Modeling method of whole-brain space-time multi-layer function connection network

Abstract

The invention provides a modeling method of a whole-brain space-time multi-layer functional connection network, which comprises the following steps: preprocessing resting state functional magnetic resonance imaging data, and extracting a blood oxygen level dependent signal of a region of interest; dividing the blood oxygen level dependent signal into a plurality of overlapped time windows by adopting a time window sliding method; based on a Pearson correlation method and an L1 sparse linear regression algorithm, estimating edges and hyperedges between the brain regions; based on a graph theory and a hypergraph theory, modeling the edge and the hyperedge into a graph and a hypergraph respectively, and constructing a spatial multi-layer function connection network; and constructing a spatial multi-layer function connection network in each time window so as to construct a whole-brain space-time multi-layer function connection network. According to the method, comprehensive description of space and time multi-layer dynamic features of the brain FCN can be realized, and the characterization capability of the brain FCN on brain dynamic information is improved.

Description

Technical Field

[0001] The present invention relates to the field of medical image technology, and in particular to a modeling method for a whole-brain spatiotemporal multi-layer functional connection network. Background Art

[0002] The functional connectivity network (FCN) based on functional magnetic resonance imaging (fMRI) is established by evaluating the temporal correlation of blood oxygen level-dependent (BOLD) signals between different brain regions. Since the concept of functional connectivity was proposed, constructing FCN to study the complex connection patterns between different brain regions (such as default mode network, visual network, attention network, etc.), as well as to explore the effects of physiological and pathological factors on brain networks, has become one of the main research directions in the field of fMRI.

[0003] At present, most studies on resting-state fMRI brain networks focus on constructing a single-layer FCN using gray matter or white matter. However, the human brain is composed of multiple highly complex networks, and a single-layer FCN can usually only describe local neural activity and single brain function characteristics of the brain, and it is difficult to capture the complex high-order functional interactions between different brain regions. In view of the limitations of single-layer FCN, researchers began to develop multi-layer FCN from the perspective of temporal dynamics or structural dynamics and achieved rich research results. For example, some studies used the sliding time window method to divide the BOLD signal of fMRI to explore the structural changes of FCN in different time periods. However, most of the current studies only use gray matter or white matter to construct FCN, and have not achieved a true whole-brain network. At the same time, most studies only focus on the dynamic characteristics of time and do not make full use of the rich structural information of the brain network. Most of the existing technologies only use the sliding time window method to construct temporal multi-layer FCN, and most of them only use gray matter data for modeling.

[0004] Existing methods lack multi-layer network construction for spatial dynamic features, such as: Application No. 202310853785.6, an attention deficit hyperactivity disorder auxiliary diagnosis method and system, Application No. 202311256227.8, a method and device for analyzing gender differences in dynamic brain networks for autism spectrum disorder, which only construct brain FCN for temporal dynamic features and only use fMRI data of gray matter, making it difficult to describe the functional interaction information between multiple regions of the whole brain. Summary of the invention

[0005] The present invention provides a modeling method for a whole-brain spatiotemporal multi-layer functional connection network, which can achieve a comprehensive description of the spatial and temporal multi-layer dynamic characteristics of brain FCN and improve the characterization ability of brain FCN for brain dynamic information.

[0006] This specification discloses a modeling method for a whole-brain spatiotemporal multi-layer functional connection network, including:

[0007] Preprocess the resting-state functional magnetic resonance imaging data and extract the blood oxygen level-dependent signal of the region of interest;

[0008] The blood oxygen level dependent signal is divided into multiple overlapping time windows using a sliding time window method;

[0009] Estimate edges and hyperedges between brain regions based on the Pearson correlation method and L1 sparse linear regression algorithm;

[0010] Based on graph theory and hypergraph theory, edges and hyperedges are modeled as graphs and hypergraphs respectively, and a spatial multi-layer functional connection network is constructed; a spatial multi-layer functional connection network is constructed within each time window to construct a whole-brain spatiotemporal multi-layer functional connection network.

[0011] In this manual, the preprocessing process includes: removing the first 10 time points of instability, time layer correction, head motion correction, gradient distortion correction, image distortion correction, spatial normalization and intensity normalization of the Montreal Neurological Institute standard template, spatial smoothing half-height width of 6 mm, bandpass filtering, and regression of 24 head motion parameters and covariates of cerebrospinal fluid signals.

