Dynamic brain network analysis system for sleep deprivation injury alertness attention mechanism

Abstract

The invention discloses a dynamic brain network analysis system for a sleep deprivation injury alertness attention mechanism, and relates to the technical field of medical image processing and analysis. The method comprises the following steps: collecting data of a sober-up state and after 24 hours of sleep deprivation of a subject; a dynamic graph theory method is adopted, dynamic brain network analysis is carried out on resting state functional magnetic resonance data, and brain network changes in a waking state and a sleep deprivation state are studied; the method specifically comprises the steps of collecting fMRI and mental movement alertness task data of a subject in different states, preprocessing image data, calculating a static function connection matrix, and analyzing time-varying function connection by using a sliding window technology; through modular analysis and a time co-occurrence matrix, time flexibility, space-time diversity and normalization centrality are calculated, dynamic space-time characteristics of a brain region are evaluated, and dynamic influences of sleep deprivation on a brain network are compared by using a statistical method.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing and analysis, and in particular to a dynamic brain network analysis system for a sleep deprivation-damaged alertness and attention mechanism. Background Art

[0002] Sleep is an important physiological phenomenon of the human body. Normal sleep is essential to maintaining a healthy life state and basic life activities. Sleep quality directly affects behavior, cognition, and physical and mental health. However, with the accelerated pace of life and the increasing pressure from external factors such as society, environment, and work style, sleep deprivation (SD) has become a common social phenomenon, but its underlying neural mechanism is still not fully understood.

[0003] Traditional fMRI-based studies on the mechanisms of sleep deprivation usually assume that brain activity is stable throughout the scan; however, many brain networks in practical applications have dynamic characteristics, and their topological structures and the connection relationships between nodes change over time; this static analysis method cannot fully capture these dynamic changes, and therefore fails to fully reflect the true state of the brain network. The emergence of dynamic brain network analysis fills this research gap, allowing us to better understand the changes in the network in time and space dimensions; this method has received increasing attention in characterizing the functional connectivity of neural networks and how network connectivity is impaired in neuropsychiatric diseases; in the brain network analysis of dynamic functional connectivity, we studied 9 networks covering the entire brain Temporal flexibility, spatiotemporal diversity and normalized centrality; temporal flexibility quantifies the frequency of interaction between a brain region and other community regions over time, and high temporal flexibility indicates that the region mainly interacts with regions outside its own community; spatiotemporal diversity quantifies the uniformity of interaction between a brain region and other community regions over a period of time, and high spatiotemporal diversity indicates that such interaction is more evenly distributed throughout the community; normalized centrality measures the importance or influence of nodes in the network, which can help understand the relative importance of each region in the brain network in the overall network without being affected by the size or structure of the network. To address the above problems, the inventors propose a dynamic brain network analysis system for the alertness and attention mechanism impaired by sleep deprivation to solve the above problems. Summary of the invention

[0004] In order to solve the problem that static analysis methods cannot fully capture these dynamic changes and therefore fail to fully reflect the actual brain network state; the purpose of the present invention is to provide a dynamic brain network analysis system for the alertness and attention mechanism damaged by sleep deprivation.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solution: a dynamic brain network analysis system for the mechanism of alertness and attention damaged by sleep deprivation, comprising the following steps:

[0006] S1: The resting-state functional magnetic resonance imaging (fMRI) data and psychomotor vigilance test (PVT) data of the subjects were collected while awake and 24 hours after sleep deprivation. The PVT data used the psychomotor vigilance task to measure sustained attention, which is the cognitive area most affected by sleep deprivation.

[0007] S2: Organize and analyze behavioral data;

[0008] S3: For image data preprocessing, the data of this study were preprocessed using statistical parametric mapping (SPM12) and the functional connectivity toolbox (CONN version 17f, https: / / www.nitrc);

[0009] S4: extract node time series and calculate static functional connectivity between nodes;

[0010] S5: Use a sliding window with a window length of 30TR and a step length of 1TR for the time series of the node, and apply an exponentially decaying weight to obtain a time-varying functional connectivity matrix; for the window length, we set three cases of 10 / 20 / 30TR to explore the impact of the window length on the results;

[0011] S6: To investigate the dynamic interactions between brain nodes, we performed a modular analysis of the functional connectivity matrix within each sliding window for each participant and calculated the temporal co-occurrence matrix;

[0012] S7: Based on the temporal co-occurrence matrix and the static functional connectivity matrix, three indices, namely temporal flexibility, spatiotemporal diversity and normalized centrality within the community, were calculated to characterize the dynamic spatiotemporal characteristics of the brain;

[0013] S8: To analyze the neural mechanisms of sleep deprivation, statistical analysis of dynamic brain networks and behavior was performed; specifically, paired T tests were first performed on behavioral data and imaging data; then, stepwise regression analysis was performed on dynamic brain network indicators in significant brain regions to explore their relationship with behavioral PVT indicators. This process aims to provide new insights into the neurobiological basis of sleep deprivation and to find potential biomarkers.

