RTMS electrical stimulation dose quantification method and device

Abstract

The invention discloses an rTMS electrical stimulation dose quantification method. The method comprises the following steps: obtaining brain structure information, neuron characteristic information and a clinical diagnosis result of a target patient; based on the clinical diagnosis result, determining the type stage of the disease of the target patient; based on the type stage of the disease of the target patient, matching with a disease brain region electrical stimulation library to obtain electrical stimulation parameters; determining a first electrical stimulation dosage of the target patient based on the brain structure information, the neuron characteristic information and the electrical stimulation parameters; performing treatment based on the first electrical stimulation dosage, and inputting symptom improvement condition data, neural electrophysiological index data and iconography index data of the target patient obtained by monitoring in the treatment into a prediction model to obtain a second electrical stimulation dosage; the prediction model is obtained by training based on sample data and a corresponding electrical stimulation dose label result. According to the invention, the electrical stimulation dosage can be accurately matched with each patient, and the treatment effect is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of repetitive transcranial magnetic stimulation, in particular to an rTMS electric stimulation dosage quantification method and device thereof. BACKGROUND

[0002] Transcranial magnetic stimulation (TMS) is a painless, non-invasive nerve modulation technology, and repetitive transcranial stimulation (rTMS) is a mode of TMS. It is widely used in clinical psychiatry, neurological diseases and rehabilitation fields.

[0003] In the conventional treatment using repetitive transcranial stimulation (rTMS), the determination of the electric stimulation dosage is mostly dependent on an empirical fixed dosage scheme or a rough dosage adjustment based on limited disease classification. Such a one-size-fits-all dosage mode leads to insufficient electric stimulation dosage for some patients, which cannot achieve the expected treatment effect. At the same time, in the implementation process of the conventional treatment scheme, once the initial dosage is determined, it is rarely changed flexibly according to the real-time treatment response of the patient throughout the treatment process. However, in actual treatment, the patient's condition and physical state are dynamically changing. With the progress of treatment, the disease-related neurophysiological indicators and symptom manifestations will change, so that the treatment effect cannot be maintained at the best level, and even it may gradually weaken over time, leading to a prolonged treatment period, a decline in patient compliance, and a waste of medical resources. SUMMARY

[0004] The purpose of the present application is to solve the problems in the prior art, and an rTMS electric stimulation dosage quantification method and device thereof are proposed.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0006] The rTMS electric stimulation dosage quantification method comprises the following steps:

[0007] Obtain the brain structure information, neuron characteristic information and clinical diagnosis result of a target patient;

[0008] Determine the type stage of the disease suffered by the target patient based on the clinical diagnosis result;

[0009] Match the type stage of the disease suffered by the target patient with a disease brain area electric stimulation library to obtain electric stimulation parameters;

[0010] Determine the first electric stimulation dosage of the target patient based on the brain structure information, the neuron characteristic information and the electric stimulation parameters;

[0011] Based on the first electric stimulation dose, treatment is carried out, and target patient symptom improvement data, neuroelectrophysiological index data and imaging index data monitored during treatment are input into a prediction model to obtain a second electric stimulation dose; the prediction model is trained based on sample data and corresponding electric stimulation dose label results.

[0012] According to the rTMS electric stimulation dose quantification method provided by the application, the neuron characteristic information includes neuron activity level and functional connection data and neuron physiological parameter data; the brain structure information, neuron characteristic information and clinical diagnosis result of the target patient are obtained, including:

[0013] The brain anatomical image is obtained based on a high-resolution MRI device; the brain cortex is subjected to thickness measurement based on the brain anatomical image to obtain brain cortex parameters; and a three-dimensional brain model is constructed based on the brain cortex parameters to obtain the brain structure information;

[0014] The neurons in different regions of the brain of the target patient are detected based on functional magnetic resonance imaging and a preset neural function task to obtain neuron activity and functional connection data; and the excitability and conduction velocity of the neurons are measured based on transcranial magnetic stimulation evoked potentials to obtain neuron physiological parameter data;

[0015] The clinical diagnosis result is determined based on a medical auxiliary diagnosis system and in combination with the medical history, symptom performance and gene detection result of the target patient.

[0016] According to the rTMS electric stimulation dose quantification method provided by the application, the electric stimulation parameters are obtained by matching the type and stage of the disease of the target patient with a disease brain region electric stimulation library, including:

[0017] The disease type index of the target patient is subjected to initial matching in the disease brain region electric stimulation library to obtain a first electric stimulation parameter record set;

[0018] The disease stage index of the target patient is subjected to secondary matching in the first electric stimulation parameter record set to obtain a second electric stimulation parameter record set;

[0019] The characteristic data of the target patient and the second electric stimulation parameter record set are subjected to similarity matching to obtain the electric stimulation parameters.

[0020] According to the rTMS electric stimulation dose quantification method provided by the application, the characteristic data of the target patient and the second electric stimulation parameter record set are subjected to similarity matching to obtain the electric stimulation parameters, including:

[0021] The characteristic data of the target patient and the second electric stimulation parameter record set are subjected to symptom similarity matching to obtain a first similarity score;

[0022] physiological index similarity matching of the characteristic data of the target patient and the second electric stimulation parameter record set, to obtain a second similarity score;

[0023] brain region function abnormality similarity matching of the characteristic data of the target patient and the second electric stimulation parameter record set, to obtain a third similarity score;

[0024] based on the importance of the symptom similarity, the physiological index similarity and the brain region function abnormality similarity, different weights are respectively given to the first similarity score, the second similarity score and the third similarity score, to obtain a first weighted score, a second weighted score and a third weighted score respectively;

[0025] the comprehensive similarity score is obtained by adding the first weighted score, the second weighted score and the third weighted score, and the parameter group with the highest comprehensive similarity score is determined as the electric stimulation parameter.

