A method for evaluating children's attention based on phase synchronization of precuneus subregions

Abstract

The present invention relates to a method for evaluating children's attention based on phase synchronization of precuneus subregions, and compares the differences in brain mechanisms between children with ADHD and normal children based on the functional connection characteristics of the precuneus subregions. At the same time, a method for evaluating children's attention based on the phase functional connection characteristics of the precuneus subregions is provided. The present invention utilizes resting state magnetic resonance imaging technology to collect brain blood oxygen level-dependent signals of healthy individuals and ADHD patients in a resting state. A brain region functional connection matrix is ​​established through phase synchronization to segment the precuneus brain region. Based on the connection characteristics of the precuneus subregions, an attention prediction model is established to achieve the function of predicting children's attention.

Description

Technical Field

[0001] The present invention relates to a method for evaluating children's attention based on phase synchronization of precuneus subregions, and in particular to a method for evaluating children's attention based on functional connection characteristics of precuneus subregions of the brain, and belongs to the technical field of medical images and magnetic resonance imaging. Background Art

[0002] Attention deficit hyperactivity disorder (ADHD) is a common neurodevelopmental disorder in children, commonly known as ADHD. Its core symptoms are attention deficit, hyperactivity, and impulsive behavior. If not treated in time, it will cause problems in learning, life, and social interaction. In today's society, attention assessment and brain mechanism research of children and adolescents with ADHD have attracted much attention.

[0003] Recent brain functional imaging studies have shown that the precuneus, located in the medial part of the cerebral hemisphere, is associated with many high-level cognitive functions, such as episodic memory, self-related information processing, and consciousness. The present invention will segment the precuneus based on the phase synchronization method, compare the differences in precuneus functional connectivity between children with ADHD and normal children, and evaluate individual attention.

[0004] The present invention uses resting state functional magnetic resonance imaging (rs-fMRI). It is an important technology for non-invasively detecting low-frequency activities of different brain regions in the central nervous system at rest. This technology can infer the state of neural network connections by analyzing the correlation of low-frequency activities in the brain. Compared with other fMRI technologies, it has the following advantages: easy signal acquisition and minimal requirements for patients, such as pediatric population.

[0005] Blood oxygen level dependent (BOLD) is the basis of functional magnetic resonance imaging. It is a magnetic resonance imaging technology that uses the changes in the ratio of oxygenated hemoglobin to deoxygenated hemoglobin in the local blood of the brain activity area to cause changes in local tissue T2, thereby reflecting the local activity function of the brain tissue on the T2-weighted image. It can be used to study the cortical activity of the brain.

[0006] At present, there are studies on attention assessment based on whole-brain functional connectivity. Research on children's attention assessment methods based on the functional connectivity characteristics within the precuneus brain region has not yet been involved. Establishing the relationship between precuneus functional connectivity and attention is the innovation of the present invention. Summary of the invention

[0007] The present invention proposes a method for evaluating children's attention based on phase synchronization of precuneus subregions. The BOLD signal in the resting state is collected by using rs-fMRI technology. The precuneus brain region is further segmented by functional connectivity analysis signals. Based on the connectivity characteristics of the precuneus subregions, a prediction model is established to predict the children's attention scores based on the functional connectivity characteristics of the precuneus subregions after analysis and segmentation.

[0008] The specific steps of the children's attention assessment method based on phase synchronization of precuneus subregions are as follows:

[0009] Step 1: Image data acquisition, collecting children's rs-fMRI image data;

[0010] Step 2: rs-fMRI data preprocessing;

[0011] Step 3: Selection of brain regions of interest. The DK brain atlas (Desikan-Killiany Atlas) was selected as a template for dividing and extracting brain regions of interest, and the 26th brain region related to attention, the precuneus, was selected as the region of interest.

[0012] Step 4: Functional connectivity feature analysis: calculate the phase synchronization coefficient of the BOLD signal at each vertex in the brain of each subject using the corresponding formula.

[0013] Step 5: Segment the precuneus brain region. Extract the BOLD signal of the vertex of the brain region of interest, use step S4 to calculate the phase synchronization coefficient matrix of the signal, and use it as the basis for further segmentation of the precuneus brain region. By comparing the modularity, use a suitable segmentation algorithm to segment the precuneus brain region. There are several calculation indicators for the functional connectivity between brain regions, which can be further divided into directed connections and undirected connections. The phase synchronization used in the present invention is an undirected connection, which describes the synchronization between two phases, that is, there is a fixed phase difference between the two phases. And phase synchronization is a nonlinear indicator that conforms to the characteristics of neural signals. Its value is between 0 and 1, which is very suitable for network weights.

