Difference component identification method and apparatus, electronic device, and medium

Abstract

Certain embodiments of the present disclosure disclose a difference component identification method and device, electronic equipment and readable storage medium. The difference component identification method comprises: obtaining two groups of directed brain network data; processing the two groups of directed brain network data to determine one or more candidate connected components with significant differences; and identifying the difference components from the candidate connected components by performing a permutation test on the two groups of directed brain network data. The technical solution avoids the problem of decreased effectiveness of multiple test correction faced by large-scale statistical test, improves the statistical effectiveness, and can realize identification of difference components of directed brain network connection.

Description

Technical Field

[0001] This invention belongs to the field of brain network technology, and in particular to a method, device, electronic device and readable storage medium for identifying differential components. Background Technology

[0002] Recent advances in neuroscience have revealed that the human brain is a connected network with non-linear, dynamic interactions between different brain regions. Resting-state functional magnetic resonance imaging (rs-fMRI) and electroencephalography (EEG) can non-invasively measure brain neural activity. By defining nodes and edges to construct brain networks, methods from network science can be applied to the analysis of functional neural data. Nodes are also known as regions of interest (ROIs), and edges are also known as connectivity. In addition to synchronicity, human brain neural activity also exhibits directionality, thus allowing the construction of directed brain networks.

[0003] A key challenge in brain network research is identifying significant differences in connectivity pathways between groups. For two sets of brain network samples, performing statistical tests on every connection results in a massive number of tests. Furthermore, multiple comparisons lead to an increased false positive rate. If the statistical significance threshold is set at 0.05, a family-wise error rate (FWER) of 0.994 occurs for 100 statistical hypothesis tests, making at least one Type I error almost unavoidable. Since brain networks typically contain thousands or tens of thousands of connections, directly performing statistical tests on each connection almost certainly results in false positives.

[0004] For multiple detection problems, the Bonferroni method or the false discovery rate (FDR) is typically used for control (Bland and Altman, 1995; Benjamini and Hochberg, 1995). Both methods essentially lower the significance threshold for statistical testing. If n tests are performed simultaneously, the Bonferroni control method lowers the significance threshold to α / n, and FDR uses a similar approach. However, for a vast number of brain network connections, the large number of connections and the large value of n mean that directly using this control method can lead to almost no significant results, failing to effectively balance the false alarm rate and statistical power.

[0005] In summary, existing methods for analyzing differences in group connectivity paths are not suitable for studying directed brain networks. New methods need to be explored. Summary of the Invention

[0006] To address the problems in the related technologies, certain embodiments of this disclosure provide a method, apparatus, electronic device, and readable storage medium for identifying differential components.

[0007] In a first aspect, certain embodiments of this disclosure provide a method for identifying differential components.

[0008] Specifically, the differential component identification method includes:

[0009] Acquire two sets of directed brain network data;

[0010] The two sets of directed brain network data are processed to identify one or more candidate connected components with significant differences;

[0011] By performing a permutation test on the two sets of directed brain network data, differential components are identified from the candidate connected components.

[0012] In conjunction with the first aspect, in a first implementation of the first aspect of this disclosure, processing the two sets of directed brain network data to determine one or more candidate connected components with significant differences includes:

[0013] Each connection of the two sets of directed brain network data was tested using statistical testing methods to obtain a first feature value for each connection, the first feature value including a significant difference statistic and / or a p-value.

[0014] A binary or weighted first difference network is constructed based on the comparison between the first feature value and a predetermined threshold;

[0015] One or more candidate connected components determined from the first differential network.

[0016] In conjunction with the first implementation of the first aspect, in the second implementation of the first aspect of this disclosure, the method further includes determining a statistical test method based on the relationship between the two sets of directed brain network data, wherein:

[0017] Given that the two sets of directed brain network data are independent samples, the statistical test method is determined to be either a two-sample t-test or a non-parametric Mann-Whitney rank-sum test.

[0018] When the two sets of directed brain network data are paired samples, the statistical test method is determined to be either the paired t-test or the nonparametric Wilcoxon signed-rank test.

[0019] In conjunction with the first aspect, in a third implementation of the first aspect, the step of identifying differential components from the candidate connected components by performing a permutation test on the two sets of directed brain network data includes:

[0020] The two groups of directed brain network data were randomly divided into two groups.

[0021] Statistical tests were used to examine each connection of the two regrouped groups of directed brain network data to obtain the third feature value of each connection.

[0022] A second difference network is constructed based on the comparison between the third feature value and the predetermined threshold;

[0023] Identify one or more connected components from the second difference network, and determine the largest connected component from among them;

[0024] After the above operation is repeated a predetermined number of times, the differential components are identified from the candidate connected components based on the maximum connected component determined in each iteration.

[0025] In conjunction with the third implementation of the first aspect, in the fourth implementation of the first aspect of this disclosure, determining one or more connected components from the second difference network includes:

[0026] Treat the directed connections in the second difference network as undirected connections, and identify one or more connected components from the second difference network; or

[0027] Based on one or more predetermined regions of interest, a diverging network and / or a converging network are determined starting from the regions of interest and along directed connections.

[0028] In conjunction with the third implementation of the first aspect, in the fifth implementation of the first aspect of this disclosure, determining one or more connected components from the second difference network includes:

[0029] Starting from any node in the second differential network, search according to directed connections, and put the searched nodes into the stack according to the deepest node priority principle;

[0030] Reverse the direction of the directed connection and mark all nodes as unvisited;

[0031] Nodes are sequentially removed from the stack. If a node is not visited, it is searched in the second differential network with the reversed direction until there are no nodes in the stack, in order to determine one or more connected components.