[0026] According to the rTMS electric stimulation dose quantification method provided by the application, the first electric stimulation dose of the target patient is determined based on the brain structure information, the neuron characteristic information and the electric stimulation parameter, comprising:

[0027] constructing a three-dimensional brain model based on the brain structure information;

[0028] simulating the electric stimulation parameter on the three-dimensional brain model to obtain an electric stimulation parameter subset;

[0029] perfecting the three-dimensional brain model based on the neuron characteristic information, and optimizing the electric stimulation parameter subset to obtain an initial parameter combination;

[0030] iterative optimization is performed on the initial parameter combination to maximize the treatment effect, to obtain the first electric stimulation dose.

[0031] According to the rTMS electric stimulation dose quantification method provided by the application, the three-dimensional brain model is constructed based on the brain structure information, comprising:

[0032] segmenting, registering and three-dimensionally reconstructing the brain structure image data in the brain structure information to obtain the individualized three-dimensional brain model of the target patient.

[0033] According to the rTMS electric stimulation dose quantification method provided by the application, the training step of the prediction model comprises:

[0034] obtaining sample data, extracting key features from the sample data to obtain first sample feature data, combining and converting the first sample feature data to obtain second sample feature data; inputting the second sample feature data into a pre-trained neural network for model training to obtain a prediction model;

[0035] Accordingly, the loss function of the prediction model is:

[0036] L ss = L sy + εL im + εL ne ;

[0037]

[0038] wherein s true (t) represents the actual symptom severity assessment value of the patient at time t; s pred (t) represents the predicted symptom severity value at time t calculated based on the predicted electrical stimulation dose; T represents the total treatment time point data; ω(t) represents the weight function at time t; V true (r) represents the true value of the gray matter volume at spatial position r in the brain measured by the actual image examination; V pred (r) represents the predicted value of the gray matter volume at spatial position r simulated based on the predicted electrical stimulation dose; I true (r) represents the true value of the white matter fiber bundle integrity index at spatial position r obtained by the actual image examination; I pred (r) represents the predicted value of the white matter fiber bundle integrity index at spatial position r simulated based on the predicted electrical stimulation dose; R represents the set of regions of interest in the brain image; FSSIM represents the feature-based structural similarity; p true (t) represents the probability distribution of n true (t); p pred (t) represents the probability distribution of n pred (t); n true (t) represents the actual index of neurons measured by experiment or indirectly estimated at time t; n pred (t) represents the corresponding neuron index vector simulated according to the predicted electrical stimulation dose by the neuron model.

[0039] The rTMS electrical stimulation dose quantification device comprises:

[0040] An acquisition unit is configured to acquire brain structure information, neuron characteristic information, and clinical diagnosis results of a target patient;

[0041] A processing unit is configured to determine the type and stage of the disease suffered by the target patient based on the clinical diagnosis results;

[0042] A matching unit is configured to match the type and stage of the disease suffered by the target patient with a disease brain area electrical stimulation library to obtain electrical stimulation parameters;

[0043] The first calculation unit is configured to determine a first electric stimulation dose of the target patient based on the brain structure information, the neuron characteristic information and the electric stimulation parameter.

[0044] The second calculation unit is configured to perform treatment based on the first electric stimulation dose, and input symptom improvement data, neuroelectrophysiological index data and imaging index data of the target patient obtained during the treatment into a prediction model to obtain a second electric stimulation dose.

[0045] The application further provides an electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the rTMS electric stimulation dose quantification method when executing the program.

[0046] The application further provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable on a processor to implement the steps of the rTMS electric stimulation dose quantification method.

[0047] Compared with the prior art, the application has the following advantages:

[0048] The rTMS electric stimulation dose quantification method and device provided by the application can obtain brain structure information, clinical diagnosis results and neuron characteristic information of a target patient, determine the type and stage of the disease according to the clinical diagnosis results, match the electric stimulation parameters with the disease brain area electric stimulation library, determine a first electric stimulation dose of the target patient based on the brain structure information, the neuron characteristic information and the electric stimulation parameter, overcome the defects of the fixed or rough classification dose mode in the conventional rTMS treatment, fully consider the differences in brain structure and neuron characteristics of individuals and the diversity and periodicity of diseases, and make the electric stimulation dose accurately adapt to each patient, thereby improving the treatment effect; the method and device can dynamically optimize the electric stimulation dose in a timely manner according to the symptom improvement of the patient, neuroelectrophysiological and imaging changes, ensure that the optimal treatment parameters are maintained throughout the treatment process, reduce the occurrence of adverse reactions, and improve the stability and continuity of long-term treatment. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and those skilled in the art can obtain other drawings according to these drawings without creative effort.

[0050] Figure 1 A schematic diagram of a flow chart of a method for quantifying rTMS electrical stimulation dosage provided in an embodiment of the present invention;

[0051] Figure 2 A schematic diagram of the structure of an rTMS electrical stimulation dose quantification device provided in an embodiment of the present invention;

[0052] Figure 3 This is a schematic structural diagram of the electronic device proposed by the present invention. DETAILED DESCRIPTION

[0053] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0054] The following combination Figure 1 - Figure 3 The present invention describes the rTMS electrical stimulation dose quantification method and device.

[0055] Figure 1 It is a flow chart of the rTMS electrical stimulation dose quantification method provided by the present invention.

[0056] like Figure 1 As shown, the method includes:

[0057] Step 101: Obtain brain structure information, neuron characteristic information, and clinical diagnosis results of a target patient.

[0058] Specifically, neuron characteristic information includes neuron activity level and functional connectivity data as well as neuron physiological parameter data.

[0059] Specifically, it includes: obtaining brain anatomical images based on high-resolution MRI equipment; measuring the thickness of the cerebral cortex based on the brain anatomical images to obtain cerebral cortical parameters; constructing a three-dimensional brain model based on the cerebral cortical parameters to obtain brain structure information.