[0014] Step 6: Analyze the differences within the subregions of the precuneus and the differences between the subregions. Use the functional connectivity method of S4 to analyze the functional connectivity characteristics of the BOLD signals within the subregions of the precuneus and between the subregions. Use statistical test methods to compare the subregional differences between the normal children group and the ADHD children group. Alternatively, further analyze the BOLD signal phase difference sequence of the precuneus based on the low-frequency amplitude value ALFF and detrended fluctuation DFA method.

[0015] Step 7: Establish a prediction model. Train the prediction model by recording the children's attention assessment scores and the phase synchronization coefficient matrix of the precuneus subregion obtained after segmentation. This allows the prediction of children's attention scores to be achieved by analyzing the functional connectivity characteristics of the precuneus subregions alone.

[0016] In step 2, the rs-fMRI technology preprocesses the data, and the process is as follows:

[0017] S2.1 Time correction (slice timing): MRI images are scanned layer by layer. To avoid the influence between adjacent layers, interval scanning is usually adopted, and the acquisition time is different. When processing and analyzing resting MRI data, the time difference of acquisition will have a great impact, so it is necessary to correct the time of each body so that all bodies have the same time point.

[0018] S2.2 Head motion correction: Generally, the first image is selected as the reference image, and the remaining images are matched with the first image by rotation or translation, and the images are resampled using the corresponding algorithm.

[0019] S2.3 Image co-registration: Co-register the individual images of the subject to make them of uniform size, such as co-registration of functional images and structural images.

[0020] S2.4 Spatial normalization: The shape and size of each person’s brain are different. The purpose of this step is to enable large-scale and unified processing and analysis of patient data in the future.

[0021] S2.5 Spatial smoothing: This step is a filtering process based on Gaussian random field theory. Since the collected BOLD signal frequency is generally 0.01-0.08Hz, the signal needs to be filtered to eliminate or suppress the high-frequency components in the signal to reduce image noise.

[0022] The DK brain map in step 3 is used as a template for dividing and extracting the precuneus brain region in the present invention, dividing the left and right hemispheres into 36 brain regions respectively, while retaining the macroscopic anatomical structure of the sulci and gyri in the human brain network group map, which is also the reason for selecting this map as the initial partition. The selection process of the precuneus brain region is as follows: given the label value of the precuneus brain region in the DK brain map, find the vertex with the same label value in the whole brain, and extract all the vertices belonging to the precuneus brain region.

[0023] The calculation process of phase synchronization in step 4 is as follows:

[0024] In step 3 of S4.1, all vertices of the precuneus brain region have been found, and the index value of each vertex has been extracted, and the BOLD signal corresponding to each vertex of the precuneus brain region is obtained from the BOLD signal data set collected from the whole brain.

[0025] S4.2 Calculation of phase synchronization coefficient of BOLD signal

[0026] Using Hilbert transformation, the BOLD signal of the precuneus brain region is transformed into an analytical signal to obtain the amplitude and instantaneous phase of each vertex signal in the brain region, and the phase difference between each vertex is calculated by the instantaneous phase. Assuming that there are n vertices in the precuneus brain region, then there are also n extracted BOLD signals. By calculating the phase difference between the signals, a total of n*(n-1) / 2 phase differences can be obtained. Mi represents the i-th vertex Mark in the brain region, the phase is represented by φ, and the phase difference is represented by Δφ.

[0027] Δφ M1-M2 =φ M1 -φ M2

[0028] Δφ M1-M3 =φ M1 -φ M3 ...

[0029]

[0030] The analytical signal can be obtained through Hilbert transformation, that is, the signal A can be expressed as:

[0031] A=|A|cosθ+|A|sinθ

[0032] =|A|e jθ

[0033] |A| represents the amplitude of the signal, e jθ Indicates the phase of the signal.

[0034] The delayed phase can be expressed as e jΔφ , the phase synchronization coefficient (ps) can be expressed as:

[0035]

[0036] The final n*(n-1) / 2 phase synchronization coefficients ps are used as the basis for segmenting the precuneus brain area.

