Elelampgenic region positioning method and system based on brain power source imaging and dynamic brain network

Abstract

The invention discloses an epilepsy region positioning method and system based on brain power supply imaging and a dynamic brain network, and the method comprises the steps: obtaining T1 weighted magnetic resonance imaging data of a user, and constructing an individual three-dimensional head model through a boundary element method; acquiring scalp electroencephalogram data of a user, and preprocessing the scalp electroencephalogram data; based on an individual three-dimensional head model, performing inverse problem solving on the preprocessed scalp electroencephalogram data by using a standardized low-resolution brain power source imaging algorithm to obtain source current density signals of 68 brain regions; decomposing into six frequency bands, calculating the power spectrum density of each brain region and carrying out normalization processing, and screening effective frequency bands; based on the source current density signals of the 68 brain regions of the effective frequency band, information flow directions and intensities of different brain regions are calculated by adopting a directional transfer function method, a directional transfer function matrix of the effective frequency band is formed, and a directed brain network is constructed; and calculating a graph theory index and / or an epilepsy index of each brain region, carrying out maximum value normalization analysis, and determining an epilepsy region positioning result.

Description

TECHNICAL FIELD

[0001] The application relates to a method and system for locating an epileptogenic zone based on electrical source imaging and dynamic brain network, and belongs to the technical field of medical instruments. BACKGROUND

[0002] Drug-Resistant Epilepsy (DRE) refers to a subtype of epilepsy in which patients cannot achieve seizure control after receiving two or more standard antiepileptic drug treatments, accounting for about one-third of all epilepsy patients. For patients with focal DRE, surgical resection of the epileptogenic zone (EZ) is one of the most effective treatment methods. However, the postoperative efficacy is highly dependent on the accurate positioning of the EZ, and this positioning process is particularly difficult in patients with magnetic resonance imaging negative (MRI-negative). MRI-negative epilepsy refers to patients who do not have clear epileptogenic lesions in conventional structural imaging, and the seizure-free rate after surgery for this type of patients is generally low. The sensitivity and specificity of traditional imaging and electroencephalogram analysis methods in positioning are limited.

[0003] In recent years, electrical source imaging (ESI) technology has become an important means to break through the positioning difficulties of MRI-negative epilepsy. ESI establishes individualized head models and source models to inversely deduce the location distribution of cortical neural activity sources using scalp electroencephalogram (EEG) signals. This technology includes two core mathematical problems: the forward problem, which is the modeling of current source to scalp potential propagation, and the inverse problem, which is to inversely deduce the optimal source distribution under biophysical constraints. Current mainstream methods such as standardized low-resolution brain electromagnetic tomography (sLORETA) can estimate current density on a whole-brain scale without assuming the number of sources, and have been verified in multiple studies to have high consistency with intracranial electrography (iEEG).

[0004] However, relying solely on sLORETA to complete static source positioning still makes it difficult to reveal the dynamic evolution process of epileptic discharges in neural networks. SUMMARY

[0005] Modern research believes that epilepsy is a "network disease", and the epileptogenic activity is not only derived from isolated cortical points, but also regulated by dynamic brain networks. Therefore, it is necessary to develop a tracing analysis strategy that integrates brain network modeling. Functional connectivity (Functional Connectivity) and effective connectivity (Effective Connectivity) technology can further depict the information interaction mode between source activities, and the directed transfer function (Directed Transfer Function, DTF) as an analysis method based on frequency domain autoregressive model can quantify the information flow intensity and direction between different brain regions.

[0006] On this basis, combined with the graph analysis method, the source space can be constructed into a dynamic directed brain network, and key indicators such as degree centrality (Degree Centrality, DC), in-degree (DCin), and out-degree (DCout) in the network topology structure can be further extracted to reflect the integration ability and driving ability of each brain region in the network. In addition, the seizure index (Seizure Index, SI) can be used as a quantitative index that integrates the power of multiple frequency bands and dynamic connection mode to evaluate the possibility of a region as an epileptogenic source. These parameters provide multi-dimensional and quantifiable diagnostic clues for non-invasive evaluation of EZ.

[0007] The prior art still lacks a set of technical solutions that integrate ESI source positioning, brain network analysis, and graph evaluation, especially in DRE patients with negative preoperative images, there is still a lack of non-invasive evaluation methods that can be stably reproduced, standardized results, and have predictive performance. Current electroencephalogram interpretation relies heavily on visual analysis and experience, and lacks objective quantitative tools, which seriously restricts the promotion and application of precise surgical intervention.

[0008] Therefore, it is urgent to develop a comprehensive positioning technology that integrates sLORETA electroencephalography source localization, multi-frequency power spectrum analysis (PSD), DTF directed network reconstruction, and multi-index graph evaluation to construct an EZ non-invasive identification method suitable for MRI-negative DRE patients. This method should be able to achieve multi-dimensional extraction and quantitative analysis of preoperative source signals and network patterns, identify potential epileptogenic core regions, and can be registered with postoperative resection regions to provide stable, reliable, and non-invasive auxiliary decision support for clinical practice.

