Brain entropy-based method and system for quantitative characterization of brain function
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-02-08
- Publication Date
- 2026-08-13
Smart Images
Figure CN2025076438_13082026_PF_FP_ABST
Abstract
Description
A method and system for quantitatively characterizing brain function based on brain entropy Technical Field
[0001] This invention belongs to the field of brain function characterization technology, and in particular relates to a method and system for quantitatively characterizing brain function based on brain entropy. Background Technology
[0002] Accurately characterizing brain functional states is a crucial issue in brain function research. Traditional methods primarily analyze interactions between brain regions based on functional connectivity strength, but neglect the complex dynamic characteristics of brain activity. To more comprehensively describe brain functional states, researchers have proposed methods based on brain entropy. However, in practical applications, organically combining brain entropy with brain functional networks to reveal the impact of brain entropy on these networks still faces numerous technical challenges.
[0003] First, designing a reasonable experimental paradigm to obtain high-quality resting-state functional magnetic resonance imaging (fMRI) data is a crucial issue during the data acquisition and preprocessing stages. Different subject states, head movement artifacts, and other factors can introduce noise, affecting the reliability of subsequent analyses. Furthermore, selecting appropriate strategies to remove various artifacts and noise during preprocessing is also a question worthy of in-depth exploration.
[0004] Secondly, in the process of brain entropy calculation and brain functional network construction, the selection of appropriate brain region partitioning schemes and functional connectivity measurement methods is crucial to the reliability and interpretability of the results. Different scales and methods of brain region partitioning may result in significant differences in the obtained brain entropy values and network characteristics. At the same time, there are numerous methods for calculating functional connectivity, and different methods have varying sensitivities to noise and their ability to capture dynamic characteristics. How to weigh and select the optimal method is also an urgent problem to be solved.
[0005] Finally, in the correlation analysis between brain entropy and brain functional networks, designing reasonable mathematical models and statistical methods to accurately characterize the intrinsic relationship between the two also presents many challenges. The complexity and dynamism of brain functional networks bring difficulties to modeling and analysis. How to ensure the interpretability of the model while making full use of brain entropy information to characterize the features of brain functional networks is an urgent problem to be explored. Summary of the Invention
[0006] This invention proposes a method and system for quantitatively characterizing brain function based on brain entropy, in order to solve the problems existing in the prior art.
[0007] To achieve the above objectives, the present invention provides a method for quantitatively characterizing brain function based on brain entropy, comprising the following steps:
[0008] Acquire resting-state functional magnetic resonance imaging (fMRI) data and perform preprocessing;
[0009] The preprocessed functional magnetic resonance imaging data are divided according to the preset brain region division scheme, and the functional activity characteristics of each brain region are determined based on the division results.
[0010] Using a pre-defined functional connectivity metric, the functional connectivity strength between the segmented brain regions is calculated to obtain the functional connectivity matrix.
[0011] Based on the preset brain entropy calculation method, the entropy value of each brain region is calculated according to the preprocessed functional magnetic resonance time series data to obtain the brain entropy distribution map;
[0012] By combining the functional connectivity matrix and the brain entropy distribution map, a brain functional network is constructed, and the characteristics of the brain functional network are analyzed.
[0013] A mathematical model was designed to correlate brain entropy values with the characteristics of the brain functional network, thereby capturing the temporal and spatial variation characteristics of the brain functional network.
[0014] Preferably, the preprocessing includes:
[0015] The head movement correction algorithm is used to correct artifacts in the data caused by the subject’s head movements.
[0016] The noise reduction process for the data after head movement correction was performed using independent component analysis.
[0017] The ICA algorithm is used to separate noise components from the data and remove them, thereby improving the signal-to-noise ratio and obtaining high-quality functional magnetic resonance imaging (fMRI) data.
[0018] Preferably, determining the functional activity characteristics of each brain region includes:
[0019] According to the preset brain region division scheme, preprocessed functional magnetic resonance imaging (fMRI) data is obtained, and the fMRI data is analyzed and processed to divide it into at least two brain regions. For each brain region after division, fMRI data within that brain region is extracted, and cluster analysis is performed on the data within that brain region using a clustering algorithm to obtain the functional activity characteristics of that brain region.
