A system and method for simultaneously solving brain source activity and noise

Abstract

The present application discloses a system and method for simultaneously solving brain source activity and noise. The system includes: a data input module for obtaining brain activity sampling data and lead field matrix L; a model building module for building a sampling data generation model, the sampling data generation model is y(t)=LS(t)+Au(t)+ε, where y(t) is the sampling data at time t, S(t) is the brain source activity at time t, A is the gain matrix of M sampling sensor noises, u(t) is the brain source activity at time t, ε is a constant, and the sampling data generation model is converted into a sampling data probability generation model; a solution module is used to solve the sampling data probability generation model using a Bayesian expectation maximum algorithm. The method can achieve the simultaneous solution of brain source activity and noise.

Description

Technical Field

[0001] The present application relates to the field of brain imaging technology, and more specifically, to a system and method for simultaneously solving brain source activity and noise. Background Art

[0002] The brain, as the most important part of the central nervous system, is the core of activities such as perception, memory, learning, and behavior. It is also a window for exploring body lesions. Brain imaging using extracerebral electromagnetic sampling data is an important method for exploring the mechanism of brain activity. It can locate and evaluate brain cognitive processes and dysfunctional brain activities. It is of great value for exploring the location of functional areas in the brain, perceiving the dynamics of brain activities, etc., and has a wide range of applications in brain research and clinical practice.

[0003] The inverse problem of brain source imaging is the process of inverting the information of neuronal activity in the brain by sampling EEG / MEG data, which is the core of brain source imaging. However, the existing tomographic source reconstruction algorithm with noise learning is still not friendly in solving the inverse problem in terms of solution flexibility and imaging robustness. The existing technology has done some research on how to estimate source activity and sensor noise, but how to estimate brain source activity and noise robustly and simultaneously has always been a challenging topic for imaging brain activity using EEG / MEG sampling data. Summary of the invention

[0004] In response to at least one defect or improvement need in the prior art, the present invention provides a system and method for simultaneously solving brain source activity and noise, which can achieve the simultaneous solution of brain source activity and noise.

[0005] To achieve the above object, according to a first aspect of the present invention, a system for simultaneously solving brain source activity and noise is provided, comprising:

[0006] A data input module is used to obtain brain activity sampling data and lead field matrix L;

[0007] A model building module is used to build a sampling data generation model, the sampling data generation model is y(t)=LS(t)+Au(t)+ε, wherein y(t) is the sampling data at time t, S(t) is the brain source activity at time t, A is the gain matrix of M sampling sensor noises, u(t) is the brain source activity at time t, ε is a constant, and the sampling data generation model is converted into a sampling data probability generation model, the sampling data probability generation model includes a prior distribution hyperparameter composed of a prior variance of N brain source activities and a prior variance of M sampling sensor noises;

[0008] The solution module is used to solve the probability generation model of sampling data using the Bayesian expectation maximum algorithm, determine the hyperparameters of the prior distribution, and solve the brain source activity data and noise data at the same time.

[0009] Furthermore, A is an M×M identity matrix, A=[e1,...,e M ].

[0010] Furthermore, before converting the sampled data generation model into a sampled data probability generation model, the sampled data generation model is converted into

[0011]

[0012] F=[L1,...L N , e1,...,e M ]

[0013]

[0014] Among them, L n is the lead field matrix of the nth brain source, S n (t) is the brain activity of the nth brain source at time t, u m (t) is the noise information of the mth sensor at time t, T is the number of sampling points, 1≤n≤N, 1≤m≤M.

[0015] Furthermore, the prior distribution hyperparameter v is

[0016]

[0017] is the prior variance of the nth brain source activity, 1≤n≤N, [v N+1 , ..., v N+M ] is the prior variance of the noise, and diag represents the construction of a diagonal matrix.

[0018] Furthermore, solving the sampling data probability generation model includes:

[0019] Step 1, determining the prior distribution of the probability model for generating the sampling data according to the brain activity sampling data obtained by the input module, and solving the mean of the posterior probability distribution of brain activity and noise by using the maximum expectation value algorithm of Bayesian theory;

[0020] Step 2, constructing a cost function for fitting the generated data covariance of the sampled data generation probability model with the covariance of the brain activity sampled data obtained by the input module, and solving the locally optimal prior distribution hyperparameters by maximizing the marginal likelihood method;

[0021] Repeat steps 1 and 2 until the cost function converges, and use the prior distribution hyperparameters at this time as the final prior distribution hyperparameters;

[0022] The brain source activity data and the noise data are solved simultaneously according to the final determined prior distribution hyperparameters.

