A brain function reconstruction method based on symbolic connection component feature pattern

Abstract

The application discloses a human brain function reconstruction method based on symbolic connection group characteristic patterns, belongs to the field of biological information, and is used for human brain function reconstruction. The method is used for the reconstruction of neural activity in the resting state and the task state, and a small number of connection group characteristic patterns can achieve high reconstruction precision, which is much better than the structural connection group characteristic pattern based on diffusion magnetic resonance imaging, and proves that dynamic cross-regional interaction and synchronization play a key role in shaping individual brain function.

Description

Technical Field

[0001] This invention relates to the field of bioinformatics, specifically to a method for reconstructing human brain function based on symbolic connectome feature patterns. Background Technology

[0002] The functions of the human brain are constrained by its anatomical network topology. Tens of thousands of neural modules with different functions work together within this network to complete perception, cognition, and behavioral activities. Even in the resting state, slow fluctuations with strong functional correlations can be observed in distant cortical regions. These fluctuations organize into spontaneous neural activity in the spatiotemporal dimensions, i.e., resting-state functional connectivity. Numerous studies have shown that a small number of relatively fixed structural connections in the human brain can generate complex and diverse functional connections. Generally, two brain regions with structural connections often have functional connections, but two brain regions with functional connections do not necessarily have structural connections. Sometimes, the structural connection between two brain regions with strong functional connections is weak or even non-existent. Currently, little is known about how functional connections in the human brain are generated, which is of great value for understanding the mechanisms of brain cognition and brain diseases.

[0003] Numerous studies have utilized graph-based network models to describe human brain function. While some extracted predefined network metrics or graph metrics can reflect functional connectivity (FC) to some extent, they do not play a dominant role in shaping the overall patterns of human brain function. Furthermore, another category of research focuses on neural field models, but often faces the challenge of insufficient determinism in physiological parameters. Inspired by the Laplace operator eigenvalue decomposition in explaining phenomena such as sound waves, light waves, and electromagnetic waves in nature, the eigenvalue decomposition of the brain connectome Laplace operator has been widely applied to modeling spontaneous cortical activity at the macroscopic scale. Numerous studies have shown that connectome feature patterns play a fundamental role in shaping neural activity during resting states or cognitive tasks, effectively capturing the correlation between structural and functional connectivity in the human brain. However, graph-based eigenvalue decomposition models can only partially predict brain function and cannot simulate negative correlations because the graph representation of the brain connectome cannot directly describe higher-order connections between structurally unconnected brain regions.

[0004] According to the principle of impulse temporal-dependent plasticity, when two brain regions are activated, if one region has a causal influence on the other, the functional connectivity between them tends to strengthen, and vice versa. This change will result in the connectivity between the two brain regions possibly increasing or decreasing, which can be described by positive or negative connectivity, thus forming a symbolic connectome in the human brain. By performing feature decomposition on the Laplace operator of this symbolic connectome, the feature patterns of the symbolic connectome can be obtained. Then, through graphical Fourier transform and inverse transform, different numbers of feature patterns are used to decompose and reconstruct neural activity signals in resting and task states. Summary of the Invention

[0005] The purpose of this invention is to provide a method for reconstructing human brain function based on symbolic connectome feature patterns, so as to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for reconstructing human brain function based on symbolic connectome feature patterns, used for reconstructing human brain neural activity signals in both resting and task-induced states, includes the following steps: S1. Based on the dynamic changes in the weight of brain regions during the propagation of neural activity signals in the human brain, a symbolic connectome of the human brain is constructed according to the principle of pulse temporal dependence plasticity. S2. Perform Laplace eigendecomposition on the graph Laplace operator of the symbolic connection group to obtain the characteristic patterns of the symbolic connection; S3. Based on graphical Fourier transform, neural activity signals are decomposed into a weighted sum of feature patterns; S4. Based on the inverse Fourier transform of the graph, the neural activity signals of individual human brains are reconstructed using different numbers of feature patterns, and the results are evaluated using the Pearson correlation coefficient.