[0032] In conjunction with the third implementation of the first aspect, in the sixth implementation of the first aspect of this disclosure, the method further includes controlling the predetermined threshold to vary within a certain range to display the difference components of the variation.

[0033] Secondly, in some embodiments of this disclosure, a differential component identification device is provided.

[0034] Specifically, the differential component identification device includes:

[0035] The acquisition module is configured to acquire two sets of directed brain network data.

[0036] The processing module is configured to process the two sets of directed brain network data to determine one or more candidate connected components with significant differences;

[0037] The identification module is configured to identify differential components from the candidate connected components by performing a permutation test on the two sets of directed brain network data.

[0038] Thirdly, certain embodiments of this disclosure provide an electronic device including a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method as described in any one of the first aspect and the first to sixth implementations of the first aspect.

[0039] Fourthly, in certain embodiments of this disclosure, a computer-readable storage medium is provided having computer instructions stored thereon, which, when executed by a processor, implement the method as described in any one of the first aspect and the first to sixth implementations of the first aspect.

[0040] According to the technical solution of this disclosure, two sets of directed brain network data are acquired; the two sets of directed brain network data are processed to determine one or more candidate connected components with significant differences; and differential components are identified from the candidate connected components by performing a permutation test on the two sets of directed brain network data. This technical solution avoids the problem of decreased power correction due to multiple tests in large-scale statistical tests, improves statistical power, and can realize the identification of differential components in directed brain network connectivity.

[0041] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0042] Other features, objects, and advantages of this disclosure will become more apparent from the following detailed description of non-limiting embodiments, taken in conjunction with the accompanying drawings. In the drawings:

[0043] Figure 1 A flowchart of a differential component identification method according to certain embodiments of the present disclosure is shown;

[0044] Figure 2 A flowchart illustrating the determination of candidate connected components according to certain embodiments of this disclosure is shown;

[0045] Figure 3 A flowchart illustrating the identification of differential components according to certain embodiments of this disclosure is shown;

[0046] Figure 4 A flowchart of a differential component identification method according to other embodiments of this disclosure is shown;

[0047] Figure 5 A flowchart of NBS significantly different connected component analysis according to certain embodiments of this disclosure is shown;

[0048] Figure 6 A partial schematic diagram of a brain network according to an embodiment of the present disclosure is shown;

[0049] Figure 7 A flowchart illustrating the search for connected components according to certain embodiments of this disclosure is shown;

[0050] Figure 8A A schematic diagram of a convergence network according to certain embodiments of the present disclosure is shown;

[0051] Figure 8B A schematic diagram of a diverging network according to certain embodiments of the present disclosure is shown;

[0052] Figure 9 A flowchart illustrating the search for connected components is shown for some other embodiments of this disclosure;

[0053] Figure 10A flowchart illustrating the search for connected components is shown for some embodiments of this disclosure;

[0054] Figure 11 A schematic diagram illustrating a differential component identification method applying certain embodiments of the present disclosure is shown;

[0055] Figures 12A-12G A schematic diagram is shown of the differential components obtained by applying the methods of certain embodiments of this disclosure;

[0056] Figure 13 A block diagram of a differential component identification device according to certain embodiments of the present disclosure is shown;

[0057] Figure 14 A block diagram of an electronic device according to certain embodiments of the present disclosure is shown;

[0058] Figure 15 A schematic diagram of the structure of a computer system suitable for implementing certain embodiments of the differential component identification method and apparatus is shown. Detailed Implementation

[0059] In the following detailed description, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings to enable those skilled in the art to readily implement them. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0060] In this disclosure, it should be understood that terms such as “comprising,” “including,” or “having” are intended to indicate the presence of features, figures, steps, behaviors, components, portions, or combinations thereof disclosed in this specification, and are not intended to exclude the possibility of the presence or addition of one or more other features, figures, steps, behaviors, components, portions, or combinations thereof.

[0061] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0062] When expressions such as "at least one of A, B, and C" are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (e.g., "a system having at least one of A, B, and C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.). When expressions such as "at least one of A, B, or C" are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (e.g., "a system having at least one of A, B, or C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.). Those skilled in the art should also understand that any conjunction and / or phrase that substantially arbitrarily indicates two or more optional items, whether in the specification, claims, or drawings, should be understood to indicate the possibility of including one of these items, either of these items, or both items. For example, the phrase “A or B” should be understood as including the possibility of “A” or “B”, or “A and B”.

[0063] It should also be noted that, unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other. This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0064] This disclosure provides a method for identifying differential components by acquiring two sets of directed brain network data; processing the two sets of directed brain network data to determine one or more candidate connected components with significant differences; and identifying differential components from the candidate connected components by performing a permutation test on the two sets of directed brain network data. This avoids the problem of reduced power of multiple test correction faced by large-scale statistical tests, improves statistical power, and enables the identification of differential components in directed brain network connectivity.

[0065] Figure 1 A flowchart of a differential component identification method according to certain embodiments is shown.

[0066] like Figure 1 As shown, the method includes steps S110-S130.

[0067] In step S110, two sets of directed brain network data are acquired;

[0068] In step S120, the two sets of directed brain network data are processed to identify one or more candidate connected components with significant differences;

[0069] In step S130, differential components are identified from the candidate connected components by performing a permutation test on the two sets of directed brain network data.