[0014] Preferably, the steps of collecting the resting-state functional magnetic resonance imaging (fMRI) data and psychomotor vigilance test (PVT) data of the subjects are as follows: first, the resting-state functional magnetic resonance imaging (rs-fMRI) data and PVT data of the subjects are collected: the subjects will have normal sleep or 24 hours of sleep deprivation in a random crossover manner, and the data will be collected at 8 o'clock in the morning of the next day; for the collection of PVT data, this study uses a psychomotor vigilance task, requiring the subjects to press a button when a red target circle appears, and the changes in their vigilance attention within 7 minutes are evaluated through real-time reaction time, with particular attention to the impact of sleep deprivation on them.

[0015] Preferably, PVT is a test tool commonly used to evaluate attention and reaction time; the main behavioral indicators used in PVT are: number of lapses (Lapse) (lapse is defined as reaction time RT>500ms); average reaction time of each trial (Mean_RT); 10% fastest reaction time (RT_fast_10); 10% slowest reaction speed (RT_slow_10_inverse); standard deviation of reaction time (sd_RT); coefficient of variation of reaction time (cv_RT).

[0016] Preferably, the steps of preprocessing the image data are as follows: 1) to maintain signal stability, remove the data of the first five brains; 2) functional image motion correction, use the SPM12 alignment program to align the remaining fMRI data to the middle brain for head motion correction; to control the influence of head motion on the results, exclude subjects whose translation exceeds 2 mm or rotation exceeds 2° during the scanning process; 3) the realigned functional data is centered at the (0,0,0) coordinate; 4) time layer correction, according to the setting of the fMRI scanning sequence parameters, the order between the fMRI layers in time is Misaligned, time-corrected using the SPM12 time slice correction program; 5) Functional image outlier detection, according to the middle setting (97th percentile in the standard sample), time points with global blood oxygen level dependent (BOLD) signal changes higher than 5 standard deviations and frame displacement higher than 0.9 mm were marked as outliers; 6) Functional image segmentation and normalization, using the SPM12 unified segmentation and normalization program, the time-slice-corrected fMRI data were segmented into gray matter, white matter, and cerebrospinal fluid (CSF), and normalized to the standard space MNI, with a voxel size of 2×2×2mm 3 ; 7) 3D T1-weighted structural images are centered at (0,0,0) coordinates; 8) Structural image segmentation and standardization; SPM12 unified segmentation and standardization procedures are used to segment the structural images into gray matter, white matter, and cerebrospinal fluid, and normalized to the standard space MNI, with a voxel size of 2×2×2mm3 ; 9) Spatial smoothing: In order to improve the BOLD signal-to-noise ratio, a Gaussian kernel with a half-width of 8 mm was used to perform spatial smoothing on the functional image data; 10) A noise correction procedure based on structural components was used to eliminate potential interference factors in the BOLD signal, and a temporal bandpass filter was performed with a frequency window of 0.008-0.09 Hz to focus on low-frequency fluctuations and minimize the influence of physiological, head movement and other noise sources; finally, linear debulking was performed to eliminate the linear trend in each functional data.

[0017] Preferably, the static functional connectivity matrix is ​​calculated based on the functional image of the subject, specifically including: using the AAL3 brain template to divide a total of 152 brain regions; calculating the Pearson correlation coefficient between the brain nodes of each subject to obtain the static functional connectivity matrix between the nodes of each subject; performing a z-transform on the static functional connectivity matrix, and calculating the average value of all subjects, representing the group average functional connectivity strength, to obtain the static functional connectivity matrix.

[0018] Preferably, a sliding window is added to the time series to calculate the time-varying functional connectivity matrix, specifically comprising: applying the sliding window to the time series with a step size of 1TR, and applying an exponential decay weight to each time point in the window; the exponential decay weight is calculated as follows: W t =W o e (t-T) / θ , t=1,…T,where W o =(1-e -1 / θ ) / (1-e -T / θ ), t is the t-th time point in the sliding window, T is the window length, and the exponent θ is usually set to one third of the window length; by calculating any two nodes X t and Y t The weighted Pearson correlation coefficient between the time series of is used to obtain the connectivity matrix under each window: in and The connectivity matrix of each window was z-transformed within the subject for subsequent analysis; three different sliding window lengths (10 / 20 / 30TR) were set to analyze their effects on sleep deprivation.

[0019] Preferably, the community structure is calculated based on the time-varying functional connectivity matrix, specifically including: the Louvain community detection algorithm in BrainConnectivityToolbox is used to identify the community structure in the static and time-varying connectivity matrices; the Louvain community detection algorithm performs community detection by optimizing the quality function Q, Where m is the number of edges in the network, A ij is the connection weight (edge ​​weight) between node i and node j, k iis the degree of node i, that is, the number of edges connecting node i, k is the degree of node j, δ(c i , c j ) is an indicator function, which takes the value of 1 when nodes i and j belong to the same community, and takes the value of 0 otherwise.