[0020] Preferably, the clustering algorithm includes K-means.
[0021] Preferably, the obtained functional connection matrix includes:
[0022] Based on a preset functional connectivity measurement method, obtain the segmented brain region information and the functional connectivity data between brain regions;
[0023] For each brain region, the functional connectivity strength between it and other brain regions is calculated to obtain the connectivity strength values between each pair of brain regions;
[0024] The calculated connection strength values between brain regions are filled into the corresponding positions in the pre-established functional connectivity matrix to form a complete functional connectivity matrix.
[0025] Preferably, obtaining the brain entropy distribution map includes:
[0026] According to the preset brain entropy calculation method, functional magnetic resonance time series data are obtained, and the time series data are preprocessed to obtain preprocessed time series data.
[0027] Wavelet decomposition algorithm is used to extract features from preprocessed time series data to obtain wavelet coefficients for each frequency band.
[0028] Based on the wavelet coefficients of each frequency band, the entropy value of each brain region is calculated. If the entropy value is greater than the preset threshold, the brain region is judged to be an active brain region.
[0029] The location coordinates of each active brain region are obtained, and the active brain regions are clustered using a clustering algorithm to obtain the distribution of brain regions with different functions;
[0030] Based on the distribution of brain regions, the support vector machine algorithm is used to classify brain regions with different functions to obtain the location range of each functional brain region;
[0031] For the location range of each functional brain region, a continuous brain entropy distribution map is generated by an interpolation algorithm to obtain a complete brain entropy distribution image.
[0032] Preferably, analyzing the characteristics of the brain functional network includes:
[0033] Based on the functional connectivity matrix, the functional connectivity strength between brain regions is obtained, and an initial brain functional network is constructed.
[0034] Based on the brain entropy distribution map, the brain entropy value of each brain region is obtained, and the brain entropy value is used as the attribute of the node to update the initial brain function network, thus obtaining the updated brain function network.
[0035] Calculate the node degree of each node in the updated brain functional network, where the node degree represents the number of edges directly connected to that node, and obtain the node degree distribution characteristics of the network.
[0036] Calculate the clustering coefficient of each node in the updated brain functional network. The clustering coefficient represents the ratio of the actual number of edges between the node's neighboring nodes to the total number of possible edges, thereby obtaining the clustering characteristics of the network.
[0037] The shortest path length between any two nodes in the updated brain functional network is calculated using Dijkstra's algorithm, and the distribution characteristics of the shortest path length of the network are obtained.
[0038] Preferably, capturing the temporal variation characteristics and spatial distribution characteristics of the brain functional network includes:
[0039] Based on the spatiotemporal characteristics of brain functional networks, a spatiotemporal network model is adopted, and brain entropy value is used as an attribute of network nodes to establish a mathematical correlation model between brain entropy value and brain functional networks.
[0040] The spatial organization characteristics of brain functional networks are obtained by calculating the topological feature parameters of brain functional networks using graph theory algorithms.
[0041] Time series analysis was used to calculate the statistical characteristics of brain entropy values and obtain the dynamic change pattern of brain entropy values over time.
[0042] The spatial feature parameters of the brain functional network and the temporal feature parameters of the brain entropy value are input into the association model. The association model is trained by machine learning algorithm to establish a nonlinear mapping relationship between the brain entropy value and the characteristics of the brain functional network.
[0043] Cross-validation was used to evaluate the generalization performance of the association model. The association model was optimized by adjusting the model hyperparameters to obtain the optimal brain entropy value-brain function network association model.
[0044] Predict brain entropy values at different times and spatial locations using an optimal brain entropy-brain function network association model.
[0045] This invention also provides a system for quantitatively characterizing brain function based on brain entropy, comprising:
[0046] The data acquisition and preprocessing module is used to acquire resting-state functional magnetic resonance data and perform preprocessing.