[0060] Specifically, in acquiring brain anatomical images, high-resolution anatomical images of the target patient's brain are acquired through magnetic resonance imaging technology (such as 7T ultra-high field MRI), including the fine structure of gray matter and white matter, and the detailed morphology of sulci and gyri. At the same time, in measuring the thickness of the cerebral cortex, the thickness of the cerebral cortex can be accurately measured by using image software based on the acquired image information, and the measurement accuracy can reach sub-millimeter level, for example, in different brain regions such as frontal lobe, parietal lobe, temporal lobe and occipital lobe, the thickness of the cortex is quantitatively analyzed, which provides key anatomical basis for subsequent evaluation of the penetration and range of magnetic field in the brain, and the cerebral cortex parameters include depth, width, curvature and other parameters, and individualized three-dimensional brain model can be constructed by using image post-processing software, so as to obtain complete brain structure information.

[0061] Based on functional magnetic resonance imaging and preset neural function tasks, neurons in different regions of the brain of the target patient are detected to obtain neuron activity and functional connection data; based on transcranial magnetic stimulation evoked potential, the excitability and conduction velocity of neurons are measured to obtain neuron physiological parameter data.

[0062] Specifically, in the scanning process using functional magnetic resonance imaging (fMRI), first, neural function tasks are designed, which cover multiple fields such as cognition, movement, sensation and emotion, for example, performing simple hand movement tasks to activate the motor cortex, performing language fluency tasks to detect the activity of the frontal lobe language-related brain area, or watching emotional pictures to stimulate the response of the limbic system, etc. During task execution, the change of blood oxygen level dependent (BOLD) signal in each region of the brain is measured to indirectly reflect the activity level of neurons. Through analysis of these signal changes, the functional activation map of the brain of the individual when performing a specific task can be drawn, and it is clear which brain areas show significant increase in neuron activity in the task, as well as the functional connection mode between these brain areas. Further, neuron activity and functional connection data are obtained.

[0063] Further, transcranial magnetic stimulation evoked potential (TMS-EP) technology is based on the principle of electromagnetic induction, by placing a stimulating coil on the scalp surface, a short high-intensity magnetic field pulse is generated, which can penetrate the skull and induce a local electric field on the surface of the cerebral cortex, thereby stimulating neurons to generate action potentials. A recording electrode is placed on the target muscle (such as the short thumb muscle of the hand), and the muscle movement evoked potential (MEP) induced by TMS stimulation is recorded. By gradually adjusting the stimulation intensity, the minimum stimulation intensity that can produce stable MEP, i.e. the motor threshold (MT), is measured, which is one of the key indicators reflecting the excitability of neurons. At the same time, by analyzing the parameters such as latency and amplitude of MEP, the conduction velocity and synaptic transmission efficiency of neurons can be calculated, and further neuron physiological parameter data is obtained.

[0064] Based on the medical auxiliary diagnosis system, and combined with the medical history, symptom performance, and genetic test results of the target patient, the clinical diagnosis result is determined.

[0065] Specifically, for the confirmation of the clinical diagnosis result, an artificial intelligence auxiliary diagnosis system can be introduced, and the medical history, symptom performance, and genetic test results of the patient as well as various scale evaluation data are comprehensively considered. Deep learning algorithms are used to make more accurate and rapid disease diagnosis, and the potential subtypes and progression trends of the disease are accurately classified to provide more targeted basis for subsequent treatment plan formulation, and further judgment is made by the doctor. The deep learning algorithm is determined by training convolutional neural network (CNN), recurrent neural network (RNN), and its variants (such as long short-term memory network LSTM, gated recurrent unit GRU), which can realize the analysis and confirmation of the clinical diagnosis result.

[0066] Step 102, based on the clinical diagnosis result, the type and stage of the disease suffered by the target patient are determined.

[0067] Specifically, for the judgment of the disease type, the collected patient symptom information is analyzed and compared according to the recognized clinical diagnosis standard of the disease. For example, according to the “Chinese Parkinson's Disease Diagnosis Standard (2016 Edition)”, if the patient has the core symptom of bradykinesia, accompanied by at least one other major symptom such as static tremor, muscle rigidity, and posture balance disorder, and other diseases that may cause similar symptoms are excluded, the patient can be preliminarily diagnosed as having Parkinson's disease. For the diagnosis of depression, the symptom criteria for depressive episode in the International Classification of Diseases (ICD-11) or the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) (such as core symptoms such as persistent low mood, decreased interest, and lack of pleasure lasting at least two weeks) are comprehensively considered. At the same time, the specific differences in symptom performance of different diseases can be compared to accurately distinguish the disease type.

[0068] Further, for the judgment of the disease stage, the disease type is determined, the symptom evolution pattern of the same disease in a large number of clinical cases is reviewed, and the current symptom performance and the order of symptom appearance of the patient are combined to determine the disease stage. Similarly, in an embodiment, in the course of Parkinson's disease, the early stage is mainly manifested as tremor and rigidity of unilateral limbs, which has little effect on daily life. As the disease progresses to the middle stage, symptoms appear in both limbs, and bradykinesia worsens, affecting the patient's daily activity ability. In the late stage, in addition to severe motor disorders, the patient may also have dysphagia, speech disorders, cognitive impairment, and other complications. By comparing the patient's actual situation, it is determined that the patient is in which stage of Parkinson's disease.

[0069] Step 103, based on the type of disease of the target patient and the disease brain area electric stimulation library are matched, the electric stimulation parameters are obtained.

[0070] Specifically, the electric stimulation parameters include electric stimulation frequency, intensity and time. The disease brain area electric stimulation library is a database established based on big data and clinical research. It includes high-quality clinical research data, medical literature and rich case data accumulated by major medical institutions worldwide, covering detailed information of patients of different races, ages and genders receiving rTMS treatment at various stages of various nervous system and mental diseases (such as depression, Parkinson's disease, epilepsy, obsessive-compulsive disorder, etc.), including treatment plan, electric stimulation parameter setting, treatment effect evaluation, adverse reaction record, etc.

