A method for identifying abnormal regions in mind map data

Abstract

This application provides a method for identifying abnormal regions in brain map data, comprising: acquiring brain map data of a target object, wherein the brain map data includes time-series data and connectivity matrices of M brain regions in the target object's brain; performing sliding window analysis on the time-series data of the M brain regions in the brain map data to determine K sets of time-varying connectivity matrices; based on the K sets of time-varying connectivity matrices, performing statistical feature extraction and dynamic feature extraction to determine the statistical characteristics and dynamic fluctuation characteristics of each brain region; determining the input data for each brain region based on the statistical characteristics and dynamic fluctuation characteristics of each brain region; and inputting the input data of each brain region into an abnormal region identification model to obtain the abnormal region identification result output by the abnormal region identification model. This solution can detect the existence of abnormal brain regions by analyzing the connectivity function of each brain region in the target object's brain map data.

Description

Technical Field

[0001] This application relates to the field of brain connectivity data processing technology, and more specifically, to a method for identifying abnormal regions in brain map data. Background Technology

[0002] In the field of neuroscience, brain region abnormality identification technology, as a cutting-edge research direction in neuroimaging, is evolving with a trend towards multimodal fusion. Currently, abnormality detection technology based on structural neuroimaging (such as sMRI and CT) has formed a mature technical path: voxel-level feature extraction is performed on high-resolution brain structural images using deep learning algorithms (such as 3D-CNN and U-Net architecture), and combined with brain atlas registration technology to achieve quantitative analysis of morphological parameters such as gray matter volume and cortical thickness, which can effectively identify structural lesions such as brain tumors and local atrophy.

[0003] However, these methods have a fundamental limitation in their sensitivity to functional network abnormalities. Studies have shown that over 60% of neuropsychiatric disorders (such as depression and bipolar disorder) do not involve significant structural abnormalities, but rather exhibit alterations in the topological characteristics of the functional connectivity set, meaning abnormal connectivity functions or patterns in certain brain regions. Existing aberration identification schemes focus primarily on structural anomalies, making them less applicable to the analysis of non-structural brain regions. Currently, a pressing issue in this field is the development of a technique capable of analyzing brain region abnormalities from a functional connectivity perspective. Summary of the Invention

[0004] The purpose of this application is to provide a method for identifying abnormal regions in brain map data, so as to detect the existence of abnormal brain regions by analyzing the connectivity of each brain region in the brain map data of the target object.

[0005] To achieve the above objectives, the embodiments of this application are implemented in the following manner:

[0006] This application provides a method for identifying abnormal regions in brain map data, comprising: acquiring brain map data of a target object, wherein the brain map data includes time-series data and connectivity strength matrices of M brain regions in the target object's brain; performing sliding window analysis on the time-series data of the M brain regions in the brain map data to determine K sets of time-varying connectivity matrices; performing statistical feature extraction and dynamic feature extraction based on the K sets of time-varying connectivity matrices to determine the statistical features and dynamic fluctuation features of each brain region; determining the input data for each brain region based on the statistical features and dynamic fluctuation features of each brain region; and inputting the input data of each brain region into an abnormal region identification model to obtain the abnormal region identification result output by the abnormal region identification model.

[0007] Furthermore, the time-series data of M brain regions in the brain map data contains data at T time points. A sliding window analysis is performed on the time-series data of the M brain regions in the brain map data to determine K sets of time-varying connectivity matrices. This includes: setting the sliding window length to 30 seconds and the step size to 10 seconds, resulting in a total of K window data points; for the k-th window data: calculating the correlation coefficient between the i-th and j-th brain regions in the window data, determining a set of correlation coefficients corresponding to the i-th and j-th brain regions in the k-th window data; and based on the set of correlation coefficients between every two brain regions in the window data, forming a set of time-varying connectivity matrices corresponding to the k-th window data.

[0008] Furthermore, the correlation coefficient between the i-th and j-th brain regions in the window data is calculated, and a set of correlation coefficients corresponding to the i-th and j-th brain regions in the k-th window data is determined, including:

[0009] For the i-th brain region in the k-th window of data:

[0010] The correlation coefficient between the i-th brain region and the j-th brain region is calculated using the following formula:

[0011]

[0012] in, This represents the correlation coefficient between the i-th and j-th brain regions in the k-th window of data, and includes four attribute data, namely the correlation coefficient at node 0. 1-node lag correlation coefficient 2-node lag correlation coefficient Correlation coefficient with 3-node lag T k The number of time points for the data in the k-th window. Let be the signal value of the i-th brain region in the window data at time node t. This represents the average signal value of the i-th brain region in the window data at each time point. Let be the signal value of the j-th brain region in the window data at time node t. This represents the average signal value of the i-th brain region in the window data at each time point. Let be the signal value of the j-th brain region in the window data at time node t+1. Let be the signal value of the j-th brain region in the window data at time node t+2. This represents the signal value of the j-th brain region in the window data at time node t+3.