[0020] Preferably, in order to study the dynamic interactions between nodes, the temporal co-occurrence matrix is ​​calculated based on the best community structure of the subjects obtained from the time-varying connectivity matrix; the community structure within each sliding window is used to construct the adjacency matrix A ijtk , if node i and node j are in the same community within time window t of participant k, then A ijtk =1, otherwise A ijtk =0; Calculate the average of the adjacency matrix over time, which is the temporal co-occurrence matrix Each element measures the proportion of times two nodes belong to the same community; higher values ​​indicate that the two corresponding nodes participate more often in the same community.

[0021] Preferably, the temporal co-occurrence matrix C is used ijk and group static communities, calculating three indices, namely temporal flexibility, spatiotemporal diversity, and normalized centrality within the community, to characterize the dynamic spatiotemporal properties of the brain.

[0022] Preferably, the dynamic brain network and behavior are statistically analyzed to explore the neural mechanism of sleep deprivation, including: first, a paired T test is performed on the six PVT behavioral indicators in the awake and sleep deprived states, and FWE multiple comparison correction is performed; secondly, the three indicators of temporal flexibility, spatiotemporal diversity, and normalized centrality are evaluated from the two aspects of brain regions and networks to evaluate the effects of sleep deprivation on brain dynamics. In terms of brain regions: the paired t-test (p<0.05 / number of nodes) is used to compare the changes in temporal flexibility, spatiotemporal diversity, and normalized centrality before and after sleep deprivation; in terms of networks: the 152 brain regions under the AAL3 atlas are divided into 9 brain networks, and the paired t-test (p<0.05 / number of networks) is used to compare the changes in the three indicators before and after sleep deprivation; finally, the stepwise regression analysis method is used to explore the correlation between the PVT indicators and the significant brain regions.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] The present invention adopts a dynamic brain network method, and through dynamic brain network analysis of resting-state functional magnetic resonance (fMRI) data and psychomotor vigilance task (PVT) data, the changes of brain networks under awake and sleep deprivation states are studied; the specific method includes collecting fMRI and PVT data of subjects in different states, preprocessing the image data, calculating the static functional connection matrix, and using the sliding window technology to analyze the time-varying functional connection; through modular analysis and temporal co-occurrence matrix, three indicators of temporal flexibility, spatiotemporal diversity, and normalized centrality are calculated to evaluate the dynamic spatiotemporal characteristics of brain regions, and statistical methods are used to compare the effects of sleep deprivation on brain regions and neural networks; the study reveals the specific effects of sleep deprivation on the spatiotemporal characteristics of brain networks, and explores the relationship between the cognitive indicator PVT and brain region changes through stepwise regression analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] 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.

[0026] Figure 1 This is a framework diagram of the dynamic brain network analysis of the present invention.

[0027] Figure 2 It is a schematic diagram of the magnetic resonance data preprocessing structure of the present invention.

[0028] Figure 3 This is a comparison result diagram of behavioral data of the present invention.

[0029] Figure 4 This is the result map of significant brain areas under different window lengths of the present invention.

[0030] Figure 5 This is a network comparison result diagram of time flexibility before and after sleep deprivation of the present invention.

[0031] Figure 6 This is a network comparison result diagram of spatiotemporal diversity before and after sleep deprivation of the present invention.

[0032] Figure 7 This is a network comparison result diagram of normalized centrality before and after sleep deprivation of the present invention.

[0033] Figure 8 This is a result diagram of the stepwise linear regression based on brain regions of the present invention.

[0034] Fig. 9 This is a result diagram of the network-based stepwise linear regression of the present invention. DETAILED DESCRIPTION

[0035] 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.

[0036] Example: Figure 1-9 As shown, the present invention provides a technical solution: a dynamic brain network analysis system for the alertness attention mechanism damaged by sleep deprivation, comprising the following steps:

[0037] S1: The resting-state functional magnetic resonance imaging (fMRI) data and psychomotor vigilance test (PVT) data of the subjects were collected while awake and 24 hours after sleep deprivation. The PVT data used the psychomotor vigilance task to measure sustained attention, which is the cognitive area most affected by sleep deprivation.

[0038] S2: Organize and analyze behavioral data;

[0039] S3: For image data preprocessing, the data of this study were preprocessed using statistical parametric mapping (SPM12) and the functional connectivity toolbox (CONN version 17f, https: / / www.nitrc);

[0040] S4: extract node time series and calculate static functional connectivity between nodes;

[0041] S5: Use a sliding window with a window length of 30TR and a step length of 1TR for the time series of the node, and apply an exponentially decaying weight to obtain a time-varying functional connectivity matrix; for the window length, we set three cases of 10 / 20 / 30TR to explore the impact of the window length on the results;

[0042] S6: To investigate the dynamic interactions between brain nodes, we performed a modular analysis of the functional connectivity matrix within each sliding window for each participant and calculated the temporal co-occurrence matrix;

[0043] S7: Based on the temporal co-occurrence matrix and the static functional connectivity matrix, three indices, namely temporal flexibility, spatiotemporal diversity and normalized centrality within the community, were calculated to characterize the dynamic spatiotemporal characteristics of the brain;

[0044] S8: To analyze the neural mechanisms of sleep deprivation, statistical analysis of dynamic brain networks and behavior was performed; specifically, paired T tests were first performed on behavioral data and imaging data; then, stepwise regression analysis was performed on dynamic brain network indicators in significant brain regions to explore their relationship with behavioral PVT indicators. This process aims to provide new insights into the neurobiological basis of sleep deprivation and to find potential biomarkers.