[0007] A further improvement to the technical solution of the present invention is that: S1 specifically includes: S101. Based on the principle of impulse temporal-dependent plasticity, when there is a causal influence between brain regions, the strength of functional connectivity increases or decreases accordingly, represented by positive or negative weights. Analyze the dynamic connectivity relationships between brain regions through linear changes, and then proceed to the next step. Overall fluctuation modeling analysis of the influence of one brain region on all other brain regions; S102. The dynamic connectivity relationships between brain regions are transformed into matrix form to obtain the adjacency matrix, degree matrix and human brain symbolic connectivity matrix of the symbolic connectivity group. The relationship is further expressed by the Laplacian operator and regularized to obtain the graph Laplacian operator that can be used for eigenvalue decomposition.

[0008] A further improvement to the technical solution of the present invention is that the specific implementation method of S1 is as follows: When two brain regions are activated, if one region has a causal influence on the other, the functional connectivity between them tends to strengthen, and vice versa. This change will lead to an increase or decrease in the connectivity between the two brain regions, which can be described as positive or negative connectivity. brain regions and the first Connection strength between brain regions and It changes linearly, that is ,in, Indicates the first The brain region and the first The amount of change in the strength of functional connectivity between brain regions; Indicates the first The brain region and the first The initial functional connectivity strength between brain regions; This represents the linear scaling factor for changes in connection strength. And it is a positive constant, because Presented in a positive or negative manner, therefore, the first... The overall fluctuation model of the influence of one brain region on all other brain regions is as follows: (1) in, Indicates the first The total fluctuation of a brain region after being affected by all other brain regions, that is, the change in its state; Indicates the first The intensity of neural activity in each brain region; Indicates the first The intensity of neural activity in each brain region; It is a symbolic variable, representing the first... The brain region and the first The direction of changes in connections between brain regions; Indicates brain regions and The connection weight between them increases, and vice versa. This indicates a decrease in connection weight; Taking all brain regions into account and ignoring constants We can obtain: (2) Transform formula (2) into matrix form: (3) in, A column vector representing the state changes of all brain regions; This represents a column vector consisting of the current states of all brain regions; Indicates the first The sum of the output connectivity strengths of all brain regions; The adjacency matrix represents the symbolic connectivity group. Since the weight variation between two brain regions is limited to less than 1, therefore... Defined as a matrix in which all elements except the diagonal elements are 1; This represents the corresponding degree matrix; This represents the symbolic connectivity matrix of the human brain, specifying positive or negative functional interactions during activation; 'Indicates the Hadamardi accumulation; The above formula (3) can be expressed using the Laplace operator, i.e. ;because Therefore, the graph Laplace operator of the symbolic connection group It can be defined as: (4) Will After regularization, we get: (5) In the formula: express n × n The identity matrix has diagonal elements of 1 and off-diagonal elements of 0.

[0009] A further improvement to the technical solution of the present invention is that: S2 specifically includes: S201. Perform Laplace eigendecomposition on the graph Laplace operator of the constructed symbolic connection group to obtain its characteristic patterns and eigenvalues. S202. Sort the decomposed feature patterns according to the size of the feature values ​​to form a feature basis for signal decomposition and reconstruction, providing a mathematical basis for subsequent graphical Fourier transform.

[0010] A further improvement to the technical solution of the present invention is that the specific implementation method of S2 is as follows: Using the following formula Laplacian eigenvalue decomposition yields the feature patterns: (6) in, express The Each feature pattern express correspond eigenvalues, where .

[0011] A further improvement to the technical solution of the present invention is that: S3 specifically includes: S301. Based on the feature pattern basis of the symbolic connection group graph Laplace operator, the human brain neural activity signal at any vertex is expanded into a graph Fourier transform, which is represented as a weighted sum of the signal on the feature pattern. S302. For each subject's magnetic resonance imaging data, the neural activity signal is decomposed into a weighted combination of each feature pattern in the time and space dimensions, where the weight reflects the contribution of each pattern to signal formation.

