An auxiliary diagnosis method for schizophrenia based on spatiotemporal dynamics

Abstract

This invention discloses a spatiotemporal dynamics-based auxiliary diagnostic method for schizophrenia, belonging to the field of biomedical image pattern recognition technology. Functional brain network images are obtained using f-MRI, and a spatiotemporal dynamic model is constructed using Markov chains to extract the convergence pattern of individual functional brain networks and the maximum information flow transmission pattern between brain networks. Three corresponding indicators are proposed to quantify the spatiotemporal dynamic pattern of functional brain networks: the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix. This invention utilizes statistical testing methods and logistic regression algorithms to construct an effective auxiliary diagnostic model for the three proposed spatiotemporal dynamic indicators of functional brain networks, enabling accurate quantitative calculation of the probability of an individual developing schizophrenia. This provides strong technical support for the early diagnosis and intervention of schizophrenia, filling a gap in existing diagnostic technologies in this field, and has significant clinical application value and scientific significance.

Description

Technical Field

[0001] This invention belongs to the field of biomedical image pattern recognition technology, and in particular relates to an auxiliary diagnostic method for schizophrenia based on spatiotemporal dynamics. Background Technology

[0002] The advancement of brain imaging technology has greatly provided technical support and convenience for disease research, and has also offered many excellent methods for the diagnosis of mental illnesses. Among these, f-MRI (Functional Magnetic Resonance Imaging) technology reflects the activity of neurons over time by measuring changes in cerebral blood flow, and can be used to detect brain activity. f-MRI technology can help doctors determine the location of important functional areas of the brain and detect functional abnormalities in specific areas, thus providing a reference for disease diagnosis and treatment.

[0003] Schizophrenia (SZ) is a severe mental disorder that affects a person's thinking, emotions, behavior, and perception. However, patients with SZ do not have obvious organic damage to their brains. Therefore, s-MRI (structural magnetic resonance imaging), which focuses on reflecting tissue structure, is difficult to provide auxiliary diagnostic and treatment options. Although traditional methods reflect real brain physiological activity through f-MRI, they often only model a single spatial or temporal domain, and the extracted information is incomplete. Summary of the Invention

[0004] The purpose of this invention is to provide an auxiliary diagnostic method for schizophrenia based on spatiotemporal dynamics. From a spatiotemporal dynamics perspective, three indicators are proposed to describe the spatiotemporal information flow patterns of an individual's functional network: a network stationary probability vector, a mean matrix of optimal steps, and a network direct connection strength matrix. Probabilistic models of information flow transmission between brain regions or networks are obtained by using f-MRI image data through Markov chains. Differences in these indicators between the schizophrenia patient group and the health control (HC) group are compared using statistical testing methods. Then, an effective diagnostic model is constructed using logistic regression to achieve accurate quantitative calculation of the probability of an individual developing schizophrenia, providing strong technical support for the early diagnosis and intervention of schizophrenia. This addresses the technical problem in existing technologies that use f-MRI technology to reflect real brain physiological activity, but often only model a single spatial or temporal domain, resulting in incomplete information extraction.

[0005] To solve the above-mentioned technical problems, the specific technical solution of the present invention is as follows:

[0006] A spatiotemporal dynamics-based auxiliary diagnostic method for schizophrenia includes the following steps:

[0007] Step S11: Collect f-MRI brain imaging data of individual subjects, preprocess the collected f-MRI brain imaging data to obtain preprocessed f-MRI brain imaging data.

[0008] Step S12: Using the cortical brain region division atlas of N ROIs based on the Yeo 7 network, the preprocessed f-MRI brain image data is mapped to the N ROIs to obtain the temporal signals of the brain regions.

[0009] Step S13: Divide the temporal signals of the brain region into sliding windows, and calculate the functional connectivity matrix in each sliding window to obtain the dynamic functional connectivity matrix.

[0010] Step S14: Calculate the brain region stationary probability vector and map it to the network to calculate the network stationary probability vector.

[0011] Step S15: Calculate the mean matrix of the optimal number of steps in the network.

[0012] Step S16: Calculate the direct connection strength matrix.

[0013] Step S17: Construct an auxiliary diagnostic model through statistical analysis of differences between groups and logistic regression.

[0014] Step S18: Acquire f-MRI images of the subject to be diagnosed. After preprocessing, calculate three indicators: the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix. Input the elements with significant differences in each indicator into the corresponding auxiliary diagnostic model to calculate the predicted probability of schizophrenia.

[0015] Further, step S14 includes the following steps:

[0016] Step S141: Calculate the corrected brain region functional connectivity matrix. Correction is achieved by setting the diagonal and negative correlation elements in the functional connectivity matrix to 0 and performing false discovery rate correction on the positive correlation values, so that the brain region functional connectivity strength value is not less than 0.

