (for identifying power interference artifacts with specific frequency peaks), and the spatial projection map of the scalp (for identifying blinking artifacts with typical forehead distribution). Finally, by comparing the extracted features with a preset artifact feature template library, or using a pre-trained classifier, each independent component is labeled as a brain signal or a specific type of artifact, and a list of identified artifact components is finally output.

[0093] Artifact removal unit 1022, configured to filter out artifact components according to a preset artifact removal threshold.

[0094] Specifically, this unit is executed immediately after artifact identification unit 1021 finishes its labeling work. Its execution process is as follows: first, this unit receives the unmixing matrix, all separated independent components, and a list of which components are labeled as artifacts. Then, for each independent component labeled as artifact in the list, a filtering operation is performed according to the preset artifact removal threshold. After finishing the processing of all specified artifact components, this unit will multiply the original mixing matrix (the inverse of the unmixing matrix) with the modified independent component matrix (some components have been set to zero or attenuated) to reconstruct a multi-channel electroencephalogram data with artifact influence removed.

[0095] Data quality assessment unit 1023, configured to score the quality of preprocessed electroencephalogram data based on signal-to-noise ratio, spectral features, and signal integrity.

[0096] Specifically, this unit is executed after artifact removal unit 1022 finishes data purification. Its workflow is as follows: first, this unit segments the purified data. For each segment of data, it calculates multiple quality indicators. For example, it estimates the signal-to-noise ratio by calculating the ratio of the average power in a specific frequency band (e.g., 1-40 Hz) to the average power in a higher frequency band (e.g., 70-100 Hz, which is usually considered as noise). At the same time, it calculates the power spectrum of each channel, detects whether there are expected physiological peaks (e.g., alpha waves), and fits the spectral slope. In addition, it scans the time series of each channel to estimate the signal integrity by counting the percentage of data points with zero or constant values. Finally, it fuses the calculation results of all channels and all indicators into a final quality score according to preset weights (for example, the weight of the occipital alpha wave can be set higher).

[0097] Preprocessing parameter optimization unit 1024, configured to adjust the preset artifact removal threshold according to the data quality score result.

[0098] Specifically, this unit is triggered after data quality assessment unit 1023 gives a score. Its workflow is as follows: after receiving a quality score less than a preset quality score threshold, if the report shows that the physiological features of the signal (such as alpha waves) become blurred, it means that the artifact removal process is too aggressive, and the optimization unit will decide to increase the preset artifact removal threshold to make it more conservative. After determining the new preset artifact removal threshold, return to the beginning of the preprocessing process and use this new preset artifact removal threshold to reprocess the original data for a complete round of preprocessing. This processing-evaluation-optimization-reprocessing cycle will continue until the quality score is greater than the preset quality score threshold or the maximum number of iterations is reached.

[0099] Referring to Figure 3 FIG. 1 is a schematic diagram of a functional network analysis module of a high-precision transcranial direct current stimulation device according to an embodiment of the present application.

[0100] The functional network analysis module 103 comprises:

[0101] The electroencephalogram signal extraction unit 1031 is configured to extract electroencephalogram signals of preset frequency bands from the electroencephalogram data, wherein the preset frequency bands at least include delta waves, theta waves, alpha waves, beta waves and gamma waves.

[0102] Specifically, the unit is executed after the data purification by the data preprocessing module. The workflow is as follows: the unit applies a set of digital band-pass filters to the preprocessed electroencephalogram signals of each channel in parallel. Each filter is precisely designed to allow only signals of a preset frequency band to pass. For example, an alpha wave filter will filter out all signal components with frequencies below 8 Hz and above 13 Hz, and only retain signals within this range. In order to prevent time delay introduced by the filtering process from affecting subsequent phase analysis, a zero-phase filter is usually used (e.g. achieved by forward and backward filtering twice). After this series of parallel filtering operations, each single-channel wideband signal input will be decomposed into multiple parallel single-channel narrowband signal streams corresponding to different frequency bands, such as delta, theta, alpha, beta and gamma, for subsequent analysis by the units respectively.

[0103] The spectral analysis unit 1032 is configured to perform frequency domain conversion on the electroencephalogram data using fast Fourier transform, calculate the power spectral density of each electrode position in each frequency band, and normalize to obtain the power proportion of each electroencephalogram wave at each electrode position.