[0047] The brain region segmentation module is used to segment the preprocessed functional magnetic resonance imaging data according to a preset brain region segmentation scheme, and to determine the functional activity characteristics of each brain region based on the segmentation results.
[0048] The functional connectivity calculation module is used to calculate the functional connectivity strength between the divided brain regions using a preset functional connectivity metric method, and obtain the functional connectivity matrix.
[0049] The brain entropy calculation module is used to calculate the entropy value of each brain region based on the preprocessed functional magnetic resonance time series data, and obtain the brain entropy distribution map, according to the preset brain entropy calculation method.
[0050] A brain functional network construction module is used to construct a brain functional network by combining the functional connectivity matrix and the brain entropy distribution map, and to analyze the characteristics of the brain functional network.
[0051] The mathematical model design module is used to design mathematical models that correlate brain entropy values with the characteristics of the brain functional network, capturing the temporal variation and spatial distribution characteristics of the brain functional network.
[0052] Compared with the prior art, the present invention has the following advantages and technical effects:
[0053] This method first acquires high-quality functional magnetic resonance imaging (fMRI) data and removes head movement artifacts through preprocessing. Then, the data is divided into multiple brain regions, and the functional connectivity strength between these regions and the entropy value of each region are calculated. Based on the functional connectivity matrix and brain entropy distribution map, a brain functional network is constructed and its characteristics are analyzed. Finally, a mathematical model is designed to correlate brain entropy values with network characteristics, capturing the spatiotemporal variation features of the network. The unique aspect of this invention lies in combining brain entropy with functional connectivity, revealing the dynamic characteristics of the brain functional network through mathematical modeling. This method not only provides a more comprehensive description of brain functional states but also offers a new perspective for studying brain functional abnormalities, which is of great significance to neuroscience and clinical diagnosis. Attached Figure Description
[0054] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:
[0055] Figure 1 is a flowchart of the method according to an embodiment of the present invention. Detailed Implementation
[0056] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.
[0057] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.
[0058] Example 1
[0059] As shown in Figure 1, this embodiment provides a method for quantitatively characterizing brain function based on brain entropy, including the following steps:
[0060] Acquire resting-state functional magnetic resonance imaging (fMRI) data and perform preprocessing;
[0061] The preprocessed functional magnetic resonance imaging data are divided according to the preset brain region division scheme, and the functional activity characteristics of each brain region are determined based on the division results.
[0062] Using a pre-defined functional connectivity metric, the functional connectivity strength between the segmented brain regions is calculated to obtain the functional connectivity matrix.
[0063] Based on the preset brain entropy calculation method, the entropy value of each brain region is calculated according to the preprocessed functional magnetic resonance time series data to obtain the brain entropy distribution map;
[0064] By combining the functional connectivity matrix and the brain entropy distribution map, a brain functional network is constructed, and the characteristics of the brain functional network are analyzed.
[0065] A mathematical model was designed to correlate brain entropy values with the characteristics of the brain functional network, thereby capturing the temporal and spatial variation characteristics of the brain functional network.
[0066] Specifically, the following steps are included:
[0067] S101. Acquire resting functional magnetic resonance imaging (fMRI) data from the subjects using a pre-defined experimental paradigm, controlling scan time and parameters to ensure the data quality meets the requirements for subsequent analysis. Preprocessing is performed on the acquired raw fMRI data. A head movement correction algorithm is used to correct artifacts introduced by the subject's head movements, reducing their impact on data quality. Independent component analysis (ICA) is used to denoise the head-movement-corrected data. The ICA algorithm separates and removes noise components from the data, improving the signal-to-noise ratio and obtaining high-quality fMRI data.