[0045] The steps for collecting the subjects' resting-state functional Magnetic Resonance Imaging (fMRI) data and psychomotorvigilance test (PVT) data are as follows: first, the subjects' resting-state functional Magnetic Resonance Imaging (rs-fMRI) data and PVT data are collected: the subjects will have normal sleep or 24-hour sleep deprivation in a random crossover manner, and the data will be collected at 8 am the next day; for the collection of PVT data, this study uses a psychomotor vigilance task, requiring the subjects to press a button when a red target circle appears, and the changes in their vigilance attention within 7 minutes are evaluated through real-time reaction time, with particular attention to the impact of sleep deprivation on them.

[0046] Participants were selected by adopting the above technical scheme, specifically recruiting right-handed healthy young subjects, aged over 18 years old; all participants had normal sleep time, 7-9 hours per night, and the sleep time was between 10 pm and 8 am; they had no history of smoking, no self-reported medical, psychiatric, neurological or sleep disorders, and no alcohol or drug abuse; all participants declared that they would not smoke or consume any stimulants, drugs, alcohol or caffeine (coffee, tea, cola, etc.) for at least 24 hours before the formal experiment; all subjects provided written informed consent before participation and were compensated for their time; all research procedures were carried out in accordance with the Declaration of Helsinki.

[0047] By adopting the above technical scheme, this study used the psychomotor vigilance test (PVT) to measure vigilant attention, which is the cognitive area most affected by sleep deprivation. First, a red fixation cross will appear in the center of the screen with a black background for 2 seconds. Then, the red cross disappears, and the screen will remain black and appear randomly for 2 to 10 seconds. Then, a red target circle appears, and the subject is required to press a button with the right index finger as quickly as possible. The subject is required to complete this action within 30 seconds. When the subject responds, the red target circle will disappear immediately, and the response time (RT) will be displayed on the screen in real time to record the subject's performance. Within 1 second after the response occurs, feedback will appear quickly to provide the subject with an immediate evaluation. If the subject fails to respond within the set time range, the real-time RT displayed is 0 milliseconds, indicating the lack of this response. The entire task process lasts about 7 minutes.

[0048] The main behavioral indicators used in PVT are: number of lapses (Lapse) (lapse is defined as reaction time RT>500ms); average reaction time of each trial (Mean_RT); 10% fastest reaction time (RT_fast_10); 10% slowest reaction speed (RT_slow_10_inverse); standard deviation of reaction time (sd_RT); coefficient of variation of reaction time (cv_RT).

[0049] The steps for image data preprocessing are as follows (e.g. Figure 2As shown in the figure, 1) to maintain signal stability, the data of the first 5 brains were removed; 2) functional image motion correction, the SPM12 alignment program was used to align the remaining fMRI data to the middle brain for head motion correction; to control the influence of head motion on the results, the subjects whose translation exceeded 2 mm or rotation exceeded 2° during the scanning process were excluded; 3) the realigned functional data were centered at (0,0,0) coordinates; 4) time layer correction, according to the setting of fMRI scanning sequence parameters, the order between fMRI layers was misaligned in time, and SPM was used to align the remaining fMRI data to the middle brain for head motion correction; to control the influence of head motion on the results, the subjects whose translation exceeded 2 mm or rotation exceeded 2° during the scanning process were excluded; 5) the realigned functional data were centered at (0,0,0) coordinates; 6) time layer correction, according to the setting of fMRI scanning sequence parameters, the order between fMRI layers was misaligned in time, and SPM was used to align the remaining fMRI data to the middle brain for head motion correction; to control the influence of head motion on the results, the subjects whose translation exceeded 2 mm or rotation exceeded 2° during the scanning process were excluded ... 12 temporal slice correction procedures were used for time correction; 5) functional image outlier detection: according to the intermediate setting (97th percentile in the standard sample), time points with global blood oxygen level-dependent (BOLD) signal changes higher than 5 standard deviations and frame displacement higher than 0.9 mm were marked as outliers; 6) functional image segmentation and normalization: using the SPM12 unified segmentation and normalization procedure, the fMRI data after temporal slice correction were segmented into gray matter, white matter and cerebrospinal fluid (CSF) and normalized to the standard space MNI with a voxel size of 2×2×2mm3; 7) 3D The T1-weighted structural image is centered at the (0,0,0) coordinate; 8) Structural image segmentation and standardization: The SPM12 unified segmentation and standardization program was used to segment the structural image into gray matter, white matter, and cerebrospinal fluid, and normalized to the standard space MNI with a voxel size of 2×2×2mm3; 9) Spatial smoothing: In order to improve the BOLD signal-to-noise ratio, a Gaussian kernel with a half-maximum full width of 8mm was used to perform spatial smoothing on the functional image data; 10) A noise correction procedure based on the structural component was used to eliminate potential interference factors in the BOLD signal, and a time bandpass filter was performed with a frequency window of 0.008-0.09 Hz to focus on low-frequency fluctuations and minimize the influence of physiological, head motion, and other noise sources; finally, linear debulking was performed to eliminate the linear trend in each functional data.