[0070] According to certain embodiments of this disclosure, the two sets of directed brain network data can be directed brain network data from two groups of individuals. The two groups typically provide a certain degree of control. Any two of the following can be selected: directed brain network data from healthy individuals, directed brain network data from patients before treatment, and directed brain network data from patients after treatment, for analyzing the physiological and pathological significance of significant connected components in the directed brain network. For example, directed brain network data from healthy individuals and directed brain network data from patients before treatment can be selected to identify connected components related to the disease. Similarly, directed brain network data from patients before treatment and directed brain network data from patients after treatment can be selected to evaluate treatment effectiveness.

[0071] According to certain embodiments of this disclosure, the method for constructing a brain network can be CCM or a delayed CCM method. Resting-state BOLD fMRI data can be preprocessed, and nodes can be defined using Brodmann atlases, Brodmann-ce atlases, AAL, AAL2, AAL3, AICHA, and Brainnetome atlases, respectively. A directed brain network can then be constructed using CCM. Alternatively, the brain network can be constructed using Granger causality analysis, tracing fiber bundles from BOLD fMRI, EEG, or DTI. Methods for defining nodes can include defining nodes based on brain regions or by defining voxels from fMRI scans. For example, selecting one or more brain regions from the atlas, or manually designating a region, and defining each voxel covered by these brain regions / regions as a node of the network; the voxel time series is then the node time series.

[0072] Figure 2 A flowchart illustrating the determination of candidate connected components according to certain embodiments of this disclosure is shown.

[0073] like Figure 2 As shown, the above operation S120 may include operations S210-S230.

[0074] In operation S210, each connection of the two sets of directed brain network data is tested by statistical testing methods to obtain the first feature value of each connection.

[0075] In operation S220, a binary or weighted first difference network is constructed based on the comparison between the first feature value and a predetermined threshold;

[0076] In operation S230, one or more candidate connection components are determined from the first differential network.

[0077] According to certain embodiments of this disclosure, statistical testing methods may include, for example, a two-sample t-test, a paired t-test, a Mann-Whitney rank-sum test, or a Wilcoxon signed-rank test. The first eigenvalue may include a significance statistic (stat_thresh) and / or a p-value (pval_thresh).

[0078] According to certain embodiments of this disclosure, by comparing the first feature value of each connection with a predetermined threshold, it can be determined whether each connection meets the threshold condition. Connections that meet the threshold condition are retained, and connections that do not meet the threshold condition are deleted, thus obtaining a difference network. If a significant difference threshold is set to a p-value, edges with p-values ​​less than the threshold are retained, and edges with p-values ​​greater than the threshold are deleted. This results in a binarized difference network, diff_net.

[0079] According to certain embodiments of this disclosure, one or more candidate connected components C* can be determined based on the difference network. The method for determining the candidate connected components can be referred to below. Figures 6 to 10 The various methods described herein will not be elaborated upon here. For example, all connected subgraphs in the differential network can be identified as candidate connected components, and the size of the subgraph can be recorded, defined as the number of nodes or edges in the subgraph (|C). * |=L * All found connective components and their sizes are denoted as...

[0080] According to certain embodiments of this disclosure, a first-weighted difference network can be constructed first, and then, during the process of determining candidate connection components, the first-weighted difference network can be transformed into a binary difference network to form an alternative equivalent to the above-described scheme.

[0081] According to the technical solution of this disclosure, each connection of the two sets of directed brain network data is tested by a statistical test method to obtain a first feature value for each connection. The first feature value includes a significant difference statistic and / or a p-value. A binary or weighted first difference network is constructed based on the comparison of the first feature value with a predetermined threshold. One or more candidate connected components are determined from the first difference network, thereby avoiding the problem of reduced power of multiple test correction faced by large-scale statistical tests, improving statistical power, and realizing the identification of differential components of directed brain network connections.

[0082] According to certain embodiments of this disclosure, the method further includes determining a statistical test method based on the relationship between the two sets of directed brain network data, wherein: when the two sets of directed brain network data are independent samples, the statistical test method is determined to be a two-sample t-test or a non-parametric Mann-Whitney rank-sum test; when the two sets of directed brain network data are paired samples, the statistical test method is determined to be a paired t-test or a non-parametric Wilcoxon signed-rank test.

[0083] For example, for independent samples, such as patients and healthy controls, a two-sample t-test or a non-parametric Mann-Whitney U test can be used. For paired samples, such as patients before and after treatment, a paired t-test or a non-parametric Wilcoxon test can be used to compare the impact of different test methods on the results.

[0084] Figure 3 A flowchart illustrating the identification of differential components according to certain embodiments of this disclosure is shown.

[0085] like Figure 3 As shown, the above operation S130 may include operations S310-S350.

[0086] In operation S310, the two groups of directed brain network data are randomly divided into two groups;

[0087] In operation S320, statistical testing methods are used to examine each connection of the two groups of regrouped directed brain network data to obtain the third feature value of each connection.

[0088] In operation S330, a second difference network is constructed based on the comparison between the third feature value and the predetermined threshold;

[0089] In operation S340, one or more connected components are determined from the second differential network, and the largest connected component is determined from there;

[0090] In operation S350, after the above operations S310-S340 are repeated a predetermined number of times, differential components are identified from the candidate connected components based on the maximum connected component determined in each iteration.

[0091] According to certain embodiments of this disclosure, the labels of two groups of samples are shuffled. For example, group 1 includes m brain networks and group 2 includes n brain networks. The order of these m+n brain networks can be randomly arranged, and the first m brain networks are taken as group A and the last n brain networks are taken as group B, resulting in regrouped group A and group B.