[0068] In this embodiment, during the data acquisition phase, researchers need to carefully design experimental paradigms, such as requiring subjects to rest with their eyes closed for 5-10 minutes, while controlling scanning parameters such as TR=2000ms and TE=30ms, to obtain high-quality raw data. Preprocessing is a crucial step in ensuring data reliability. Head movement correction can employ rigid body transformation algorithms to align the image at each time point with the reference image, thereby eliminating the influence of head movements. For example, if a subject's head undergoes a 2mm translation and a 1° rotation during the scan, head movement correction can minimize the errors caused by these minute movements. Independent component analysis (ICA) is a powerful noise reduction tool. It can decompose the fMRI signal into multiple independent components. Researchers can identify components representing physiological noise (such as respiration and heartbeat) based on the temporal and spatial distribution characteristics of these components and remove them from the raw signal. This process can significantly improve the signal-to-noise ratio of the data, laying the foundation for subsequent analysis.
[0069] S102. According to a preset brain region division scheme, preprocessed functional magnetic resonance imaging (fMRI) data is acquired. The fMRI data is analyzed and processed to divide it into at least two brain regions. For each divided brain region, fMRI data within that region is extracted, and cluster analysis is performed on the data within that region using a clustering algorithm to obtain the functional activity characteristics of that brain region.
[0070] In this embodiment, the brain can be divided into multiple functional regions, such as the prefrontal cortex, parietal lobe, occipital lobe, and temporal lobe, according to a preset brain region segmentation scheme. Taking the prefrontal cortex as an example, it is mainly responsible for higher cognitive functions, such as decision-making and executive control. When acquiring preprocessed functional magnetic resonance imaging (fMRI) data, automated brain region segmentation algorithms, such as image intensity-based watershed algorithms or anatomical structure-based template matching methods, can be used to accurately segment the brain image into different functional regions. For each segmented brain region, fMRI data within that region is extracted and analyzed using clustering algorithms. Taking K-means clustering as an example, voxels in the prefrontal cortex region can be divided into several classes based on their temporal series similarity. Assuming the prefrontal cortex region is divided into three classes, it may be found that these three classes correspond to different cognitive functions, such as working memory, emotion regulation, and attention control. This clustering result reflects the functional differentiation within the prefrontal cortex, providing a new perspective for understanding its complex functions. When constructing brain region functional activity feature vectors, multiple dimensions of features can be considered. For example, for the prefrontal cortex, this can include average activation intensity, spatial range of activation, and temporal variation patterns. These features together form a multidimensional vector that comprehensively describes the functional activity characteristics of this brain region.
[0071] S103. According to the preset functional connectivity measurement method, obtain the segmented brain region information and the functional connectivity data between brain regions. For each brain region, calculate the functional connectivity strength between it and other brain regions to obtain the connectivity strength values between each pair of brain regions. Fill the calculated connectivity strength values between brain regions into the corresponding positions in the pre-established functional connectivity matrix to form a complete functional connectivity matrix.
[0072] Specifically, functional connectivity metrics are important tools for assessing interactions between brain regions. For example, the Pearson correlation coefficient can be used to quantify the similarity of time-series signals from two brain regions. For segmented brain regions, such as the prefrontal and temporal lobes, the correlation coefficient of their functional magnetic resonance imaging (fMRI) time-series signals is calculated, yielding a value between -1 and 1. Positive values indicate a positive correlation, negative values indicate a negative correlation, and larger absolute values indicate a stronger correlation. When constructing the functional connectivity matrix, the connectivity strength of each pair of brain regions is filled into the corresponding positions. Assuming there are 10 brain regions, a 10x10 symmetric matrix is obtained. The diagonal elements of the matrix are usually set to 1, indicating that the brain region is perfectly correlated with itself. The other elements of the matrix are filled with the connectivity strength values of the corresponding pair of brain regions.
[0073] S104. Based on a preset brain entropy calculation method, the entropy value of each brain region is calculated using the preprocessed functional magnetic resonance time series data to obtain a brain entropy distribution map.