[0050] According to the functional image of the subject, the static functional connectivity matrix (such as Figure 1 The method includes: using the AAL3 brain template to divide a total of 152 brain regions; calculating the Pearson correlation coefficient between the brain nodes of each subject to obtain the static functional connectivity matrix between the nodes of each subject; performing a z-transform on the static functional connectivity matrix, and calculating the average value of all subjects to represent the group average functional connectivity strength, and obtaining the static functional connectivity matrix.

[0051] A sliding window is added to the time series to calculate the time-varying functional connectivity matrix, which includes: applying the sliding window to the time series with a step size of 1TR, and applying an exponential decay weight to each time point in the window; the exponential decay weight is calculated as follows: W t =W o e (t-T) / θ , t=1,…T, where W o=(1-e -1 / θ ) / (1-e -T / θ ), t is the t-th time point in the sliding window, T is the window length, and the exponent θ is usually set to one third of the window length; by calculating any two nodes X t and Y t The weighted Pearson correlation coefficient between the time series of is used to obtain the connectivity matrix under each window: in and The connectivity matrix of each window was z-transformed within the subject for subsequent analysis; three different sliding window lengths (10 / 20 / 30TR) were set to analyze their effects on sleep deprivation.

[0052] The community structure is calculated based on the time-varying functional connectivity matrix, including: the Louvain community detection algorithm in BrainConnectivityToolbox is used to identify the community structure in static and time-varying connectivity matrices; the Louvain community detection algorithm performs community detection by optimizing the quality function Q. Where m is the number of edges in the network, A ij is the connection weight (edge ​​weight) between node i and node j, k i is the degree of node i, that is, the number of edges connecting node i, k is the degree of node j, δ(c i , c j ) is an indicator function, which takes the value of 1 when nodes i and j belong to the same community, and takes the value of 0 otherwise.

[0053] In order to study the dynamic interactions between nodes, the temporal co-occurrence matrix is ​​calculated based on the best community structure of the subjects obtained from the time-varying connectivity matrix; the community structure within each sliding window is used to construct the adjacency matrix A ijtk , if node i and node j are in the same community within time window t of participant k, then A ijtk =1, otherwise A ijtk =0; Calculate the average of the adjacency matrix over time, which is the temporal co-occurrence matrix Each element measures the proportion of times two nodes belong to the same community; higher values ​​indicate that the two corresponding nodes participate more often in the same community.

[0054] By adopting the above technical solution, using the time co-occurrence matrix C ijk The temporal flexibility, spatiotemporal diversity, and normalized centrality within the community are calculated to characterize the dynamic spatiotemporal properties of the brain. The temporal flexibility of node i and subject k is calculated as: Among them C ijk is the temporal co-occurrence matrix of individual k, u iis the community to which node i belongs, measures the total frequency of node i's interactions with nodes outside its local community, Measures the total interactions with all nodes. The temporal flexibility of a node captures the tendency of a node to deviate from its local community and interact with nodes in other communities.

[0055] The spatiotemporal diversity of node i and participant k is calculated as: in s ik is the strength of node i of participant k in all communities, s ik (u), is the strength of node i in community u for participant k, m is the total number of communities, and M is the set of communities. Nodes with high spatiotemporal diversity scores are those whose distribution of time-varying interactions with all communities varies relatively spatially.

[0056] The normalized intra-community centrality of node i and participant k is calculated as: Where Z ik is the normalized community centrality of node i and participant k, m i is the community containing node i, s ik (m i ) is the community m i The strength of internal node i for participant k, It is a community i The average strength of all nodes in for individual k, and It is a community i The standard deviation of inlier strength for participant k.

[0057] By adopting the above technical solution, the present invention evaluates the changes in the whole brain and network scales under three different indicators of temporal flexibility, spatiotemporal diversity and normalized centrality at two time points of wakefulness and sleep deprivation, and explores the impact of different window lengths on the results. Figure 4 A, B, and C are the comparison results of significant brain regions under three indicators with window lengths of 10 / 20 / 30TR. The number of significant brain regions under temporal flexibility is 146 / 136 / 126, the number of significant brain regions under spatiotemporal diversity is 135 / 110 / 87, and the number of significant brain regions under normalized centrality is 59 / 58 / 56. By comparing the results of significant brain regions under the three window lengths, it is found that the window length has little effect on the results, so the window length used in this study is 30TR.

[0058] By adopting the above technical scheme, the dynamic brain network and behavior were statistically analyzed to explore the neural mechanism of sleep deprivation, including: firstly, the six PVT behavioral indicators in the awake and sleep deprivation states were subjected to paired T test, and FWE multiple comparison correction was performed. Then, the significant brain area data were normalized with the PVT data, and then a stepwise regression analysis was performed, in which the independent variable was the PVT data and the dependent variable was the brain area data.