[0013] Furthermore, based on K sets of time-varying connectivity matrices, statistical and dynamic feature extraction is performed to determine the statistical and dynamic fluctuation characteristics of each brain region, including: For the i-th brain region: based on the (M-1) sets of correlation coefficients between brain region i and other brain regions in the k-th time-varying connectivity matrix, the mean connectivity strength within the time window of brain region i in the k-th time-varying connectivity matrix is ​​calculated; based on the mean connectivity strength within the time window of brain region i in the k-th time-varying connectivity matrix, the variance of connectivity strength corresponding to brain region i in the k-th time-varying connectivity matrix is ​​calculated; based on the mean connectivity strength within the time window of brain region i in each time-varying connectivity matrix, the mean connectivity strength across time windows of brain region i is calculated; based on the mean connectivity strength within the time window and the mean connectivity strength across time windows of brain region i, the dynamic fluctuation characteristics of brain region i are calculated.

[0014] Furthermore, based on the (M-1) group correlation coefficients between brain region i and other brain regions in the k-th time-varying connectivity matrix, the mean connectivity strength within the time window of brain region i in the k-th time-varying connectivity matrix is ​​calculated, including: For the k-th time-varying connectivity matrix, the mean connectivity strength within the time window of brain region i is calculated using the following formula:

[0015]

[0016] in, This represents the mean connectivity strength within the time window of brain region i in the k-th window of data. It contains four mean data points, representing the mean connectivity strength within the lag time window of node 0. Mean connection strength within the 1-node lag time window Mean connection strength within the 2-node lag time window Mean connection strength within the 3-node lag time window Based on the 0-node lag correlation coefficient 1-node lag correlation coefficient 2-node lag correlation coefficient Correlation coefficient with 3-node lag Calculated.

[0017] Furthermore, based on the mean connection strength of brain region i within the time window in the k-th time-varying connectivity matrix, the variance of the connection strength corresponding to brain region i in the k-th time-varying connectivity matrix is ​​calculated, including: For the k-th time-varying connectivity matrix, the variance of the connection strength of brain region i is calculated using the following formula:

[0018]

[0019] in, This represents the variance of the connection strength corresponding to brain region i in the k-th time-varying connectivity matrix, containing four variance data points: the variance of the lag connection strength at node 0. 1-node hysteresis connection strength variance 2-node hysteresis connection strength variance 3-node hysteresis connection strength variance Based on the 0-node lag correlation coefficient Mean connection strength within the lag time window of node 0 1-node lag correlation coefficient Mean connection strength within the lag time window of node 1 2-node lag correlation coefficient Mean connection strength within the lag time window of 2 nodes 3-node lag correlation coefficient Mean connection strength within the 3-node lag time window Calculated.

[0020] Furthermore, based on the mean connectivity strength within the time window of brain region i in each time-varying connectivity matrix, the mean connectivity strength across the time window of brain region i is calculated, including: calculating the mean connectivity strength across the time window of brain region i using the following formula:

[0021]

[0022] in, This represents the mean connectivity strength across a time window in brain region i, containing four mean data points: the mean connectivity strength across a time window at node 0 lag. Mean of connection strength across time window with 1-node lag Mean of 2-node lag cross-time window connection strength Mean of 3-node lag cross-time window connection strength Based on the mean connection strength within the 0-node lag time window Mean connection strength within the 1-node lag time window Mean connection strength within the 2-node lag time window Mean connection strength within the 3-node lag time window The calculation shows that K is the total number of data points in the window.