[0009] Objective: In view of at least one of the above technical problems, the present application provides a method and system for locating epileptogenic zones based on electroencephalography source imaging and dynamic brain networks to improve the accuracy of epileptogenic zone localization.

[0010] The technical solution adopted by the present application is: In a first aspect, the present application provides a system for locating epileptogenic zones based on electroencephalography source imaging and dynamic brain networks, comprising: an individual head model construction module, configured to: acquire T1-weighted magnetic resonance imaging data of a user; and construct an individual three-dimensional head model according to the T1-weighted magnetic resonance imaging data by using a boundary element method; a scalp electroencephalogram data acquisition module, configured to: acquire scalp electroencephalogram data of the user, and preprocess the scalp electroencephalogram data; a reverse problem solving and source imaging module, configured to: perform reverse problem solving on the preprocessed scalp electroencephalogram data by using a standardized low-resolution electroencephalogram source imaging algorithm based on the individual three-dimensional head model, to obtain source current density signals of 68 brain regions; a multi-frequency band signal power analysis module, configured to: decompose the source current density signals of the 68 brain regions into six frequency bands of δ, θ, α, β, low γ, and high γ waves, calculate power spectral densities of the brain regions and perform normalization processing, and screen effective frequency bands; a directed brain network construction module, configured to: calculate information flow directions and intensities between different brain regions by using a directed transfer function method based on the source current density signals of the 68 brain regions in the effective frequency bands, form a directed transfer function matrix of the effective frequency bands, and construct a directed brain network; an index extraction module, configured to: calculate graph theory indexes and / or epilepsy indexes of each brain region based on the directed brain network; the graph theory indexes include out-degree centrality; an epileptogenic zone determination module, configured to: perform maximum value normalization analysis on the graph theory indexes and / or the epilepsy indexes of the brain regions, to determine an epileptogenic zone positioning result.

[0011] In a second aspect, the application provides an epileptogenic zone positioning method based on electroencephalogram source imaging and dynamic brain networks, comprising: acquiring T1-weighted magnetic resonance imaging data of a user; and constructing an individual three-dimensional head model according to the T1-weighted magnetic resonance imaging data by using a boundary element method; acquiring scalp electroencephalogram data of the user, and preprocessing the scalp electroencephalogram data; performing reverse problem solving on the preprocessed scalp electroencephalogram data by using a standardized low-resolution electroencephalogram source imaging algorithm based on the individual three-dimensional head model, to obtain source current density signals of 68 brain regions; decomposing the source current density signals of the 68 brain regions into six frequency bands of δ, θ, α, β, low γ, and high γ waves, calculating power spectral densities of the brain regions and performing normalization processing, and screening effective frequency bands; calculating information flow directions and intensities between different brain regions by using a directed transfer function method based on the source current density signals of the 68 brain regions in the effective frequency bands, forming a directed transfer function matrix of the effective frequency bands, and constructing a directed brain network; calculating graph theory indexes and / or epilepsy indexes of each brain region based on the directed brain network; the graph theory indexes include out-degree centrality; The graph theory index and / or the epilepsy index of each brain region are subjected to maximum value normalization analysis to determine the localization result of the epileptogenic zone.

[0012] In a third aspect, the application provides an epileptogenic zone localization device based on electro source imaging and dynamic brain network, comprising a processor and a storage medium. The storage medium is configured to store instructions. The processor is configured to operate according to the instructions to perform the method according to the first aspect.

[0013] In a fourth aspect, the application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method according to the first aspect.

[0014] In a fifth aspect, the application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method according to the first aspect.

[0015] Advantages: The epileptogenic zone localization method and system based on electro source imaging and dynamic brain network provided by the application have the following advantages: The application proposes an innovative non-invasive electro source localization method to solve the diagnosis difficulty in preoperative epileptogenic zone localization of MRI-negative focal drug-resistant epilepsy patients. The technology combines standardized low-resolution electro source imaging (sLORETA), multi-band power spectral density (PSD) and directional transfer function (DTF) to construct a dynamic directed brain network, and uses graph theory algorithm to extract multi-dimensional indexes including out-degree centrality (DCout), in-degree centrality (DCin), degree centrality (DC) and seizure index (SI), to comprehensively quantify and identify the potential epileptogenic core region.

[0016] Compared with the traditional evaluation method relying on imaging and artificial interpretation, the application realizes dynamic tracking and multi-dimensional analysis of the epilepsy network, has the advantages of strong non-invasiveness, high automation, strong adaptability to MRI-negative patients, etc. The method not only improves the accuracy of epileptogenic focus localization, but also provides new objective quantitative indexes for preoperative evaluation, which can be spatially registered and verified with the resection area after surgery, and provides strong auxiliary decision support for epilepsy surgery.