[0037] The principle of precuneus segmentation in step 5:

[0038] S5.1 selects an appropriate clustering analysis algorithm and clusters the precuneus brain region based on the phase synchronization coefficient of the BOLD signal in the precuneus brain region to segment the brain region. Based on the obtained results, a complex brain network can also be constructed and features calculated.

[0039] The clustering method uses three community detection algorithms in the igraph package: eigenvalue, multilevel, and greedy. By comparing the modularity values, the algorithm with the best classification effect (multilevel community) is selected to cluster and segment the brain regions. After clustering, the connections between nodes within each community are relatively tight, and the connections between communities are relatively sparse.

[0040] S5.2 Basic principles of multi-level clustering algorithm:

[0041] Hierarchical clustering is a bottom-up algorithm. At the initialization, each node is defined as belonging to a separate community. According to the principle of maximizing the modularity of the overall community partition, the local contribution brought by the movement of each node community is maximized in an iterative manner.

[0042] S5.2.1 Treat each node as a community.

[0043] S5.2.2 decides which neighbors should be merged based on the modularity criterion. After one iteration, some communities are merged into one community.

[0044] S5.2.3 Each community is treated as a node, and its degree and connection information are counted. The next iterative merge is performed based on the statistical results.

[0045] S5.2.4 Repeat step 3 until the overall modularity corresponding to the network community division cannot be increased any further.

[0046] S5.3 The modularity calculation process is as follows:

[0047] Modularity is a commonly used method to measure the strength of a network's community structure. Its size mainly depends on the community distribution of nodes in the network, that is, the community division of the network. It can be used to quantitatively measure the quality of network community division. The closer its value is to 1, the stronger the strength of the community structure divided by the network, that is, the better the division quality.

[0048] The basic idea is to compare the network after community division with the corresponding null model to measure the quality of community division. The null model refers to a random graph model that has some of the same properties as the given actual network but is completely random in other aspects.

[0049] For example, for a given brain network, after community division, the sum of the edges within all communities is:

[0050]

[0051] Among them A ij is the adjacency matrix of the actual network, B i and B j They represent the communities to which node i and node j belong in the network respectively. If the two nodes belong to the same community, δ is assigned a value of 1, otherwise it is 0.

[0052] For a null model of the same size as the brain network, if the same community division is used, the expected value of the sum of the edges within all communities is:

[0053]

[0054] Among them C ij is the expected value of the number of edges between node i and node j in the null model. u i with u j are the degrees of node i and node j in the original network respectively.

[0055] Since the modularity of a network is defined as the ratio of the difference between the number of edges within the community of the network and the number of edges within the community of the corresponding null model to the total number of edges M of the network, it can be expressed as:

[0056]

[0057] Therefore, the optimal network community division can be obtained by maximizing the modularity E.

[0058] The above-mentioned brain region segmentation method is used to obtain a unified segmentation template of the precuneus brain region.

[0059] The steps are as follows: collect BOLD signals from the precuneus of a large number of normal children. The BOLD signal here is an x*y matrix, where x represents the number of vertices in the precuneus, and y represents the length of the collected time series signal, that is, the number of time points. Analyze the functional connection characteristics of the vertices in each subject's brain region, that is, the phase synchronization coefficient matrix. Average the phase synchronization coefficient matrix of the children tested and use it as the basis for segmenting the precuneus. A multi-level community segmentation algorithm can be selected to segment the precuneus. The segmentation of the left precuneus region results in p subregions, and the segmentation of the right precuneus region results in q subregions.

[0060] When a new child is added to the study, the precuneus brain region will be segmented according to the unified segmentation template. Based on the segmentation results, the BOLD signal of the precuneus subregions will be analyzed.

[0061] In step six, different methods are used to study the BOLD signal of the segmented precuneus brain region.

[0062] S6.1 obtains the segmented precuneus subregions from step S5, and based on this, calculates the average BOLD signal of each subregion. Then, according to step S4, calculates the functional connectivity features between subregions and compares the differences between subregions. Or, calculates the average BOLD signal of other brain regions, and compares the differences between subregions and other brain regions by calculating the functional connectivity features of precuneus subregions and other brain regions. A comparative analysis is performed between the disease group and the normal group using statistical test methods.