[0017] Step S142: Calculate the single-step transition probability matrix and the 10-step transition probability matrix of the brain region. The single-step brain region transition probability is calculated by standardizing the brain region functional connectivity matrix using the L1-norm. Before calculating the 10-step transition probability matrix, a time window is set, and the matrix is ​​obtained by continuously multiplying the 10 matrices within the time window.

[0018] Step S143: Calculate the mean matrix of 10-step transition probabilities, and obtain the stationary probability vector of the brain region through time arithmetic mean and column mean. Then calculate the stationary probability vector of the network through the brain region network mapping relationship.

[0019] Further, step S15 includes the following steps:

[0020] Step S151: Calculate the network functional connectivity matrix, which is obtained by mapping the brain region functional connectivity matrix.

[0021] Step S152: Calculate the network single-step transition probability matrix, which is calculated by normalizing the network functional connectivity matrix L1-norm.

[0022] Step S153: Calculate the optimal step matrix of the network. Before calculation, set a time window and calculate the matrix elements by comparing the maximum value of the elements of the multi-step transition probability matrix within the time window.

[0023] Step S154: Calculate the mean matrix of the network's optimal steps, which is obtained by the time arithmetic mean of the network's optimal steps matrix.

[0024] Further, step S16 includes the following steps:

[0025] Step S161: Calculate the network direct connection label matrix. If there is a direct connection between networks, mark it as 1 at the corresponding position; otherwise, mark it as 0.

[0026] Step S162: Calculate the network direct connection strength matrix, which is the time arithmetic mean of the network direct connection label values.

[0027] Further, step S17 includes the following steps:

[0028] Step S171: Perform t-tests on the three index elements of the modeling sample group—the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix—to analyze the differences between groups and identify the significant difference elements for each index.

[0029] Step S172: Construct the logistic regression hypothesis function and estimate the regression coefficients using the maximum likelihood estimation method.

[0030] Step S173: Construct the network stationary probability vector to assist the diagnostic model in predicting probabilities, the network optimal step mean matrix to assist the diagnostic model in predicting probabilities, and the network stationary probability vector to assist the diagnostic model in predicting probabilities.

[0031] Compared to existing technologies, this invention offers the following advantages: Utilizing the principles of information dynamics, this invention proposes using the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix as indicators to measure the functional information flow characteristics of brain networks. By employing statistical testing methods and logistic regression algorithms, this invention constructs an effective auxiliary diagnostic model for these three network spatiotemporal dynamic indicators—the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix—to achieve precise quantitative calculation of the probability of an individual developing schizophrenia. This provides strong technical support for the early diagnosis and intervention of schizophrenia, filling a gap in existing diagnostic technologies in this field and possessing significant clinical application value and scientific significance. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a schematic diagram of network stationary vector calculation according to the present invention.

[0034] Figure 2 This is a schematic diagram illustrating the calculation of the mean matrix of the optimal number of steps in the network and the network direct connection strength matrix according to the present invention.

[0035] Figure 3 This is a schematic diagram illustrating the construction of the auxiliary diagnostic model of the present invention.

[0036] Figure 4 A schematic diagram illustrating the construction and use of an auxiliary diagnostic model for a single indicator. Detailed Implementation

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0038] This invention utilizes Markov chains to represent the transitions between different states at a given time using probabilities. In this invention, the state is represented as an interest region (ROI) or a brain network.

[0039] This invention proposes an auxiliary diagnostic method for schizophrenia based on spatiotemporal dynamics, comprising the following steps:

[0040] Step S11: Collect f-MRI brain imaging data of individual subjects, preprocess the collected f-MRI brain imaging data to obtain preprocessed f-MRI brain imaging data.

[0041] Furthermore, preprocessing was performed as follows: data from the first four time points of the acquired f-MRI brain imaging data were removed, with the time length after removing the first four time points being T, to eliminate interference from unstable factors that might exist in the initial stage; next, slice time correction was performed to ensure precise synchronization of the acquisition time of each slice; then, head movement correction was performed to accurately compensate for the subject's minor head movements and reduce head movement artifacts; further, image edges were registered to structural images; covariate regression was used to remove the potential influence of physiological covariates such as white matter signals, ventricular signals, and head movements on the brain imaging data; finally, bandpass filtering was used to screen out effective signals in the frequency range of 0.01Hz–0.08Hz.

[0042] Step S12: Using the cortical brain region division atlas of N ROIs based on the Yeo 7 network, the preprocessed f-MRI brain image data is mapped to the N ROIs to obtain the temporal signals of the brain regions.

[0043] To distinguish them from each other, each brain region is represented by a number, and the i-th brain region is denoted as ROI. i The ROI time signal is represented by a matrix, denoted as Y, with dimensions N×T, where N is the number of ROIs in the partition map and T is the time length of the time signal.