[0023] Furthermore, ε is zero or 1e -20 .

[0024] According to a second aspect of the present invention, a method for simultaneously solving brain source activity and noise is also provided, comprising:

[0025] Obtain brain activity sampling data and lead field matrix L;

[0026] Constructing a sampling data generation model, the sampling data generation model is y(t)=LS(t)+Au(t)+ε, wherein y(t) is the sampling data at time t, S(t) is the brain source activity at time t, A is the gain matrix of M sampling sensor noises, u(t) is the brain source activity at time t, ε is a constant, and converting the sampling data generation model into a sampling data probability generation model, the sampling data probability generation model includes a prior distribution hyperparameter composed of a prior variance of N brain source activities and a prior variance of M sampling sensor noises;

[0027] The Bayesian expectation-maximization algorithm is used to solve the probability generation model of the sampling data, determine the hyperparameters of the prior distribution, and solve the brain source activity data and noise data at the same time.

[0028] Furthermore, A is an M×M identity matrix, A=[e1,...,e M ].

[0029] Furthermore, before converting the sampled data generation model into a sampled data probability generation model, the sampled data generation model is converted into

[0030]

[0031] F=[L1,...L N , e1,...,e M ]

[0032]

[0033] Among them, L n is the lead field matrix of the nth brain source, S n (t) is the brain activity of the nth brain source at time t, u m (t) is the noise information of the mth sensor at time t, T is the number of sampling points, 1≤n≤N, 1≤m≤M.

[0034] Furthermore, the prior distribution hyperparameter v is

[0035]

[0036] is the prior variance of the nth brain source activity, 1≤n≤N, [v N+1 , ..., v N+M ] is the prior variance of the noise, diag represents the construction of a diagonal matrix

[0037] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0038] (1) The present invention can image a variety of complex brain source activities under different noise and interference conditions. Based on the idea of ​​augmented brain source lead field matrix, it can realize the simultaneous estimation of brain source and noise, and accurately estimate low signal-to-noise ratio and strongly correlated multi-source activities; and it is simpler in theoretical derivation and has better robustness to different types of interference noise.

[0039] (2) The present invention is based on the theory of empirical Bayes and maximizes the marginal likelihood of data according to a principled cost function, thereby generating a fast and convergent update rule.

[0040] (3) This patent uses the maximum expectation value algorithm to estimate source activity and noise expectation, and further optimizes the cost function to estimate hyperparameters. It has strong robustness to hyperparameter initialization and fast convergence computational efficiency.

[0041] (4) Simulation results show that the algorithm only needs a small amount of experimental sampling data to achieve complex brain source activity imaging and estimation of sensor noise interference, regardless of Gaussian noise or real noise conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 is a schematic diagram of a system for simultaneously solving brain source activity and noise according to an embodiment of the present invention; DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0045] The terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or modules is not limited to the listed steps or modules, but may optionally include steps or modules that are not listed, or may optionally include other steps or modules that are inherent to these processes, methods, products or devices.

[0046] like Figure 1 As shown, a system for simultaneously solving brain source activity and noise according to an embodiment of the present invention includes a data input module, a model building module and a parameter estimation module.

[0047] (1) Data input module: used to obtain brain activity sampling data and lead field matrix.

[0048] The brain activity sampling data is electroencephalogram (EEG) or magnetoencephalogram (MEG) sampling data.

[0049] Furthermore, the brain electromagnetic sampling data is valid data that has been preprocessed.

[0050] The EEG / MEG sampling data at time t is recorded as y(t) is the preprocessed EEG / MEG sampling data, and M is the total number of sampling sensors.

[0051] The lead field matrix is ​​based on segmenting human brain MRI images and solving the correlation matrix between brain source activity and sampling data through the brain imaging forward problem model.

[0052] The lead field matrix is ​​determined based on brain structure information and sampling sensor information.

[0053] The lead field matrix is ​​defined as L = [L1, ..., L N ], N represents the number of brain sources, L n is the lead field matrix of the nth brain source.

[0054] (2) Model construction module: used to construct a sampling data generation model from the preprocessed EEG / MEG sampling data and convert the sampling data generation model into a sampling data generation probability model.

[0055] When it is assumed that brain activities and different interference noises all satisfy Gaussian distribution, a new EEG sampling data generation model that satisfies Gaussian distribution is constructed, and the prior distribution of the sampling data generation probability model is determined according to the brain sampling data.