[0071] Specifically, it includes: based on the disease type index of the target patient, the first electric stimulation parameter record set is obtained by performing initial matching in the disease brain area electric stimulation library.

[0072] Specifically, the disease brain area electric stimulation library uses a structured database management system and constructs a detailed index directory according to the disease type. When the disease type of the target patient is obtained, the database query language (such as SQL) is used to quickly locate all record entries (i.e. the first electric stimulation parameter record set) corresponding to the disease type in the index by using the exact matching algorithm, including various electric stimulation parameter combinations used for rTMS treatment for the disease type in the past, as well as the corresponding treatment effect evaluation, patient feature description and other detailed information. It can quickly narrow the search range and filter out potential effective parameter combinations related to the target disease type from the massive electric stimulation library data, avoiding blind search in the entire database and greatly improving the matching efficiency.

[0073] Based on the disease stage index of the target patient, the second electric stimulation parameter record set is obtained by performing secondary matching in the first electric stimulation parameter record set.

[0074] Specifically, in the first electric stimulation parameter record set, each record is attached with identification information of the disease stage, such as mild, moderate and severe stages of depression, or early, medium and late stages of Parkinson's disease, etc. Again, by using database query technology, the records matched with the disease stage of the target patient (i.e. the second electric stimulation parameter record set) are further filtered out from the first electric stimulation parameter record set. By introducing the disease stage information for secondary screening, the matching process of the electric stimulation parameters is further refined, and the specificity of the parameters is improved.

[0075] The feature data of the target patient and the second electric stimulation parameter record set are matched in similarity, and the electric stimulation parameters are obtained.

[0076] Specifically, the feature data of the target patient is extracted, including age, gender, brain structure information (such as gray matter volume of specific brain regions, white matter fiber integrity, etc.), neuron characteristic information (such as neuron excitability, conduction velocity, etc.), and other clinical features (such as the presence or absence of comorbid conditions, physical basic conditions, etc.). Then the similarity matching is performed in turn. The specific matching process is described in steps 1031-1035.

[0077] Step 104, determining the first electric stimulation dose of the target patient based on the brain structure information, neuron characteristic information and electric stimulation parameters.

[0078] Specifically, it includes: constructing a three-dimensional brain model based on the brain structure information.

[0079] Segmenting, registering and three-dimensionally reconstructing the brain structure image data in the brain structure information to obtain the individualized three-dimensional brain model of the target patient.

[0080] Specifically, the construction of the three-dimensional brain model is shown in step 101, using three-dimensional reconstruction software such as 3DSlicer, Mimics, to convert the two-dimensional image data after image segmentation into a three-dimensional model, and to construct an accurate three-dimensional structural model of the brain of the target patient. The constructed three-dimensional brain model provides an intuitive and accurate anatomical basis for subsequent electric stimulation parameter simulation and dose determination. Through the visual three-dimensional model, the structural characteristics of the brain can be clearly understood, and the target point and range of electric stimulation can be accurately determined, avoiding unnecessary impact on the surrounding normal tissues during electric stimulation, and improving the precision and safety of treatment.

[0081] Simulating the electric stimulation parameters on the three-dimensional brain model to obtain a subset of electric stimulation parameters.

[0082] Specifically, on the constructed three-dimensional brain model, according to the electric stimulation parameters obtained from the disease brain area electric stimulation library, the distribution of magnetic field in the brain is simulated by using finite element analysis (FEA) software such as COMSOL Multiphysics. Based on Maxwell's equations, the three-dimensional brain model is divided into a plurality of small finite element units, considering the magnetic permeability, electrical conductivity and other physical parameters of the brain tissue, the magnetic field intensity B(x, y, z) and the electric field intensity E(x, y, z) in each unit under given electric stimulation parameters are calculated. By analyzing the distribution of magnetic field and electric field, the effective action area and penetration depth of electric stimulation in the brain are determined, and whether the electric stimulation can accurately act on the target brain area and whether it will cause excessive stimulation to the surrounding non-target area are evaluated. According to the simulation results, the electric stimulation parameter combination that can meet the basic magnetic field action requirements (such as the magnetic field intensity of the target brain area reaching a certain threshold, and the magnetic field intensity of the non-target brain area being within a safe range) is selected to form a subset of electric stimulation parameters.

[0083] Furthermore, by simulating electrical stimulation parameters on a three-dimensional brain model, the distribution of the magnetic field in the brain under different parameter combinations can be predicted in advance, avoiding the blind use of inappropriate electrical stimulation parameters for treatment.

[0084] The three-dimensional brain model is improved based on neuron characteristic information, and a subset of electrical stimulation parameters is optimized to obtain an initial parameter combination.

[0085] Specifically, the neuronal characteristic information such as neuronal excitability E[subjective marker] and neuronal stimuli-evoked potential (TMS-EP) technology, electroencephalogram (EEG) spectrum analysis, and patch clamp technology were combined to measure the neuronal characteristics. n , conduction velocity C n , synaptic plasticity P s et al. further refined and functionalized the 3D brain model. Within the model, neuronal functional parameters were assigned to each brain region based on their distribution and characteristics, reflecting not only anatomical information but also the electrophysiological properties of neurons. For example, in brain regions known to have high neuronal excitability, the model parameters for the response to electrical stimulation were adjusted accordingly to more accurately simulate neuronal activity under electrical stimulation. Then, using computational neuroscience models such as the Hodgkin-Huxley neuron model and its extensions, the neuronal electrical activity response was simulated for each parameter combination within the refined 3D brain model. Considering the principle of electromagnetic induction, changes in the magnetic field induce currents on neuronal membranes, which in turn affect the neuronal membrane potential and excitability. By simulating the generation of action potentials, the conduction of nerve impulses, and synaptic transmission, the effect of each parameter combination on the activation of neurons in the target brain region and its regulatory effect on neural network function were evaluated. Based on the simulation results, optimization algorithms such as genetic algorithms (GA) and particle swarm optimization algorithms (PSO) are used to optimize the subset of electrical stimulation parameters with the objective function of maximizing the degree of neuronal activation or improving specific neural function indicators to obtain an initial parameter combination that performs best in neuronal electrophysiological responses.