[0023] Furthermore, based on the mean connectivity strength within and across time windows of brain region i, the dynamic fluctuation characteristics of brain region i are calculated, including:

[0024] The local dynamic variability of brain region i is calculated using the following formula:

[0025]

[0026] Among them, V i The local dynamic variability of brain region i is represented by four sets of data, including the local dynamic variability V with lag at node 0. i(0), 1-node lag local dynamic fluctuation V i (1) Local dynamic fluctuations with 2-node lag V i (2) Local dynamic fluctuations with 3-node lag V i (3);

[0027] The overall dynamic variability of brain region i is calculated using the following formula:

[0028]

[0029] Among them, V' i The overall dynamic variability of brain region i is represented by four sets of data, namely the overall dynamic variability V' with lag at node 0. i (0), 1-node lag overall dynamic volatility V' i (1) The overall dynamic volatility V' of the two-node lag i (2) The overall dynamic fluctuation of the three-node lag V' i (3), μ i The mean overall connectivity strength of brain region i. This represents the correlation coefficient between brain region i and brain region j in the connection strength matrix of brain map data.

[0030] Furthermore, based on the statistical and dynamic fluctuation characteristics of each brain region, the input data for each brain region is determined, including: for brain region i: the mean connection strength of brain region i within each window of data over a time window. Connection strength variance and the local dynamic fluctuations of brain region i V i and overall dynamic volatility V' i This forms a 4×(2K+2) feature matrix; the feature matrix is ​​flattened to form a feature vector of length (8K+8), which is used as the input data for brain region i.

[0031] Furthermore, the abnormal region identification model is built based on a multilayer perceptron model.

[0032] Beneficial effects:

[0033] This approach utilizes brain mapping data (including time-series data and connectivity strength matrices of M brain regions) for sliding window analysis to identify K sets of time-varying connectivity matrices. Further statistical feature extraction (window-level features) and dynamic feature extraction (including local and global variability) are performed, followed by multi-scale feature fusion to form input data for each brain region. An anomaly region identification model (based on MLP) is then used to obtain the anomaly region identification results. This approach overcomes the limitations of traditional static functional connectivity analysis by constructing time-varying connectivity matrices through multi-time-lag cross-correlation calculations (0-3 order lags). This effectively reflects the time-varying characteristics of functional connectivity and captures the directional characteristics of information transmission between brain regions, complementing the principles of causal analysis with lower computational complexity. Furthermore, by calculating the mean of cross-window connectivity strength and local dynamic variability, it simultaneously reflects the temporal stability and instantaneous fluctuation characteristics of functional connectivity, which is highly beneficial for identifying mental illnesses such as schizophrenia accompanied by dynamic connectivity disorders. The extracted window-level features reflect the strength distribution characteristics of brain region connections within a single sliding window, thereby revealing the instantaneous state of the functional network. Dynamic features, by calculating cross-window variability indices (including local and global variability), quantify the temporal variability of brain region connections. This index is beneficial for identifying default mode network anomalies in depression. Finally, the features are fused and flattened to form the input data. An anomaly region identification model based on MLP (because the sample size available for training in this scheme is not very large, but the feature processing is relatively good, containing sufficient multi-level features, further spatial feature extraction through convolutional kernels is not required; MLP's parameter efficiency is higher with a relatively small sample size) is used to identify anomaly regions, maximizing identification accuracy. Therefore, this scheme can detect the presence of abnormal brain regions by analyzing the connectivity functions of various brain regions in the target subject's brain mapping data.

[0034] Creatively designed extraction of multi-stage lag correlation features effectively reduces high-frequency noise interference in signals and improves the signal-to-noise ratio. Furthermore, the extraction of multi-stage lag correlation features reflects the direction of information transmission between brain regions and the causal patterns between brain regions, enhancing the information-reflecting ability of the features. Based on this, window-level features (i.e., statistical features of window data) can reflect the intensity distribution characteristics of brain region connections within a single time window, reflecting the instantaneous state of the functional network. In the further extracted dynamic features, overall dynamic volatility and local dynamic volatility complement each other. Local dynamic volatility reflects the degree of variation in brain region connection strength over time, while overall dynamic volatility reflects the degree to which brain region connection patterns deviate from population norms. Therefore, it is more sensitive to diseases accompanied by decreased functional connectivity stability (such as Alzheimer's disease) and diseases with abnormal connection patterns (such as autism), which is beneficial for improving the effectiveness of abnormal region identification. The input data constructed accordingly (feature vectors of length (8K+8)) has low overall computational complexity, ensuring that the system maintains high operating efficiency.

[0035] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0036] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating a method for identifying abnormal regions in mind map data, as provided in an embodiment of this application.