[0063] S6.2 The BOLD signal phase difference sequence is also a time series. The present invention is based on the low-frequency amplitude method and detrending method commonly used to analyze brain signals, and can further study the phase difference sequence. The low-frequency amplitude (Amplitude of LowFrequency Fluctuations, ALFF) technology is an important indicator that reflects the changes in the intensity of spontaneous brain activity by detecting the average amplitude value of the BOLD low-frequency oscillation signal of the subject deviating from the baseline in a short period of time. Usually the low frequency range is 0.01~0.1Hz, and the signal in this range can reflect the activity of neurons. The present invention analyzes the BOLD signal phase difference sequence based on the ALFF technology.

[0064] In order to obtain the low-frequency component of the BOLD signal, the signal is first low-pass filtered to calculate the ALFF value corresponding to the signal. Each ALFF value represents the activity intensity of the voxel in the brain, and the signal activity intensity between brain regions is compared by comparing the difference in values. In the present invention, the ALFF values ​​of the BOLD signal phase difference sequence are compared. For the phase difference sequence, the amplitude of the sequence is the phase difference value. By extracting the average BOLD signal of each precuneus subregion obtained after segmentation, the phase difference sequence between subregions is calculated, and the low-frequency phase difference value of the subregion signal of each subject is calculated, and the statistical test method is used to find the subregions with significant differences between the children's disease group and the normal group.

[0065] Based on the ALFF method, the average BOLD signal of each subregion is Hilbert transformed to obtain the phase value at each time point. Each signal is a matrix of size 1*tn, and the matrix element is the phase of the low-frequency BOLD signal. ti Represents the phase value at time point t i.

[0066] [θ t1 ,θ t2 ,,θ t3 ,,θ t4 ,,…,θ tn ]

[0067] The phase difference of the BOLD signal between sub-regions is calculated to obtain a phase difference sequence. If the left precuneus is divided into p sub-regions, there are a total of p*(p-1) / 2 phase difference sequences.

[0068] [Δθ t1 ,Δθ t2 ,,Δθ t3 ,,…,Δθ tn ] i,j

[0069] The above formula represents the phase difference sequence of BOLD signals between the i-th subregion and the j-th subregion of the left precuneus.

[0070] Calculate the mean value of the phase difference value at each time point of the phase difference signal:

[0071]

[0072] The square root of the mean phase difference of the signal at each time point is obtained by taking the square root.

[0073]

[0074] Through the above calculation method, the average phase difference value of each sub-region of the left precuneus has been obtained. On this basis, the average phase difference value of the precuneus brain region can be further obtained.

[0075]

[0076] The average phase difference value of each subregion was then divided by the average phase difference value of the precuneus brain region to obtain a standardized low-frequency phase difference value divided by the mean.

[0077]

[0078] By calculating the standardized difference Δθ between the phases of different subregions normalized , an intergroup comparison was conducted between the disease group and the control group to find out the precuneus subregion with the largest phase difference between children with ADHD and normal children.

[0079] S6.3 For phase difference sequences, detrended flux analysis (DFA) can also be performed. For example, in the process of collecting signals, polynomial trend signals such as noise may be mixed in, making it impossible to clearly find the internal connection of the signal. Or unstable components in the signal may cause pseudo-correlation of the signal. DFA can filter out trend components of various orders in the sequence and find the internal connection of the signal.

[0080] The principle of detrending is to eliminate the trend change of the data by subtracting the corresponding fitting straight line from the point image of the original data, so as to obtain the essential characteristics of the data. In the present invention, the internal connection can be found by performing detrending analysis on the BOLD signal phase difference time series.

[0081] S6.3.1 The Hilbert transform results in the time series of the BOLD signal phase difference between the i-th subregion and the j-th subregion of the precuneus:

[0082] [Δθ t1 ,Δθ t2 ,,Δθ t3 ,,Δθ t4 ,,…,Δθ tn ] i,j

[0083] S6.3.2 Based on the phase difference sequence, the fitting curve formula is calculated as follows:

[0084] y=Δθ ti *t+b

[0085] S6.3.3 Subtract the phase difference at each time point from the value of the fitting curve at the corresponding time point to obtain the detrended data:

[0086] [Δθ t1 -y,Δθ t2 -y,Δθ t3 -y,…,Δθ tn -y] i,j

[0087] S6.3.4 Statistical analysis was used to analyze the precuneus subregions with significant differences by comparing the detrended phase difference data between the disease group and the normal group.