[0059] Statistical analysis of dynamic brain networks and behaviors was conducted to explore the neural mechanisms of sleep deprivation, including: first, paired T tests were performed on the six PVT behavioral indicators in the awake and sleep deprived states, and FWE multiple comparison correction was performed (P<0.05 / 6); secondly, the three indicators of temporal flexibility, spatiotemporal diversity, and normalized centrality were used to evaluate the effects of sleep deprivation on brain dynamics from two aspects: brain regions and networks. In terms of brain regions: paired t tests (p<0.05 / number of nodes) were used to compare the changes in temporal flexibility, spatiotemporal diversity, and normalized centrality before and after sleep deprivation; in terms of networks: 152 brain regions under the AAL3 atlas were divided into It was divided into 9 brain networks, including visual network (VN), sensorimotor network (SMN), dorsal attention network (DAN), ventral attention network (VAN), limbic network (LN), fronto-parietal network (FPN), default mode network (DMN), subcortical network (SUB) and cerebellar network (CN). The paired t-test (p<0.05 / number of networks) was used to compare the changes of the three indicators before and after sleep deprivation; finally, the data of significant brain regions and PVT data were normalized, and then a stepwise regression analysis was performed, in which the independent variable was PVT data and the dependent variable was brain region data, thereby exploring the correlation between PVT indicators and significant brain regions.

[0060] By adopting the above technical solution, it was found that the behavioral performance was significantly worse after sleep deprivation ( Figure 3 ). The specific performance is: rt_fast_10 slows down (P = 1.34x10 -6 , Figure 3 (a)), cv_RT increased significantly (P = 3.7x10 -3 , Figure 3 (b)), MeanRT increased significantly (P = 1.43x10 -9 , Figure 3 (c)), Lapse_ratio increased significantly (P = 3.16x10 -10 , Figure 3 (d)), sd_rt increased significantly (P = 2.73x10 -6 , Figure 3 (e)), rt_slow_10_inverse speed is significantly slower (P = 2.70x10 -6 , Figure 3 (f)). In terms of dynamic brain networks, after paired T test (P<0.05 / 152=0.00033), the temporal and spatial diversity of all brain regions decreased after sleep deprivation compared with the awake state; the temporal flexibility increased in all regions except for 36 regions such as the caudate nucleus, the right superior temporal pole, and some thalamic regions; the normalized centrality decreased except for the left and right medial nuclei and the central lateral nucleus of the thalamus. The average values ​​of the 152 brain regions in the AAL3 template were divided into networks, and the average temporal flexibility between each network node was calculated. A total of 9 brain networks were calculated. We used the P value of 0.05 / 9=0.0056 (FWE correction) to evaluate the results of the paired T test. The results showed that compared with RW, the temporal flexibility of the SD group, except for the subcortical network, was increased in other neural networks. The temporal and spatial diversity of neural networks was reduced ( Figure 5-7 ). Except for the dorsal attention network (DAN) and the subcortical network (SUB), the normalized centrality was significantly reduced after sleep deprivation. The following is a stepwise regression of the screened significant brain regions and PVT data. In terms of brain regions, by applying the stepwise regression method, the effects of Δcv_rt (coefficient of variation of reaction time) and Δrt_fast_10 (10% fastest reaction time) on the right anterior gyrus parietal cortex under the AAL3 template were explored. It was found that the right anterior gyrus parietal cortex was negatively correlated with the coefficient of variation of reaction time and the 10% fastest reaction time ( Figure 8 ); In terms of network, by applying the stepwise regression method, the effects of Δcv_rt and Δrt_slow_10_inverse on the cerebellar network were explored, and it was found that the cerebellar network was positively correlated with the coefficient of variation of reaction time and the inverse of the 10% slowest reaction time ( Fig. 9 ).

[0061] Working principle: The present invention evaluates the changes in brain regions and networks before and after sleep deprivation by using sliding windows and community detection methods, explores the effects of different window lengths on the results, and calculates temporal flexibility, spatiotemporal diversity, and normalized centrality to understand the time-varying properties of the brain; the results show that compared with wakefulness, in terms of brain regions, the spatiotemporal diversity of all brain regions is reduced after sleep deprivation; the temporal flexibility is increased except for 36 brain regions such as the caudate nucleus, the right upper temporal pole, and some thalamus regions; the normalized centrality is reduced except for the left and right medial nuclei and the central lateral nucleus of the thalamus; in terms of networks, the temporal flexibility of other neural networks is increased after sleep deprivation except for the subcortical network; the spatiotemporal diversity of neural networks is reduced; except for the dorsal attention network (DAN) and the subcortical network (SUB), the normalized centrality is significantly reduced after sleep deprivation; the above results show that the reduction in spatiotemporal diversity indicates that after sleep deprivation, the activity patterns of various brain regions become more single and lack diversity; The general increase in temporal flexibility means that the brain's activity pattern switching speed increases when responding to different tasks or stimuli, which may be an adaptive adjustment made by the brain to compensate for the decline in cognitive function caused by sleep deprivation; the general decrease in normalized centrality indicates that under sleep deprivation, the connectivity and influence of various brain regions are generally weakened; after stepwise regression analysis, it was found that the right anterior gyrus parietal cortex was negatively correlated with the coefficient of variation of reaction time and the 10% fastest reaction time; the cerebellar network was positively correlated with the coefficient of variation of reaction time and the reciprocal of the 10% slowest reaction time; the above results indicate that the right anterior gyrus parietal cortex may be the central brain area for behavioral studies of sleep deprivation damage and may provide new evidence for revealing the neural mechanism of sleep deprivation; and the cerebellar network is positively correlated with PVT, indicating that after sleep deprivation, the cerebellar network may play a role in maintaining a slower but more stable reaction; this may be related to the basic function of the cerebellar network in coordinating movement and maintaining body balance.