[0086] The initial parameter combination is iteratively optimized to obtain the first electrical stimulation dose with the goal of maximizing the therapeutic effect.

[0087] Specifically, quantitative indicators of treatment effect are defined, such as the degree of regulation of specific neural oscillation patterns in the target brain area (such as power changes of alpha waves, beta waves, etc.), improvement of disease-related neurological function scores (such as changes in Hamilton Depression Rating Scale scores for patients with depression, changes in Unified Parkinson's Disease Rating Scale scores for patients with Parkinson's disease, etc.), changes in the release level of specific neurotransmitters, etc., and E tRepresents the treatment effect index. Using the improved brain 3D model and computational neuroscience model, the initial parameter combination is iteratively optimized. In each iteration, by fine-tuning the electrical stimulation parameters (such as changing the stimulation intensity and frequency within a certain range), the treatment effect index E is re-simulated and calculated. t The changes in the parameters are then compared with the results of the previous iteration. If the new parameter combination leads to an improvement in the treatment effect index, this parameter combination is retained as the basis for the next iteration; if the treatment effect index decreases, this parameter combination is discarded and the previous parameter settings are restored. Through multiple iterations, the parameter space is continuously explored, and the electrostimulation parameter combination that maximizes the treatment effect is gradually approached, ultimately obtaining the first electrostimulation dose. Through iterative optimization, the electrostimulation parameters can be further fine-tuned based on the initial parameter combination to maximize the treatment effect.

[0088] In step 105, treatment is performed based on the first electrical stimulation dose. The target patient's symptom improvement data, neurophysiological indicators, and imaging indicators monitored during treatment are input into a prediction model to determine a second electrical stimulation dose. The prediction model is trained based on the sample data and the corresponding electrical stimulation dose label results. Based on the monitoring results, the prediction model, built using a machine learning algorithm, can adjust the electrical stimulation dose in real time to maintain optimal treatment efficacy and avoid adverse reactions.

[0089] Specifically, the training steps of the prediction model include:

[0090] Obtain sample data, extract key features of the sample data to obtain first sample feature data, perform combination transformation on the first sample feature data to obtain second sample feature data; input the second sample feature data into a pre-trained neural network for model training to obtain a prediction model; the pre-trained neural network can be a deep neural network, a recurrent neural network, a long short-term memory neural network, etc.

[0091] Correspondingly, the loss function of the prediction model is:

[0092] L ss =L sy +εL im +∈L ne ;

[0093] Among them, L ss Represents the total loss function, L sy Indicates symptom improvement loss, L im Indicates the image index loss, L nedenotes the neuron indicator loss, ε, ∈ denote hyperparameters, and comprehensively considers the information of symptom improvement, image indicators and neuron indicators, and comprehensively reflects the influence of electrical stimulation treatment on patients. Compared with the loss function of a single indicator, it can more comprehensively evaluate the treatment effect and guide the optimization of electrical stimulation dose, avoid the problem of only focusing on a certain aspect and ignoring other important factors, and thus improve the overall quality and effectiveness of the treatment plan.

[0094] Further,

[0095]

[0096] wherein s true (t) denotes the actual symptom severity evaluation value of the patient at time t; s pred (t) denotes the symptom severity prediction value at time t calculated based on the predicted electrical stimulation dose; T denotes the total treatment time point data; ω(t) denotes the weight function at time t, ω(t) = exp(-λ(T-t)), wherein λ is a decay coefficient, and the size of λ controls the decay speed of the weight over time. When λ is larger, the weight decays faster over time, which means that the model pays more attention to the recent treatment effect; when λ is smaller, the weight is relatively more uniform over time, and the symptom evaluation at the past time point has a relatively greater influence on the loss function; for the symptom improvement of the patient, through L sy The loss function can intuitively reflect the deviation degree between the prediction value and the true value.

[0097] Further,

[0098]

[0099] wherein V true (r) denotes the true value of the gray matter volume at the space r in the brain measured through the actual image examination, V pred (r) denotes the gray matter volume prediction value at the spatial position r simulated based on the predicted electrical stimulation dose; I true (r) denotes the true value of the white matter fiber bundle integrity indicator at the spatial position r obtained through the actual image examination; I pred(r) represents the predicted value of the white matter fiber bundle integrity index at spatial location r simulated by the predicted electric stimulation dose; R represents the set of regions of interest in the brain image; FSSIM represents the feature-based structural similarity; not just a simple comparison at the pixel level, which is very important for complex brain image data. Because the structure and function of the brain are highly correlated, by comparing the similarity of these image features, the effect of electric stimulation on brain structure can be more accurately evaluated, and the predictive ability of the model for this effect can be evaluated. The 1-FSSIM is calculated for each region of interest r within the gray matter and white matter image indicators, respectively, and they are accumulated to obtain the total image indicator loss. In this way, the image changes in different regions of the brain can be considered comprehensively, ensuring that the model can consider the effects on the entire brain structure when predicting the electric stimulation dose, rather than just focusing on local areas, thereby improving the safety and effectiveness of treatment and avoiding adverse consequences caused by excessive or insufficient local stimulation.

[0100]

[0101] where p true (t) represents the probability distribution of n true (t), p pred (t) represents the probability distribution of n pred (t); n true (t) represents the actual indicator of neurons measured by experiment or indirectly estimated at time t; n pred (t) represents the corresponding neuron indicator vector simulated by the neuron model according to the predicted electric stimulation dose. By calculating the KL divergence and accumulating the differences within the entire treatment time T, the loss function can guide the model to continuously optimize the prediction of the electric stimulation dose to better match the actual physiological response of neurons, thereby improving the targeting and effectiveness of treatment, improving the effect of neural function regulation, and promoting the rehabilitation of patients.