[0074] According to a preset brain entropy calculation method, functional magnetic resonance imaging (fMRI) time-series data is acquired. The time-series data is preprocessed to obtain preprocessed time-series data. For the preprocessed time-series data, wavelet decomposition is used for feature extraction to obtain wavelet coefficients for each frequency band. Based on the wavelet coefficients of each frequency band, the entropy value of each brain region is calculated. If the entropy value is greater than a preset threshold, the brain region is determined to be an active brain region. The location coordinates of each active brain region are obtained, and a clustering algorithm is used to cluster the active brain regions to obtain the distribution of brain regions with different functions. Based on the brain region distribution, a support vector machine (SVM) algorithm is used to classify the brain regions with different functions to obtain the location range of each functional brain region. For the location range of each functional brain region, an interpolation algorithm is used to generate a continuous brain entropy distribution map, resulting in a complete brain entropy distribution image.
[0075] Specifically, brain entropy calculation is an important method for assessing the functional state of the brain. First, functional magnetic resonance imaging (fMRI) time-series data is acquired, reflecting the activity intensity of different brain regions over time. The time-series data is preprocessed, including noise removal and head movement correction, to improve data quality. Wavelet decomposition, a powerful signal processing tool, decomposes the time-series data into wavelet coefficients of different frequency bands. These coefficients reflect the characteristics of brain activity at different time scales. For example, low-frequency coefficients may correspond to slow-wave activity in the brain, while high-frequency coefficients may reflect rapid neuronal firing. The entropy value of each brain region is calculated based on the wavelet coefficients. The entropy value reflects the complexity and uncertainty of brain region activity. A higher entropy value generally indicates that the brain region is active. Assuming an entropy threshold of 0.8 is set, a brain region is considered active when its entropy value exceeds 0.8. This method helps identify the most active areas of the brain. After obtaining the location coordinates of the active brain regions, clustering algorithms (such as K-means) are used to perform cluster analysis on these coordinates. The purpose of this step is to group functionally similar brain regions together. For example, multiple active points in the prefrontal cortex might be clustered together, which could be related to higher cognitive functions. Support Vector Machine (SVM) algorithms are used to classify brain regions with different functions. SVM can find the optimal classification hyperplane in high-dimensional space, thus accurately dividing brain regions into different functions. The result of this step is to obtain the precise location range of each functional brain region. Next, a continuous brain entropy distribution map is generated using interpolation algorithms. Commonly used interpolation methods include linear interpolation and spline interpolation. The purpose of this step is to transform discrete brain entropy values into a continuous distribution map, allowing for a visual observation of the entropy distribution across the entire brain.
[0076] S105. Based on the functional connectivity matrix, obtain the functional connectivity strength between brain regions and construct an initial brain functional network; based on the brain entropy distribution map, obtain the brain entropy value of each brain region, use the brain entropy value as the attribute of the node, update the initial brain functional network, and obtain an updated brain functional network; for the updated brain functional network, calculate the node degree of each node, where the node degree represents the number of edges directly connected to the node, and obtain the node degree distribution characteristics of the network; for the updated brain functional network, calculate the clustering coefficient of each node, where the clustering coefficient represents the ratio of the actual number of edges between the node's neighboring nodes to the total number of possible edges, and obtain the clustering characteristics of the network; for the updated brain functional network, use Dijkstra's algorithm to calculate the shortest path length between any two nodes, and obtain the shortest path length distribution characteristics of the network.
[0077] Specifically, the functional connectivity matrix reflects the strength of interactions between different brain regions. Analyzing these connections allows for the construction of an initial brain functional network. For example, for a functional connectivity matrix containing 100 brain regions, pairs of regions with a connectivity strength greater than 0.5 can be considered functionally connected, thus obtaining a preliminary brain functional network topology. The brain entropy distribution map provides information on the activity level of each brain region. Incorporating brain entropy values as node attributes into the network provides a more comprehensive description of the functional state of brain regions. Assuming the entropy value of a brain region is 0.8, this value can be assigned to the corresponding network node, making the network reflect not only connectivity relationships but also brain region activity information. Node degree is a key indicator of the importance of a brain region. In the updated brain functional network, if a brain region has direct connections with the other 20 brain regions, its node degree is 20. High node degree often indicates that the brain region plays an important role in information processing. The clustering coefficient reflects the local connectivity density of the network. For example, if a node has 10 actual edges between its neighbors, while theoretically there could be a maximum of 15 edges, then the clustering coefficient of that node is approximately 10 / 15 ≈ 0.67. A high clustering coefficient indicates that the local network structure of that brain region is compact. The distribution characteristics of the shortest path length reflect the global efficiency of the network. Using Dijkstra's algorithm, the shortest connection path between any two brain regions can be calculated. For example, if the shortest path between two distant brain regions only requires passing through 3 intermediate nodes, this indicates that the network has efficient long-distance information transmission capabilities.