[0044] The cortical brain region mapping described herein is a more refined division based on Yeo's 7-network mapping. Each ROI can be mapped to one of Yeo's 7 networks. The 7-network designation indicates that there are a total of 7 networks, with the k-th network denoted as net. k , where k∈{1,2,…,7}.

[0045] Step S13: Divide the temporal signals of the brain region into sliding windows, and calculate the functional connectivity matrix in each sliding window to obtain the dynamic functional connectivity matrix.

[0046] Using L as the window length and 1 as the step size, the time signal of the ROI is divided into overlapping matrices over a time length T. This is called a sliding window, and the t-th sliding window is denoted as W. t The corresponding time is t, where t∈{1,2,…,T-L+1}, and the dimension of each sliding window is N×L. The t-th sliding window W... t The i-th row vector (i.e., the brain region ROI) iIn the sliding window W t The time signal in the middle) is denoted as The t-th sliding window W t The j-th row vector (i.e., the brain region ROI) j In the sliding window W t The time signal in the middle) is denoted as

[0047] Calculate the functional connectivity (FC) matrix of the brain region at time t, denoted as FC. ROI (t). Where t∈{1,2,…,T-(L-1)}, matrix elements This indicates the brain region ROI at time t. i and brain region ROI j The strength of the functional connections between them, where i,j∈{1,2,…,N}, This refers to the brain region ROI. i with brain region ROI j Pearson correlation coefficient at time t Right now Pearson correlation coefficient Depend on and The calculation yielded:

[0048]

[0049] in express The sample mean; express The sample mean; W represents the t-th sliding window. t The kth term of the i-th row vector; The t-th sliding window W t The kth term of the j-th row vector.

[0050] Brain region functional connectivity matrix FC at time t ROI (t) has an N×N dimension and is FC ROI (t) is represented as:

[0051]

[0052] Calculating the functional connectivity matrix at all time points yields a total of T-(L-1) functional connectivity matrices, which are called dynamic functional connectivity (d-FC).

[0053] Step S14: Calculate the brain region stationary probability vector and map it to the network to calculate the network stationary probability vector.

[0054] Step S141: Calculate the corrected brain region functional connectivity matrix.

[0055] The functional connectivity matrix of brain regions at time t is FC ROI In the matrix (t), the diagonal and negatively correlated elements are set to 0. The positively correlated values ​​are further corrected for false discovery rate (fdr) to ensure that the brain region functional connectivity strength is not less than 0. This yields the corrected brain region functional connectivity matrix at time t, denoted as FC. ROI′ (t), t∈{1,2,…,T-(L-1)}, matrix elements Indicates the corrected brain region ROI i and brain region ROI j The functional connection strength, where i∈[1,N], j∈[1,N], and has Matrix FC ROI′ (t) is in the form of:

[0056]

[0057] Step S142: Calculate the single-step transition probability matrix and the 10-step transition probability matrix of the brain region.

[0058] Calculate the single-step transition probability matrix A of the brain region at time t. (t) , where t∈{1,2,…,T-(L-1)}, matrix elements This indicates the brain region ROI at time t. i to brain region ROI j The single-step transition probability can be simply written as: i,j∈{1,2,…,N}. Its elements are calculated using the L1-norm normalization of the brain region functional connectivity matrix. The formula for calculating the L1-norm normalization of the brain region functional connectivity matrix is ​​as follows:

[0059]

[0060] Note that due to the corrected nonnegative matrix FC of brain region functional connectivity. ROI′ (t) diagonal elements so The brain region single-step transition probability matrix A at time t (t) With dimensions of N×N, probability normalization is achieved, and self-loops are not considered. Matrix A (t) Represented as:

[0061]

[0062] Let A (τ,n) Let be the brain region transition probability matrix at time τ, where τ∈{1,2,…T-(L-1)-(n-1)}, and the matrix elements are... This indicates that at time τ, the region is in the ROI. i Under the condition that the transition to ROI occurs at time τ+n j The probability of the time τ from the ROI, or simply the probability of the time τ from the ROI i To ROI j The n-step transition probabilities, i,j∈{1,2,…,N}, The calculation method is as follows:

[0063]

[0064] in, This indicates that at time τ, from the ROI i To ROI k1 The single-step transition probability, Indicates time τ+1 from ROI k1 To ROI k2 The single-step transition probability, and so on, Indicates time τ+n-2 from arrive The single-step transition probability, Indicates time τ+n-1 from To ROI j The single-step transition probability.