[0038] Figure 2 This is a schematic diagram illustrating the conditions for introducing visualized abnormal regions into the mind mapping software used in our organization. Detailed Implementation

[0039] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0040] Since the abnormal region identification model used in this embodiment relies on MLP (Multilayer Perceptron) construction, and the model construction process is relatively simple, it is only necessary to define the model using TensorFlow and set the relevant parameters. The key point is the processing of feature data. The feature processing process of training data is similar. Therefore, this embodiment does not introduce the processing process of training data separately, but introduces the process of abnormal region identification on the mind map data of a target object.

[0041] Please see Figure 1 , Figure 1 This is a flowchart of a method for identifying abnormal regions in mind map data provided in an embodiment of this application. In this embodiment, the method for identifying abnormal regions in mind map data may include steps S10, S20, S30, S40, and S50.

[0042] In this embodiment, step S10 can be run first.

[0043] Step S10: Obtain the brain map data of the target object, wherein the brain map data includes time series data and connectivity strength matrix of M brain regions in the target object's brain.

[0044] In this embodiment, brain map data of the target object can be obtained. The brain map data includes time-series data and connectivity matrices of M brain regions in the target object. Currently, many software programs can generate a brain network map (containing M brain regions and their inter-brain connections) using the target object's fMRI data (combined with sMRI data or T1 structural imaging, etc.), thereby obtaining the time-series data and connectivity matrices of the target object's M brain regions. These details will not be elaborated here.

[0045] After obtaining the mind map data of the target object, step S20 can be further run.

[0046] Step S20: Perform sliding window analysis on the time series data of M brain regions in the brain map data to determine K sets of time-varying connectivity matrices.

[0047] In this embodiment, in order to meet the minimum time unit requirement of functional connection dynamic analysis, the length of the sliding window is set to 30 seconds, the step size is 10 seconds, and a total of K window data are determined.

[0048] For example, if the time-series data of a brain region is 10 minutes long (signals are collected every 2 seconds), then it contains 300 nodes, and each sliding window contains 15 nodes (corresponding to a step size of 5 nodes), resulting in 58 window data.

[0049] For each window of data, taking the k-th window of data as an example, the following processing is required:

[0050] Calculate the correlation coefficient between the i-th and j-th brain regions in the window data, and determine a set of correlation coefficients between the i-th and j-th brain regions in the k-th window data.

[0051] For example, for the i-th brain region in the k-th window of data, the following formula is used to calculate a set of correlation coefficients between the i-th brain region and the j-th brain region:

[0052]

[0053]

[0054] in, This represents the correlation coefficient between the i-th and j-th brain regions in the k-th window of data, and includes four attribute data, namely the correlation coefficient at node 0. 1-node lag correlation coefficient 2-node lag correlation coefficient Correlation coefficient with 3-node lag T k The number of time points for the data in the k-th window. Let be the signal value of the i-th brain region in the window data at time node t. This represents the average signal value of the i-th brain region in the window data at each time point. Let be the signal value of the j-th brain region in the window data at time node t. This represents the average signal value of the i-th brain region in the window data at each time point. Let be the signal value of the j-th brain region in the window data at time node t+1. Let be the signal value of the j-th brain region in the window data at time node t+2. This represents the signal value of the j-th brain region in the window data at time node t+3.

[0055] Unlike traditional correlation coefficient calculation methods, this approach introduces multi-order lags for subsequent processing to extract features that reflect causal patterns (i.e., connection directions), thereby improving the detection performance of abnormal regions.

[0056] Based on this, a set of correlation coefficients between every two brain regions in each window of data is calculated (the correlation coefficient between brain region i and itself can be recorded as 1), thus forming a set of time-varying connection matrices corresponding to the k-th window of data (actually 4 time-varying connection matrices, each corresponding to a time-varying connection matrix formed by different numbers of lag nodes).

[0057] After obtaining a set of time-varying connection matrices corresponding to each window of data (a total of K sets of time-varying connection matrices are obtained), step S30 can be further run.

[0058] Step S30: Based on the K sets of time-varying connectivity matrices, perform statistical feature extraction and dynamic feature extraction to determine the statistical features and dynamic fluctuation features of each brain region.

[0059] In this embodiment, feature extraction is required for each brain region in each time-varying connectivity matrix (i.e., corresponding to each window of data). Here, we take the i-th brain region in the k-th time-varying connectivity matrix as an example:

[0060] Specifically, the mean connection strength of brain region i within the time window can be calculated based on the (M-1) group correlation coefficients between brain region i and other brain regions in the k-th time-varying connectivity matrix.