[0092] According to certain embodiments of this disclosure, the two reorganized directed brain networks are the same size as the original two directed brain networks. For example, the size of group A is the same as that of group 1, and the size of group B is the same as that of group 2.

[0093] According to certain embodiments of this disclosure, operations S320-S340 are basically the same as operations S210-S230 described above, especially the statistical test methods used are required to be the same as those in S210 described above, which will not be repeated here.

[0094] In operation S350, differential components can be identified from the candidate connected components based on the largest connected component among the connected components.

[0095] According to certain embodiments of this disclosure, the connectivity component and its size obtained from the i-th permutation test are denoted as C. i With L i ,|C i |=L i Permutation tests can be repeated up to 5000 times, and the size of the largest connection component obtained from these 5000 permutations is recorded; this represents the empirical distribution of the null hypothesis. The empirical p-value (emp_pval) given by NBS is equal to the number of connection components whose size is greater than the original connection component size in the permutation test divided by the number of repetitions. Whether the connection component is significant is determined by whether the empirical p-value is less than the significance level (α). For connection components... and its size The empirical p-value is calculated as follows:

[0096]

[0097] Where N is the number of permutation tests, and the numerator represents the number of permutation tests. The number of terms is counted, that is, in N permutation tests, how many permutation tests yield a maximum connection component size greater than [the specified value]. Component size

[0098] According to the basic principles of statistical testing, a sample is considered to be randomly drawn from a population with a certain distribution. Whether two groups of samples are significant is essentially determining whether, within a certain tolerance range of error, the two groups of samples were drawn from the same population with the same distribution. Here, the "certain tolerance range of error" is the significance level α. The principle is that if the two groups of samples are indeed drawn from the same population with the same distribution (i.e., there is no significant difference), what is the probability that random sampling errors or noise disturbances would cause such a large statistical difference between the two groups of samples? If this probability is less than the significance level, then the two groups of samples are not considered to be drawn from the same population, because the probability of such a large difference if they were drawn from the same population is too small. Generally, α = 0.05.

[0099] However, in reality, the overall distribution cannot be explicitly expressed. For example, in the brain network application scope of this embodiment, the distribution patterns of brain network connections in patients and healthy individuals are difficult to clearly represent. In this case, the permutation test technique is used to generate an empirical null hypothesis from the data. The process of shuffling the labels described above assumes that group 1 and group 2 are drawn from the same overall distribution (i.e., the null hypothesis). If this hypothesis holds, the test results with and without shuffling the labels should be quite similar.

[0100] Shuffle the labels multiple times and record the maximum connection component size L each time. i This allows us to obtain the empirical distribution of the largest connectivity component size identified after difference testing between the two groups, assuming the null hypothesis is true (i.e., there are no significant differences). Finally, based on this empirical distribution, we can determine all differential connectivity components obtained from the original group 1 and group 2. Is the size of the difference rare in the empirical distribution? In other words, what is the probability of a connection component with such a small difference? This probability is expressed as the count ratio or frequency ratio, which is the formula for calculating the empirical p-value mentioned above. If this probability is small, less than the given significance α, then there is reason to believe that group 1 and group 2 are not samples drawn from the same overall distribution; that is, group 1 and group 2 are significantly different.

[0101] According to the technical solution of this disclosure, the two groups of directed brain network data are randomly regrouped into two groups; each connection of the regrouped two groups of directed brain network data is tested using a statistical test method to obtain a third feature value for each connection; a second difference network is constructed based on the comparison between the third feature value and a predetermined threshold; one or more connected components are determined from the second difference network, and the largest connected component is determined from it; after the above operation is repeated a predetermined number of times, the difference components are identified from the candidate connected components based on the largest connected component determined each time, thereby avoiding the problem of decreased power of multiple test correction faced by large-scale statistical tests, improving statistical power, and realizing the identification of difference components in directed brain network connections.

[0102] The above process will be described below with reference to certain specific embodiments.

[0103] Figure 4 A flowchart of a differential component identification method according to other embodiments of this disclosure is shown.

[0104] like Figure 4 As shown, the method includes operations S410-S440.

[0105] First, two directed brain networks were obtained, group 1 and group 2.

[0106] In operation S410, the network-based statistical analysis method (NBS) is executed to identify significantly different connected components, obtaining candidate connected components and their magnitudes. The network-based statistical analysis method is as described above. Figure 2 The operations described are S210-S230.

[0107] When operating S420, shuffle the group labels, as described above. Figure 3 The described operation S310 shuffles the group labels to obtain group A and group B.

[0108] In operation S430, the network-based statistical analysis method NBS is executed to identify significantly different connected components, obtaining the connected components and their sizes. This operation is described above. Figure 3 The operations S320-S340 described are similar to operation S410. After operation S430 is completed, it returns to S420, and S420 and S430 are repeated until a predetermined number of times is reached, such as 5000 times.

[0109] In operation S440, based on the identification results of a predetermined number of times, and combined with the results obtained in operation S410, one or more significant connected components (i.e., differential components) and their p-values ​​are determined from the candidate connected components.

[0110] Figure 5 A flowchart of NBS significant difference connectivity component analysis according to certain embodiments of this disclosure is shown.

[0111] like Figure 5 As shown, operation S410 or S430 can include operations S510-S530. The two groups of brain networks shown in the figure are group A and group B, which are related to S430; for operation S410, they can be replaced with group 1 and group 2.

[0112] In the S510 operation, based on statistical analysis, difference tests are performed on each connection to obtain the group difference statistics network and the group difference p-value network.