[0078] S106. Based on the spatiotemporal characteristics of brain functional networks, a spatiotemporal network model is adopted, using brain entropy as an attribute of network nodes to establish a mathematical correlation model between brain entropy and brain functional networks. Graph theory algorithms are used to calculate the topological characteristic parameters of the brain functional network, including clustering coefficients, shortest path lengths, and small-world properties, revealing the spatial organization characteristics of the brain functional network. Time series analysis is used to calculate the statistical characteristics of brain entropy, including mean, variance, and autocorrelation coefficient, characterizing the temporal dynamic changes of brain entropy. The spatial characteristic parameters of the brain functional network and the temporal characteristic parameters of brain entropy are input into the correlation model. Machine learning algorithms are used to train the correlation model, establishing a nonlinear mapping relationship between brain entropy and brain functional network characteristics. Cross-validation is used to evaluate the generalization performance of the correlation model. By adjusting the model hyperparameters, the correlation model is optimized to obtain the optimal brain entropy-brain functional network correlation model. Using the trained correlation model, brain entropy values at different times and spatial locations are predicted, and abnormal changes in the brain functional network are judged based on the predicted brain entropy values, achieving dynamic monitoring and early warning of brain function. The results of the correlation analysis between brain entropy values and brain functional networks are visualized, generating a spatiotemporal distribution map of brain functional states, providing an intuitive reference for neuroscience research and clinical diagnosis.
[0079] Specifically, the spatiotemporal network model is an analytical method that combines temporal and spatial information, suitable for studying the dynamic changes of brain functional networks. In this model, brain entropy is used as an attribute of network nodes, providing a more comprehensive description of the brain's functional state. For example, in a network containing 100 brain regions, each node has a corresponding brain entropy value, which may vary between 0 and 1, reflecting the complexity of information processing in that brain region. Graph theory algorithms play a crucial role in analyzing the topological characteristics of brain functional networks. The clustering coefficient reflects the degree of clustering of nodes in the network, typically ranging from 0 to 1. A higher clustering coefficient indicates the presence of tightly connected local structures within the network. The shortest path length measures the efficiency of information transmission in the network, usually expressed as the average shortest path length. The small-world property refers to a network that simultaneously possesses a high clustering coefficient and a short average path length; this characteristic is prevalent in healthy brains. Time series analysis methods can reveal the dynamic patterns of brain entropy changes. For example, calculating the average brain entropy value at different times of day may reveal that brain entropy values are higher in the early morning and evening, and slightly decrease in the afternoon. Variance reflects the fluctuation of brain entropy values; a large variance may indicate unstable brain function. The autocorrelation coefficient reveals periodic changes in brain entropy, such as the possibility of similar patterns appearing every 24 hours. Combining spatial and temporal features allows for the construction of a correlation model. This model can be a deep neural network, with the input layer containing network topology features and time-series features, and the output layer predicting future brain entropy distributions. Using backpropagation, this model can be trained to accurately capture the non-linear relationship between brain entropy and network characteristics. Cross-validation is an important method for evaluating model generalization performance. For example, 5-fold cross-validation can be used, dividing the dataset into five parts, using four parts for training and one part for testing, repeated five times. By adjusting the model's hyperparameters, such as the number of layers in the neural network and the number of neurons per layer, the optimal model configuration can be found. A trained model can be used to predict and monitor brain function. For instance, if the model predicts a sudden decrease in entropy in a brain region, this may indicate functional abnormalities in that region, providing clues for early diagnosis. Through continuous monitoring, spatiotemporal distribution maps of brain functional states can be created, visually demonstrating the dynamic changes in brain activity. This visualization not only helps researchers understand how the brain works but also provides clinicians with a reference for diagnosis and treatment. The advantage of this comprehensive analytical approach lies in its ability to fully capture the complexity of brain function. By combining spatial and temporal information, it can more accurately describe and predict the functional state of the brain, providing strong support for neuroscience research and the diagnosis of brain diseases.