[0062] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A dynamic brain network analysis system for the mechanism of alertness and attention damaged by sleep deprivation, characterized in that: The following steps are involved: S1: The resting-state functional magnetic resonance imaging (fMRI) data and psychomotor vigilance test (PVT) data of the subjects were collected while awake and 24 hours after sleep deprivation. The PVT data used a psychomotor vigilance task to measure sustained attention, which is the cognitive area most affected by sleep deprivation. S2: Organize and analyze behavioral data; S3: For image data preprocessing, the data of this study were preprocessed using statistical parametric mapping (SPM12) and the functional connectivity toolbox (CONN version 17f, https: / / www.nitrc); S4: extract node time series and calculate static functional connectivity between nodes; S5: Use a sliding window with a window length of 30TR and a step length of 1TR for the time series of the node, and apply an exponentially decaying weight to obtain a time-varying functional connectivity matrix; for the window length, we set three cases of 10 / 20 / 30TR to explore the impact of the window length on the results; S6: To investigate the dynamic interactions between brain nodes, we performed a modular analysis of the functional connectivity matrix within each sliding window for each participant and calculated the temporal co-occurrence matrix; S7: Based on the temporal co-occurrence matrix and the static functional connectivity matrix, three indices, namely temporal flexibility, spatiotemporal diversity and normalized centrality within the community, were calculated to characterize the dynamic spatiotemporal characteristics of the brain; S8: To analyze the neural mechanisms of sleep deprivation, statistical analysis of dynamic brain networks and behavior was performed; specifically, paired T tests were first performed on behavioral data and imaging data; then, stepwise regression analysis was performed on dynamic brain network indicators in significant brain regions to explore their relationship with behavioral PVT indicators. This process aims to provide new insights into the neurobiological basis of sleep deprivation and to find potential biomarkers.

2. A dynamic brain network analysis system for the mechanism of alertness and attention damaged by sleep deprivation as claimed in claim 1, characterized in that: The steps for collecting the subjects' resting-state functional Magnetic Resonance Imaging (fMRI) data and psychomotor vigilance test (PVT) data are as follows: first, the subjects' resting-state functional Magnetic Resonance Imaging (rs-fMRI) data and PVT data are collected: the subjects will have normal sleep or 24-hour sleep deprivation in a random crossover manner, and the data will be collected at 8 am the next day; for the collection of PVT data, this study uses a psychomotor vigilance task, requiring the subjects to press a button when a red target circle appears, and the changes in their vigilance attention within 7 minutes are evaluated through real-time reaction time, with particular attention to the impact of sleep deprivation on them.

3. A dynamic brain network analysis system for the mechanism of alertness and attention damaged by sleep deprivation as claimed in claim 1, characterized in that: Among them, PVT is a test tool commonly used to evaluate attention and reaction time; the main behavioral indicators used in PVT are: number of errors (Lapse) (error is defined as reaction time RT>500ms); average reaction time of each trial (Mean_RT); 10% fastest reaction time (RT_fast_10); 10% slowest reaction speed (RT_slow_10_inverse); standard deviation of reaction time (sd_RT); coefficient of variation of reaction time (cv_RT).

4. A dynamic brain network analysis system for sleep deprivation-induced damage to alertness and attention mechanism as claimed in claim 1, characterized in that: The steps of image data preprocessing are as follows: 1) to maintain signal stability, remove the data of the first 5 brains; 2) functional image motion correction, use the SPM12 alignment program to align the remaining fMRI data to the middle brain for head motion correction; to control the influence of head motion on the results, exclude subjects whose translation exceeds 2 mm or rotation exceeds 2° during the scanning process; 3) realign the functional data with the (0,0,0) coordinate as the center; 4) time layer correction, according to the setting of fMRI scanning sequence parameters, the order between fMRI layers is misaligned in time 5) Functional image outlier detection: according to the middle setting (97th percentile in the standard sample), the time points with global blood oxygen level dependent (BOLD) signal changes higher than 5 standard deviations and frame displacement higher than 0.9 mm were marked as outliers; 6) Functional image segmentation and normalization: using the SPM12 unified segmentation and normalization program, the fMRI data after time layer correction were segmented into gray matter, white matter and cerebrospinal fluid (CSF), and normalized to the standard space MNI, with a voxel size of 2×2×2mm 3 ;7) 3D T1-weighted structural image is centered at (0,0,0) coordinates; 8) Structural image segmentation and standardization: SPM12 unified segmentation and standardization procedures were used to segment the structural images into gray matter, white matter, and cerebrospinal fluid, and normalized to the standard space MNI with a voxel size of 2×2×2 mm 3 ; 9) Spatial smoothing: In order to improve the BOLD signal-to-noise ratio, a Gaussian kernel with a half-width of 8 mm was used to perform spatial smoothing on the functional image data; 10) A noise correction procedure based on structural components was used to eliminate potential interference factors in the BOLD signal, and a temporal bandpass filter was performed with a frequency window of 0.008-0.09 Hz to focus on low-frequency fluctuations and minimize the influence of physiological, head movement and other noise sources; finally, linear debulking was performed to eliminate the linear trend in each functional data.