[0113] In operation S520, the group difference statistics network or group difference p-value network is processed by a significant difference threshold (predetermined threshold). The connections in the network are binarized, the connections that meet the threshold conditions are retained, and the connections that do not meet the threshold conditions are deleted, resulting in a binarized difference network.

[0114] In operation S530, the connected components and their sizes are determined based on the difference network.

[0115] According to certain embodiments of this disclosure, determining connected components based on the difference network presents certain obstacles, whether operating the first difference network in S410 or the second difference network in S430. This is because the difference network is a directed network, and the edges of a directed network have directionality. The network adjacency matrix is ​​asymmetric, making it more complex than an undirected network and possessing directional characteristics that undirected networks lack.

[0116] The following is combined with Figures 6-10 The method for determining connected components according to embodiments of this disclosure will be described.

[0117] Figure 6 A partial schematic diagram of a brain network according to certain embodiments of the present disclosure is shown.

[0118] like Figure 6 As shown, among nodes 1-5, there are unidirectional connections from node 1 to node 2, unidirectional connections from node 2 to nodes 3 and 4 respectively, a connection from node 3 to node 1, and a unidirectional connection from node 4 to node 5.

[0119] In some embodiments of this disclosure, determining one or more connected components from the second difference network includes treating directed connections in the second difference network as undirected connections and identifying one or more connected components from the second difference network.

[0120] According to certain embodiments of this disclosure, directed networks are treated as undirected networks, meaning that as long as there is a connection in at least one direction between two nodes, they are considered to be related. Figure 6 In this approach, all nodes in the network are considered to form a single connected component. This type of connected component is also known as a weakly connected component, and it can be found using depth-first search (DFS) of the graph. (See also: [link to relevant documentation]). Figure 7 .

[0121] Figure 7 A flowchart illustrating the search for connected components according to certain embodiments of this disclosure is shown.

[0122] like Figure 7 As shown, the method includes operations S710-S790.

[0123] When operating the S710, mark all nodes in the network as unvisited.

[0124] When operating S720, traverse from a node, which can be any node.

[0125] In operation S730, determine whether the node is a visited node. If it is a visited node, execute operation S740; otherwise, execute operation S750.

[0126] In operation S740, skip the node and return to operation S720 to continue processing the next node.

[0127] In operation S750, the node is added to the current connected component and marked as visited.

[0128] In operations S760-S780, find other nodes connected to this node. If they exist, return to operation S730 and repeat the above steps until all connected nodes have been traversed to obtain the current connected component.

[0129] In operation S790, it is determined whether all nodes in the network have been traversed. If not, it returns to operation S720 to continue traversing the remaining unvisited nodes until all nodes in the network have been visited. This yields all connected components of the network. The size of a connected component is defined as the number of nodes in that component.

[0130] According to the technical solution of the present disclosure, by treating the directed connections in the second difference network as undirected connections, one or more connected components are identified from the second difference network, thereby enabling the convenient processing of directed networks using the connected component identification method of undirected networks.

[0131] In other embodiments of this disclosure, determining one or more connected components from the second differential network includes determining a diverging network and / or a converging network defined along directed connections from one or more predetermined regions of interest (ROIs). The diverging network refers to the nodes and edges reachable from the ROI, and the converging network refers to the nodes and edges that ultimately reach the ROI from other nodes.

[0132] According to certain embodiments of this disclosure, in order to limit the network size and avoid loops, a first-level or second-level divergent and convergent network is generally sought. A first-level network refers to a subnetwork including a region of interest (ROI) and nodes directly connected to the ROI. Directly connected edges are called first-level connections, and directly connected nodes are called first-level nodes. A second-level network, based on the first-level network, includes nodes directly connected to first-level nodes (excluding cases where first-level nodes are directly connected to each other). Newly added edges are called second-level connections, and newly added nodes are called second-level nodes. Figure 8A A schematic diagram of a two-level convergence network according to certain embodiments of the present disclosure is shown. Figure 8B A schematic diagram of a two-level divergent network according to certain embodiments of the present disclosure is shown.

[0133] According to certain embodiments of this disclosure, there can be multiple regions of interest. Depth-first search can be used to find the interconnected network between these nodes, serving as a connectivity component. For example, it can be... Figure 7 Or below Figure 9 and Figure 10 Based on the illustrated embodiment, additional filtering conditions are added to require that the identified connected components contain the given region of interest.

[0134] According to the technical solution of the present disclosure, by determining a divergent network and / or convergent network starting from one or more predetermined regions of interest along directed connections, the limitation of the general linear model is eliminated, the calculation is greatly simplified, and it is beneficial for processing the topological path analysis of complex brain networks.

[0135] In some embodiments of this disclosure, strongly connected components of the graph can be identified. There is a directed reachable path between any two nodes in a strongly connected component, such as... Figure 6 As shown, nodes 1-3 represent a strongly connected component. This method utilizes information about the direction of directed network edges, placing higher demands on the connectivity of the components.

[0136] Figure 9 A flowchart illustrating the search for connected components is shown for some other embodiments of this disclosure.

[0137] like Figure 9 As shown, determining one or more connected components from the second difference network includes operations S910-S930.

[0138] In operation S910, starting from any node in the second differential network, a search is performed according to the directed connection, and the searched nodes are put into the stack according to the deepest node priority principle;

[0139] In operation S920, the direction of the directed connection is reversed, and all nodes are marked as unvisited.

[0140] In operation S930, nodes are sequentially removed from the stack. If a node is not visited, it is searched in the second differential network with the reversed direction until there are no nodes in the stack, in order to determine one or more connected components.