[0061] For example, for the k-th time-varying connectivity matrix, the mean connectivity strength within the time window of brain region i is calculated using the following formula:

[0062]

[0063] in, This represents the mean connectivity strength within the time window of brain region i in the k-th window of data. It contains four mean data points, representing the mean connectivity strength within the lag time window of node 0. Mean connection strength within the 1-node lag time window Mean connection strength within the 2-node lag time window Mean connection strength within the 3-node lag time window Based on the 0-node lag correlation coefficient 1-node lag correlation coefficient 2-node lag correlation coefficient Correlation coefficient with 3-node lag The result is obtained by substituting the formula (6) into the calculation, which is not shown here.

[0064] Then, based on the mean connection strength of brain region i within the time window in the k-th time-varying connection matrix, the variance of the connection strength corresponding to brain region i in the k-th time-varying connection matrix can be calculated.

[0065] For the k-th time-varying connectivity matrix, the variance of connectivity strength in brain region i is calculated using the following formula:

[0066]

[0067] in, This represents the variance of the connection strength corresponding to brain region i in the k-th time-varying connectivity matrix, containing four variance data points: the variance of the lag connection strength at node 0. 1-node hysteresis connection strength variance 2-node hysteresis connection strength variance 3-node hysteresis connection strength variance Based on the 0-node lag correlation coefficient Mean connection strength within the lag time window of node 0 1-node lag correlation coefficient Mean connection strength within the lag time window of node 1 2-node lag correlation coefficient Mean connection strength within the lag time window of 2 nodes 3-node lag correlation coefficient Mean connection strength within the 3-node lag time window The result is obtained by substituting the formula (7) into the calculation, which is not shown here.

[0068] Furthermore, the mean connectivity strength of brain region i across time windows can be calculated based on the mean connectivity strength within the time window of brain region i in each time-varying connectivity matrix.

[0069] For brain region i, the mean connectivity strength across the time window is calculated using the following formula:

[0070]

[0071] in, This represents the mean connectivity strength across a time window in brain region i, containing four mean data points: the mean connectivity strength across a time window at node 0 lag. Mean of connection strength across time window with 1-node lag Mean of 2-node lag cross-time window connection strength Mean of 3-node lag cross-time window connection strength Based on the mean connection strength within the 0-node lag time window Mean connection strength within the 1-node lag time window Mean connection strength within the 2-node lag time window Mean connection strength within the 3-node lag time window The result is obtained by substituting the formula (8) into the calculation, which is not shown here. K is the total number of window data.

[0072] Furthermore, the dynamic fluctuation characteristics of brain region i can be calculated based on the mean connectivity strength within the time window and the mean connectivity strength across the time window. In this embodiment, the dynamic fluctuation characteristics include two types of fluctuation calculations: local dynamic fluctuation and global dynamic fluctuation.

[0073] First, the local dynamic variability of brain region i is calculated using the following formula:

[0074]

[0075] Among them, V i The local dynamic variability of brain region i is represented by four sets of data, including the local dynamic variability V with lag at node 0. i (0), 1-node lag local dynamic fluctuation V i (1) Local dynamic fluctuations with 2-node lag V i (2) Local dynamic fluctuations with 3-node lag V i (3).

[0076] Secondly, the overall dynamic variability of brain region i is calculated using the following formula:

[0077]

[0078] Among them, V' i The overall dynamic variability of brain region i is represented by four sets of data, namely the overall dynamic variability V' with lag at node 0. i (0), 1-node lag overall dynamic volatility V' i (1) The overall dynamic volatility V' of the two-node lag i (2) The overall dynamic fluctuation of the three-node lag V' i (3), μ i The mean overall connectivity strength of brain region i. This represents the correlation coefficient between brain region i and brain region j in the connection strength matrix of brain map data.

[0079] After calculating the statistical characteristics of each window data in each brain region and the dynamic fluctuation characteristics of each brain region, step S40 can be run.

[0080] Step S40: Based on the statistical characteristics and dynamic fluctuation characteristics of each brain region, determine the input data for each brain region.