[0065] In fact, according to the definition and rules of matrix multiplication, This can be calculated more conveniently using matrix multiplication. By successively multiplying all the single-step probability matrices from time τ to time τ+n-1, the n-step probability transition matrix A of the brain region at time τ can be obtained. (τ,n) The element at position (i,j) is The calculation formula is as follows:

[0066]

[0067] To calculate the 10-step transition probability matrix of the brain region, the single-step transition probability matrix A of the brain region at 10 consecutive time points is used. (t) These can form a sequence, which is called a time window. Then the τ-th time window Ω... τ =[A (τ) A (τ+1) ,…,A (τ+9) ], where τ∈{1,2,…,(TL-8)}, because at most [T-(L-1)-(10-1)] time windows can be obtained, i.e., (TL-8) time windows. For time window Ω τ Calculate the 10-step transition probability matrix A of the corresponding brain region. (τ,10) The dimension is N×N, and by applying Ω τ The calculation is obtained by multiplying all the single-step transition probability matrices sequentially, and the formula is:

[0068]

[0069] in, That is, at time τ, from ROI i To ROI j The 10-step transition probability.

[0070] Step S143: Calculate the mean matrix of 10-step transition probabilities, and obtain the stationary probability vector of the brain region through time arithmetic mean and column mean. Then calculate the stationary probability vector of the network through the brain region network mapping relationship.

[0071] Further calculate the mean matrix of 10-step transition probabilities Matrix elements Indicates from ROI i To ROI j The time arithmetic mean of the 10-step transition probabilities, where i,j∈{1,2,…,N}, The calculation formula is:

[0072]

[0073] 10-step transition probability mean matrix A matrix with dimensions of N×N The calculation formula is:

[0074]

[0075] Calculate the stationary probability vector of brain regions u = (u1, ..., u2) j ,…,u N ), where j∈{1,2,…,N}. u j brain region ROI j The stationary probability, which is calculated by taking the mean matrix of the 10-step transition probabilities. The arithmetic mean of the elements in column j, u j The calculation method is as follows:

[0076]

[0077] Brain Region ROI j stationary probability u j For a brain region to reach a steady state, convergence from any brain region to the ROI is required. j The transition probability is such that the sum of the stationary probabilities of all brain regions is 1, which satisfies the condition that...

[0078] A stationary state refers to a state distribution of a Markov chain that gradually stabilizes after a large number of transitions. The similarity of elements in the transition probability matrix of each row (corresponding to different initial states) becomes higher and higher, and the final probability distribution hardly depends on the specific state at the start, but only on the target state to be reached.

[0079] Here, a stationary state refers to the characteristic that, over a sufficiently long period, the information flow model exhibits a probability distribution of information transfer between brain regions (networks) that is independent of the originating brain region (network) and depends only on the target brain region (network). In other words, when the number of steps n is sufficiently large (according to research, n ≥ 7), the similarity of the row vectors in the n-step transition probability matrix increases sharply and even approaches uniformity. Therefore, the stationary probability vector u reflects the probability distribution of each brain region as the target brain region for information flow in a stable state.

[0080] The k-th network is net k Where k∈{1,2,…,7}, now calculate the network stationary probability vector v=(v1,…,v k ,…,v7),v k That is, network (net) k The stationary probability.

[0081] For any ROI, it can be mapped to a network. A network contains several ROIs. Let P be an example. k Indicates belonging to the network (net) k The set of ROI indices, if ROI i Belongs to the network (net) k Let i∈P k .

[0082] network k The sum of the stationary probabilities of all included brain regions yields the network net. k The stationary probability of the network net k stationary probability v k The calculation formula is:

[0083]

[0084] j∈P k Indicates ROI j Belongs to the network (net) k ;net k stationary probability v k To ensure the network reaches a stable state, starting from any network and going to net k The transition probability, and satisfying

[0085] The network stationary probability vector v = (v1, ..., v2)k ,…,v N This reflects the probability that each network will be the target network for information flow in a stable state.

[0086] Step S15: Calculate the mean matrix of the optimal number of steps in the network.

[0087] Step S151: Calculate the network functional connectivity matrix at time t. The network functional connectivity matrix is ​​obtained by mapping the brain region functional connectivity matrix.

[0088] The network function connection matrix at time t is represented as FC net (t), where t∈{1,2,…,T-(L-1)}, matrix elements This indicates that at time t, the network net k and network net m The functional connection strength between networks is defined as k,m∈{1,2,…,7}. Furthermore, the functional connection strength between networks is set to 0. When k≠m, via network net k Midbrain Region ROI i (i∈P k ) and network net m Midbrain Region ROI j (j∈P m Functional connectivity strength after correction between ) To achieve normalization, a double summation is performed, considering that different brain networks contain different numbers of ROIs. The sum must be divided by the product of the number of ROIs in the two brain networks. The calculation formula is as follows:

[0089]

[0090] Where P k and P m They represent the network (net). k and network net m The set of ROI indices included, |P k | and | P m | represent the network net k and network net m The number of ROIs included. Matrix FC net (t) has a dimension of 7×7, represented as:

[0091]

[0092] Step S152: Calculate the network single-step transition probability matrix at time t, which is calculated by normalizing the network functional connectivity matrix L1-norm.