[0012] In this specification, the process of extracting the blood oxygen level dependent signal of the region of interest includes:

[0013] Regions of interest in gray matter were delineated using the functional grouping template provided by the Stanford Neuropsychiatric Functional Imaging Laboratory;

[0014] Regions of interest in the white matter were delineated using the white matter tractography template provided by Johns Hopkins University;

[0015] The blood oxygen level dependence signal of each region of interest is the average of the blood oxygen level dependence signals of all voxels in the region of interest, and the blood oxygen level dependence signal of each large-scale resting-state network is the average of the blood oxygen level dependence signals of all regions of interest in the large-scale resting-state network.

[0016] In this specification, the process of dividing the blood oxygen level dependent signal into multiple overlapping time windows using the sliding time window method includes:

[0017] The number of time windows T depends on the length of the blood oxygen level dependent time series signal, the length of the sliding window, and the sliding step. The calculation formula is as follows:

[0018]

[0019] Where L is the length of the original blood oxygen level dependent time series signal, W is the length of the sliding window, and S is the sliding step of adjacent sliding windows.

[0020] In this specification, the sliding time window method constructs multiple layers in time scale, and the spatial multi-layer functional connection network combines the hypergraph and the graph to construct multiple layers in space, and at the same time introduces high-order associations and low-order associations into the representation of the brain network. The mathematical representation method of the hypergraph and the graph is as follows:

[0021] The hypergraph G = (ν, ε) consists of a set of nodes ν = (v1, v2, ..., v N-1 , v N ) and the hyperedge set ε=(e1,e2,…,e M-1 , e M ); the brain regions of interest are represented as nodes in the hypergraph, and the hyperedges represent the high-order functional connections between these regions of interest; v N is the nth node; e M is the mth hyperedge.

[0022] In this specification, the hypergraph uses the incidence matrix H ij It is represented and defined as follows:

[0023]

[0024] v i is the i-th node; e j is the jth hyperedge.

[0025] In this specification, the L1 sparse linear regression algorithm is as follows:

[0026]

[0027] Where X i Indicates the blood oxygen level-dependent time series signal of the region of interest corresponding to the central response vector; B i =(X1,…,X i-1 , X i+1 , …, X N ) is composed of nodes except node v i The matrix composed of the time series corresponding to all nodes except α; i is the regression coefficient vector; λ is the regularization coefficient.

[0028] In this specification, the graph in the spatial multi-layer functional connection network is represented by a thresholded binary adjacency matrix, and the functional connection between two nodes is obtained by calculating the Pearson correlation of their blood oxygen level dependent time series signals, which is defined as follows:

[0029]

[0030] Where r is the Pearson correlation coefficient; X i , Y i Represent the value of the i-th point of the blood oxygen level dependent time series signal X and Y respectively, Represent the mean of the blood oxygen level dependent time series signals X and Y respectively.

[0031] In this specification, a three-layer functional connection network is constructed at the spatial scale. The first layer of the functional connection network consists of the association matrix H of the hypergraph ij To describe the high-order associations between large-scale resting-state networks in gray matter; the second-layer functional connection network is represented by the adjacency matrix To describe the low-order associations between the regions of interest within each large-scale resting-state network; the third-layer functional connection network is represented by the adjacency matrix To describe the low-order correlation between gray matter regions of interest and white matter regions of interest, the adjacency matrix A used is ij It is expressed as:

[0032]

[0033] r ij is the Pearson correlation coefficient between the blood oxygen level-dependent time series signals of node i and node j, which is regarded as the functional connection strength here; the threshold of each brain FCN is determined by the average value of all functional connection strengths of the current layer plus one standard deviation.

[0034] In summary, the present invention can at least achieve the following beneficial effects:

[0035] At the same time, the gray matter and white matter of the brain are introduced into the construction of the brain network, realizing the construction of a true whole-brain network. This model introduces the concept of spatial multi-layer into the multi-layer construction of FCN, improving the FCN's ability to represent the spatiotemporal dynamic characteristics of brain region functional interactions. Brain white matter data is added to the modeling process, and multi-layer FCN is constructed on a spatial scale. Low-order and high-order associations between brain regions are introduced into the FCN construction, which significantly improves the multi-layer FCN's ability to represent the functional interaction capabilities of brain networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0037] Figure 1 It is a schematic diagram of the modeling method of the whole-brain spatiotemporal multi-layer functional connection network involved in the present invention.