[0104] Specifically, the unit is executed after the electroencephalogram signals are extracted into each frequency band (or directly analyzes the wideband signals). The process is as follows: first, the electroencephalogram signals of each channel are segmented into multiple overlapping data segments. Second, a window function (such as a Hanning window) is applied to each data segment to reduce spectral leakage. Then, FFT is performed on each windowed data segment to calculate its power spectrum. Finally, the power spectra of all data segments are averaged to obtain the average power spectral density of the channel. Based on this PSD, the absolute power of each preset frequency band, such as delta, theta, alpha, beta and gamma, can be calculated by integrating the frequency range of each frequency band. Dividing the absolute power of each frequency band by the sum of the absolute powers of all frequency bands gives the power proportion of the frequency band. This process is repeated for all electrode positions.

[0105] The functional connection calculation unit 1033 is configured to calculate the functional connection strength between brain regions in each frequency band using coherence analysis method, and construct a multi-frequency band functional connection network.

[0106] Among them, the coherence analysis method (Coherence) refers to an analysis technique in the frequency domain, which is used to quantify the linear correlation degree of two time series at a specific frequency, and the value is between 0 (completely unrelated) and 1 (completely linearly related). The functional connection strength refers to the value calculated by the coherence analysis method and the like, which represents the synchronization of neural activities of two brain regions. The multi-band functional connection network refers to a set composed of multiple networks, each network corresponding to a frequency band, and the nodes in the network represent brain regions (electrodes), and the weight of the connection (edge) between the nodes is defined by the functional connection strength of the frequency band.

[0107] Specifically, the unit is executed after the electroencephalogram signal is decomposed into each frequency band. Its workflow is: taking the alpha band as an example, the unit will traverse all possible electrode pairs in the brain (for example, for 16 electrodes, there are 120 non-repeated electrode pairs). For each electrode pair (such as F3-P3), the unit will take out their respective alpha band signals. Then, it calculates the cross power spectral density and the respective self-power spectral density of the two signals. Finally, according to the formula: coherence = |cross power spectral density|^2 / (self-power spectral density of signal 1 x self-power spectral density of signal 2), the coherence value of F3 and P3 in the alpha band is calculated. This value serves as the functional connection strength between F3 and P3 in the alpha band. Fill in the calculation results of all electrode pairs into a 16x16 symmetric matrix, and the functional connection network of the alpha band is constructed. This process is repeated for all other frequency bands such as delta, theta, beta, gamma, and the like, and finally a set of multi-band functional connection networks is generated.

[0108] The deviation calculation unit 1034 is configured to statistically compare the power ratio and the multi-band functional connection network of the subject with a preset standard electroencephalogram spectrum of the same age group and a preset functional network mode of the same age group, and calculate the deviation.

[0109] Specifically, the unit is executed after the spectrum analysis and functional connection calculation are completed. Its core working mechanism is: the unit receives the complete power ratio and functional connection strength data of the subject. For each index (for example, the alpha wave power ratio of the O1 electrode, or the functional connection strength of F3-P4 in the beta band), it will initiate a query to a large, age-limited standard database. The database will return the mean and standard deviation of the index in the subject's same age group of healthy people. Then, the unit uses the standard score (Z-score) calculation method to operate the subject's index value with the mean and standard deviation of the population, so as to convert each original index into a standardized deviation score. This process will systematically cover all power and connection indicators, and finally generate a comprehensive deviation score matrix that can reflect the deviation of the subject's brain function indicators from the normal condition.

[0110] The abnormal function sub-network identification unit 1035 is configured to identify the abnormal function sub-network based on the deviation degree, and determine the abnormal function sub-network according to the abnormal power distribution and the abnormal connection mode of the brain region combination.

[0111] Specifically, the unit is executed after the deviation degree is calculated. The workflow is as follows: first, the unit receives the overall deviation score matrix. Then, a significance level threshold is set, which usually corresponds to a statistical p value (for example, the absolute value of the deviation score is greater than 1.96, which corresponds to a two-sided p<0.05). Then, all the deviation scores are traversed: all the electrode positions with the power proportion deviation score exceeding the threshold are marked as abnormal nodes, and all the connections with the functional connection strength deviation score exceeding the threshold are marked as abnormal edges. Finally, the graph structure composed of all the marked abnormal nodes and abnormal edges is defined as the abnormal function sub-network of the subject. The sub-network intuitively shows the core of the subject's brain function abnormality.

[0112] Please refer to Figure 4 The deviation degree calculation unit of the high-precision transcranial direct current stimulation device in the embodiment of the present application.

[0113] The deviation degree calculation unit 1034 includes:

[0114] The power proportion standard score calculation sub-unit 10341 is configured to calculate the power proportion standard score of each electrode position and each frequency band of the subject relative to the power proportion of the preset standard electroencephalogram of the same age group.