[0093] The network single-step transition probability matrix at time t is represented as B. (t) , where t∈{1,2,…,T-(L-1)}, matrix elements This indicates that at time t, the network net k to network net m The single-step transition probabilities are obtained from the L1-norm of the network functional connectivity matrix, where k,m∈{1,2,…,7}. The formula for calculating the network single-step transition probability matrix is:

[0094]

[0095] Due to the network function connection matrix FC net (t) diagonal elements so The network single-step transition probability matrix B at time t (t) The dimension is 7×7, which normalizes the probabilities and does not consider self-loops. The network single-step transition probability matrix B at time t is... (t) Represented as:

[0096]

[0097] Step S153: Calculate the optimal step matrix of the network. Before calculation, set a time window and calculate the matrix elements by comparing the maximum value of the elements of the multi-step transition probability matrix within the time window.

[0098] To calculate the optimal step matrix of the network, the single-step transition probability matrix B of the network at 10 consecutive time points is used. (t) Let φ be the τth time window. τ =[B (τ) ,…,B (τ+i) ,…,B (τ+9) ].

[0099] Time window φ τ The optimal step matrix of the network is denoted as S. * (τ), where τ∈{1,2,…,TL-8}, and matrix element s km (τ) represents the time window τ from the network net k to network net m The optimal number of steps, where k,m∈{1,2,…,7}.

[0100] The optimal number of steps refers to determining a specific number of steps within a given time window such that the sum of probabilities of all reachable paths from the starting network to the target network reaches its maximum and is then minimized. Furthermore, the optimal number of steps between the same network is defined as 0. Therefore, s km (τ)∈{0,1,2,…,10},s kmThe formula for calculating (τ) is:

[0101]

[0102] in For the network net at time τ k to network net m The s-step transition probability, i.e., matrix B (τ,s) The element at position (k,m), where matrix B (τ,s) It is the s-step network transition probability matrix, with Among them B (τ+i) It is the network single-step transition probability matrix at time τ+i; Its purpose is to find out why The set s that yields the maximum value. The role of the min function here is to determine if... If the resulting set of s values ​​contains at least two elements, then the smallest s value is selected.

[0103] Then time window φ τ The optimal step matrix S * (τ) has a dimension of 7×7, and its expression is:

[0104]

[0105] Step S154: Calculate the mean matrix of the network's optimal steps, which is obtained by the time arithmetic mean of the network's optimal steps matrix.

[0106] Calculate the mean matrix of the optimal number of steps in the network. Matrix elements Represents network (net) k to network net m The optimal step-time arithmetic mean, k,m∈{1,2,…,7}, The calculation method is as follows:

[0107]

[0108] The network's optimal step mean matrix The characteristics of the spatiotemporal information path distance between individual networks were quantified from a spatiotemporal dynamics perspective, and the matrix... The dimension is 7×7, and the expression is:

[0109]

[0110] Step S16: Calculate the direct connection strength matrix.

[0111] Step S161: Calculate the network direct connection label matrix. If there is a direct connection between networks, mark it as 1 at the corresponding position; otherwise, mark it as 0.

[0112] Calculate the time window φ τ The network direct connection label matrix in the image is denoted as I. direct (τ), where τ∈{1,2,…,TL-8}, and the matrix elements are... If within the time window φ τ China Network k to network net m There is a direct connection. otherwise Right now Where k,m∈{1,2,…,7}

[0113] Among them, if the network net is in the time window τ k to network net m The optimal number of steps s km If (τ) = 1, then it is said that within the time window φ τ China Network k and network net m There are direct connections between them, where k,m∈{1,2,…,7}. Calculation method:

[0114]

[0115] Matrix I direct (τ) has the following form:

[0116]

[0117] Step S162: Calculate the network direct connection strength matrix, which is the time arithmetic mean of the network direct connection label values.

[0118] Calculate the direct connection strength matrix S direct Matrix elements Represents network (net) k to network net m The direct connection strength of the network (net) k to network net m The time arithmetic mean of the direct join marker values, and The calculation method is as follows:

[0119]

[0120] Direct connection strength matrix S direct The format is:

[0121]

[0122] The network direct connection strength Sdirect It quantifies the average degree of direct connection between individual networks over time.

[0123] Step S17: Construct an auxiliary diagnostic model through statistical analysis of differences between groups and logistic regression.

[0124] Step S171: Perform t-tests on the three index elements of the modeling sample group—the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix—to analyze the differences between groups and identify the significant difference elements for each index.