[0088] There are many methods for attention assessment. The present invention studies an attention assessment method based on phase synchronization of the precuneus subregions, which is mainly designed for children. The functional connectivity characteristics of the precuneus subregions of children's brains and the scores of children's attention are obtained statistically, and a prediction model is established. The function of predicting the attention score of the corresponding child is achieved by only analyzing the functional connectivity characteristics of the BOLD signals between the precuneus subregions of children.

[0089] Attention evaluation method in step 7:

[0090] S7.1 There are many ways to obtain attention scores, which can be scored through professional attention assessment mechanisms. For example, fill out an attention assessment questionnaire, score the child's attention based on the answers, and use it as the dependent variable y of the prediction model.

[0091] The prediction model building method in step seven:

[0092] S7.2 Collect a large number of children's attention assessment scores and functional connectivity characteristics of the precuneus brain area, use Lasso or SVR method to optimize the prediction model, and obtain the optimal a ij and b parameters. The prediction model is as follows:

[0093]

[0094] As described in the formula: y represents the attention evaluation score; x ij It represents the functional connection characteristics between the segmented precuneus subregion and the j subregion, that is, the phase synchronization coefficient. Assuming that there are p subregions in the left precuneus, there are p*(p-1) / 2 phase synchronization coefficients between the subregions.

[0095] After successfully establishing the prediction model, when it is necessary to predict the attention assessment scores of other children, the BOLD signal of the children's precuneus can be collected, the phase synchronization coefficient between sub-regions after brain segmentation can be calculated, and the numerical value can be substituted into the model for calculation and prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0097] Figure 1 The flowchart of the experimental process of the child attention assessment method based on the phase synchronization of the precuneus subregions of the present invention includes specific steps from raising questions, determining solutions to solution demonstration;

[0098] Figure 2 The present invention is a flow chart for obtaining a precuneus segmentation template for a children's attention assessment method based on phase synchronization of precuneus subregions. DETAILED DESCRIPTION

[0099] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0100] 1.1 Subject selection

[0101] ADHD-200 is an international neuroimaging data sharing program (INDI) alliance, and its ADHD-200 sample is obtained by the cooperation of 8 international imaging websites. These websites collect and publicly share neuroimaging data of 362 children and adolescents with ADHD and 585 normal controls. These 947 data sets consist of structural and resting state fMRI data and phenotypic information. For example, the KKI (Kennedy Krieger Institute) site provides 69 normal subjects (28 males, age range 8-12 years, average age 10.3 years; 41 females, age range 8-13 years, average age 10.4 years), and 25 children with ADHD (10 males, age range 8-13 years, average age 9.9 years; 15 females, age range 8-13 years, average age 10.1 years). For phenotypic data, there was no significant difference in gender, age, and handedness between the two groups (p>0.05). In addition to obtaining data from the above sites, actual clinical data can also be collected.

[0102] 1.2 Image data acquisition

[0103] Collect rs-fMRI image data of the children. The children were required to lie flat and keep their heads stable during data collection, try not to be disturbed by the outside world, and stay relaxed. The scanning parameters provided by each website are different. Taking the KKI site as an example, the magnetic resonance scanner is Phillips 3T, and the scanning parameters are: 47 slices, repetition time (TR) 2500ms, echo time (TE) 30ms, resolution (Resolution, Res) 3.0*3.0mm^2, layer thickness (Thickness) 3mm, rotation angle (Flip Angle) 75°.

[0104] 1.3 rs-fMRI data preprocessing

[0105] The rs-fMRI data of the children were preprocessed using FreeSurfer or CAT12 software based on SPM in MATLAB. The steps included: converting dicom data into recognizable nii format, removing the first 10 data that may have disturbances, time correction, selecting image reference layers, head motion correction, skull removal, image registration, segmentation, spatial normalization, spatial smoothing, etc.

[0106] 1.4 Selection of brain regions of interest

[0107] The DK brain atlas (Desikan-KillianyAtlas), which can preserve the anatomical structure of sulci and gyri at a macroscopic level, was selected as a template for dividing and extracting brain regions. The template divides the left and right brain into 36 brain regions each. Each brain region has a corresponding identification number, and according to the number, the 26th brain region, the precuneus brain region, was extracted.