[0141] According to certain embodiments of this disclosure, a depth-first search can be performed on the graph, placing the found nodes into a stack according to the principle of prioritizing the deepest node. That is, if a node has not been visited, all child nodes reachable from that node are visited first, and then the node is placed into the stack. The stack is a first-in-last-out (FILO) data structure, where nodes placed in the stack first are removed later. Then, the direction of the edges in the directed graph is reversed, all nodes are marked as unvisited, and nodes are removed from the stack sequentially. If a node has not been visited, a depth-first search is performed on that node in the reversed network to find its connected components; if the node has already been visited, it is skipped. The above steps are repeated until there are no nodes in the stack. This yields all strongly connected components of the network, where the size of a component is defined as the number of nodes.

[0142] The following is combined with Figure 10 The illustrated embodiment illustrates the method.

[0143] Figure 10 A flowchart for finding connected components is shown, representing some embodiments of this disclosure.

[0144] like Figure 10 As shown, the method may include operations S1001-S1016.

[0145] In operation S1001, all nodes in the network are marked as unvisited.

[0146] In operation S1002, select a node in sequence. The order can be arbitrary, meaning you can start from any node and continue until the traversal is complete.

[0147] In operation S1003, determine whether the node has been visited. If it has been visited, execute operation S1004; otherwise, execute operation S1005.

[0148] In operation S1004, skip this node and return to operation S1002.

[0149] In operation S1005, mark the node as visited.

[0150] In operation S1006, it is determined whether the node is connected to other nodes, where "connection" here refers to a directed connection. If it is not connected to other nodes, operation S1007 is executed; otherwise, operation S1008 is executed.

[0151] In operation S1007, the node is added to the stack.

[0152] In operation S1008, if the traversal is not complete, the next node is selected in sequence and the process returns to execute operation S1003; if the traversal is complete, the node stack is obtained.

[0153] In operation S1009, the connection direction of the directed network is reversed to obtain a reversed network.

[0154] In operation S1010, all nodes are marked as unvisited.

[0155] In operation S1011, a node is retrieved from the aforementioned node stack.

[0156] In operation S1012, determine whether the node has been visited. If it has been visited, execute operation S1013; otherwise, execute operation S1014.

[0157] In operation S1013, skip the node and return to operation S1011 to continue processing the next node.

[0158] In operation S1014, the node is added to the current connected component and marked as visited.

[0159] In operation S1015, if there are other unvisited nodes connected to this node, return to operation S1011 and select any other unvisited node connected to this node until the traversal is complete; if there are no other unvisited nodes connected to this node, it means that the traversal has been completed and the current connected component is obtained.

[0160] In operation S1016, it is determined whether the node stack is empty. If it is not empty, the process returns to operation S1011 to continue processing until the node stack is empty, thus obtaining all connected components.

[0161] According to the technical solution of this disclosure embodiment, by starting from any node in the second difference network and searching according to the directed connection, the searched nodes are put into the stack according to the deepest node priority principle; the direction of the directed connection is reversed, and all nodes are marked as unvisited; nodes are taken out from the stack in turn, and if a node is unvisited, the node is searched in the second difference network with the reversed direction until there are no nodes in the stack, thereby effectively determining one or more strongly connected components.

[0162] According to certain embodiments of this disclosure, the method further includes controlling the predetermined threshold to vary within a certain range to display the changing difference components, thereby facilitating the observation of the changing patterns of the identified connection components and whether their significance is affected by the threshold. When a certain threshold is used for binarization such that the binarized difference network is still a fully connected graph, the threshold is determined as the lower limit; when a certain threshold is used for binarization such that the binarized difference network is a completely disconnected graph (i.e., there is no connection between any two nodes), the threshold is determined as the upper limit.

[0163] Figure 11A schematic diagram illustrating a differential component identification method applying certain embodiments of the present disclosure is shown.

[0164] like Figure 11 As shown, after preprocessing the BOLD fMRI data of patients before and after treatment, as well as the healthy control group, this method can define nodes using Brodmann atlases, Brodmann_ce atlases, AAL, AAL2, AAL3, AICHA, and Brainnetome atlases, and construct a directed brain network using CCM. After obtaining the whole-brain directed network, the patients before treatment and the healthy control group, the patients after treatment and the healthy control group, and the patients after treatment and the patients before treatment are respectively designated as group 1 and group 2. Using the differential component identification method of this embodiment, significant connected components and their sizes are found. Furthermore, combined with the physiological functions of brain regions, the physiological, pathological, and other medical significance implied by the significant connected components is analyzed.

[0165] For the two combinations of patients before treatment and healthy controls, and patients after treatment and healthy controls, two-sample t-tests and nonparametric Mann-Whitney U tests were used, respectively. For patients after treatment and patients before treatment, paired t-tests and nonparametric Wilcoxon tests were used, respectively. pval_thresh was used as the significance threshold, with values ​​ranging from pval_thresh∈{0.001,0.002,…,0.01,0.02,…,0.05}. For each pval_thresh value, the connected components, the magnitude of the components, and whether they were significant were obtained.

[0166] According to certain embodiments of this disclosure, the analysis results can be stored in a database jointly constructed using MongoDB and Redis. Features of each person, each graph, and each method of brain network computation are stored in the MongoDB database. When needed, features are first read from MongoDB, stored in the Redis database, and then returned to the user from the Redis database for computation. Redis serves as a high-speed cache during use, while MongoDB is used as a long-term storage database.