[0125] The pre-collected image data of the schizophrenia patient group (SZ group) and the healthy control group (HC group) were converted into three indicators: network stationary probability vector, network optimal step mean matrix, and network direct connection strength matrix.

[0126] For ease of explanation, the three indicators mentioned above—the network stationary probability vector, the network optimal step mean matrix, and the network direct connection strength matrix—are all regarded as vectors (if they were originally matrices, the elements are sorted by column to form a vector).

[0127] Two-sample t-tests were used to analyze the differences between groups for all indicator elements, and multiple hypothesis testing was performed using FDR. If the p-value of an element after FDR correction is still not greater than 0.05, the element is considered to have a significant difference.

[0128] The significantly different elements selected by the network stationary probability vector are denoted as x1,…,x i1 ,…,x n1 n1 represents the number of significantly different elements in the network stationary probability vector; x i1 Let i represent the significantly different element of the i1th network stationary probability vector, i1∈{1,…,n1}, and n1≤7.

[0129] The significantly different elements selected from the mean matrix of the optimal number of steps in the network are denoted as x1′,…,x i ′2,…,x′ n2 n² represents the number of significantly different elements in the mean matrix of the network's optimal steps; x i ′2 represents the significantly different element in the mean matrix of the optimal steps of the i2th network, i2∈{1,…,n2}, and n2≤49.

[0130] The significantly different elements selected from the network direct connection strength matrix are denoted as x1″,…,x i ′3′,…,x′ n ′3, n3 represents the number of significantly different elements in the network direct connection strength matrix; x i ′3′ represents the significant difference element in the direct connection strength matrix of the i3th network, i3∈{1,…,n3}, and n3≤49.

[0131] Step S172: Construct the logistic regression hypothesis function and estimate the regression coefficients using the maximum likelihood estimation method.

[0132] Let z, z′, z″ represent the linear prediction values.

[0133] The formula for constructing a linear combination of stationary probability vectors of the network is as follows:

[0134] z = β0 + β1x1 + ... + β i1 x i1 +…+β n1 x n1

[0135] Where β0,β1…,β i1 ,…β n1 β is the regression coefficient of the network stationary probability vector, and β0 is the intercept term.

[0136] The formula for constructing a linear combination of the mean matrix of the optimal number of steps in the network is as follows:

[0137] z′=β0′+β1′x1′+…+β i '2x i ′2+…+β′ n2 x′ n2

[0138] Where β0′,β1′…,β i ′2,…,β′ n2 β0′ represents the regression coefficients of the mean matrix of the optimal steps in the network, and β0′ is the intercept term.

[0139] The linear combination of the network direct connection strength matrix is ​​constructed, and its calculation formula is as follows:

[0140] z″=β0″+β1″x1″+…+β i ′3′x i ′3′+…+β′ n '3x' n '3

[0141] Where β0″,β0″…,β i ′3′,…,β′ n ′3 represents the regression coefficients of the network direct connection matrix, and β0″ is the intercept term.

[0142] Step S173: Construct network stationary probability vector to assist the diagnostic model in predicting probabilities, network optimal step mean matrix to assist the diagnostic model in predicting probabilities, and network stationary probability vector to assist the diagnostic model in predicting probabilities. Transform z, z′, z″ using the Sigmoid function to obtain the probability that the subject to be diagnosed is a patient, with a value between 0 and 1, which can be interpreted as the probability of being a patient with schizophrenia. If the probability is greater than 0.75, the subject can be diagnosed with schizophrenia.

[0143] The predicted probability of the network stationary probability vector-assisted diagnostic model is:

[0144]

[0145] Where e is the natural constant.

[0146] The prediction probability of the network optimal step mean matrix-assisted diagnostic model is:

[0147]

[0148] The predicted probability of the network stationary probability vector-assisted diagnostic model is:

[0149]

[0150] In this model, the regression coefficients of each indicator were estimated using the maximum likelihood estimation method to ensure that the auxiliary diagnostic model can predict the known patient and healthy populations as accurately as possible. The independent variables for training the logistic regression model are the significantly different values ​​of each indicator for the subjects, and the dependent variable is the label value y. i , where y i Let y be the label value of the i-th subject. If the i-th subject is a patient, let y be... i =1, otherwise record y i =0.

[0151] Step S18: Acquire f-MRI images of the subject to be diagnosed. After preprocessing, calculate the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix. Input the elements with significant differences in each indicator into the corresponding auxiliary diagnostic model to calculate the probability of predicting schizophrenia.

[0152] For the stationary probability vector of the network, extract the elements with significant differences, input the extracted elements with significant differences into the stationary probability vector-assisted diagnostic model, and calculate the predicted probability of the stationary probability vector of the network.