[0115] Specifically, the sub-unit is executed after the power proportion data of the subject is obtained after the spectrum analysis is completed. The execution process is as follows: first, the sub-unit receives a matrix representing the power proportion of the subject (for example, the dimension is 16 electrodes x 6 frequency bands). Then, for each element in the matrix (for example, the power proportion of the alpha wave of the O1 electrode position), it queries the standard electroencephalogram database, and extracts the power proportion mean (μ) and standard deviation (σ) of the normal population corresponding to the specific position and frequency band according to the age of the subject. Then, the standard score (Z-score) calculation formula is applied: Z=(subject observation value-population mean) / population standard deviation. This calculation generates a corresponding Z-score for each element in the matrix. Finally, the sub-unit outputs a matrix with the same dimension as the input matrix, but the numerical content has been converted to power proportion standard scores.

[0116] The functional connection standard score calculation sub-unit 10342 is configured to calculate the functional connection standard score of the functional connection strength between brain regions and the corresponding connection of the preset standard functional network mode of the same age group.

[0117] Specifically, the sub-unit is executed after the functional connectivity analysis is completed and the functional connectivity strength data of the subject is obtained. The execution process is as follows: first, the sub-unit receives a set of matrices representing the functional connectivity strength of the subject (for example, a 16x16 connectivity matrix for each frequency band). Then, all non-repeated brain region connections in each matrix are traversed (for example, the connection of Fp1-Fp2, the connection of Fp1-F7, etc.). For each specific connection (for example, the connection strength of P3-P4 in the beta band), it queries the standard functional network mode database to extract the mean (μ) and standard deviation (σ) of the normal population corresponding to the specific connection in the frequency band according to the subject's age. Then, the Z-score of the connection is calculated by applying the standard score formula. This process calculates a corresponding Z-score for all brain region connections and all frequency bands, and finally outputs a set of matrices with the same dimension as the input matrix, but the numerical content has been completely converted to functional connectivity standard scores.

[0118] The comprehensive deviation matrix sub-unit 10343 is used to combine the power proportion standard scores and the functional connectivity standard scores to form a comprehensive deviation matrix, and the comprehensive deviation matrix is used as the deviation.

[0119] Specifically, the sub-unit is the last step of the deviation calculation process and is executed after the power proportion and functional connectivity standard scores have been calculated. The execution process is as follows: first, the sub-unit receives the power proportion standard score matrix (for example, 16x6 dimensions) and the functional connectivity standard score matrix (for example, 6 16x16 matrices). Then, it performs a predefined "flattening and splicing" operation. For example, it first flattens the 16x6 power proportion standard score matrix by column or by row to form a vector with a length of 96. Then, for each functional connectivity standard score matrix, it extracts the non-repeated connection scores in the upper triangular or lower triangular part (for 16 electrodes, there are 120 non-repeated connections), and flattens and splices the connection scores of all 6 frequency bands to form a vector with a length of 720 (120x6). Finally, it connects the two vectors (the power deviation vector with a length of 96 and the connection deviation vector with a length of 720) end to end to form a single vector with a length of 816. This long vector containing all the deviation information is defined as the comprehensive deviation matrix (or vector).

[0120] Please refer to Figure 5 , the individual difference adaptation module schematic diagram of the high-precision transcranial direct current stimulation device in the embodiment of the application.

[0121] The individual difference adaptation module 106 includes:

[0122] The head geometry feature extraction unit 1061 is configured to measure the geometric parameters from the brain structure image data of the subject, the geometric parameters at least including the head circumference, skull thickness, cerebrospinal fluid thickness and gray matter volume.

[0123] Specifically, the unit is executed after the navigation positioning module constructs the preliminary three-dimensional model and before the electric field simulation. The workflow is as follows: first, the unit receives the MRI data of the individual and calls the automatic segmentation algorithm in the mature neuroimaging processing software (such as FreeSurfer, SPM, etc.) to classify each voxel in the three-dimensional image and accurately divide the boundaries of different tissues such as scalp, skull, cerebrospinal fluid, gray matter and white matter. Based on the segmentation results, the unit will perform a series of measurements: the head circumference is calculated by extracting the contour of a specific plane on the scalp surface model; the skull thickness is obtained by calculating the average distance between the inner and outer surfaces of the skull; the cerebrospinal fluid thickness is calculated in the same way; the total gray matter volume is calculated by counting the number of voxels labeled as gray matter and multiplying by the unit voxel volume. Finally, the unit outputs a set of quantitative geometric parameters that accurately describe the unique anatomical morphology of the subject's head.