[0056] The specific implementation process is as follows:

[0057] The sampling data generation model is constructed as follows:

[0058]

[0059] Where M is the number of channels measured, S(t) is the brain source activity at time t, and S n (t) is the brain activity of the nth brain source at time t, assuming it has d c directions, A=[a1,...,a M ] is the noise gain matrix, which can be estimated from the baseline data or fixed. For the sake of simplicity, this patent assumes that there is no prior noise information and sets the gain matrix to be an M×M unit matrix, that is, A=I=[e1,...,e M ], u(t) is the noise information at time t, and ε is a constant, which is also the residual term.

[0060] Assuming that brain activity and interference noise both satisfy Gaussian distribution, the sampling data generation model can be unified as:

[0061]

[0062] F=[L1,...L N , e1,...,e M ]=[F1,...,F N+M ], (3)

[0063]

[0064] Where F is the augmented steering matrix, y(t) is the time course of the augmented time series at time t, and Y = [y(1), ..., y(T)], X = [x(1), ..., x(T)], T is the number of sampling points, and T represents the matrix transpose.

[0065] Determine the prior distribution of the probability generation model of the sampled data based on the empirical Bayesian theory and the sampled data. Assume that the prior distribution of X(t) is:

[0066]

[0067] In the above formula, K = N + M, p(X|v) represents the prior probability distribution of x(t), and v k is the prior variance of the kth source activity or noise, and the prior distribution hyperparameter v of brain source and noise activities is defined as

[0068]

[0069] is the prior variance of the activity of the nth brain source, [v N+1 , ..., v N+M ] is the prior variance of the noise.

[0070] Based on the probability generation model of sampled data, the conditional probability can be divided into

[0071]

[0072] βI M×M Is the residual information. Then the marginal distribution is constructed as follows,

[0073]

[0074] The above-mentioned sampling data generation model and probability model constitute a sampling data probability generation model.

[0075] (3) Parameter solution module: The probability model generated by the sampled data is solved based on the Bayesian expectation maximum algorithm. The core idea is to fit the covariance matrix of the structured estimation model with the covariance matrix of the sampled data, that is, to define the corresponding cost function, optimize the cost function until convergence, and realize the simultaneous estimation of brain power activity and interference noise information;

[0076] Furthermore, the parameter solving module includes a cost function optimization module based on an EM (expected maximum) algorithm, and a simultaneous solving module for brain electrical activity and interference noise.

[0077] (3.1) Cost function optimization module based on EM (expected maximum) algorithm: used to solve the expected value of the sampled data using the empirical Bayes method, then define the cost function, iteratively optimize the cost function, calculate and determine the hyperparameters of brain source activity and sampling noise variance, and iterate to optimize the cost function. Specifically, it includes two steps: Step 1 (E-Step), constructing the solution of brain source activity and noise expectation estimation; Step 2 (M-Step), cost function optimization estimation of hyperparameters;

[0078] Step 1 (E-Step): Estimation of the expected value of brain source activity and noise interference.

[0079] The prior distribution of the probability model for generating the sampling data is determined according to the brain activity sampling data obtained by the input module, and the maximum expectation value algorithm of Bayesian theory is used to solve the mean of the posterior probability distribution of brain activity and noise.

[0080] Here we assume that the hyperparameters are known and the posterior distribution p(X|Y) also obeys a Gaussian distribution, with the mean and covariance expressed as:

[0081]

[0082]

[0083] ∑ y is the model covariance matrix, where ∑ y =βIM×M +FvF T ,∑ x is the variance of the posterior distribution of X.

[0084] In order to estimate unknown brain sources and noise, v must be a known hyperparameter. A principled way to achieve this is to use the so-called maximization of marginal likelihood, which requires the following M-step.

[0085] Step 2 (M-Step): Construct a cost function for fitting the generated data covariance of the sampled data generation probability model with the covariance of the brain activity sampled data obtained by the input module, and use the maximum marginal likelihood method to solve the locally optimal prior distribution hyperparameters.

[0086] The constructed cost function is:

[0087]

[0088] It can be seen that minimizing this cost function is equivalent to maximizing the marginal likelihood. The cost function can be understood as a fitting problem, which can be regarded as estimating the model data covariance ∑ y =βI M×M +FvF T Covariance with empirical data Perform the fitting.