[0012] A further improvement to the technical solution of the present invention is that the specific implementation method of S3 is as follows: Connecting human brain neural activity signals to any vertex on a symbolic graph network The graphical Fourier transform of the signal is defined as follows: Expansion on the feature pattern basis associated with the Graph Laplacian operator: (7) in, Represents a feature pattern in the basis. Representation of feature patterns In the i The components of each brain region; This indicates the neural activity signals in the human brain at a certain moment. In the i Signal values ​​of individual brain regions; The magnetic resonance imaging data of each subject reflects the signals of human brain neural activity. Formula (7) can be expressed as: (8) in, Indicates the corresponding brain region. Indicates time and location. ( y ) indicates in y Time of the first The amplitude or weight of a feature pattern in the formation of neural activity signals. Indicates the first The feature pattern in the first The proportion of each brain region This indicates the total number of brain regions.

[0013] A further improvement to the technical solution of the present invention is that: S4 specifically includes: S401. Based on the inverse Fourier transform of the graph, select the first... M Each symbolic connection group features a characteristic pattern, reconstructing neural activity signals into a weighted linear combination of these characteristic patterns. During the reconstruction process, adjustments are made to... M The value controls the number of feature patterns used, in order to observe the impact of different numbers of patterns on the reconstruction effect; S402. Calculate the cross-correlation between the time series of the reconstructed signals to generate the reconstructed functional connectivity matrix, and calculate the Pearson correlation coefficient between it and the measured functional connectivity matrix. The Pearson correlation coefficient is used as a quantitative indicator to evaluate the reconstruction accuracy and verify the effectiveness of the symbolic connectivity group feature pattern in neural activity reconstruction.

[0014] A further improvement to the technical solution of the present invention is that the specific implementation method of S4 is as follows: Based on the inverse graph Fourier transform, neural activity signals are reconstructed: (9) in M Indicates the number of feature patterns used in the reconstruction process; The functional connectivity matrix is ​​obtained by calculating the cross-correlation between the reconstructed signal time series. The Pearson correlation coefficient obtained by correlating the reconstructed matrix with the measured functional connectivity matrix is ​​used as the evaluation criterion to verify the reconstruction performance.

[0015] Compared with existing technologies, the beneficial effects of this invention include: Based on the dynamic changes in the connection weights between brain regions during the propagation of neural activity signals in the human brain, this invention constructs a symbolic connectome of the human brain, capable of describing the dynamic interactions between brain regions. Simultaneously, based on graphical Fourier transform and inverse transform, it achieves the decomposition and reconstruction of individual human brain neural activity signals in both resting and task states, and achieves high reconstruction accuracy using only a small number of connectome feature patterns, far superior to structural connectome feature patterns based on diffusion magnetic resonance imaging. This demonstrates the crucial role of dynamic cross-regional interactions and synchronization in shaping individual brain function. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method disclosed in this invention; Figure 2 This is a schematic diagram of the feature patterns of the first five symbol connection groups in one embodiment; Figure 3 The example shows the accuracy results of individual reconstruction using different numbers of symbolic connection group feature patterns in the resting state, i.e., the reconstruction accuracy and average result curves of individuals in the resting state. Figure 4 The example plot shows the accuracy results of individual reconstruction using different numbers of symbolic connection group feature patterns under the relationship judgment task activation state, i.e., the reconstruction accuracy and average result curve of individuals under the relationship judgment task activation state. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figure 1-4 As shown, this invention provides a method for reconstructing human brain function based on symbolic connectome feature patterns, used for reconstructing human brain neural activity signals in both resting and task-triggered states, comprising the following steps: S1. Based on the dynamic changes in the weight of brain regions during the propagation of neural activity signals in the human brain, a symbolic connectome of the human brain is constructed according to the principle of pulse temporal dependence plasticity. S2. Perform Laplace eigendecomposition on the graph Laplace operator of the symbolic connection group to obtain the characteristic patterns of the symbolic connection; S3. Based on graphical Fourier transform, neural activity signals are decomposed into a weighted sum of feature patterns; S4. Based on the inverse Fourier transform of the graph, the neural activity signals of individual human brains are reconstructed using different numbers of feature patterns, and the results are evaluated using the Pearson correlation coefficient.