[0017] In terms of application prospects, the technology can be integrated into the conventional preoperative evaluation process of epilepsy centers as a precise positioning tool for MRI-negative refractory epilepsy patients, improving the operability rate and long-term efficacy of patients. At the same time, the method has good computational compatibility and scalability, and can be further developed into an intelligent electroencephalogram analysis platform to build an epilepsy surgery decision support system combined with artificial intelligence algorithms, providing a new path for precision medicine and neurological disease diagnosis, and has wide clinical popularization value and industrial transformation potential. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 Fig. 1 is a schematic diagram of an epileptogenic zone positioning system based on electro source imaging and dynamic brain network according to an embodiment of the present application; Figure 2 Fig. 2 is a flowchart of an epileptogenic zone positioning method based on electro source imaging and dynamic brain network according to an embodiment of the present application. DETAILED DESCRIPTION

[0019] The present application will be further described below in conjunction with the drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present application, and cannot be used to limit the protection scope of the present application.

[0020] In the description of the present application, the meaning of several is more than one, the meaning of multiple is more than two, greater than, less than, more than, etc. are understood as not including the number, above, below, etc. are understood as including the number. If it is described as first, second, it is only used for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features or the sequence of indicated technical features.

[0021] In the description of the present application, the description of the terms "one embodiment", "some embodiments", "exemplary embodiment", "example", "specific example" or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the exemplary description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the described specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0022] The term "and / or", only describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. In addition, the character " / ", generally represents that the front and rear associated objects are in an "or" relationship.

[0023] Embodiment 1: The embodiment provides a localization system of epileptogenic zone based on electroencephalography and dynamic brain network, as shown in the figure, comprising an individual head model construction module 110, a scalp electroencephalogram data acquisition module 120, a reverse solving and source imaging module 130, a multi-band signal power analysis module 140, a directed brain network construction module 150, an index extraction module 160 and an epileptogenic zone determination module 170. Figure 1 The individual head model construction module 110 is used for: acquiring T1-weighted magnetic resonance imaging (T1WI) data of a user; and constructing an individual three-dimensional head model by using a boundary element method according to the T1-weighted magnetic resonance imaging data. The individual head model construction module 110 is used for: acquiring T1-weighted magnetic resonance imaging (T1WI) data of a user; and constructing an individual three-dimensional head model by using a boundary element method according to the T1-weighted magnetic resonance imaging data.

[0024] More specifically, the individual head model construction: firstly, based on the preoperative T1-weighted magnetic resonance imaging (T1WI) data of the subject, an individualized three-layer head model is constructed by using a boundary element method (BEM). The three-layer structure is respectively a scalp surface, an outer surface of the skull and an inner surface of the skull, and each layer of surface is composed of 1922 vertices for accurately representing the head anatomical structure.

[0025] The scalp electroencephalogram data acquisition module 120 is used for: acquiring scalp electroencephalogram data of a user, and pre-processing the scalp electroencephalogram data. Optionally, in some embodiments, the pre-processing steps from the original signal to the analysis input of the scalp electroencephalogram data include filtering, ICA, re-referencing and other processes. The method is based on the recognition of electroencephalogram characteristics in each stage of the seizure, and realizes the extraction and standardized preprocessing of the electroencephalogram signals with high signal-to-noise ratio and high repeatability through a multi-step automatic and semi-automatic processing flow.

[0026] The technical process mainly includes the following: (1) Seizure stage division and sample selection: according to the clinical diagnosis standard and the electroencephalogram characteristics, the seizure process is divided into four stages: interictal, preictal, ictal and postictal. Among them, the preictal stage is defined as the stage in which the electroencephalogram appears abnormal discharge activity within about 10 minutes before the onset of the seizure, which is the key warning window of the epilepsy activity. The data of the patient experiencing complete seizure process is selected, and the preictal electroencephalogram signal is extracted for subsequent analysis.

[0027] (2) Electroencephalogram data acquisition specification: using Nihon Kohden EEG-1200C electroencephalogram system, performing standardized electrode distribution according to the international 10-20 electrode system, collecting 21 channels or more EEG data, including conventional scalp electrodes (such as Fp1, Fp2, F3, F4, etc.), electrocardiogram (ECG) and electromyogram channels (EMG_1, EMG_2), to ensure wide coverage and comprehensive sampling of the signal.

[0028] ⑶Data format standardization and import: Import the raw EEG data into a professional signal processing platform and convert it to EDF format to achieve cross-device data compatibility and unified management.

[0029] ⑷Electrode channel identification and calibration: Automatically identify and locate all electrode channels based on the device configuration file to ensure compliance with the standard coordinate system of the international 10-20 system.

[0030] ⑸Irrelevant electrode rejection: Exclude electrode data that is not directly related to the research target to improve data processing efficiency.

[0031] ⑹Signal filtering: Use a band-pass filter (0.5-80 Hz) to denoise the original signal, focusing on eliminating power frequency interference (48-52 Hz) and low-frequency drift, and enhancing the effective neural signal components.

[0032] ⑺Downsampling: To optimize computing efficiency and preserve key information, reduce the original sampling rate to 250 Hz, balancing data volume and information density.

[0033] ⑻Reference electrode reconstruction: Use the Average Reference reconstruction technique to unify the reference baseline of the whole brain signal and reduce the influence of artifacts.

[0034] ⑼Time window segmentation: Segment the continuous EEG data into equal time segments of 2 seconds for subsequent time-frequency analysis and batch processing.