[0124] The individualized conductivity calculation unit 1062 is configured to calculate the electrical conductivity of each layer of tissue in the brain of the subject based on the statistical relationship between age and brain tissue electrical conductivity.

[0125] Specifically, the unit is executed after the head geometry feature extraction and before the electric field simulation. The execution process is as follows: first, the unit receives the exact age of the subject. Then, access an internal parameter database that stores the functional relationship between tissue conductivity and age. For skull tissue, which has the most significant change in conductivity with age, the unit will substitute the subject's age into the pre-set regression equation (for example: σ_skull=-0.0001*age+0.018) to calculate an age-adapted skull conductivity value. For other tissues such as scalp, cerebrospinal fluid, gray matter and white matter, their conductivity changes less with age, and this unit directly uses the standard values widely accepted in the literature, or also uses an age-based fine-tuning function. Finally, the unit outputs a complete set of tissue conductivity parameters tailored for the subject.

[0126] The electric field conduction efficiency evaluation unit 1063 is configured to calculate the electric field conduction efficiency coefficient from the electrode to each target brain region based on the individualized conductivity and head geometry features using the finite element method.

[0127] Among them, the finite element method (Finite Element Method, FEM) refers to a powerful numerical calculation technology, which discretizes complex continuous geometry (such as the head) into millions of tiny, regular-shaped units (such as tetrahedrons), and then approximates the physical field distribution of the entire object by solving the physical equation (here, the electrical equation) on each unit. The target brain area refers to the specific cerebral cortex region where the abnormal functional sub-network determined by the functional network analysis module as the neuroregulation target. The electric field conduction efficiency coefficient is a key quantitative index, which represents how strong the electric field (in volts / meter) can be generated in the target brain area per unit input current (such as per milliampere), directly reflecting the conduction performance of the individual's head structure to the stimulating current.

[0128] Specifically, this unit is executed after the preliminary stimulation scheme is generated and before the final parameter adjustment. Its detailed process is as follows: First, this unit integrates the outputs of the previous two units, that is, based on the individual's geometric characteristics, a high-precision head finite element mesh model is constructed, and individualized conductivity values corresponding to the units in the model belonging to different tissues are assigned. Then, the electrode positions and current intensities defined in the preliminary stimulation scheme are taken as the boundary conditions of the model. Next, the finite element solver is started to calculate the electric potential distribution in the entire head model under these conditions. By taking the negative gradient of the electric potential distribution, a three-dimensional electric field vector field is obtained. Finally, the unit takes the target brain area as the region of interest (ROI) and calculates the average value of the electric field intensity in this area. Divide this average value by the total current applied to obtain the final electric field conduction efficiency coefficient.

[0129] The stimulation parameter individualization adjustment unit 1064 is configured to individually adjust the multi-target point collaborative stimulation scheme output by the preset machine learning model according to the electric field conduction efficiency coefficient, and increase the current intensity of the individual whose conduction efficiency is less than the preset efficiency threshold by a preset adjustment ratio.

[0130] Specifically, this unit is executed immediately after the electric field conduction efficiency is evaluated. Its workflow is as follows: First, the unit receives the calculated individual electric field conduction efficiency coefficient and compares it with the preset efficiency threshold. If the efficiency coefficient of the individual is greater than or equal to the threshold, it means that the preliminary scheme is expected to produce a strong enough electric field on this individual, so the scheme is not modified and is passed directly. Conversely, if the efficiency coefficient of the individual is less than the threshold, the unit will start the adjustment program. According to the preset adjustment ratio (for example, a simple inverse relationship: adjusted current = original current x (efficiency threshold / individual efficiency coefficient)), an electric current gain factor is calculated. Then, multiply the current intensity of all electrodes in the preliminary scheme by this gain factor.

[0131] Please refer to Figure 6Fig. 1 is a schematic diagram of an electric field conduction efficiency evaluation unit of a high-precision transcranial direct current stimulation device in the embodiments of the present application.

[0132] The electric field conduction efficiency evaluation unit 1063 comprises:

[0133] The finite element mesh generation subunit 10631 is configured to divide the three-dimensional navigation model into tetrahedral mesh units with a side length not greater than a preset mesh size.