[0019] S1 includes the principle of impulse-time dependent plasticity, which states that when two brain regions are activated, if one region has a causal influence on the other, the functional connectivity between them tends to strengthen, and vice versa. This change will result in either an increase or decrease in the connectivity between the two brain regions, which can be described as positive or negative connectivity; brain regions and the first Connection strength between brain regions Usually with It changes linearly, that is ,in, Indicates the first The brain region and the first The amount of change in the strength of functional connectivity between brain regions; Indicates the first The brain region and the first The initial functional connectivity strength between brain regions; This represents the linear scaling factor for changes in connection strength. And it is a positive constant. Because It could be positive or negative, therefore the first The total fluctuation of the influence of one brain region on all other brain regions can be modeled as follows: (1) in, Indicates the first The total fluctuation of a brain region after being affected by all other brain regions, that is, the change in its state; Indicates the first The intensity of neural activity in each brain region; Indicates the first The intensity of neural activity in each brain region; It is a symbolic variable, representing the first... The brain region and the first The direction of changes in connections between brain regions; Indicates brain regions and The connection weight between them increases, and vice versa. This indicates a reduction in connection weights, taking into account all brain regions and ignoring constants. We can obtain: (2) Transform formula (2) into matrix form: (3) in, A column vector representing the state changes of all brain regions; This represents a column vector consisting of the current states of all brain regions; Indicates the first The sum of the output connectivity strengths of all brain regions; The adjacency matrix represents the symbolic connectivity group. Since the weight variation between two brain regions is limited to less than 1, therefore... Defined as a matrix in which all elements except the diagonal elements are 1; This represents the corresponding degree matrix; This represents the symbolic connectivity matrix of the human brain, specifying positive or negative functional interactions during activation; 'Indicates the Hadamardi accumulation; The above formula (3) can be expressed using the Laplace operator, i.e. .because Therefore, the graph Laplace operator of the symbolic connection group It can be defined as: (4) Will After regularization, we get: (5) In the formula: express n × n The identity matrix has diagonal elements of 1 and off-diagonal elements of 0.

[0020] S2 includes the following formula for... Laplacian eigenvalue decomposition yields the feature patterns: (6) in, express The Each feature pattern express correspond eigenvalues, where .

[0021] S3 includes human brain neural activity signals that connect any vertex on a symbolically connected graph network. The graphical Fourier transform of the signal is defined as follows: Expansion on the feature pattern basis associated with the Graph Laplacian operator: (7) in, Represents a feature pattern in the basis. Representation of feature patterns In the i The components of each brain region; This indicates the neural activity signals in the human brain at a certain moment. In the i Signal values ​​of individual brain regions; The magnetic resonance imaging data of each subject reflects the signals of human brain neural activity. Formula (7) can be expressed as: (8) in, Indicates the corresponding brain region. Indicates time and location. ( y ) indicates in y Time of the first The amplitude or weight of a feature pattern in the formation of neural activity signals. Indicates the first The feature pattern in the first The proportion of each brain region This indicates the total number of brain regions.

[0022] S4 includes an inverse transform based on the graphical Fourier transform to reconstruct neural activity signals: (9) in M This indicates the number of feature patterns used in the reconstruction process.

[0023] The functional connectivity matrix is ​​obtained by calculating the cross-correlation between the reconstructed signal time series. The Pearson correlation coefficient obtained by correlating the reconstructed matrix with the measured functional connectivity matrix is ​​used as the evaluation criterion to verify the reconstruction performance.

[0024] In this embodiment, functional magnetic resonance imaging (fMRI) data from 309 subjects in the HCP-S1200 database, under resting state and relational task excitation states, were selected, and the brain was divided into 360 regions of interest (ROIs). First, based on the dynamic changes in the weights of brain inter-brain connections during the propagation of neural activity signals, a symbolic connectome was constructed according to the principle of impulse temporal-dependent plasticity. The graph Laplace operator of the symbolic connectome was then applied. Perform Laplace eigenvalue decomposition to obtain the feature patterns arranged in ascending order of frequency. Figure 2 The image shows a schematic diagram of the first five feature patterns of one of the subjects. Based on graphical Fourier transform and inverse Fourier transform, neural activity signals in the resting state and task-triggered state are decomposed and reconstructed. The functional connectivity matrix is ​​obtained by calculating the cross-correlation between the reconstructed signal time series. The Pearson correlation coefficient obtained by correlating this matrix with the measured functional connectivity matrix is ​​used as the reconstruction accuracy to measure reconstruction performance. Figure 3 and Figure 4 As shown.