[0081] In this embodiment, processing is also performed on a per-brain-region basis, specifically for brain region i:

[0082] The mean connection strength of brain region i within each window of data can be used to calculate the connection strength within each window. Connection strength variance and the local dynamic fluctuations of brain region i V i and overall dynamic volatility V' i This forms a 4×(2K+2) feature matrix:

[0083]

[0084] in, and If it has K elements (taking 58 window data as an example, then...), and Both have 58 elements), while V i and V' i Since each element is a single element, the number of columns in the matrix is ​​(2K+2).

[0085] Next, the feature matrix can be flattened to form a feature vector of length (8K+8), which will be used as the input data for brain region i. For example, it can be concatenated into... The feature vector. Of course, the actual processing can also be other forms of feature fusion. This embodiment considers an abnormal region recognition model built on MLP. Therefore, the features of each brain region at each level are fused and processed into a feature vector of length (8K+8), which is beneficial for model processing.

[0086] After obtaining the input data corresponding to each brain region, step S50 can be run.

[0087] Step S50: Input the input data of each brain region into the abnormal region identification model to obtain the abnormal region identification result output by the abnormal region identification model.

[0088] In this embodiment, the input data of each brain region can be input into the abnormal region identification model, and the abnormal region identification result can be output by the abnormal region identification model. Considering that if the input data of all brain regions are combined to construct a whole-brain model, it would be difficult to achieve a good result with the current amount of available training data, this embodiment temporarily adopts a multi-level parallel single brain region model scheme.

[0089] That is, M sub-models need to be built (each sub-model is based on MLP and can be done using TensorFlow; taking 3 hidden layers as an example, the number of neurons in each layer decreases, the activation function for the hidden layers is ReLU, the activation function for the output layer is Sigmoid, Dropout is set to 0.5, and the L2 regularization weight decay is set to e). -4 The optimizer is Adam, and the learning rate is set to 0.001. Each sub-model corresponds to a brain region and is numbered. Brain region-level training data is used (also employing the same feature processing method, but with labels). Currently, data labels in this field are usually object-level; therefore, the labeling of this part of the brain region-level labeled data (as training data) needs to be done manually. Alternatively, a pseudo-label strategy can be used. Abnormal regions are identified and labeled using object-level labels, while the labels for other regions are designed to be "trustworthy." For normal subjects, all regions are labeled as "normal," and different weights are assigned to the three types of labels. For example, abnormal labels are weighted 1.5, normal labels 1.0, and trustworthy labels 0.8. This process completes the training data processing, forming a training set to train the corresponding sub-model.

[0090] Ultimately, the abnormal region identification model can be used to determine the abnormal region identification results at the brain region level, thereby achieving abnormal region identification.

[0091] The results of abnormal region identification can be fed back into the mind map data and visualized through coloring (different colors), for example, using our organization's mind mapping software (although we haven't integrated with this software's functionality yet, but it may be introduced in a future version). Figure 2 As shown.

[0092] In summary, this application provides a method for identifying abnormal regions in brain map data. It utilizes brain map data (including time-series data and connectivity matrices of M brain regions) for sliding window analysis to determine K sets of time-varying connectivity matrices. Further statistical feature extraction (window-level features) and dynamic feature extraction (including local and global variability) are performed, followed by multi-scale feature fusion to form input data for each brain region. An abnormal region identification model (based on MLP) is then used to obtain the abnormal region identification result. This approach overcomes the limitations of traditional static functional connectivity analysis by constructing time-varying connectivity matrices through multi-time-lag cross-correlation calculations (0-3 order lags). This effectively reflects the time-varying characteristics of functional connectivity and captures the directional characteristics of information transmission between brain regions, complementing the principles of causal analysis with lower computational complexity. Furthermore, by calculating the mean cross-window connectivity strength and local dynamic variability, it simultaneously reflects the temporal stability and instantaneous fluctuation characteristics of functional connectivity, which is highly beneficial for identifying mental illnesses such as schizophrenia accompanied by dynamic connectivity disorders. The extracted window-level features reflect the strength distribution characteristics of brain region connections within a single sliding window, thus revealing the instantaneous state of the functional network. Dynamic features, by calculating cross-window variability indices (including local and global variability), quantify the temporal variability of brain region connections; this index is beneficial for identifying default mode network anomalies in depression. Finally, the features are fused and flattened to form the input data. An anomaly region identification model based on MLP (because the sample size available for training in this scheme is not currently very large, but the feature processing is relatively good, containing sufficient multi-level features, further spatial feature extraction through convolutional kernels is not required; MLP's parameter efficiency is higher with a relatively small sample size) is used to identify anomaly regions, maximizing identification accuracy. Therefore, this scheme can detect the presence of abnormal brain regions by analyzing the connectivity functions of various brain regions in the target subject's brain mapping data.