[0035] ⑽Artifact channel and segment rejection: Identify poorly connected, motion interference, or other non-brain-derived noise electrodes and segments through manual review or automatic detection algorithms, and effectively reject them to significantly improve signal reliability.

[0036] ⑾Independent Component Analysis (ICA) artifact rejection: Perform independent component analysis on the denoised data to identify and reject typical interference sources such as eye movement, ECG, and EMG, further purifying the neural signal.

[0037] ⑿Frequency band decomposition and feature extraction: Decompose the preprocessed signal into six standard frequency bands: δ wave (1-4 Hz), θ wave (4-8 Hz), α wave (8-13 Hz), β wave (13-30 Hz), low γ wave (30-48 Hz), and high γ wave (52-80 Hz). Perform time domain, frequency domain, and spatial distribution analysis on each frequency band to reveal the characteristics of epilepsy dynamic activity in multiple frequency bands.

[0038] The inverse problem solving and source imaging module 130 is configured to: based on the individual three-dimensional head model, using the standardized low-resolution brain electrical source imaging (sLORETA) algorithm to solve the inverse problem of the preprocessed scalp EEG data, and obtaining the source current density signals of 68 brain regions; Further, in the step, the source localization based on the individual three-dimensional head model and the preprocessed scalp EEG data is further included, and the key steps of the source localization method include: (1) Individual head model construction.

[0039] (2) Cortex reconstruction and resampling: using the FreeSurfer software to process the T1WI image, reconstructing the individual gray matter cortex surface, and resampling the same to 15002 vertices to construct a high-density cortex grid for subsequent dipole distribution setting.

[0040] (3) Electrode spatial registration: registering the actual collected scalp electroencephalogram (EEG) electrode position information to the scalp surface of the individual head model to ensure the consistency of the electrode space and the structural anatomy space, and providing a basis for the forward modeling.

[0041] (4) Forward model generation: generating a source model through the OpenMEEG plug-in, setting 15002 dipoles at the vertices of the reconstructed cortex surface, and limiting the direction of the dipoles to be perpendicular to the cortex surface, thereby constructing a current transfer matrix (ForwardMatrix) for simulating the process of current conduction from the cortex to the scalp.

[0042] (5) Inverse problem solving and source imaging: using the standardized low-resolution electroencephalogram source reconstruction algorithm (sLORETA) to inverse solve the scalp EEG data, and estimating the source current density at each dipole position, thereby completing the neural source localization of the electroencephalogram signal.

[0043] (6) Brain region division and data extraction: dividing the cortex grid into 68 standard brain regions based on the Desikan-Killiany anatomical atlas, and calculating the average value of the dipole source current density in each brain region, thereby realizing the functional conversion from the scalp signal to the structured brain region.

[0044] The source imaging process can realize the quantitative analysis of different anatomical brain regions of the gray matter cortex on the basis of individual modeling, and can be used in neural network research, epilepsy source localization and other neural disease mechanism research, and has good popularization value and clinical application prospect.

[0045] The multi-band signal power analysis module 140 is configured to: decompose the source current density signals of the 68 brain regions into six frequency bands of δ, θ, α, β, low γ, and high γ, calculate the power spectral density of each brain region and normalize the same, and screen the effective frequency band.

[0046] In this embodiment, the effective frequency band is the delta wave frequency band.

[0047] In this embodiment, the effective frequency band is the delta wave frequency band.

[0048] More specifically, the electroencephalogram frequency domain feature extraction method based on power spectral density analysis (Power Spectral Density, PSD) aims to obtain the energy distribution characteristics in the frequency dimension from the signals of the brain regions after source localization, for seizure detection, brain function state assessment and abnormal signal pattern recognition. The method combines fast Fourier transform (FFT) and multi-band decomposition strategy, and can realize quantitative analysis of the neural activity intensity of multiple functional areas of the brain.

[0049] The technical process includes the following steps: (1) Source signal acquisition and brain region division: first, based on source imaging technology, the source current density signals of 68 anatomical brain regions of the whole brain cortex are obtained. Each brain region signal is divided and extracted by a standard atlas (Desikan-Killiany atlas), providing regionalized data support for subsequent spectral analysis.

[0050] (2) Spectrum calculation and logarithmic conversion: the source current density signals of each brain region are analyzed by fast Fourier transform (FFT). The signal is segmented by 2 seconds, and each segment is transformed by FFT using a rectangular window function. The power value corresponding to each frequency component in the frequency domain is calculated. All power values are logarithmically converted (10 × log10), and the converted unit is dB / Hz, which enhances the visual effect of the spectrum and the intuitiveness of the comparative analysis.

[0051] (3) Brain region average power estimation: the PSD results of multiple time periods of the same brain region are averaged to obtain the stable frequency domain characteristics of each brain region. At the same time, the power values of the 68 brain regions are further averaged to calculate the overall brain average power level, reflecting the overall neural activity intensity of the whole brain.

[0052] (4) Frequency band division and feature extraction: the spectrum is divided into six typical neural oscillation frequency bands, namely: delta wave band (1~4 Hz), theta wave band (4~8 Hz), alpha wave band (8~13 Hz), beta wave band (13~30 Hz), low gamma wave band (30~48 Hz), and high gamma wave band (52~80 Hz). The power distribution of each frequency band in different brain regions can be independently extracted, providing high-resolution frequency domain markers for identifying abnormal activities related to epilepsy.