[0025] Depend on Figure 3 and Figure 4 It can be seen that significant reconstruction accuracy can be achieved using only a small number of symbolic connectome feature patterns, both in the resting state and under task-induced conditions. This demonstrates that symbolic connectome feature patterns play a crucial role in shaping individual functional brain structure, meeting the expected goals. This invention can reconstruct human brain function relatively accurately and proves the key role of dynamic cross-regional interactions and synchronization in shaping individual brain function.

[0026] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for reconstructing human brain function based on symbolic connectome feature patterns, used for reconstructing human brain neural activity signals in resting and task-induced states, characterized in that, Includes the following steps: S1. Based on the dynamic changes in the weight of brain regions during the propagation of neural activity signals in the human brain, a symbolic connectome of the human brain is constructed according to the principle of pulse temporal dependence plasticity. S2. Perform Laplace eigendecomposition on the graph Laplace operator of the symbolic connection group to obtain the characteristic patterns of the symbolic connection; S3. Based on graphical Fourier transform, neural activity signals are decomposed into a weighted sum of feature patterns; S4. Based on the inverse Fourier transform of the graph, the neural activity signals of individual human brains are reconstructed using different numbers of feature patterns, and the results are evaluated using the Pearson correlation coefficient.

2. The method for reconstructing human brain function based on symbolic connectome feature patterns according to claim 1, characterized in that: S1 specifically includes: S101. Based on the principle of impulse temporal-dependent plasticity, when there is a causal influence between brain regions, the strength of functional connectivity increases or decreases accordingly, represented by positive or negative weights. Analyze the dynamic connectivity relationships between brain regions through linear changes, and then proceed to the next step. Overall fluctuation modeling analysis of the influence of one brain region on all other brain regions; S102. The dynamic connectivity relationships between brain regions are transformed into matrix form to obtain the adjacency matrix, degree matrix and human brain symbolic connectivity matrix of the symbolic connectivity group. The relationship is further expressed by the Laplacian operator and regularized to obtain the graph Laplacian operator that can be used for eigenvalue decomposition.

3. The method for reconstructing human brain function based on symbolic connectome feature patterns according to claim 2, characterized in that: The specific implementation method of S1 is as follows: When two brain regions are activated, if one region has a causal influence on the other, the functional connectivity between them tends to strengthen, and vice versa. This change will lead to an increase or decrease in the connectivity between the two brain regions, which can be described as positive or negative connectivity. brain regions and the first Connection strength between brain regions and It changes linearly, that is ,in, Indicates the first The brain region and the first The amount of change in the strength of functional connectivity between brain regions; Indicates the first The brain region and the first The initial functional connectivity strength between brain regions; This represents the linear scaling factor for changes in connection strength. And it is a positive constant, because Presented in a positive or negative manner, therefore, the first... The overall fluctuation model of the influence of one brain region on all other brain regions is as follows: (1) in, Indicates the first The total fluctuation of a brain region after being affected by all other brain regions, that is, the change in its state; Indicates the first The intensity of neural activity in each brain region; Indicates the first The intensity of neural activity in each brain region; It is a symbolic variable, representing the first... The brain region and the first The direction of changes in connections between brain regions; Indicates brain regions and The connection weight between them increases, and vice versa. This indicates a decrease in connection weight; Taking all brain regions into account and ignoring constants We can obtain: (2) Transform formula (2) into matrix form: (3) in, A column vector representing the state changes of all brain regions; This represents a column vector consisting of the current states of all brain regions; Indicates the first The sum of the output connectivity strengths of all brain regions; The adjacency matrix represents the symbolic connectivity group. Since the weight variation between two brain regions is limited to less than 1, therefore... Defined as a matrix in which all elements except the diagonal elements are 1; This represents the corresponding degree matrix; This represents the symbolic connectivity matrix of the human brain, specifying positive or negative functional interactions during activation; 'Indicates the Hadamardi accumulation; The above formula (3) can be expressed using the Laplace operator, i.e. ;because Therefore, the graph Laplace operator of the symbolic connection group It can be defined as: (4) Will After regularization, we get: (5) In the formula: express n×n The identity matrix has diagonal elements of 1 and off-diagonal elements of 0.