[0134] Specifically, the subunit is executed immediately after obtaining the three-dimensional navigation model composed of multiple closed surfaces representing the individual's anatomical structure. The workflow is as follows: first, the subunit receives surface mesh data describing the boundaries of tissues such as the scalp, skull, and brain. Then, a mesh generation algorithm (such as the constrained Delaunay triangulation algorithm) is called to fill and subdivide the volume space enclosed by the surfaces. During the subdivision process, the algorithm adheres to the preset mesh size constraint, ensuring that the longest side of any tetrahedral unit generated does not exceed the size. In addition, to ensure the stability of numerical calculations, the algorithm also optimizes the mesh quality, such as avoiding the generation of excessively flat or elongated distorted units. Finally, the subunit outputs a volume mesh model composed of millions of tetrahedral units that can fully describe the multi-layer structure of the head, with each unit labeled with its corresponding tissue type.

[0135] The boundary condition setting subunit 10632 is configured to map the electrode positions generated by the transcranial direct current stimulation scheme module to the tetrahedral mesh units constituting the scalp surface, setting the electrode positions as voltage boundary conditions, and setting the other areas of the scalp as zero current density boundary conditions.

[0136] Specifically, the subunit is executed after the finite element mesh generation and before the solver is started. The execution process is as follows: first, the subunit identifies all the tetrahedral mesh units covered by the anode and cathode electrodes on the scalp surface mesh according to the electrode center and size determined by the navigation positioning module. Then, all tetrahedral mesh units belonging to the anode are assigned a preset potential value (such as +V), and all tetrahedral mesh units belonging to the cathode are assigned another potential value (such as -V or 0). For the remaining tetrahedral mesh units on the scalp surface that do not belong to any electrode, the subunit applies a zero normal current density constraint. These set boundary conditions, together with the Laplace equation describing the internal physical laws of the model, form a complete mathematical problem with a unique solution.

[0137] The potential distribution solving subunit 10633 is configured to solve the Laplace equation using the conjugate gradient method, calculate the potential values of each mesh node by the constraint condition that the divergence of conductivity and potential gradient is zero.

[0138] Specifically, the subunit is executed after the mesh generation and boundary condition setting are completed. The workflow is as follows: the finite element method first converts the Laplace equation acting on each micro tetrahedron unit into a linear algebraic equation. Then, through an assembly process, the equations of millions of units are combined into a huge linear equation group in the form of [K] {V} = {F}. Where [K] is the global stiffness matrix, which is determined by the grid geometry and the conductivity. {V} is the unknown vector to be solved, which contains all the node potential values. {F} is the load vector determined by the boundary conditions. Since the [K] matrix is huge, direct inversion is not feasible, at which time the conjugate gradient method is used, starting from an initial guess solution, iteratively searching along a series of mutually conjugate directions, each step making the solution closer to the true value, until the difference between the results of two iterations is less than a small convergence tolerance. Finally, the subunit outputs the converged vector {V}, which is the accurate potential value of all nodes.

[0139] The conduction efficiency calculation subunit 10634 is used to calculate the conduction efficiency coefficient by the ratio of the abnormal function sub-network electric field gradient amplitude to the electrode current.

[0140] Specifically, the subunit is executed after the potential distribution is solved. The workflow is as follows: first, the subunit receives the vector {V} containing all the node potential values. Then, all the grid nodes located in the predefined abnormal function sub-network are identified. For each tetrahedron unit in the abnormal function sub-network, the approximate uniform electric field vector E in the unit is obtained by calculating the spatial gradient of the potential values of its four vertices. Then, the amplitude (i.e. the length |E|) of the electric field vector E of each unit is calculated. Subsequently, the electric field amplitudes of all units in the abnormal function sub-network are weighted and averaged (usually weighted by unit volume) to obtain a value representing the average electric field strength of the abnormal function sub-network. Finally, the average electric field strength value is divided by the total stimulation current set in the boundary conditions to obtain the final electric field conduction efficiency coefficient.

[0141] Please refer to Figure 7 , the stimulation safety guarantee module of the high-precision transcranial direct current stimulation device in the embodiment of the application.

[0142] The stimulation safety guarantee module 108 comprises:

[0143] The current density monitoring unit 1081 is used to monitor the current density of the skin under each electrode.

[0144] The impedance change detection unit 1082 is used to measure the impedance of the electrode in contact with the skin at a preset detection frequency, and determine that the electrode is off or in poor contact when the impedance change value is greater than a preset change threshold.

[0145] The temperature monitoring unit 1083 is configured to monitor the skin temperature change of the electrode site, and reduce the current intensity by a preset adjustment ratio when the temperature is greater than a first preset temperature threshold.