[0053] The directed brain network construction module 150 is configured to: based on the source current density signals of the 68 brain regions in the effective frequency band, calculate the information flow direction and intensity between different brain regions by using a direct transfer function (DTF) method, form a direct transfer function matrix of the effective frequency band, and construct a directed brain network.

[0054] The frequency domain direct transfer function (DTF) based brain function network construction and quantitative analysis method aims to reveal the information flow direction and intensity between brain regions by using the multi-brain region neural signals after source localization, and further depict the dynamic directed connection mode related to neurological diseases such as epilepsy. The method combines a multivariate autoregressive model (MVAR) and a graph theory network analysis framework, can realize frequency band decomposition, topological index calculation and high-value brain region identification of brain network connection characteristics, and has the advantages of quantitative analysis, directionality and multi-frequency domain analysis.

[0055] The index extraction module 160 is configured to: based on the directed brain network, calculate the graph theory index and / or epilepsy index of each brain region; the graph theory index includes the out-degree centrality.

[0056] Exemplarily, the calculation of the graph theory index includes: (1) source signal input and DTF calculation: taking the source current density time series data of the 68 anatomical brain regions obtained by source imaging as input, the DTF calculation is realized by using the FieldTrip toolbox. The DTF is based on the MVAR model to construct a frequency domain transfer function matrix, which is used to quantitatively depict the information flow intensity and directionality between any two brain regions.

[0057] (2) frequency band decomposition: the calculated full connection DTF matrix is divided into six standard neural oscillation frequency bands, which are: δ band (1-4 Hz), θ band (4-8 Hz), α band (8-13 Hz), β band (13-30 Hz), low γ band (30-48 Hz), and high γ band (52-80 Hz), so as to reveal the information interaction mode between brain regions in different frequency ranges.

[0058] (3) sparsification processing and binary network construction: in order to improve the accuracy and stability of network topology analysis, a threshold sparsification strategy is used to generate a binary directed brain network. The sparsity lower limit is set to 0.07, the range is set to 0.17-0.5, and the step is 0.01. At each sparsity level, the corresponding network structure is constructed, and three types of degree centrality indexes of each node in the network are calculated: Degree centrality (DC): reflects the total number of connections between nodes and other nodes, and measures the activity level of the node in the whole network; In-degree centrality (DCin): represents the information input intensity of all nodes pointing to the node, and reflects the information integration ability of the node; DCout: The number and strength of connections from the node, measuring its ability to drive other nodes.

[0059] ⑷Index normalization and high-value brain region identification: Normalize the index values of each brain region in each frequency band for each subject to the maximum value, and extract brain regions with normalized values greater than or equal to 0.95 as high-weight nodes. The results are displayed in the form of bar charts (Bar Plot) and brain maps (Brain Map), which intuitively reflect the frequency domain distribution and network action characteristics of the functional core brain regions in epilepsy or other neuropathological states.

[0060] ⑸Index comprehensive evaluation: To improve the stability and comparability of network analysis, integrate the curves of each network index at multiple sparsity levels, calculate the area under the curve (Area Under Curve, AUC), and use it as a comprehensive representative value for each type of index to support system evaluation at multiple scales, multiple frequency bands, and multiple brain regions.

[0061] Seizure Index (SI) calculation method based on bistable dynamics model, used to quantify the spatial distribution, intensity and possibility of seizure activity. The method combines electroencephalography, frequency domain connectivity analysis and nonlinear dynamics modeling to establish a highly sensitive identification mechanism for significant regions of seizure activity, allowing individual-scale assessment of the epileptogenicity of each brain region. It is widely applicable to seizure onset zone auxiliary positioning, brain function assessment and seizure prediction research.

[0062] The epileptogenic zone determination module 170 is configured to: perform maximum value normalization analysis on the graph theory index and / or the seizure index of each brain region to determine the epileptogenic zone positioning result.

[0063] In some embodiments, the graph theory index further includes degree centrality, in-degree centrality.

[0064] In this embodiment, the seizure index (SI) is defined as the reciprocal of the escape time of each brain region; the calculation method of the seizure index includes: first, Hilbert transform is performed on the source current density signals of the 68 brain regions to extract the instantaneous phase and amplitude, then based on the directed brain network, the escape time of each brain region is simulated and calculated using a bistable model, and the seizure index is obtained according to the reciprocal of the escape time.

[0065] (1) Seizure index calculation process (1) Frequency-domain signal feature extraction: Based on the source current density data, the whole cerebral cortex was divided into 68 anatomical brain regions. The instantaneous amplitude and phase information of each brain region in six typical frequency bands (delta: 1-4 Hz, theta: 4-8 Hz, alpha: 8-13 Hz, beta: 13-30 Hz, low gamma: 30-48 Hz, high gamma: 52-80 Hz) were extracted using Hilbert transform for subsequent coupling modeling.