[0080] This embodiment also provides a system for quantitatively characterizing brain function based on brain entropy, including:
[0081] The data acquisition and preprocessing module is used to acquire resting-state functional magnetic resonance data and perform preprocessing.
[0082] The brain region segmentation module is used to segment the preprocessed functional magnetic resonance imaging data according to a preset brain region segmentation scheme, and to determine the functional activity characteristics of each brain region based on the segmentation results.
[0083] The functional connectivity calculation module is used to calculate the functional connectivity strength between the divided brain regions using a preset functional connectivity metric method, and obtain the functional connectivity matrix.
[0084] The brain entropy calculation module is used to calculate the entropy value of each brain region based on the preprocessed functional magnetic resonance time series data, and obtain the brain entropy distribution map, according to the preset brain entropy calculation method.
[0085] A brain functional network construction module is used to construct a brain functional network by combining the functional connectivity matrix and the brain entropy distribution map, and to analyze the characteristics of the brain functional network.
[0086] The mathematical model design module is used to design mathematical models that correlate brain entropy values with the characteristics of the brain functional network, capturing the temporal variation and spatial distribution characteristics of the brain functional network.
[0087] The above are merely preferred embodiments 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 quantitatively characterizing brain function based on brain entropy, characterized in that, Includes the following steps: Acquire resting-state functional magnetic resonance imaging (fMRI) data and perform preprocessing; The preprocessed functional magnetic resonance imaging data are divided according to the preset brain region division scheme, and the functional activity characteristics of each brain region are determined based on the division results. Using a pre-defined functional connectivity metric, the functional connectivity strength between the segmented brain regions is calculated to obtain the functional connectivity matrix. Based on the preset brain entropy calculation method, the entropy value of each brain region is calculated according to the preprocessed functional magnetic resonance time series data to obtain the brain entropy distribution map; By combining the functional connectivity matrix and the brain entropy distribution map, a brain functional network is constructed, and the characteristics of the brain functional network are analyzed. A mathematical model was designed to correlate brain entropy values with the characteristics of the brain functional network, thereby capturing the temporal and spatial variation characteristics of the brain functional network.
2. The method according to claim 1, characterized in that, The preprocessing includes: The head movement correction algorithm is used to correct artifacts in the data caused by the subject’s head movements. The noise reduction process for the data after head movement correction was performed using independent component analysis. The ICA algorithm is used to separate noise components from the data and remove them, thereby improving the signal-to-noise ratio and obtaining high-quality functional magnetic resonance imaging (fMRI) data.
3. The method according to claim 1, characterized in that, The determination of the functional activity characteristics of each brain region includes: According to the preset brain region division scheme, preprocessed functional magnetic resonance imaging (fMRI) data is obtained, and the fMRI data is analyzed and processed to divide it into at least two brain regions. For each brain region after division, fMRI data within that brain region is extracted, and cluster analysis is performed on the data within that brain region using a clustering algorithm to obtain the functional activity characteristics of that brain region.
4. The method according to claim 3, characterized in that, The clustering algorithm includes K-means.
5. The method according to claim 1, characterized in that, The obtained functional connection matrix includes: Based on a preset functional connectivity measurement method, obtain the segmented brain region information and the functional connectivity data between brain regions; For each brain region, the functional connectivity strength between it and other brain regions is calculated to obtain the connectivity strength values between each pair of brain regions; The calculated connection strength values between brain regions are filled into the corresponding positions in the pre-established functional connectivity matrix to form a complete functional connectivity matrix.