[0146] The emergency stop execution unit 1084 is configured to cut off the power supply of all electrodes within a preset response time when the current density is greater than a preset safety current density threshold, the impedance change value is greater than a preset change threshold, or the temperature is greater than a second preset temperature threshold.

[0147] The safety event recording unit 1085 is configured to record the safety event information each time, and establish a safety file.

[0148] Please refer to Figure 8 , which is a schematic diagram of the change amount calculation module of the high-precision transcranial direct current stimulation device in the embodiment of the present application.

[0149] The change amount calculation module 109 includes:

[0150] The post-stimulation data acquisition unit 1091 is configured to acquire the EEG data of the subject at a preset time point after the stimulation ends.

[0151] The correspondence analysis processing unit 1092 processes the post-stimulation EEG data through the same digital filtering processing, spectral analysis, and functional connectivity analysis process as before the stimulation.

[0152] Specifically, the unit is executed after the post-stimulation data acquisition is completed. The workflow is as follows: first, the unit retrieves all parameter configuration files used when processing the pre-stimulation data of the subject from the system. This file records the specific settings of each step in the data preprocessing module 102 and the functional network analysis module 103 in detail. Then, the unit applies this parameter configuration file to the newly acquired post-stimulation EEG data. The output result is an analysis result consistent with the format, dimension, and index type of the pre-stimulation data, such as a post-stimulation power proportion matrix and a set of post-stimulation multi-band functional connectivity network matrices. These two results can now be directly compared point by point and connection by connection with the pre-stimulation results.

[0153] The waveband power change calculation unit 1093 is configured to calculate the difference in power proportion before and after stimulation for each electrode and each frequency band, and quantify the power change degree of each brain region and each frequency band.

[0154] Specifically, this unit is executed after the correspondence analysis processing unit 1092 generates two sets of power spectrum data before and after stimulation. Its workflow is: this unit receives two power proportion matrices with the same dimensions: P_pre (electrode number x frequency band number) and P_post (electrode number x frequency band number). Then, a simple matrix subtraction operation is performed: ΔP = P_post - P_pre. Each element in the resulting difference matrix ΔP represents the change in power proportion of a specific frequency band on a specific electrode. For example, ΔP(i, j) is positive, indicating that the brain wave activity in the jth frequency band on the ith electrode has relatively increased after stimulation. If it is negative, it means that it has weakened. This unit finally outputs this difference matrix, providing direct evidence of changes in local brain activity for subsequent comprehensive evaluation.

[0155] The network connection change calculation unit 1094 is used to calculate the intensity change of the corresponding connection of the functional connection network before and after stimulation, and identify the connection mode of connection intensity enhancement, weakening or recovery.

[0156] Specifically, this unit is executed after the correspondence analysis processing unit 1092 generates two sets of functional connection network matrices before and after stimulation. Its workflow is: for each frequency band, the unit receives two functional connection matrices C_pre and C_post. First, calculate the intensity change matrix by matrix subtraction ΔC = C_post - C_pre. Then, it classifies each element in ΔC: if the value is greater than a small positive threshold, it is marked as enhanced. If it is less than a negative threshold, it is marked as weakened. For the identification of normal recovery, the unit will also call the deviation matrix of the subject before stimulation. If the connection has a high deviation score before stimulation (for example, |Z|>2, indicating significant abnormalities), and the connection intensity value after stimulation falls near the mean of the normal population distribution (for example, the deviation score |Z|<1 after stimulation), the connection is specially marked as normal.

[0157] The comprehensive change amount integration unit 1095 is used to integrate the wave band power change and the network connection change into a comprehensive neuroregulation effect evaluation index.

[0158] In particular, this unit is executed after all the specific changes have been calculated. Its workflow is as follows: the unit receives the power change matrix ΔΡ and the connection change matrix ΔC. One way of integrating is as follows: first, the nodes and connections that are identified as abnormal functional subnetworks before stimulation are identified. Then, a targeting effect score is calculated, which is a weighted sum of the degree to which these target regions' power and connections are restored to normal. For example, a connection that is abnormally enhanced and is attenuated after stimulation will receive a positive score. At the same time, a non-target region impact score is calculated to assess possible unintended side effects. Finally, these scores can be combined into a single, comprehensive evaluation metric, or presented to the operator in the form of a dashboard.

[0159] The above-described embodiments are merely intended for describing and illustrating, but not limiting the technical solutions of the present application; even though the present application is described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced equivalently; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0160] In the above-described embodiments, according to the context, the term "when" can be interpreted as meaning "if" or "after" or "in response to determining" or "in response to detecting". Similarly, according to the context, the phrase "upon determining" or "if detecting (the stated condition or event)" can be interpreted as meaning "if determining" or "in response to determining" or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)".