[0102] The present application relates to the field of repetitive transcranial magnetic stimulation (rTMS), and provides an rTMS electric stimulation dose quantification method. In the present application, the proposed rTMS electric stimulation dose quantification method overcomes the defects of fixed or rough classification dose mode in traditional rTMS treatment, fully considers the differences in individual brain structure and neuron characteristics, as well as the diversity and stage of diseases, so that the electric stimulation dose can be accurately adapted to each patient, improving the treatment effect; by continuously monitoring various indicators during treatment and using a prediction model trained based on big data, the electric stimulation dose can be dynamically optimized in a timely manner according to the improvement of the patient's symptoms, neuroelectrophysiology and imaging changes, ensuring that the optimal treatment parameters are maintained throughout the treatment process, reducing the occurrence of adverse reactions, and improving the stability and continuity of long-term treatment.

[0103] In an embodiment, steps 1031-1035 are as follows:

[0104] Step 1031, performing symptom similarity matching between the characteristic data of the target patient and the second set of electrical stimulation parameter records to obtain a first similarity score.

[0105] Specifically, the symptom manifestations of the target patient are collected in detail, including the type, severity, frequency of onset, duration, and sequence of symptoms, etc. The symptom information is compared with the patient symptom information corresponding to each record in the second set of electrical stimulation parameter records. A method based on semantic analysis and quantitative scoring is used to calculate the similarity.

[0106] Step 1032, performing physiological index similarity matching between the characteristic data of the target patient and the second set of electrical stimulation parameter records to obtain a second similarity score.

[0107] Specifically, various physiological index data of the target patient are collected, such as the concentration of relevant markers in blood tests (such as neurotransmitter metabolites, inflammatory factors, etc.), cerebrospinal fluid test indicators (specific proteins, cytokines, etc.), neuroelectrophysiological indicators (specific frequency band power of electroencephalogram, latency and amplitude of transcranial magnetic stimulation evoked potential, etc.), and imaging indicators (gray matter volume in brain magnetic resonance imaging, white matter fiber bundle integrity). For each physiological indicator, it is compared with the corresponding indicator in the second set of electrical stimulation parameter records. For numerical indicators such as neurotransmitter metabolite concentration, the relative difference is calculated, and the difference is converted to a similarity score by a standardization function; for imaging indicators, methods based on image feature extraction and matching can be used, such as calculating the difference ratio of gray matter volume, the similarity index of white matter fiber bundle (by comparing the characteristics of fiber bundle such as direction and density), etc. After obtaining the similarity score of each physiological indicator, the weight is assigned according to the importance of each indicator (for example, neuroelectrophysiological indicators are more critical for neuron function evaluation, and may be given a higher weight), and the second similarity score is obtained by weighted summation, reflecting the similarity of the target patient and the patients in the record set in terms of physiological indicators.

[0108] Step 1033, performing brain region functional abnormality similarity matching between the characteristic data of the target patient and the second set of electrical stimulation parameter records to obtain a third similarity score.

[0109] Specifically, the functional abnormal brain regions and their abnormality degrees in the target patient's brain are determined by functional magnetic resonance imaging (fMRI), positron emission tomography (PET) and other technologies. The functional abnormality information of these brain regions is compared with the cases in the second electric stimulation parameter record set. The coincidence of the functional abnormal brain regions of the target patient and the similarity of the abnormality degrees of the patients in the record set are analyzed. For the coincident brain regions, the specific characteristics of the functional abnormalities (such as the degree of metabolic reduction, the level of neuron activity inhibition, etc.) are further compared, and the corresponding similarity scores are given by quantitative method. For multiple brain regions, the importance of each brain region is considered (determined according to the neurobiological mechanism of the disease, such as the importance of the temporal lobe in epilepsy), and the third similarity score is calculated to measure the similarity degree of the target patient and the patients in the record set in terms of brain region functional abnormality

[0110] In step 1034, different weights are given to the first similarity score, the second similarity score and the third similarity score based on the importance of the symptom similarity, the physiological indicator similarity and the brain region functional abnormality similarity, respectively, to obtain the first weighted score, the second weighted score and the third weighted score, respectively.

[0111] Specifically, the weights of the symptom similarity, the physiological indicator similarity and the brain region functional abnormality similarity are determined according to their relative importance to the electric stimulation dose selection. They can be determined in the following ways: one is based on statistical analysis of a large number of clinical case data to observe the influence degree of different factors on the treatment effect, so as to determine their weights; the other is to combine the experience judgment of experts in the field, and the experts give reasonable weights to each factor according to their understanding and treatment experience of the disease. For example, for some diseases dominated by symptoms, such as depression, the weight of the symptom similarity may be higher; while for some neurodegenerative diseases, the weights of the brain region functional abnormality similarity and the physiological indicator similarity may be relatively larger. The first similarity score, the second similarity score and the third similarity score are multiplied by their respective weights to obtain the first weighted score, the second weighted score and the third weighted score. Finally, the three weighted scores are added to obtain the comprehensive similarity score S 总 = ω1*S1 + ω2*S2 + ω3*S3, where S 总 represents the comprehensive similarity score, S1, S2, S3 are the first similarity score, the second similarity score and the third similarity score respectively, ω1, ω2, ω3 are the corresponding weights, and ω1 + ω2 + ω3 = 1.

[0112] In step 1035, the comprehensive similarity score is obtained by adding the first weighted score, the second weighted score and the third weighted score, and the electric stimulation parameter with the highest comprehensive similarity score is determined as the electric stimulation parameter.

[0113] Specifically, in the second electric stimulation parameter record set, after calculating the comprehensive similarity score of each group of records with the target patient, the electric stimulation parameter corresponding to the group of records with the highest score is selected as the final determined electric stimulation parameter. If there are multiple records with similar scores, other factors (such as the number of successful treatment cases, the stability of treatment, etc.) can be further considered in combination, or these parameters can be taken as a candidate set, and the clinician can further screen and fine-tune according to the actual situation to ensure that the selected electric stimulation parameter is most suitable for the individual situation of the target patient.