[0066] (2) Directional connection sparsification processing: Based on the pre-calculated DTF connection matrix, the sparsity threshold was set to 0.3, and the top 30% of directional edges in terms of strength were retained to reduce noise interference and retain key information pathways, and a sparse information transfer network was constructed.

[0067] (3) Bistable dynamics modeling and SI calculation: A bistable nonlinear dynamics model was introduced to simulate the state of each brain region, and the escape time of the node was calculated to reflect its ability to transition from a normal state to a pathological oscillation state (epileptic state). The seizure index (SI) was defined as the inverse of the escape time, and the larger the value, the easier it was to enter an abnormal state.

[0068] (4) Normalization and visualization analysis: The SI values of each subject in each frequency band and each brain region were normalized by the maximum value. The brain regions with normalized values ≥ 0.95 were extracted as high epileptic susceptibility regions, and bar plots and brain maps were used for visualization.

[0069] (II) Bistable model parameter setting and simulation method (1) Model parameter setting: Excitability parameter (λ): set to -0.25 to control the ability of the node to enter the excited state (parameter range: -1 to 0).

[0070] Oscillation frequency (ω): set to 3 × 2π (i.e., 3 Hz) to simulate slow-frequency epileptic oscillations.

[0071] Model constants: set a = -1, b = 2 as standard bistable dynamics parameters.

[0072] Coupling strength (β): set to 0.5 to control information exchange between different brain regions.

[0073] Random disturbance term (η(t)): set to mean 0.0003, standard deviation 0.05 to introduce system noise.

[0074] Pathological oscillation judgment rule: when the modulus of complex variable z is greater than 0.5, it is judged that the brain area enters the pathological oscillation state of epilepsy (2). The time when the condition is first met is recorded as the escape time.

[0075] (4) Simulation process and method: 1000 times of Monte Carlo simulation are performed on each sparse connection matrix, Euler-Maruyama numerical integration method is adopted, and the time step is dt = 0.0001. The escape time of all brain areas is recorded in each simulation, and the SI value is calculated accordingly.

[0076] (5) Epilepsy index output result: the average SI of each brain area in 1000 simulations is taken as the final epilepsy index result for quantitative analysis and spatial positioning.

[0077] The epilepsy index analysis method provided by the application can effectively identify potential epileptogenic areas, mine the dynamic characteristics of high-risk brain areas, and help realize non-invasive and quantitative epilepsy activity evaluation. The method has the advantages of directionality information, frequency band specificity and individualized dynamics simulation, and has good clinical application value and popularization prospect.

[0078] In some embodiments, the scalp EEG data of the user is acquired, and the scalp EEG data is preprocessed, including: The scalp EEG data of the user is collected, 10-minute signals before the onset of seizures are selected, and the signals are filtered, re-referenced, and independent component analysis is performed to remove artifacts to obtain preprocessed scalp EEG data.

[0079] In some embodiments, the multi-band signal power analysis module is specifically used for: The source current density signals of the 68 brain areas are subjected to frequency spectrum analysis by using fast Fourier transform to obtain the power spectrum density of the 68 brain areas; The power spectrum densities of the 68 brain areas in multiple time periods are averaged to obtain the stable frequency domain characteristics of each brain area, and the power distribution of the delta band (1-4 Hz), theta band (4-8 Hz), alpha band (8-13 Hz), beta band (13-30 Hz), low gamma band (30-48 Hz) and high gamma band (52-80 Hz) in different brain areas is decomposed.

[0080] In some embodiments, the graph theory index and / or epilepsy index of each brain area are subjected to maximum value normalization analysis to determine the epileptogenic area positioning result, including: The graph theory index and / or epilepsy index of each brain area in the effective frequency band are subjected to maximum value normalization, and the brain area with a normalized value greater than or equal to 0.95 is extracted as a potential epileptogenic core area.

[0081] Embodiment 2: Based on Embodiment 1, the application also provides a method for locating an epileptogenic zone based on dynamic brain network and electroencephalography, as shown in Figure 2 , comprising: acquiring T1-weighted magnetic resonance imaging data of a user; constructing a three-dimensional head model of the user by using a boundary element method according to the T1-weighted magnetic resonance imaging data; acquiring scalp electroencephalography data of the user, and preprocessing the scalp electroencephalography data; based on the three-dimensional head model of the user, inversely solving the preprocessed scalp electroencephalography data by using a standardized low-resolution electroencephalography (sLORETA) algorithm to obtain source current density signals of 68 brain regions; decomposing the source current density signals of the 68 brain regions into six frequency bands of delta, theta, alpha, beta, low gamma and high gamma waves, calculating power spectral densities of the brain regions and performing normalization processing, and screening effective frequency bands; based on the source current density signals of the 68 brain regions in the effective frequency bands, calculating information flow directions and intensities between different brain regions by using a direct transfer function method, forming a direct transfer function matrix of the effective frequency bands, and constructing a directed brain network; based on the directed brain network, calculating graph theory indexes and / or epilepsy indexes of each brain region; the graph theory indexes include out-degree centrality; performing maximum value normalization analysis on the graph theory indexes and / or the epilepsy indexes of the brain regions to determine a locating result of the epileptogenic zone.