[0167] The method of this disclosure can be used to perform connectivity path difference analysis on grouped directed brain networks, study changes in human brain connectivity under disease states, identify disease biomarkers, assist in disease diagnosis, and also study changes in human brain connectivity paths after treatment, and combine changes in clinical scores to assist in treatment and evaluate treatment effects.

[0168] The methods of certain embodiments of this disclosure are applied to resting-state functional magnetic resonance imaging (fMRI) data of patients with spinal cord injuries. A directed brain network is constructed using CCM and Brodmann+AAL cerebellar atlases. A differential network is constructed using a two-sample t-test, with pval_thresh used as a threshold, varying within a certain range. Figures 12A-12G As shown, where, Figures 12A-12G The pval_thresh values ​​were 0.0002, 0.0003, 0.0004, 0.0007, 0.0008, 0.0009, and 0.0020, respectively. Using this invention, the significantly different connectivity components and their empirical p-values ​​at each threshold were obtained. Figures 12A-12G The empirical p-values ​​given were 0.0030, 0.0046, 0.0060, 0.0034, 0.0038, 0.0032, and 0.0254, respectively. These connectivity components showed significant differences in weak connectivity components between spinal cord injury patients and healthy controls.

[0169] Figure 13 A block diagram of a differential component identification device 1300 according to certain embodiments is shown. This device can be implemented as part or all of an electronic device through software, hardware, or a combination of both.

[0170] like Figure 13 As shown, the differential component identification device 1300 includes an acquisition module 1310, a processing module 1320, and an identification module 1330.

[0171] The acquisition module 1310 is configured to acquire two sets of directed brain network data;

[0172] Processing module 1320 is configured to process the two sets of directed brain network data to determine one or more candidate connected components with significant differences;

[0173] The identification module 1330 is configured to identify differential components from the candidate connected components by performing a permutation test on the two sets of directed brain network data.

[0174] According to the technical solution provided in this embodiment, the acquisition module is configured to acquire two sets of directed brain network data; the processing module is configured to process the two sets of directed brain network data to determine one or more candidate connected components with significant differences; and the identification module is configured to identify the differential components from the candidate connected components by performing a permutation test on the two sets of directed brain network data, thereby avoiding the problem of decreased power of multiple test correction faced by large-scale statistical tests, improving statistical power, and realizing the identification of differential components in directed brain network connectivity.

[0175] This disclosure also discloses an electronic device. Figure 14A block diagram of an electronic device 1400 according to certain embodiments is shown.

[0176] like Figure 14 As shown, the electronic device 1400 includes a memory 1401 and a processor 1402, wherein the memory 1401 is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor 1402 to perform the following operations:

[0177] Acquire two sets of directed brain network data;

[0178] The two sets of directed brain network data are processed to identify one or more candidate connected components with significant differences;

[0179] By performing a permutation test on the two sets of directed brain network data, differential components are identified from the candidate connected components.

[0180] According to certain embodiments of this disclosure, processing the two sets of directed brain network data to determine one or more candidate connected components with significant differences includes:

[0181] Each connection of the two sets of directed brain network data was tested using statistical testing methods to obtain a first feature value for each connection, the first feature value including a significant difference statistic and / or a p-value.

[0182] A binary or weighted first difference network is constructed based on the comparison between the first feature value and a predetermined threshold;

[0183] One or more candidate connectivity components determined from the first differential network.

[0184] According to certain embodiments of this disclosure, the processor is further configured to execute a statistical test method based on the relationship between the two sets of directed brain network data, wherein:

[0185] Given that the two sets of directed brain network data are independent samples, the statistical test method is determined to be either a two-sample t-test or a non-parametric Mann-Whitney rank-sum test.

[0186] When the two sets of directed brain network data are paired samples, the statistical test method is determined to be either the paired t-test or the nonparametric Wilcoxon signed-rank test.

[0187] According to certain embodiments of this disclosure, the processor is further configured to control the predetermined threshold to vary within a certain range in order to reveal the differential components of the variation.

[0188] According to certain embodiments of this disclosure, identifying differential components from the candidate connected components by performing a permutation test on the two sets of directed brain network data includes:

[0189] The two groups of directed brain network data were randomly divided into two groups.

[0190] Statistical tests were used to examine each connection of the two regrouped groups of directed brain network data to obtain the third feature value of each connection.

[0191] A second difference network is constructed based on the comparison between the third feature value and the predetermined threshold;

[0192] Identify one or more connected components from the second difference network, and determine the largest connected component from among them;

[0193] After the above operation is repeated a predetermined number of times, the differential components are identified from the candidate connected components based on the maximum connected component determined in each iteration.

[0194] According to certain embodiments of this disclosure, determining one or more connected components from the second difference network includes:

[0195] Treat the directed connections in the second difference network as undirected connections, and identify one or more connected components from the second difference network; or

[0196] Based on one or more predetermined regions of interest, a diverging network and / or a converging network are determined starting from the regions of interest and along directed connections.

[0197] According to certain embodiments of this disclosure, determining one or more connected components from the second difference network includes:

[0198] Starting from any node in the second differential network, search according to directed connections, and put the searched nodes into the stack according to the deepest node priority principle;

[0199] Reverse the direction of the directed connection and mark all nodes as unvisited;

[0200] Nodes are sequentially removed from the stack. If a node is not visited, it is searched in the second differential network with the reversed direction until there are no nodes in the stack, in order to determine one or more connected components.

[0201] Figure 15 A schematic diagram of the structure of a computer system 1500 suitable for implementing the differential component identification method and apparatus of certain embodiments is shown.