[0161] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line) or wireless (such as infrared, wireless, microwave, etc.) manner. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk) and the like.

[0162] Those of ordinary skill in the art understand that all or part of the processes in the above embodiments can be implemented by a computer program to instruct the relevant hardware, which can be stored in a computer readable storage medium. The program can include the processes of the above method embodiments when executed. The aforementioned storage medium includes ROM or random access memory (RAM), magnetic disk or optical disk, and various media that can store program codes.

Claims

1. A high-precision transcranial direct current stimulation device based on a navigation system, characterized by, The method comprises the following steps: A brain electrical data acquisition module is used to acquire brain electrical data of a subject; A functional network analysis module is used to perform spectral analysis and functional connection analysis on the brain electrical data, wherein the spectral analysis obtains the power proportion of each brain wave at each electrode position, and the functional connection analysis constructs a functional connection network by calculating the functional connection strength between brain regions, and compares the power proportion and the functional connection network with a preset standard electroencephalogram spectrum and a functional network mode of the same age group to identify abnormal functional sub-networks and calculate the deviation degree of the power proportion and the functional connection network; A navigation positioning module is used to determine the positions of multi-target transcranial direct current stimulation based on a three-dimensional navigation model constructed based on individual brain structure image data of the subject and the abnormal functional sub-networks; A transcranial direct current stimulation scheme generation module includes a preset machine learning model, which is used to generate a multi-target collaborative stimulation scheme by taking the age of the subject, the deviation degree, and the positions of multi-target transcranial direct current stimulation as input features; The transcranial direct current stimulation scheme generation module specifically includes: a decision tree ensemble construction unit, which is used to construct a gradient boosting decision tree model by taking the age of the subject, the deviation degree, the positions of multi-target transcranial direct current stimulation, and effective historical multi-target collaborative stimulation schemes as input features; a gradient boosting training unit, which is used to perform iterative training by taking the negative gradient as an approximation of the residual, and in each iteration, a new decision tree is trained to fit the residual generated by the last round of prediction; a feature importance evaluation unit, which is used to determine the weight contribution of the age of the subject, the deviation degree, and the positions of multi-target transcranial direct current stimulation in prediction by calculating the number of splits and information gain of each input feature in all decision trees; and a stimulation parameter output unit, which is used to weight and sum the prediction results of all decision trees according to a preset learning rate, and output a multi-target collaborative stimulation scheme composed of current intensity, polarity, and stimulation duration parameters of each electrode; A transcranial direct current stimulation execution module is used to implement the multi-target collaborative stimulation scheme on the subject.

2. The high-precision transcranial direct current stimulation device according to claim 1, characterized in that, It also includes an individual difference adaptation module, which comprises: A head geometric feature extraction unit is used to measure geometric parameters from brain structure image data of the subject, wherein the geometric parameters at least include head circumference, skull thickness, cerebrospinal fluid thickness, and gray matter volume; An individualized conductivity calculation unit is used to calculate the conductivity of each layer of the subject's brain based on the statistical relationship between age and brain tissue conductivity; An electric field conduction efficiency evaluation unit is used to calculate the electric field conduction efficiency coefficient of each target brain region from the electrode based on the conductivity of each layer of the subject's brain and the head geometric features by using the finite element method; A stimulation parameter individualized adjustment unit is used to individually adjust the multi-target collaborative stimulation scheme output by the preset machine learning model according to the electric field conduction efficiency coefficient, and increase the current intensity of individuals with a conduction efficiency less than a preset efficiency threshold by a preset adjustment ratio.

3. The high-precision transcranial direct current stimulation device according to claim 2, characterized in that, The electric field conduction efficiency evaluation unit comprises: A finite element mesh generation subunit is configured to divide the three-dimensional navigation model into tetrahedral mesh units with a side length not greater than a preset mesh size; A boundary condition setting subunit is configured to map the electrode positions generated by the transcranial direct current stimulation scheme module to the tetrahedral mesh units constituting the scalp surface as voltage boundary conditions, and set other regions of the scalp as zero current density boundary conditions, the other regions being regions other than the electrode positions; A potential distribution solving subunit is configured to solve the Laplace equation by using a conjugate gradient method, and calculate the potential values of each mesh node by using the constraint condition that the divergence of the conductivity and the potential gradient is zero; A conduction efficiency calculation subunit is configured to calculate a conduction efficiency coefficient by using the ratio of the abnormal function sub-network electric field gradient amplitude to the electrode current.