[0114] Figure 2 is a structural schematic diagram of an rTMS electric stimulation dosage quantification device provided by the present application, as shown in Figure 2 The device includes: an acquisition unit 10 for acquiring brain structure information, neuron characteristic information and clinical diagnosis results of a target patient; a processing unit 20 for determining the type and stage of the disease suffered by the target patient based on the clinical diagnosis results; a matching unit 30 for matching the type and stage of the disease suffered by the target patient with a disease brain area electric stimulation library to obtain electric stimulation parameters; a first calculation unit 40 for determining the first electric stimulation dosage of the target patient based on the brain structure information, neuron characteristic information and electric stimulation parameters; a second calculation unit 50 for performing treatment based on the first electric stimulation dosage, and inputting the monitored data of the target patient's symptom improvement, neuroelectrophysiological index and imaging index in the treatment into a prediction model to obtain the second electric stimulation dosage; and the prediction model is trained based on sample data and corresponding electric stimulation dosage label results.

[0115] Figure 3 is a structural schematic diagram of an electronic device provided by the present application, as shown in Figure 3As shown, the electronic device can include a processor 310, a communications interface 320, a memory 330, and a communications bus 340, wherein the processor 310, the communications interface 320, and the memory 330 complete communications with each other through the communications bus 340. The processor 310 can invoke a logical instruction in the memory 330 to execute an rTMS electric stimulation dosage quantification method, which includes: acquiring brain structure information, neuron characteristic information, and a clinical diagnosis result of a target patient; determining a type and stage of a disease suffered by the target patient based on the clinical diagnosis result; matching the type and stage of the disease suffered by the target patient with a disease brain area electric stimulation library to obtain an electric stimulation parameter; determining a first electric stimulation dosage of the target patient based on the brain structure information, the neuron characteristic information, and the electric stimulation parameter; performing treatment based on the first electric stimulation dosage, and inputting symptom improvement data, neuroelectrophysiological index data, and imaging index data of the target patient monitored during the treatment into a prediction model to obtain a second electric stimulation dosage; and the prediction model is trained based on sample data and corresponding electric stimulation dosage label results.

[0116] In addition, the logical instruction in the memory 330 described above can be implemented in the form of a software function unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the embodiments of the method of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0117] In another aspect, the present application also provides a computer program product, which comprises a computer program stored on a non-transitory computer-readable storage medium, and the computer program comprises program instructions, when the program instructions are executed by a computer, the computer can execute the rTMS electric stimulation dosage quantification method provided by the above-mentioned methods, and the method comprises the following steps: obtaining brain structure information, neuron characteristic information and clinical diagnosis results of a target patient; determining a type and stage of a disease suffered by the target patient based on the clinical diagnosis results; matching the type and stage of the disease suffered by the target patient with a disease brain area electric stimulation library to obtain electric stimulation parameters; determining a first electric stimulation dosage of the target patient based on the brain structure information, the neuron characteristic information and the electric stimulation parameters; performing treatment based on the first electric stimulation dosage, and inputting symptom improvement data, neuroelectrophysiological index data and imaging index data of the target patient monitored during the treatment into a prediction model to obtain a second electric stimulation dosage; and the prediction model is trained based on sample data and corresponding electric stimulation dosage label results.

[0118] In another aspect, the present application also provides a non-transitory computer-readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the rTMS electric stimulation dosage quantification method provided by the above-mentioned methods, and the method comprises the following steps: obtaining brain structure information, neuron characteristic information and clinical diagnosis results of a target patient; determining a type and stage of a disease suffered by the target patient based on the clinical diagnosis results; matching the type and stage of the disease suffered by the target patient with a disease brain area electric stimulation library to obtain electric stimulation parameters; determining a first electric stimulation dosage of the target patient based on the brain structure information, the neuron characteristic information and the electric stimulation parameters; performing treatment based on the first electric stimulation dosage, and inputting symptom improvement data, neuroelectrophysiological index data and imaging index data of the target patient monitored during the treatment into a prediction model to obtain a second electric stimulation dosage; and the prediction model is trained based on sample data and corresponding electric stimulation dosage label results.

[0119] The device embodiments described above are only schematic, wherein the units shown as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment scheme. Those skilled in the art can understand and implement without creative labor.

[0120] Those skilled in the art can clearly understand the implementation of the various embodiments by means of software and necessary general hardware platforms through the description of the above embodiments, and of course, the implementation can also be through hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, and the computer software product can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the method of each embodiment or some parts of the embodiment.

[0121] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can make equivalent replacements or changes within the technical range disclosed by the present application according to the technical solutions and inventive concepts of the present application, which should be covered within the protection scope of the present application.

Claims

1. rTMS electrical stimulation dose quantification method, characterized in that, The following steps are involved: Obtain brain structure information, neuronal characteristics information, and clinical diagnosis results of target patients; Determining the type and stage of the disease suffered by the target patient based on the clinical diagnosis results; Obtaining electrical stimulation parameters based on matching the type and stage of the diseased brain region of the target patient with the diseased brain region electrical stimulation library; determining a first electrical stimulation dose for a target patient based on the brain structure information, the neuron characteristic information, and the electrical stimulation parameters; Treatment is performed based on the first electrical stimulation dose, and the target patient symptom improvement data, neuroelectrophysiological index data and imaging index data monitored during the treatment are input into the prediction model to obtain a second electrical stimulation dose; the prediction model is trained based on the sample data and the corresponding electrical stimulation dose label results.