[0082] In some embodiments, the method further comprises: visualizing and evaluating the epileptogenic zone EZ registration: mapping the locating result of the epileptogenic zone to the individual MRI space or the MNI template to perform three-dimensional positioning and visualization of the brain regions. After the operation, it is judged whether the resection range covers the located area by CT registration.

[0083] The method has the characteristics of non-invasiveness, high spatial resolution and multi-dimensional index fusion, can significantly improve the identification accuracy of the epileptogenic focus of the MRI-negative epilepsy patient, provides a scientific basis for preoperative evaluation, and has strong clinical practicability and popularization prospect.

[0084] Specific application example: In order to better realize the "non-invasive epileptogenic zone locating method based on dynamic brain network and electroencephalography" proposed in the application, the embodiment effectively identifies the epileptogenic zone (EZ) by actually collecting and processing electroencephalography data of epilepsy patients, and combines multi-frequency band network analysis and brain source positioning means, and compares and verifies the clinical accuracy of the resection area after the operation.

[0085] I. Electroencephalography power spectral density analysis (Power Spectral Density, PSD) In the example, 15 patients with refractory epilepsy with complete preoperative video-EEG monitoring records of seizures and MRI negative were selected, and the scalp EEG data of 10 minutes before the onset of the disease was extracted. After all the data were preprocessed (including filtering, artifact removal, re-reference and other steps), power spectral density analysis was performed using fast Fourier transform (FFT) algorithm. Analysis found that in all patients, the power of δ band (1-4 Hz) increased significantly in the early stage of the disease, suggesting that δ wave activity played an important role in the epilepsy network, so the subsequent EEG network analysis focused on this frequency band.

[0086] II. Directed Transfer Function (DTF) Based on the 68 brain region signals after source reconstruction, the DTF connection strength matrix in each frequency band (δ, θ, α, β, low-γ, high-γ) was calculated using the FieldTrip toolbox. To enhance specificity, the network sparsification range was set to 0.17-0.5. The graph theory indicators DC (degree centrality), DCin (in-degree centrality) and DCout (out-degree centrality) were calculated, and the normalized values of each patient in each frequency band were normalized. The brain regions with normalized values ≥0.95 were selected as potential EZ candidate regions.

[0087] Taking patient sub01 as an example, the DCout located brain region in the δ band was highly matched with the postoperative resection region, suggesting that this indicator could effectively reflect the origin area of abnormal discharge of epilepsy. In comparison, the DCin located region was wider, but the specificity was slightly worse, and the DC identified additional possible EZ in some frequency bands (such as high-γ).

[0088] In all 15 patients, 7 patients had DCout localization results matched with the surgical resection region in the δ band, of which 4 were temporal lobe epilepsy and 3 were frontal lobe epilepsy, showing that this method was adaptable to different types of epilepsy.

[0089] III. Seizure Index (SI) Further based on the bistable model, the seizure index (SI) of each patient in each frequency band was calculated. First, the Hilbert transform was performed on the source signal to extract the instantaneous phase and amplitude; then based on the DTF network, the model parameters (such as λ =-0.25, β = 0.5, etc.) were set, the escape time of each brain region was simulated, and the reciprocal of the escape time was defined as the SI value. The normalized SI values of each patient in each frequency band were selected, and the brain regions with normalized values ≥0.95 were selected as candidate EZ.

[0090] The SI results of sub01 patients showed that the SI localization results were highly consistent with the surgical area in the delta to low-gamma band, and the SI results in the high-gamma band were inconsistent with the resection area. This method identified 8 cases of EZ matching the postoperative resection area in all 15 patients, of which 5 were temporal lobe epilepsy and 3 were frontal lobe epilepsy, with a slightly higher accuracy than the DCout results.

[0091] The present embodiment utilizes PSD, DTF, graph theory indicators, and SI and other multi-dimensional analysis methods to establish a set of EZ non-invasive localization process integrating EEG signal feature extraction, brain source reconstruction, and network-driven analysis. This method has high accuracy and strong adaptability, especially for preoperative evaluation and individualized treatment plan for MRI-negative epilepsy.

[0092] Embodiment 3: Based on embodiment 1, the present embodiment provides an apparatus for locating an epileptogenic zone based on electroencephalography and dynamic brain network, comprising a processor and a storage medium. The storage medium is used to store instructions. The processor is used to operate according to the instructions to execute the method according to embodiment 1.

[0093] Embodiment 4: Based on embodiment 1, the present embodiment provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method of embodiment 1.

[0094] Embodiment 5: Based on embodiment 1, the present embodiment provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method of embodiment 1.

[0095] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0096] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart or flowsheet block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0097] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart or flowsheet block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0098] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart or flowsheet block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0099] The above only is the preferred embodiment of the present application, it should be pointed out that: for the ordinary skilled in the art, without departing from the principles of the present application, can also make a number of improvements and refinements, these improvements and refinements should also be considered as the protection scope of the present application.