4. The high-precision transcranial direct current stimulation device of claim 1, wherein, Further comprising a data preprocessing module, comprising: An artifact identification unit is configured to identify artifact components of the electroencephalogram data by using an independent component analysis method, the artifact components at least including artifact components caused by blinking, electromyography, and power supply interference; An artifact removal unit is configured to filter out the artifact components according to a preset artifact removal threshold; A data quality assessment unit is configured to score the quality of the preprocessed electroencephalogram data based on signal-to-noise ratio, spectral characteristics, and signal integrity; A preprocessing parameter optimization unit is configured to adjust the preset artifact removal threshold according to the data quality score result.

5. The high-precision transcranial direct current stimulation device according to claim 1, characterized in that, The functional network analysis module comprises: An electroencephalogram signal extraction unit is configured to extract electroencephalogram signals of a preset frequency band from the electroencephalogram data, the preset frequency band at least including delta waves, theta waves, alpha waves, beta waves, and gamma waves; A spectral analysis unit is configured to perform frequency domain conversion on the electroencephalogram data by using a fast Fourier transform, calculate the power spectral density of each electrode position in each frequency band, and normalize to obtain the power proportion of each electroencephalogram wave at each electrode position; A functional connection calculation unit is configured to calculate the functional connection strength between brain regions in each frequency band by using a coherence analysis method, and construct a multi-frequency band functional connection network; A deviation degree calculation unit is configured to statistically compare the power proportion and the multi-frequency band functional connection network of the subject with a preset standard electroencephalogram spectrum of the same age group and a preset functional network mode of the same age group, and calculate a deviation degree; An abnormal function sub-network identification unit is configured to identify a combination of brain regions with abnormal power distribution and abnormal connection mode based on the deviation degree, and determine an abnormal function sub-network.

6. The high-precision transcranial direct current stimulation device according to claim 5, characterized in that, The deviation degree calculation unit comprises: A power proportion standard score calculation subunit is configured to calculate the power proportion of each electrode position in each frequency band and the power proportion standard score of the preset standard electroencephalogram spectrum of the same age group; A functional connection standard score calculation subunit is configured to calculate the functional connection strength between brain regions in each frequency band and the functional connection standard score of the corresponding connection of the preset standard functional network mode of the same age group; A comprehensive deviation degree matrix subunit is configured to combine the power proportion standard score and the functional connection standard score to form a comprehensive deviation degree matrix, and use the comprehensive deviation degree matrix as the deviation degree.

7. The high-precision transcranial direct current stimulation device of claim 1, wherein, Further comprising a change amount calculation module, comprising: A post-stimulation data acquisition unit is configured to acquire electroencephalogram data of the subject at a preset time point after stimulation ends; The correspondence analysis processing unit processes the brain electrical data after stimulation through the same digital filtering processing, spectral analysis and functional connection analysis process as before stimulation; The wave band power change calculation unit is used to calculate the difference of power proportion before and after stimulation for each electrode and each frequency band, and to quantify the power change degree of each brain region and each frequency band; The network connection change calculation unit is used to calculate the intensity change of the corresponding connection of the functional connection network before and after stimulation, and to identify the connection mode of connection strength enhancement, weakening or recovery to normal; The comprehensive change amount integration unit is used to integrate the wave band power change and the network connection change into a comprehensive neuroregulation effect evaluation index.

8. The high-precision transcranial direct current stimulation device according to claim 7, characterized in that, It also includes a feedback optimization module for updating the machine learning model with the neuroregulation effect evaluation index as feedback data.

9. The high-precision transcranial direct current stimulation device of claim 1, wherein, It also includes a stimulation safety guarantee module, which includes: A current density monitoring unit for monitoring the current density of the skin under each electrode; An impedance change detection unit for measuring the impedance of the electrode and skin contact at a preset detection frequency, and determining electrode shedding or poor contact when the impedance change value is greater than a preset change threshold; A temperature monitoring unit for monitoring the skin temperature change at the electrode site, and reducing the current intensity by a preset adjustment ratio when the temperature is greater than a first preset temperature threshold; An emergency stop execution unit for cutting off the power supply of all electrodes within a preset response time when the current density is greater than a preset safe current density threshold, or the impedance change value is greater than the preset change threshold, or the temperature is greater than a second preset temperature threshold; A safety event recording unit for recording each safety event information and establishing a safety file.

Images