[0108] 1.5 Analysis of functional connectivity characteristics

[0109] There is a BOLD signal data set collected from the whole brain, and the preprocessed BOLD signal of each vertex in the precuneus is extracted. Using the Hilbert transform, the BOLD signal is transformed into an analytical signal to obtain the amplitude and instantaneous phase of the signal. The phase difference between each vertex can be calculated from the instantaneous phase. Through functional connectivity feature analysis, the phase synchronization coefficient between the BOLD signals of each vertex in the precuneus of each child is calculated. If there are n vertices in the precuneus of the left brain, the phase synchronization coefficient that can be obtained is n*(n-1) / 2. The obtained phase synchronization coefficient ps is used as the basis for segmenting the precuneus.

[0110] 1.6 Segmentation of the precuneus brain region.

[0111] The phase synchronization coefficient matrix obtained in the previous step will be used as the basis for further segmentation of the precuneus brain region. The present invention uses three community segmentation algorithms: eigenvalue, multi-level, and greedy. Modularity is used as a method to determine the effect of network community segmentation. By comparing the modularity values, the brain region segmentation algorithm with the best classification effect can be selected to segment the precuneus brain region. The segmentation results of the left and right brains are not necessarily the same.

[0112] The process of obtaining a unified template for the segmentation of the precuneus brain region is to collect the BOLD signals of the precuneus brain region of a large number of normal children, analyze the functional connectivity characteristics of the vertices in each subject's brain region, that is, the phase synchronization coefficient matrix of each person. The phase synchronization coefficient matrix of the test children is averaged and used as the basis for the segmentation of the precuneus brain region. The community segmentation algorithm with the best classification effect is selected to segment the precuneus brain region. The segmentation of the left precuneus brain region obtains p subregions, and the segmentation of the right precuneus brain region obtains q subregions.

[0113] When a child subject is added in the specific embodiment, the precuneus brain region needs to be segmented according to this template. Based on the segmentation results obtained, the BOLD signal is further analyzed.

[0114] 1.7 Analyze differences within and between brain regions.

[0115] After the precuneus brain region was segmented, the average BOLD signal of each subregion was calculated, and the functional connection characteristics between subregions were analyzed. The average BOLD signal of other brain regions can also be calculated to analyze the functional connection characteristics of each subregion with other brain regions. Statistical test methods were used to compare and analyze the differences between the control group of normal children and the experimental group of ADHD children to find out the precuneus subregions with significant differences.

[0116] Or based on the low-frequency amplitude ALFF and detrended DFA analysis methods that are commonly used to analyze brain signals, the phase difference sequence of the BOLD signal in the precuneus brain region is analyzed. The low-frequency amplitude method requires low-pass filtering and Hilbert transform of the average BOLD signal of each precuneus subregion to obtain the phase value at each time point. The phase difference Δθ of the BOLD signal between subregions is calculated, and this value is used as the "amplitude" of the low-frequency amplitude method analysis. Then the average phase difference of each subregion is Divide by the average phase difference of the precuneus brain area to obtain the standardized low-frequency phase difference value Δθ divided by the mean normalized By calculating the standardized difference of phase between different subregions, the intergroup differences between the ADHD children experimental group and the healthy children control group were compared to find out the precuneus subregions with significant differences between the two groups.

[0117] Or, the internal connection can be found by analyzing the BOLD signal phase difference time series based on the detrending method. The phase difference time series is obtained by Hilbert transforming and calculating the phase difference value of the BOLD signal between the precuneus subregions. The detrended data can be obtained by subtracting the phase difference value of each time point from the value of the fitting curve at the corresponding time point. The statistical analysis method is used to analyze the precuneus subregions with significant differences by comparing the detrended phase difference data between the disease group and the normal group.

[0118] 1.8 Establish a prediction model.

[0119] Children's attention is assessed in the form of a professional questionnaire, and their attention assessment scores are recorded. This score is used as the dependent variable y in the prediction model. The phase synchronization coefficient matrix between the precuneus subregions in step five is extracted, and this coefficient is used as the independent variable x in the prediction model. Based on the known dependent variable y and independent variable x, the prediction model is trained using methods such as Lasso and SVR to obtain the optimal model parameters. In this way, the function of predicting children's attention assessment scores can be achieved by only measuring the BOLD signal in the precuneus brain area of ​​the child and calculating the phase synchronization coefficient matrix. This function can be used to analyze the information of the precuneus subregions that mainly affect children's attention.