2. The rTMS electrical stimulation dose quantification method according to claim 1, characterized in that: The neuron characteristic information includes neuron activity level and functional connectivity data as well as neuron physiological parameter data; The acquisition of the target patient's brain structure information, neuron characteristic information, and clinical diagnosis results includes: Acquire brain anatomical images based on a high-resolution MRI device; measure the thickness of the cerebral cortex based on the brain anatomical images to obtain cerebral cortical parameters; construct a three-dimensional brain model based on the cerebral cortical parameters to obtain brain structural information; Based on functional magnetic resonance imaging and preset neurological function tasks, neurons in different areas of the target patient's brain are detected to obtain neuronal activity and functional connectivity data; based on transcranial magnetic stimulation-evoked potentials, neuronal excitability and conduction velocity are measured to obtain neuronal physiological parameter data; Based on the medical auxiliary diagnosis system and combined with the target patient's medical history, symptoms, and genetic test results, the clinical diagnosis results are determined.

3. The rTMS electrical stimulation dose quantification method according to claim 1, characterized in that: The electrical stimulation parameters are obtained by matching the type and stage of the diseased brain region of the target patient with the diseased brain region electrical stimulation library, including: Performing an initial match in the diseased brain region electrical stimulation library based on the disease type index of the target patient to obtain a first electrical stimulation parameter record set; Performing secondary matching in the first electrical stimulation parameter record set based on the disease stage index of the target patient to obtain a second electrical stimulation parameter record set; Similarity matching is performed on the characteristic data of the target patient and the second electrical stimulation parameter record set to obtain the electrical stimulation parameters.

4. The rTMS electrical stimulation dose quantification method according to claim 3, characterized in that: The performing similarity matching between the characteristic data of the target patient and the second electrical stimulation parameter record set to obtain the electrical stimulation parameters includes: performing symptom similarity matching between the characteristic data of the target patient and the second electrical stimulation parameter record set to obtain a first similarity score; performing physiological indicator similarity matching between the characteristic data of the target patient and the second electrical stimulation parameter record set to obtain a second similarity score; performing similarity matching of brain region functional abnormalities between the characteristic data of the target patient and the second electrical stimulation parameter record set to obtain a third similarity score; Based on the importance of symptom similarity, physiological index similarity, and brain region functional abnormality similarity, different weights are assigned to the first similarity score, the second similarity score, and the third similarity score to obtain the first weighted score, the second weighted score, and the third weighted score, respectively; A comprehensive similarity score is obtained by adding the first weighted score, the second weighted score, and the third weighted score, and a group of parameters with the highest comprehensive similarity score is determined as the electrical stimulation parameters.

5. The rTMS electrical stimulation dose quantification method according to claim 1, characterized in that: The determining a first electrical stimulation dose for the target patient based on the brain structure information, the neuron characteristic information, and the electrical stimulation parameters includes: constructing a three-dimensional brain model based on the brain structure information; simulating the electrical stimulation parameters on the three-dimensional brain model to obtain an electrical stimulation parameter subset; improving the three-dimensional brain model based on the neuron characteristic information, and optimizing the subset of electrical stimulation parameters to obtain an initial parameter combination; The initial parameter combination is iteratively optimized to obtain the first electrical stimulation dose with the goal of maximizing the therapeutic effect.

6. The rTMS electrical stimulation dose quantification method according to claim 5, characterized in that: The constructing of a three-dimensional brain model based on the brain structure information includes: The brain structure image data in the brain structure information is segmented, registered and three-dimensionally reconstructed to obtain a three-dimensional brain model individualized to the target patient.

7. The rTMS electrical stimulation dose quantification method according to claim 1, characterized in that: The training step of the prediction model includes: Obtaining sample data, extracting key features from the sample data to obtain first sample feature data, performing combination conversion on the first sample feature data to obtain second sample feature data; inputting the second sample feature data into a pre-trained neural network for model training to obtain a prediction model; Accordingly, the loss function of the prediction model is: L ss =L sy +εL im +∈L ne ; Among them, s true (t) represents the patient's actual symptom severity assessment value at time t; s pred (t) represents the predicted value of symptom severity at time t calculated based on the predicted electrical stimulation dose; T represents the total treatment time point data; ω(t) represents the weight function of time t; V true (r) represents the true value of the gray matter volume measured by actual imaging examination at the space r in the brain, V pred (r) represents the predicted value of gray matter volume at spatial position r obtained by simulation based on the predicted electrical stimulation dose; I true (r) represents the true value of the white matter fiber bundle integrity index obtained by actual imaging examination at the spatial position r; I pred (r) represents the predicted value of the white matter fiber bundle integrity index at the spatial position r obtained by the simulation of the predicted electrical stimulation dose; R represents the set of regions of interest in the brain image; FSSIM represents feature-based structural similarity; p true (t) represents n true The probability distribution of (t), p pred (t) represents n pred The probability distribution of (t); n true (t) represents the actual neuron index obtained by experimental measurement or indirect estimation at time t; n pred (t) represents the corresponding neuron index vector obtained by simulating the neuron model based on the predicted electrical stimulation dose.

8. rTMS electrical stimulation dose quantification device, characterized in that: The method for quantifying rTMS electrical stimulation dose according to any one of claims 1 to 7; The rTMS electrical stimulation dose quantification device comprises: Acquisition unit: used to obtain brain structure information, neuron characteristic information and clinical diagnosis results of target patients; Processing unit: used to determine the type and stage of the disease suffered by the target patient based on the clinical diagnosis result; Matching unit: used for matching the type and stage of the diseased patient's disease with the diseased brain region electrical stimulation library to obtain electrical stimulation parameters; a first calculation unit, configured to determine a first electrical stimulation dose for a target patient based on the brain structure information, the neuron characteristic information, and the electrical stimulation parameters; The second calculation unit is used to perform treatment based on the first electrical stimulation dose, and input the target patient's symptom improvement data, neuroelectrophysiological index data and imaging index data monitored during the treatment into the prediction model to obtain a second electrical stimulation dose; the prediction model is trained based on sample data and corresponding electrical stimulation dose label results.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the rTMS electrical stimulation dose quantification method according to any one of claims 1 to 7 are implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the rTMS electrical stimulation dose quantification method according to any one of claims 1 to 7 are implemented.

Images