4. The method for reconstructing human brain function based on symbolic connectome feature patterns according to claim 3, characterized in that: S2 specifically includes: S201. Perform Laplace eigendecomposition on the graph Laplace operator of the constructed symbolic connection group to obtain its characteristic patterns and eigenvalues. S202. Sort the decomposed feature patterns according to the size of the feature values ​​to form a feature basis for signal decomposition and reconstruction.

5. A method for reconstructing human brain function based on symbolic connectome feature patterns according to claim 4, characterized in that: The specific implementation method of S2 is as follows: Using the following formula Laplacian eigenvalue decomposition yields the feature patterns: (6) in, express The Each feature pattern express correspond eigenvalues, where .

6. The method for reconstructing human brain function based on symbolic connectome feature patterns according to claim 5, characterized in that: S3 specifically includes: S301. Based on the feature pattern basis of the symbolic connection group graph Laplace operator, the human brain neural activity signal at any vertex is expanded into a graph Fourier transform, which is represented as a weighted sum of the signal on the feature pattern. S302. For each subject's magnetic resonance imaging data, the neural activity signal is decomposed into a weighted combination of each feature pattern in the time and space dimensions, where the weight reflects the contribution of each pattern to signal formation.

7. A method for reconstructing human brain function based on symbolic connectome feature patterns according to claim 6, characterized in that: The specific implementation method of S3 is as follows: Connecting human brain neural activity signals to any vertex on a symbolic graph network The graphical Fourier transform of the signal is defined as follows: Expansion on the feature pattern basis associated with the Graph Laplacian operator: (7) in, Represents a feature pattern in the basis. Representation of feature patterns In the i The components of each brain region; This indicates the neural activity signals in the human brain at a certain moment. In the i Signal values ​​of individual brain regions; The magnetic resonance imaging data of each subject reflects the signals of human brain neural activity. Formula (7) can be expressed as: (8) in, Indicates the corresponding brain region. Indicates time and location. ( y ) indicates in y Time of the first The amplitude or weight of a feature pattern in the formation of neural activity signals. Indicates the first The feature pattern in the first The proportion of each brain region This indicates the total number of brain regions.

8. A method for reconstructing human brain function based on symbolic connectome feature patterns according to claim 7, characterized in that: S4 specifically includes: S401. Based on the inverse Fourier transform of the graph, select the first... M Each symbolic connection group features a characteristic pattern, reconstructing neural activity signals into a weighted linear combination of these characteristic patterns. During the reconstruction process, adjustments are made to... M The value controls the number of feature patterns used, in order to observe the impact of different numbers of patterns on the reconstruction effect; S402. Calculate the cross-correlation between the time series of the reconstructed signals to generate the reconstructed functional connectivity matrix, and calculate the Pearson correlation coefficient between it and the measured functional connectivity matrix. The Pearson correlation coefficient is used as a quantitative indicator to evaluate the reconstruction accuracy and verify the effectiveness of the symbolic connectivity group feature pattern in neural activity reconstruction.

9. A method for reconstructing human brain function based on symbolic connectome feature patterns according to claim 8, characterized in that: The specific implementation method of S4 is as follows: Based on the inverse graph Fourier transform, neural activity signals are reconstructed: (9) in M Indicates the number of feature patterns used in the reconstruction process; The functional connectivity matrix is ​​obtained by calculating the cross-correlation between the reconstructed signal time series. The Pearson correlation coefficient obtained by correlating the reconstructed matrix with the measured functional connectivity matrix is ​​used as the evaluation criterion to verify the reconstruction performance.

Images