[0153] For the mean matrix of the network's optimal steps, extract the elements with significant differences, input the extracted elements with significant differences into the network's optimal step mean matrix auxiliary diagnostic model, and calculate the predicted probability of the network's optimal step mean matrix.

[0154] For the network direct connection strength matrix, extract the elements with significant differences, input the extracted elements with significant differences into the network stationary probability vector-assisted diagnostic model, and calculate the predicted probability of the network direct connection strength matrix.

[0155] Experts further analyzed the predictive probabilities of the three indicators and finally arrived at a comprehensive diagnostic result.

[0156] The following describes a specific implementation of the spatiotemporal dynamics-based auxiliary diagnostic method for schizophrenia of the present invention.

[0157] The data used in this embodiment of the invention are all resting-state data collected by a Siemens machine (magnetic field strength 3.0T). The scanning parameters are: echo time (TE) 0.03s, repetition time (TR) 2.5s, flip angle 90°, layer thickness 3mm, spatial resolution 3.75mm × 3.75mm × 4mm, and 212 time points were collected.

[0158] The preprocessing of f-MRI images was based on the CBIG workflow tool (based on FSL and FreeSurfer). The processing steps included removing the first four time points, slice time correction, motion correction, registering image edges to structural images, covariate regression (white matter signal, ventricular signal, head motion), and bandpass filtering (0.01–0.08 Hz).

[0159] Among them, participants whose original data had the following characteristics were excluded: severe visual segmentation errors, intra-individual registration cost exceeding 0.7, abnormal FD (framewise displacement) index, excessively high proportion of subject data, and unsatisfactory registration between functional and anatomical images.

[0160] The network stationary probability vector is calculated as follows:

[0161] like Figure 1 As shown, for each subject, a 400-ROI partitioning atlas based on the Yeo 7 network was used to map the preprocessed f-MRI brain imaging data to the brain region (ROI) level (see...). Figure 1 In a), the f-MRI image data is thus mapped to functional image signals at the brain region level, with a signal matrix dimension of 400×208.

[0162] With a window length of 50 and a step size of 1, 159 sliding windows were created, each with a dimension of 50×208, corresponding to times 1 to 159. Within each sliding window, the Pearson correlation values ​​between ROIs were calculated to obtain the dynamic functional connectivity matrix of the brain region (see...). Figure 1As shown in b), this resulted in 159 functional connectivity matrices, each with a dimension of 400×400. Negative values ​​in each functional connectivity matrix were set to zero, and positive correlation values ​​underwent further false discovery rate (FDR) correction (see [link to documentation]). Figure 1 c) is then normalized to obtain a single-step transition probability matrix for 159 brain regions (see c). Figure 1 In d), the dimensions are all 400×400.

[0163] The single-step probability transition matrices at 10 consecutive time points are combined to form a time window, resulting in a total of 150 time windows. The single-step probability matrices within each time window are then multiplied sequentially to obtain a 10-step transition probability matrix (see...). Figure 1 In step e), calculate the mean matrix of 10-step transition probabilities, and then average it column-wise to obtain the brain region stationary probability vector (see...). Figure 1 f in the text is then mapped to a stationary probability vector of the network based on the cortical partitioning map of the Yeo7 network (see f in the text). Figure 1 g in the middle.

[0164] The mean matrix of the optimal number of steps in the network and the direct connection strength matrix of the network are calculated as follows:

[0165] like Figure 2 As shown, the brain region-corrected functional connectivity matrix is ​​mapped to the network functional connectivity matrix (see...). Figure 2 In section a), there are 159 elements with a dimension of 7×7, and the single-step transition probability matrix of the network is calculated using the L1-norm (see...). Figure 2 b) has 159 elements, with a dimension of 7×7, and is divided into 150 time windows using the same method. Based on the direct connection labeling rule, the direct connection label matrix for each time window is calculated (see [reference]). Figure 2 In section d), calculate the mean of the optimal step matrix and the direct connection strength matrix (see d). Figure 2 In the example, e), the dimensions are all 7×7.

[0166] The auxiliary diagnostic model is constructed in the following way:

[0167] like Figure 3 As shown, index A represents the network stationary probability vector, index B represents the mean matrix of the network's optimal steps, and index C represents the network's direct connection strength matrix. For each element of the three indices in the existing sample set of participants, a t-test was performed to analyze differences between groups, and statistically significant differences in each index were screened after FDR correction (see...). Figure 3 Step a). Using the existing sample set, estimate the corresponding regression coefficients using the maximum likelihood estimation method (see...). Figure 3 Step b) involves using a transformation formula to convert the predicted value into a predicted probability (see...). Figure 3 Step c).