[0202] like Figure 15As shown, the computer system 1500 includes a processing unit 1501, which can execute various processes described in the above embodiments according to a program stored in a read-only memory (ROM) 1502 or a program loaded from a storage portion 1508 into a random access memory (RAM) 1503. The RAM 1503 also stores various programs and data required for the operation of the system 1500. The processing unit 1501, ROM 1502, and RAM 1503 are interconnected via a bus 1504. An input / output (I / O) interface 1505 is also connected to the bus 1504.

[0203] The following components are connected to I / O interface 1505: an input section 1506 including a keyboard, mouse, etc.; an output section 1507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1508 including a hard disk, etc.; and a communication section 1509 including a network interface card such as a LAN card, modem, etc. The communication section 1509 performs communication processing via a network such as the Internet. A drive 1510 is also connected to I / O interface 1505 as needed. A removable medium 1511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 1510 as needed so that computer programs read from it can be installed into storage section 1508 as needed. The processing unit 1501 can be implemented as a CPU, GPU, TPU, FPGA, NPU, etc.

[0204] In particular, according to certain embodiments, the methods described above can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program tangibly embodied on a readable medium thereof, the computer program containing program code for performing the methods described above. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1509, and / or installed from removable medium 1511.

[0205] According to the technical solutions provided in certain embodiments, brain network models can be constructed by converging cross-mapping or delayed converging cross-mapping, which can more accurately simulate the connection relationships between nodes in the brain network model and improve the accuracy of the brain network model.

[0206] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0207] The units or modules described in some embodiments may be implemented in software or programmable hardware. The described units or modules may also be located in a processor, and the names of these units or modules do not necessarily limit the unit or module itself.

[0208] In another aspect, this disclosure also provides a computer-readable storage medium, which may be a computer-readable storage medium included in the electronic device or computer system described above; or it may be a standalone computer-readable storage medium not assembled into a device. The computer-readable storage medium stores one or more programs, which are used by one or more processors to perform the methods described in this disclosure.

[0209] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

Claims

1. A method for identifying differentially expressed components, comprising: Two sets of directed brain network data were acquired, wherein the directed brain network was constructed by defining nodes and edges based on measured brain neural activity signals; The two sets of directed brain network data are processed to determine one or more candidate connected components with significant differences. The one or more candidate connected components with significant differences are all connected subgraphs of the differential networks of the two sets of directed brain network data, and the connected subgraphs include nodes and edges. By performing a permutation test on the two sets of directed brain network data, differential components are identified from the candidate connected components.

2. The method according to claim 1, wherein, The process of processing the two sets of directed brain network data to determine one or more candidate connected components with significant differences includes: Each connection of the two sets of directed brain network data was tested using statistical testing methods to obtain a first feature value for each connection, the first feature value including a significant difference statistic and / or a p-value. A binary or weighted first difference network is constructed based on the comparison between the first feature value and a predetermined threshold; One or more candidate connected components determined from the first differential network.

3. The method according to claim 2, further comprising determining a statistical test method based on the relationship between the two sets of directed brain network data, wherein: Given that the two sets of directed brain network data are independent samples, the statistical test method is determined to be either a two-sample t-test or a non-parametric Mann-Whitney rank-sum test. When the two sets of directed brain network data are paired samples, the statistical test method is determined to be either the paired t-test or the nonparametric Wilcoxon signed-rank test.

4. The method according to claim 2, wherein, The step of identifying differential components from the candidate connected components by performing a permutation test on the two sets of directed brain network data includes: The two groups of directed brain network data were randomly divided into two groups. Statistical tests were used to examine each connection of the two regrouped groups of directed brain network data to obtain the third feature value of each connection. A second difference network is constructed based on the comparison between the third feature value and the predetermined threshold; Identify one or more connected components from the second difference network, and determine the largest connected component from among them; After the above operation is repeated a predetermined number of times, the differential components are identified from the candidate connected components based on the maximum connected component determined in each iteration.

5. The method according to claim 4, wherein, Determining one or more connected components from the second difference network includes: Treat the directed connections in the second difference network as undirected connections, and identify one or more connected components from the second difference network; or Based on one or more predetermined regions of interest, a diverging network and / or a converging network are determined starting from the regions of interest and along directed connections.

6. The method according to claim 4, wherein, Determining one or more connected components from the second difference network includes: Starting from any node in the second differential network, search according to directed connections, and put the searched nodes into the stack according to the deepest node priority principle; Reverse the direction of the directed connection and mark all nodes as unvisited; Nodes are sequentially removed from the stack. If a node is not visited, it is searched in the second differential network with the reversed direction until there are no nodes in the stack, in order to determine one or more connected components.

7. The method according to claim 4, further comprising: The predetermined threshold is controlled to vary within a certain range to reveal the differences in the components of the variation.

8. A differential component identification device, comprising: The acquisition module is configured to acquire two sets of directed brain network data, wherein the directed brain network is constructed by defining nodes and edges based on measured brain neural activity signals; The processing module is configured to process the two sets of directed brain network data to determine one or more candidate connected components with significant differences, wherein the one or more candidate connected components with significant differences are all connected subgraphs of the differential networks of the two sets of directed brain network data, and the connected subgraphs include nodes and edges; The identification module is configured to identify differential components from the candidate connected components by performing a permutation test on the two sets of directed brain network data.

9. An electronic device, characterized in that, The method includes a memory and a processor; wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the steps of the method according to any one of claims 1 to 7.

10. A readable storage medium having computer instructions stored thereon, characterized in that, When executed by a processor, the computer instructions implement the steps of the method described in any one of claims 1 to 7.

Images