5. A dynamic brain network analysis system for the mechanism of alertness and attention damaged by sleep deprivation as claimed in claim 1, characterized in that: The static functional connectivity matrix was calculated based on the functional images of the subjects, specifically including: using the AAL3 brain template to divide a total of 152 brain regions; calculating the Pearson correlation coefficient between the brain nodes of each subject to obtain the static functional connectivity matrix between the nodes of each subject; performing a z-transform on the static functional connectivity matrix, and calculating the average value of all subjects to represent the group average functional connectivity strength, to obtain the static functional connectivity matrix.

6. A dynamic brain network analysis system for sleep deprivation-induced damage to alertness and attention mechanism as claimed in claim 1, characterized in that: A sliding window is added to the time series to calculate the time-varying functional connectivity matrix, which includes: applying the sliding window to the time series with a step size of 1TR, and applying an exponential decay weight to each time point in the window; the exponential decay weight is calculated as follows: W t =W o e (t-T) / θ , t=1,…T,where W o =(1-e -1 / θ ) / (1-e -T / θ ), t is the t-th time point in the sliding window, T is the window length, and the exponent θ is usually set to one third of the window length; by calculating any two nodes X t and Y t The weighted Pearson correlation coefficient between the time series of is used to obtain the connectivity matrix under each window: in and The connectivity matrix of each window was z-transformed within the subject for subsequent analysis; three different sliding window lengths (10 / 20 / 30TR) were set to analyze their effects on sleep deprivation.

7. A dynamic brain network analysis system for sleep deprivation-induced damage to alertness and attention mechanism as claimed in claim 1, characterized in that: The community structure is calculated based on the time-varying functional connectivity matrix, including: the Louvain community detection algorithm in the BrainConnectivity Toolbox is used to identify the community structure in the static and time-varying connectivity matrices; the Louvain community detection algorithm performs community detection by optimizing the quality function Q. Where m is the number of edges in the network, A ij is the connection weight (edge ​​weight) between node i and node j, k i is the degree of node i, that is, the number of edges connecting node i, k is the degree of node j, δ(c i , c j ) is an indicator function, which takes the value of 1 when nodes i and j belong to the same community, and takes the value of 0 otherwise.

8. A dynamic brain network analysis system for sleep deprivation-induced damage to alertness and attention mechanism as claimed in claim 1, characterized in that: In order to study the dynamic interactions between nodes, the temporal co-occurrence matrix is ​​calculated based on the best community structure of the subjects obtained from the time-varying connectivity matrix; the community structure within each sliding window is used to construct the adjacency matrix A ijtk , if node i and node j are in the same community within time window t of participant k, then A ijtk =1, otherwise A ijtk =0; Calculate the average of the adjacency matrix over time, which is the temporal co-occurrence matrix Each element measures the proportion of times two nodes belong to the same community; higher values ​​indicate that the two corresponding nodes participate more often in the same community.

9. A dynamic brain network analysis system for sleep deprivation-induced damage to alertness and attention mechanism as claimed in claim 1, characterized in that: The temporal co-occurrence matrix C is used ijk and group static communities, calculating three indices, namely temporal flexibility, spatiotemporal diversity, and normalized centrality within the community, to characterize the dynamic spatiotemporal properties of the brain.

10. A dynamic brain network analysis system for sleep deprivation-induced damage to alertness and attention mechanism as claimed in claim 1, characterized in that: Among them, statistical analysis of dynamic brain networks and behaviors was conducted to explore the neural mechanisms of sleep deprivation, including: first, paired T tests were performed on the six PVT behavioral indicators in the awake and sleep deprived states, and FWE multiple comparison correction was performed; secondly, the three indicators of temporal flexibility, spatiotemporal diversity, and normalized centrality were used to evaluate the impact of sleep deprivation on brain dynamics from two aspects: brain regions and networks. In terms of brain regions: the paired t-test (p < 0.05 / number of nodes) was used to compare the changes in temporal flexibility, spatiotemporal diversity, and normalized centrality before and after sleep deprivation; in terms of networks: the 152 brain regions under the AAL3 atlas were divided into 9 brain networks, and the paired t-test (p < 0.05 / number of networks) was used to compare the changes in the three indicators before and after sleep deprivation; finally, the stepwise regression analysis method was used to explore the correlation between PVT indicators and significant brain regions.

Images