[0120] When the model is needed to predict the attention score of a new subject, the BOLD signal time series of the precuneus region of the subject at rest after preprocessing is obtained according to the above steps. The precuneus region is segmented according to a unified segmentation template. The average BOLD signal time series of each subregion after segmentation and the phase synchronization coefficient between subregions are calculated and input into the prediction model to predict the attention score of the subject.

[0121] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions and variations of these embodiments are made without departing from the principles and spirit of the present invention, and still fall within the scope of protection of the present invention.

Claims

1. A method for evaluating children's attention based on phase synchronization of precuneus subregions, characterized by: The following steps are involved: Step 1: Image acquisition, collecting fMRI image data of the children under study; Step 2: fMRI data preprocessing, preprocessing the fMRI data obtained in step 1; Step 3: Selection of brain regions of interest: Using the DK brain atlas as a template for dividing and extracting brain regions, the precuneus was selected as the brain region of interest; Step 4: Functional connectivity feature analysis, calculate the phase synchronization coefficient of the BOLD signal of the precuneus brain area of ​​interest; Step 5: Segment the precuneus brain region, using the result obtained in step 4 as the condition for segmenting the brain region of interest; Step 6: Analyze the differences within the subregions of the precuneus and between the subregions, analyze the functional connectivity characteristics of the BOLD signals within the subregions of the precuneus and between the subregions, and use different methods to compare the differences between the disease group and the normal group; Step 7: Establish a prediction model by recording the children's attention assessment scores and the phase synchronization coefficient matrix of the precuneus subregions obtained after segmentation, train the prediction model, and synchronously predict the children's attention scores; In the step 6, the functional connectivity characteristics of BOLD signals between the subregions of the precuneus and between the subregions and other brain regions are analyzed, or the BOLD signal phase difference sequence is analyzed based on the low-frequency amplitude method and the detrending method, and the significant differences in brain regions between the children's disease group and the normal group are analyzed by statistical test methods; In step 7, the BOLD signal phase synchronization value x between the precuneus subregions ij The prediction model for the child's attention score y is: Where: y represents the attention evaluation score; x ij It represents the functional connection characteristics between the segmented precuneus subregion and the j subregion, that is, the phase synchronization coefficient.

2. The method for evaluating children's attention based on phase synchronization of precuneus subregions according to claim 1, characterized in that: The data preprocessing in step 2 is to use FreeSurfer or CAT12 software based on SPM in MATLAB to preprocess the rs-fMRI data of the test children.

3. The method for evaluating children's attention based on phase synchronization of precuneus subregions according to claim 2, characterized in that: In the step 2, the fMRI data is collected using a gradient echo-planar echo imaging sequence, with the following specific parameters: 47 layers, a repetition time TR of 2500 ms, an echo time TE of 30 ms, a resolution Res of 3.0*3.0 mm^2, a slice thickness slice thickness of 3.0 mm, and a rotation angle FA of 75°.

4. The method for evaluating children's attention based on phase synchronization of precuneus subregions according to claim 3, characterized in that: The fMRI data preprocessing in step 2 specifically includes the following steps: S2.1 Time correction: Using the interval scanning method of magnetic resonance images, the time of each body is corrected during the processing and analysis of resting state magnetic resonance data so that all bodies have the same time point; S2.2 Head motion correction: Select the first image as the reference image, match the positions of the remaining images with the first image by rotation or translation, and resample these images using the corresponding algorithm; S2.3 Image registration: Register the individual images of the subjects to make them uniform in size; S2.4 Spatial standardization: large-scale unified processing and analysis of patient data; S2.5 spatial smoothing: A filtering process based on Gaussian random field theory, collecting BOLD signal frequencies at 0.01-0.08 Hz. Through filtering, the high-frequency components in the signal are eliminated or suppressed to reduce image noise.

5. The method for evaluating children's attention based on phase synchronization of precuneus subregions according to claim 1, characterized in that: In the step three, the DK brain atlas is used as a template for dividing and extracting brain regions, and the precuneus brain region is selected as the region of interest.

6. The method for evaluating children's attention based on phase synchronization of precuneus subregions according to claim 1, characterized in that: In the step 4, the BOLD signal phase synchronization coefficient between vertices in the precuneus brain region is calculated.

Images