[0093] Creatively designed extraction of multi-stage lag correlation features effectively reduces high-frequency noise interference in signals and improves the signal-to-noise ratio. Furthermore, the extraction of multi-stage lag correlation features reflects the direction of information transmission between brain regions and the causal patterns between brain regions, enhancing the information-reflecting ability of the features. Based on this, window-level features (i.e., statistical features of window data) can reflect the intensity distribution characteristics of brain region connections within a single time window, reflecting the instantaneous state of the functional network. In the further extracted dynamic features, overall dynamic volatility and local dynamic volatility complement each other. Local dynamic volatility reflects the degree of variation in brain region connection strength over time, while overall dynamic volatility reflects the degree to which brain region connection patterns deviate from population norms. Therefore, it is more sensitive to diseases accompanied by decreased functional connectivity stability (such as Alzheimer's disease) and diseases with abnormal connection patterns (such as autism), which is beneficial for improving the effectiveness of abnormal region identification. The input data constructed accordingly (feature vectors of length (8K+8)) has low overall computational complexity, ensuring that the system maintains high operating efficiency.

[0094] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for identifying abnormal regions in mind map data, characterized in that, include: Obtain the brain map data of the target object, wherein the brain map data includes time series data of M brain regions in the target object's brain and a connectivity strength matrix, the connectivity strength matrix being used to determine dynamic fluctuation characteristics; Sliding window analysis was performed on the time series data of M brain regions in the brain map data to determine K sets of time-varying connectivity matrices; Based on K sets of time-varying connectivity matrices, statistical and dynamic features are extracted to determine the statistical and dynamic fluctuation features of each brain region. Based on the statistical and dynamic fluctuation characteristics of each brain region, the input data for each brain region is determined; Input data for each brain region is fed into the abnormal region identification model to obtain the abnormal region identification results output by the abnormal region identification model. The time-series data of M brain regions in the brain map data contains data at T time points. Sliding window analysis is performed on the time-series data of the M brain regions in the brain map data to determine K sets of time-varying connectivity matrices, including: The length of the sliding window is set to 30 seconds, and the step size is 10 seconds, resulting in a total of K window data points. Regarding the first Data for each window: The first in the calculation window data The brain regions and the first The correlation coefficient of each brain region was used to determine the first... The data in the nth window The brain regions and the first A set of correlation coefficients corresponding to each brain region; Based on a set of correlation coefficients between every two brain regions in the window data, the first... A set of time-varying connection matrices corresponding to each window of data; The first in the calculation window data The brain regions and the first The correlation coefficient of each brain region was used to determine the first... The data in the nth window The brain regions and the first A set of correlation coefficients corresponding to each brain region includes: Regarding the first The data in the nth window Brain regions: The following formula is used to calculate the first... The brain region and the first A set of correlation coefficients for individual brain regions: ， in, Indicates the first The data in the nth window The brain regions and the first The correlation coefficients of individual brain regions contain four attribute data, including the correlation coefficient of the 0-node lag. 1-node lag correlation coefficient 2-node lag correlation coefficient Correlation coefficient with 3-node lag , For the first The number of time points for each window's data. For the first data in the window individual brain regions at time points The signal value at that time, For the first data in the window The average signal value of each brain region at each time point. For the first data in the window individual brain regions at time points The signal value at that time, For the first data in the window The average signal value of each brain region at each time point For the first data in the window individual brain regions at time points The signal value at that time, For the first data in the window individual brain regions at time points The signal value at that time, For the first data in the window individual brain regions at time points The signal value at that time.

2. The method for identifying abnormal regions in mind map data according to claim 1, characterized in that, Based on K sets of time-varying connectivity matrices, statistical and dynamic feature extraction is performed to determine the statistical and dynamic fluctuation characteristics of each brain region, including: Regarding the first Brain regions: Based on the Group time-varying connectivity matrix midbrain regions The correlation coefficient with other brain regions (M-1 group) was calculated. Group time-varying connectivity matrix midbrain regions The mean connection strength within the time window; Based on the Group time-varying connectivity matrix midbrain regions The mean connection strength within the time window is used to calculate the first... Group time-varying connectivity matrix midbrain regions The corresponding connection strength variance; Based on brain regions in each time-varying connectivity matrix The mean connection strength within the time window is used to calculate the brain region. The mean connection strength across the time window; Based on brain regions The mean connectivity strength within and across time windows is used to calculate the brain region. Its dynamic fluctuation characteristics.