Claims

1. An epileptogenic zone localization system based on electroencephalography and dynamic brain network, characterized in that, Comprise: An individual head model construction module for: obtaining T1-weighted magnetic resonance imaging data of a user; Constructing an individual three-dimensional head model according to the T1-weighted magnetic resonance imaging data by using a boundary element method; A scalp electroencephalogram data acquisition module for: obtaining scalp electroencephalogram data of a user, and preprocessing the scalp electroencephalogram data; An inverse problem solving and source imaging module for: based on the individual three-dimensional head model, using a standardized low-resolution electroencephalogram source imaging algorithm to solve the inverse problem of the preprocessed scalp electroencephalogram data, and obtaining source current density signals of 68 brain regions; A multi-band signal power analysis module for: decomposing the source current density signals of the 68 brain regions into six frequency bands of δ, θ, α, β, low γ, and high γ waves, calculating the power spectral density of each brain region and normalizing the power spectral density, and screening effective frequency bands; A directed brain network construction module for: based on the source current density signals of the 68 brain regions in the effective frequency bands, using a directional transfer function method to calculate the information flow direction and intensity between different brain regions, forming a directional transfer function matrix of the effective frequency bands, and constructing a directed brain network; An index extraction module for: based on the directed brain network, calculating graph theory indexes and / or epilepsy indexes of each brain region; the graph theory indexes include out-degree centrality; An epileptogenic zone determination module for: performing maximum value normalization analysis on the graph theory indexes and / or epilepsy indexes of each brain region to determine an epileptogenic zone positioning result.

2. The system of claim 1, wherein, The δ wave frequency band is 1-4 Hz, the θ wave frequency band is 4-8 Hz, the α wave frequency band is 8-13 Hz, the β wave frequency band is 13-30 Hz, the low γ wave frequency band is 30-48 Hz, and the high γ wave frequency band is 52-80 Hz.

3. The system of claim 1, wherein, The effective frequency band is the δ wave frequency band.

4. The system of claim 1, wherein, The graph theory indexes also include degree centrality and in-degree centrality.

5. The system of claim 1, wherein, The epilepsy index is defined as the reciprocal of the escape time of each brain region; the calculation method of the epilepsy index includes: first, performing Hilbert transform on the source current density signals of the 68 brain regions to extract instantaneous phase and amplitude, then based on the directed brain network, using a bistable model to simulate and calculate the escape time of each brain region, and obtaining the epilepsy index according to the reciprocal of the escape time.

6. The system of claim 1, wherein, The scalp electroencephalogram data of the user is obtained, and the scalp electroencephalogram data is preprocessed, including: The scalp electroencephalogram data of the user is collected, 10-minute signals before a seizure are selected, and the preprocessed scalp electroencephalogram data is obtained through filtering, re-referencing, and independent component analysis to remove artifacts.

7. The system of claim 1, wherein, The multi-band signal power analysis module is specifically used for: Performing frequency spectrum analysis on the source current density signals of the 68 brain regions by using fast Fourier transform to obtain the power spectral density of the 68 brain regions; Taking the mean of the power spectral density of the 68 brain regions in multiple time periods to obtain stable frequency domain features of each brain region, and decomposing the stable frequency domain features into power distribution of δ waves, θ waves, α waves, β waves, low γ waves, and high γ waves in different brain regions.

8. The system of claim 1, wherein, The maximum value normalization analysis is performed on the graph theory indexes and / or epilepsy indexes of each brain region to determine the epileptogenic zone positioning result, including: The maximum value normalization is performed on the graph theory indexes and / or epilepsy indexes of each brain region in the effective frequency band, and brain regions with normalized values greater than or equal to 0.95 are extracted as potential epileptogenic core regions.

9. A method for locating an epileptogenic zone based on electroencephalography and dynamic brain network, characterized in that, Comprise: Obtaining T1-weighted magnetic resonance imaging data of a user; A boundary element method is used to construct an individual three-dimensional head model according to T1-weighted magnetic resonance imaging data; Obtain scalp electroencephalogram data of a user, and preprocess the scalp electroencephalogram data; Based on the individual three-dimensional head model, a standardized low-resolution electroencephalography source imaging algorithm is used to solve the inverse problem of the preprocessed scalp electroencephalogram data, and source current density signals of 68 brain regions are obtained; The source current density signals of the 68 brain regions are decomposed into six frequency bands of delta, theta, alpha, beta, low gamma and high gamma waves, the power spectral density of each brain region is calculated and normalized, and effective frequency bands are screened; Based on the source current density signals of the 68 brain regions in the effective frequency bands, the information flow direction and intensity between different brain regions are calculated by using a directional transfer function method, a directional transfer function matrix of the effective frequency bands is formed, and a directed brain network is constructed; Based on the directed brain network, graph theory indexes and / or epilepsy indexes of each brain region are calculated; the graph theory indexes include out-degree centrality; Maximum normalization analysis is performed on the graph theory indexes and / or epilepsy indexes of each brain region to determine the localization result of an epileptogenic zone.

10. An apparatus for locating an epileptogenic zone based on electroencephalography and dynamic brain network, characterized in that, It comprises a processor and a storage medium. The storage medium is used to store instructions. The processor is used to operate according to the instructions to execute the method according to claim 9.

Images