[0038] Figure 2 It is a flowchart of the modeling method of the whole-brain spatiotemporal multi-layer functional connection network involved in the present invention. DETAILED DESCRIPTION

[0039] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the embodiments of the present invention. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.

[0040] The disclosure below provides many different embodiments or examples to implement different structures of the embodiments of the present invention. In order to simplify the disclosure of the embodiments of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the embodiments of the present invention. In addition, the embodiments of the present invention can repeat reference numbers and / or reference letters in different examples, and this repetition is for the purpose of simplification and clarity, which itself does not indicate the relationship between the various embodiments and / or settings discussed.

[0041] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0042] like Figure 1 and Figure 2 As shown, the embodiment of this specification provides a modeling method for a whole-brain spatiotemporal multi-layer functional connection network, which is as follows:

[0043] Step 1: Preprocess the resting-state fMRI data and extract the BOLD signal of the region of interest (ROI) from the resting-state fMRI data. The specific process includes:

[0044] The first 10 time points of instability were removed, and the temporal slices were corrected, head motion was corrected, gradient distortion was corrected, image distortion was corrected, spatial normalization and intensity normalization were performed using the Montreal Neurological Institute standard template, spatial smoothing with a half-width of 6 mm, bandpass filtering (0.01-0.8 Hz), and regression of 24 head motion parameters and cerebrospinal fluid signal as covariates.

[0045] The gray matter ROI division uses the functional grouping template provided by the Stanford University Neuropsychiatric Functional Imaging Laboratory, which consists of 90 gray matter ROIs, and these ROIs belong to 14 macroscale resting networks (MRSNs). Each MRSN represents a brain area responsible for a specific function in the human brain. For example, the auditory network consists of all brain areas that affect auditory function. The 14 MRSNs are Anterior Salience Network (ASN), Auditory Network (AN), Basal Ganglia Network (BGN), Dorsal Default Mode Network (DDMN), Higher Visual Network (HVN), Language Network (LN), Left Executive Control Network (LECN), Posterior Salience Network (PSN), Precuneus Network (PN), Primary Visual Network (PVN), Right Executive Control Network (RECN), Sensorimotor Network (SN), Ventral Default Mode Network (VDMN), and Visuospatial Network (VN). The ROI division of the white matter area used the white matter fiber bundle imaging template provided by Johns Hopkins University, which divides the white matter into 48 fiber bundles. The 48 white matter fiber bundles are considered as one MRSN, representing the entire white matter area. The BOLD signal of each ROI is the average of the BOLD signals of all voxels in the ROI, and the BOLD signal of each MRSN is the average of the BOLD signals of all ROIs in the MRSN. Finally, the whole-brain fMRI data was divided into 138 ROIs and 15 MRSNs.

[0046] Step 2: Use the sliding window method to divide the BOLD signal into multiple overlapping time windows, including:

[0047] The number of time windows depends on the length of the BOLD time series signal, the size of the sliding window, and the sliding step. The calculation formula is as follows:

[0048]

[0049] Where L is the length of the original BOLD time series signal, W is the length of the sliding window, and S is the sliding step of adjacent sliding windows.

[0050] Step 4: Construction of a multi-layer FCN for the whole brain, including:

[0051] The sliding time window method constructs multiple layers in time on the time scale, while the spatial multi-layer FCN combines hypergraphs and graphs to construct multiple layers in space, while introducing high-order and low-order associations into the representation of brain networks. The mathematical representation methods of hypergraphs and graphs are as follows:

[0052] The hypergraph G = (ν, ε) consists of a set of nodes ν = (v1, v2, ..., v N-1 , v N ) and the hyperedge set ε=(e1,e2,…,e M-1 , e M ). Brain ROIs are represented as nodes in a hypergraph, and hyperedges represent high-order functional connections between these ROIs. The hypergraph can be represented by an association matrix, which is defined as follows:

[0053]

[0054] The L1 sparse linear regression algorithm is used to estimate the hyperedges to construct the incidence matrix, which is defined as follows:

[0055]

[0056] Where X i represents the BOLD time series signal of the ROI corresponding to the central response vector, B i =(X1,…,X i-1 , X i+1 , …, X N ) is composed of nodes except node v i The matrix consisting of the time series corresponding to all nodes except α i is the regression coefficient vector, where the larger the value of the element is, the more the corresponding node is closer to the central node v i The stronger the correlation is. It is generally believed that α i The corresponding nodes with non-zero values ​​in are considered to belong to the hyperedge ε iλ is the regularization coefficient, which is used to control the sparsity of the regression results. The graph in the spatial multilayer network is represented by a thresholded binary adjacency matrix, and the functional connection between two nodes is obtained by calculating the Pearson correlation of their BOLD time series signals, which is defined as follows:

[0057]

[0058] Where Xi and Yi represent the values ​​of the i-th point of the BOLD time series signal X and Y respectively. Represent the mean of BOLD time series signals X and Y respectively. The threshold of each adjacency matrix is ​​determined by the mean value of all functional connection strengths in the current layer plus a standard deviation. This method adaptively filters weaker functional connections and ensures the effectiveness of the adjacency matrix in representing functional connection information.

[0059] By combining the pre-segmented ROI and MRSN with the above method, a three-layer FCN was constructed at the spatial scale. The first layer of FCN is represented by the hypergraph association matrix H, which is used to describe the high-order associations between gray matter MRSNs. Specifically, each gray matter MRSN is successively used as a central node, and 14 hyperedges are estimated using the L1 sparse linear regression algorithm. The association matrix is ​​constructed based on the association matrix definition method. The second layer FCN focuses on characterizing the low-order associations between ROIs within each MRSN, represented by the adjacency matrix The third layer FCN represents the low-order association between gray matter ROI and white matter, using the adjacency matrix To express.

[0060] The above-described embodiments are used to illustrate the present invention, not to limit the present invention, so changes in the exemplified values ​​or replacement of equivalent elements should still fall within the scope of the present invention.

[0061] From the above detailed description, it can be understood by those skilled in the art that the present invention can indeed achieve the aforementioned objectives and is in compliance with the provisions of the Patent Law.

[0062] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. It should be pointed out that any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0063] It should be noted that the above description of the relevant process is only for example and explanation, and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to the process under the guidance of this specification. However, these modifications and changes are still within the scope of this specification.

[0064] The basic concepts have been described above. Obviously, for those of ordinary skill in the art who have read this application, the above invention disclosure is only for example and does not constitute a limitation of this application. Although not explicitly stated here, those of ordinary skill in the art may make various modifications, improvements and amendments to this application. Such modifications, improvements and amendments are suggested in this application, so such modifications, improvements and amendments still belong to the spirit and scope of the exemplary embodiments of this application.

[0065] At the same time, the present application uses specific words to describe the embodiments of the present application. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of the present application. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, some features, structures or characteristics in one or more embodiments of the present application can be appropriately combined.

[0066] In addition, it will be appreciated by those skilled in the art that various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful combination of processes, machines, products or substances, or any new and useful improvements thereto. Therefore, various aspects of the present application may be implemented entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "units", "modules" or "systems". In addition, various aspects of the present application may take the form of a computer program product embodied in one or more computer-readable media, wherein computer-readable program code is contained therein.

[0067] The computer program code required for the operation of each part of the application can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, etc., such as C programming language, VisualBasic, Fortran2103, Perl, COBOL2102, PHP, ABAP, such as Python, Ruby and Groovy dynamic programming language or other programming languages. The program code can be run completely on the user's computer, or run on the user's computer as an independent software package, or run partly on the user's computer and partly on the remote computer, or run completely on the remote computer or server. In the latter case, the remote computer can be connected to the user's computer through any network form, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., via the Internet), or in a cloud computing environment, or used as a service such as software as a service (SaaS).

[0068] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this application, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this application. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this application. For example, although the implementation of the various components described above can be embodied in a hardware device, it can also be implemented as a pure software solution, for example, installation on an existing server or mobile device.

[0069] Similarly, it should be noted that in order to simplify the description disclosed in this application and thus help understand one or more embodiments of the invention, in the above description of the embodiments of the application, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this method of the application should not be interpreted as reflecting the intention that the claimed object requires more features than those explicitly stated in each claim. On the contrary, the subject of the invention should have fewer features than the above single embodiment.