3. The method for identifying abnormal regions in mind map data according to claim 2, characterized in that, Based on the Group time-varying connectivity matrix midbrain regions The correlation coefficient with other brain regions (M-1 group) was calculated. Group time-varying connectivity matrix midbrain regions The mean connection strength within the time window, including: Regarding the first Group time-varying connection matrix: Brain regions are calculated using the following formula. Mean connection strength within the time window: ， in, For the first Brain regions in the data of each window The mean connection strength within the time window, containing four mean data points, representing the mean connection strength within the time window after the lag at node 0. Mean connection strength within the lag time window of node 1 Mean connection strength within the lag time window of 2 nodes Mean connection strength within the lag time window of 3 nodes Based on the 0-node lag correlation coefficient 1-node lag correlation coefficient 2-node lag correlation coefficient Correlation coefficient with 3-node lag Calculated.

4. The method for identifying abnormal regions in mind map data according to claim 3, characterized in that, Based on the Group time-varying connectivity matrix midbrain regions The mean connection strength within the time window is used to calculate the first... Group time-varying connectivity matrix midbrain regions The corresponding connection strength variance includes: Regarding the first Group time-varying connection matrix: Brain regions are calculated using the following formula. Variance of connection strength: ， in, For the first Group time-varying connectivity matrix midbrain regions The corresponding connection strength variance contains four variance data points, namely the variance of connection strength at 0-node lag.

1. Node lag connection strength variance 2-node lag connection strength variance 3-node hysteresis connection strength variance Based on the 0-node lag correlation coefficient Mean connection strength within the lag time window of node 0 1-node lag correlation coefficient Mean connection strength within the lag time window of node 1 2-node lag correlation coefficient Mean connection strength within the lag time window of 2 nodes 3-node lag correlation coefficient Mean connection strength within the 3-node lag time window Calculated.

5. The method for identifying abnormal regions in mind map data according to claim 3, characterized in that, Based on brain regions in each time-varying connectivity matrix The mean connection strength within the time window is used to calculate the brain region. The mean connection strength across time windows includes: Brain regions are calculated using the following formula. Mean connection strength across time windows: ， in, brain region The mean connection strength across time windows contains four mean data points, representing the mean connection strength across time windows with 0-node lag. Mean of connection strength across time window with 1 node lag Mean of 2-node lag cross-time window connection strength 3-node lag cross-time window mean connection strength Based on the mean connection strength within the 0-node lag time window, respectively Mean connection strength within the lag time window of node 1 Mean connection strength within the lag time window of 2 nodes Mean connection strength within the lag time window of 3 nodes Calculations show that This represents the total number of data points in the window.

6. The method for identifying abnormal regions in mind map data according to claim 5, characterized in that, Based on brain regions The mean connectivity strength within and across time windows is used to calculate the brain region. The dynamic fluctuation characteristics include: Brain regions are calculated using the following formula. Local dynamic fluctuations: ， in, brain region The local dynamic volatility includes four sets of data, namely the local dynamic volatility with 0-node lag.

1. Node-lagging local dynamic fluctuations 2. Local dynamic fluctuations with lag at nodes 3-node lag local dynamic fluctuations ; Brain regions are calculated using the following formula. Overall dynamic volatility: ， ， in, brain region The overall dynamic volatility also includes four sets of data, namely the overall dynamic volatility with 0-node lag.

1. Node lag and overall dynamic volatility 2. Node lag overall dynamic fluctuation 3-node lag overall dynamic fluctuation , brain region The overall average connection strength, The connection strength matrix of brain map data represents the brain regions. With brain regions The correlation coefficient.

7. The method for identifying abnormal regions in mind map data according to claim 3, characterized in that, Based on the statistical and dynamic fluctuation characteristics of each brain region, the input data for each brain region is determined, including: Targeting brain regions : brain regions Mean connection strength within the time window in each window of data Connection strength variance and brain regions Local dynamic fluctuations and overall dynamic volatility , forming a The feature matrix; Flatten the feature matrix to form a matrix of length [missing information]. The feature vectors of brain regions Input data.

8. The method for identifying abnormal regions in mind map data according to claim 1, characterized in that, The abnormal region identification model is built based on the multilayer perceptron model.

Images