6. The method according to claim 1, characterized in that, The obtained brain entropy distribution map includes: According to the preset brain entropy calculation method, functional magnetic resonance time series data are obtained, and the time series data are preprocessed to obtain preprocessed time series data. Wavelet decomposition algorithm is used to extract features from preprocessed time series data to obtain wavelet coefficients for each frequency band. Based on the wavelet coefficients of each frequency band, the entropy value of each brain region is calculated. If the entropy value is greater than the preset threshold, the brain region is judged to be an active brain region. The location coordinates of each active brain region are obtained, and the active brain regions are clustered using a clustering algorithm to obtain the distribution of brain regions with different functions; Based on the distribution of brain regions, the support vector machine algorithm is used to classify brain regions with different functions to obtain the location range of each functional brain region; For the location range of each functional brain region, a continuous brain entropy distribution map is generated by an interpolation algorithm to obtain a complete brain entropy distribution image.
7. The method according to claim 1, characterized in that, The characteristics of the brain functional network analyzed include: Based on the functional connectivity matrix, the functional connectivity strength between brain regions is obtained, and an initial brain functional network is constructed. Based on the brain entropy distribution map, the brain entropy value of each brain region is obtained, and the brain entropy value is used as the attribute of the node to update the initial brain function network, thus obtaining the updated brain function network. Calculate the node degree of each node in the updated brain functional network, where the node degree represents the number of edges directly connected to that node, and obtain the node degree distribution characteristics of the network. Calculate the clustering coefficient of each node in the updated brain functional network. The clustering coefficient represents the ratio of the actual number of edges between the node's neighboring nodes to the total number of possible edges, thereby obtaining the clustering characteristics of the network. The shortest path length between any two nodes in the updated brain functional network is calculated using Dijkstra's algorithm, and the distribution characteristics of the shortest path length of the network are obtained.
8. The method according to claim 1, characterized in that, Capturing the temporal variation and spatial distribution characteristics of the brain functional network includes: Based on the spatiotemporal characteristics of brain functional networks, a spatiotemporal network model is adopted, and brain entropy value is used as an attribute of network nodes to establish a mathematical correlation model between brain entropy value and brain functional networks. The spatial organization characteristics of brain functional networks are obtained by calculating the topological feature parameters of brain functional networks using graph theory algorithms. Time series analysis was used to calculate the statistical characteristics of brain entropy values and obtain the dynamic change pattern of brain entropy values over time. The spatial feature parameters of the brain functional network and the temporal feature parameters of the brain entropy value are input into the association model. The association model is trained by machine learning algorithm to establish a nonlinear mapping relationship between the brain entropy value and the characteristics of the brain functional network. Cross-validation was used to evaluate the generalization performance of the association model. The association model was optimized by adjusting the model hyperparameters to obtain the optimal brain entropy value-brain function network association model. Predict brain entropy values at different times and spatial locations using an optimal brain entropy-brain function network association model.
9. A system for quantitatively characterizing brain function based on brain entropy, characterized in that, include: The data acquisition and preprocessing module is used to acquire resting-state functional magnetic resonance data and perform preprocessing. The brain region segmentation module is used to segment the preprocessed functional magnetic resonance imaging data according to a preset brain region segmentation scheme, and to determine the functional activity characteristics of each brain region based on the segmentation results. The functional connectivity calculation module is used to calculate the functional connectivity strength between the divided brain regions using a preset functional connectivity metric method, and obtain the functional connectivity matrix. The brain entropy calculation module is used to calculate the entropy value of each brain region based on the preprocessed functional magnetic resonance time series data, and obtain the brain entropy distribution map, according to the preset brain entropy calculation method. A brain functional network construction module is used to construct a brain functional network by combining the functional connectivity matrix and the brain entropy distribution map, and to analyze the characteristics of the brain functional network. The mathematical model design module is used to design mathematical models that correlate brain entropy values with the characteristics of the brain functional network, capturing the temporal variation and spatial distribution characteristics of the brain functional network.