[0089] EM (expected maximum) algorithm is used for optimization: Using the standard determinant identity, the cost function is regarded as the inverse function of the hyperparameters, and the cost function can be expressed as:

[0090]

[0091] C represents the term without v. Since the second term is in every The upper bound of the cost function is

[0092]

[0093] tr(A) represents the addition of the diagonal elements of matrix A, p = ∑ x =v-vF T ∑ y -1 Fv, calculate the above formula for v i The derivative of , set it to 0, and the update rule is:

[0094]

[0095] Repeat steps 1 and 2. The final estimate of the hyperparameter v is obtained by iteratively updating the mean (9), covariance (10) and (14) of the posterior distribution until the cost function (11) defined at the beginning converges. The prior distribution hyperparameter at this time is used as the final prior distribution hyperparameter.

[0096] (3.2) The module for simultaneously solving brain source activity and interference noise is used to solve the mean of the posterior probability of brain source activity and interference noise based on the calculated brain source activity and sampling noise variance based on the Bayesian theory, that is, the brain source activity and interference noise information solved simultaneously.

[0097] Furthermore, the captured residual term ε is properly processed according to the fit between the model data covariance and the empirical data covariance to enhance the robustness / stability of the estimation results.

[0098] In this paper, the above module describes a new method for simultaneously estimating brain temporal activity and noise sequences, that is, the structured model data covariance ∑ y =βI M×M +FvF T Covariance with empirical data Fitting, in previous work, assuming ∑ y The second part of captures the brain signal of interest, while the first part βI contains all the confounding factors and can access the pre-stimulus data (i.e., the data assumed to not contain the signal of interest, the source). The stimulus-evoked partitioning factor provides a powerful means to transform the data covariance matrix ∑ y decomposed into signal and interference components. There are two possible ways to perform the decomposition: fix the noise covariance using an estimate of the pre-stimulus data (baseline), or estimate it separately from the signal component in iterations. However, both methods require separate estimates of the brain activity of interest and the noise information. Here we propose the idea of ​​an augmented lead field matrix, where ∑ y The second part of reflects the brain signal of interest and all interference factors, such as spontaneous brain activity, sensor noise, muscle artifacts, etc., while the first part captures the residual. In this embodiment of the present invention, for robustness / stability, we set β to zero or a small constant variable (e.g., β = 1e -20 ); from experience, this approach works successfully in either setting.

[0099] In general, compared with the prior art, the present invention has the following beneficial effects:

[0100] The present invention is based on the theory of empirical Bayes and maximizes the marginal likelihood of the data according to a principled cost function, thereby generating a fast and convergent update rule. The algorithm further optimizes the cost function and uses a robust method to learn hyperparameters, further proving that the algorithm has strong robustness to hyperparameter initialization and fast convergence computational efficiency.

[0101] The new algorithm is able to process point brain sources of various configurations under high noise and interference conditions. Calculations show that augmenting the brain source lead field matrix with a unit matrix corresponding to the sensor noise can simultaneously update the covariance of the brain source and noise.

[0102] Simulation results show that the algorithm only needs a small amount of sensor data to reconstruct complex brain source activities and sensor noise under both Gaussian noise and real noise conditions.

[0103] A method for simultaneously solving brain source activity and noise according to an embodiment of the present invention comprises the steps of:

[0104] Obtain brain activity sampling data and lead field matrix L;

[0105] Constructing a sampling data generation model, the sampling data generation model is y(t)=LS(t)+Au(t)+ε, wherein y(t) is the sampling data at time t, S(t) is the brain source activity at time t, A is the gain matrix of M sampling sensor noises, u(t) is the brain source activity at time t, ε is a constant, and converting the sampling data generation model into a sampling data probability generation model, the sampling data probability generation model includes a prior distribution hyperparameter composed of a prior variance of N brain source activities and a prior variance of M sampling sensor noises;

[0106] The Bayesian expectation-maximization algorithm is used to solve the probability generation model of the sampling data, determine the hyperparameters of the prior distribution, and solve the brain source activity data and noise data at the same time.

[0107] The specific implementation of the method is the same as the above system and will not be repeated here.

[0108] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0109] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0110] In the several embodiments provided in the present application, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are only schematic, such as the division of the modules, which is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, indirect coupling or communication connection of systems or modules, which can be electrical or other forms.

[0111] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0112] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or software functional modules.

[0113] If the integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, disk or optical disk and other media that can store program codes.