[0168] Specifically, Figure 4 The construction of the prediction model for indicator A will be explained in detail as an example. The prediction probability formula is obtained through t-test, FDR correction, and maximum likelihood estimation. This indicator is then calculated using the imaging data of the subjects to be diagnosed. Significantly different elements are selected, and the prediction probability is calculated using the estimated regression coefficients (see...). Figure 4 Step d). The predicted probabilities of the three indicators are calculated separately, and experts conduct a comprehensive analysis to determine whether a diagnosis has been made.

[0169] Compared to other auxiliary diagnostic methods for schizophrenia, this method proposes three indicators from a spatiotemporal dynamics perspective: the network stationary probability vector, the mean matrix of the network's optimal steps, and the strength of direct connections in the network. Using statistical tests and logistic regression, a formula is derived to calculate the probability of an individual being diagnosed with schizophrenia.

[0170] It is understood that the present invention has been described through some embodiments, and those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A spatiotemporal dynamics-based auxiliary diagnostic method for schizophrenia, characterized in that, Includes the following steps: Step S11: Collect f-MRI brain imaging data of individual subjects, preprocess the collected f-MRI brain imaging data to obtain preprocessed f-MRI brain imaging data; Step S12: Using the cortical brain region division atlas of N ROIs based on the Yeo 7 network, the preprocessed f-MRI brain image data is mapped to the N ROIs to obtain the temporal signals of the brain regions; Step S13: Divide the temporal signal of the brain region into sliding windows, and calculate the functional connectivity matrix in each sliding window to obtain the dynamic functional connectivity matrix; Step S14: Calculate the stationary probability vector of the brain region and map it to the network to calculate the stationary probability vector of the network; Step S15: Calculate the mean matrix of the optimal number of steps in the network; Step S16: Calculate the direct connection strength matrix; Step S17: Construct an auxiliary diagnostic model through statistical analysis of inter-group differences and logistic regression; Step S171: Perform t-tests on the three index elements of the modeling sample group—the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix—to analyze the differences between groups and identify the significant difference elements for each index. Step S172: Construct the logistic regression hypothesis function and estimate the regression coefficients using the maximum likelihood estimation method; Step S173: Construct the network stationary probability vector to assist the diagnostic model in predicting probabilities, the network optimal step mean matrix to assist the diagnostic model in predicting probabilities, and the network stationary probability vector to assist the diagnostic model in predicting probabilities. Step S18: Acquire f-MRI images of the subject to be diagnosed. After preprocessing, calculate three indicators: the network stationary probability vector, the mean matrix of the network's optimal steps, and the network direct connection strength matrix. Input the elements with significant differences in each indicator into the corresponding auxiliary diagnostic model to calculate the probability of predicting schizophrenia.

2. The method for auxiliary diagnosis of schizophrenia based on spatiotemporal dynamics according to claim 1, characterized in that, Step S14 includes the following steps: Step S141: Calculate the corrected brain region functional connectivity matrix. Correction is achieved by setting the diagonal and negative correlation elements in the functional connectivity matrix to 0 and performing false discovery rate correction on the positive correlation values, so that the brain region functional connectivity strength value is not less than 0. Step S142: Calculate the single-step transition probability matrix and the 10-step transition probability matrix of the brain region. The single-step brain region transition probability is calculated by standardizing the brain region functional connectivity matrix using the L1-norm. Before calculating the 10-step transition probability matrix, a time window is set, and the matrix is ​​obtained by continuously multiplying the 10 matrices within the time window. Step S143: Calculate the mean matrix of 10-step transition probabilities, and obtain the stationary probability vector of the brain region through time arithmetic mean and column mean. Then calculate the stationary probability vector of the network through the brain region network mapping relationship.

3. The method for auxiliary diagnosis of schizophrenia based on spatiotemporal dynamics according to claim 2, characterized in that, Step S15 includes the following steps: Step S151: Calculate the network functional connectivity matrix, which is obtained by mapping the brain region functional connectivity matrix; Step S152: Calculate the network single-step transition probability matrix, which is calculated by normalizing the network functional connectivity matrix L1-norm. Step S153: Calculate the optimal step matrix of the network. Before calculation, set a time window and calculate the matrix elements by comparing the maximum value of the elements of the multi-step transition probability matrix within the time window. Step S154: Calculate the mean matrix of the network's optimal steps, which is obtained by the time arithmetic mean of the network's optimal steps matrix.

4. The method for auxiliary diagnosis of schizophrenia based on spatiotemporal dynamics according to claim 3, characterized in that, Step S16 includes the following steps: Step S161: Calculate the network direct connection label matrix. If there is a direct connection between networks, mark it as 1 at the corresponding position; otherwise, mark it as 0. Step S162: Calculate the network direct connection strength matrix, which is the time arithmetic mean of the network direct connection label values.

Images