[0114] A person skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by entering a program to instruct related hardware. The program can be stored in a computer-readable memory, and the memory can include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0115] The above is only an exemplary embodiment of the present disclosure, and the scope of the present disclosure cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure here, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field not recorded in the present disclosure. The description and examples are regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

[0116] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0117] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A system for simultaneously solving brain source activity and noise, characterized in that: include: A data input module is used to obtain brain activity sampling data and lead field matrix L; A model building module is used to build a sampling data generation model, the sampling data generation model is y(t)=LS(t)+Au(t)+ε, wherein y(t) is the sampling data at time t, S(t) is the brain source activity at time t, A is the gain matrix of M sampling sensor noises, u(t) is the brain source activity at time t, ε is a constant, and the sampling data generation model is converted into a sampling data probability generation model, the sampling data probability generation model includes a prior distribution hyperparameter composed of a prior variance of N brain source activities and a prior variance of M sampling sensor noises; The solution module is used to solve the probability generation model of sampling data using the Bayesian expectation maximum algorithm, determine the hyperparameters of the prior distribution, and solve the brain source activity data and noise data at the same time.

2. The system for simultaneously solving brain source activity and noise as claimed in claim 1, characterized in that: A is an M×M identity matrix.

3. The system for simultaneously solving brain source activity and noise as claimed in claim 2, characterized in that: Before converting the sampled data generation model into a sampled data probability generation model, convert the sampled data generation model into F=[L1,…L N ,e1,…,e M ] Among them, L n is the lead field matrix of the nth brain source, S n (t) is the brain activity of the nth brain source at time t, u m (t) is the noise information of the mth sensor at time t, T is the number of sampling points, 1≤n≤N, 1≤m≤M.

4. The system for simultaneously solving brain source activity and noise as claimed in claim 1, characterized in that: The prior distribution hyperparameter v is is the prior variance of the nth brain source activity, 1≤n≤N, [v N+1 , ..., v N+M ] is the prior variance of the noise, and diag represents the construction of a diagonal matrix.

5. The system for simultaneously solving brain source activity and noise as claimed in claim 1, characterized in that: The solving of the sampling data probability generation model comprises: Step 1, determining the prior distribution of the probability model for generating the sampling data according to the brain activity sampling data obtained by the input module, and solving the mean of the posterior probability distribution of brain activity and noise by using the maximum expectation value algorithm of Bayesian theory; Step 2, constructing a cost function for fitting the generated data covariance of the sampled data generation probability model with the covariance of the brain activity sampled data obtained by the input module, and solving the locally optimal prior distribution hyperparameters by maximizing the marginal likelihood method; Repeat steps 1 and 2 until the cost function converges, and use the prior distribution hyperparameters at this time as the final prior distribution hyperparameters; The brain source activity data and the noise data are solved simultaneously according to the final determined prior distribution hyperparameters.

6. The system for simultaneously solving brain source activity and noise as claimed in claim 1, characterized in that: ε is zero or 1e -20 .

7. A method for simultaneously solving brain source activity and noise, characterized in that: include: Obtain brain activity sampling data and lead field matrix L; Constructing a sampling data generation model, the sampling data generation model is y(t)=LS(t)+Au(t)+ε, wherein y(t) is the sampling data at time t, S(t) is the brain source activity at time t, A is the gain matrix of M sampling sensor noises, u(t) is the brain source activity at time t, ε is a constant, and converting the sampling data generation model into a sampling data probability generation model, the sampling data probability generation model includes a prior distribution hyperparameter composed of a prior variance of N brain source activities and a prior variance of M sampling sensor noises; The Bayesian expectation-maximization algorithm is used to solve the probability generation model of the sampling data, determine the hyperparameters of the prior distribution, and solve the brain source activity data and noise data at the same time.

8. The method for simultaneously solving brain source activity and noise according to claim 7, characterized in that: A is an M×M identity matrix.

9. The method for simultaneously solving brain source activity and noise according to claim 8, characterized in that: Before converting the sampled data generation model into a sampled data probability generation model, convert the sampled data generation model into F=[L1,…L N ,e1,…,e M ] Among them, L n is the lead field matrix of the nth brain source, S n (t) is the brain activity of the nth brain source at time t, u m (t) is the noise information of the mth sensor at time t, T is the number of sampling points, 1≤n≤N, 1≤m≤M.

10. The method for simultaneously solving brain source activity and noise according to claim 7, characterized in that: The prior distribution hyperparameter v is is the prior variance of the nth brain source activity, 1≤n≤N, [v N+1 , ..., v N+M ] is the prior variance of the noise, and diag represents the construction of a diagonal matrix.

Images