Claims

1. A modeling method for a whole-brain spatiotemporal multi-layer functional connection network, characterized in that: include: Preprocess the resting-state functional magnetic resonance imaging data and extract the blood oxygen level-dependent signal of the region of interest; The blood oxygen level dependent signal is divided into multiple overlapping time windows using a sliding time window method; The edges and hyperedges between brain regions were estimated based on the Pearson correlation method and the L1 sparse linear regression algorithm. The edges and hyperedges were modeled as graphs and hypergraphs respectively based on graph theory and hypergraph theory, and a spatial multi-layer functional connection network was constructed. A spatial multi-layer functional connection network was constructed within each time window to construct a whole-brain spatiotemporal multi-layer functional connection network.

2. The modeling method of the whole-brain spatiotemporal multi-layer functional connection network according to claim 1 is characterized in that: The preprocessing process included: removing the first 10 time points of instability, time slice correction, head motion correction, gradient distortion correction, image distortion correction, spatial normalization and intensity normalization to the Montreal Neurological Institute standard template, spatial smoothing with a half-height width of 6 mm, bandpass filtering, and regression of 24 head motion parameters and cerebrospinal fluid signal covariates.

3. The modeling method of the whole-brain spatiotemporal multi-layer functional connection network according to claim 1 is characterized in that: The process of extracting the blood oxygen level dependent signal of the region of interest includes: Delineate regions of interest in gray matter using functional grouping templates; Delineation of regions of interest in the white matter using a white matter tractography template; The blood oxygen level dependence signal of each region of interest is the average of the blood oxygen level dependence signals of all voxels in the region of interest, and the blood oxygen level dependence signal of each large-scale resting-state network is the average of the blood oxygen level dependence signals of all regions of interest in the large-scale resting-state network.

4. The modeling method of the whole-brain spatiotemporal multi-layer functional connection network according to claim 1 is characterized in that: The process of dividing the blood oxygen level dependent signal into multiple overlapping time windows using the sliding time window method includes: The number of time windows T depends on the length of the blood oxygen level dependent time series signal, the length of the sliding window, and the sliding step. The calculation formula is as follows: Where L is the length of the original blood oxygen level dependent time series signal, W is the length of the sliding window, and S is the sliding step of adjacent sliding windows.

5. The modeling method of the whole-brain spatiotemporal multi-layer functional connection network according to claim 1 is characterized in that: The sliding time window method constructs multiple layers in time on the time scale. The spatial multi-layer functional connection network combines the hypergraph and the graph to construct multiple layers in space. At the same time, high-order associations and low-order associations are introduced into the representation of the brain network. The mathematical representation method of the hypergraph and the graph is as follows: The hypergraph G = (ν, ε) consists of a node set ν = (v1, v2, ..., v N-1 , v N ) and the hyperedge set ε=(e1,e2,…,e M-1 , e M ); the brain regions of interest are represented as nodes in the hypergraph, and the hyperedges represent the high-order functional connections between these regions of interest; v N is the Nth node; e M is the Mth hyperedge.

6. The modeling method of the whole-brain spatiotemporal multi-layer functional connection network according to claim 5 is characterized in that: The hypergraph uses the incidence matrix H ij It is represented and defined as follows: v i is the i-th node; e j is the jth hyperedge.

7. The modeling method of the whole-brain spatiotemporal multi-layer functional connection network according to claim 6 is characterized in that: The L1 sparse linear regression algorithm is as follows: Where X i Indicates the blood oxygen level-dependent time series signal of the region of interest corresponding to the central response vector; B i =(X1,…,X i-1 , X i+1 , …, X N ) is composed of nodes except node v i The matrix composed of the time series corresponding to all nodes except α; i is the regression coefficient vector; λ is the regularization coefficient.

8. The modeling method of the whole-brain spatiotemporal multi-layer functional connection network according to claim 7 is characterized in that: The graph in the spatial multi-layer functional connection network is represented by a thresholded binary adjacency matrix, and the functional connection between two nodes is obtained by calculating the Pearson correlation of their blood oxygen level dependent time series signals, which is defined as follows: Where r is the Pearson correlation coefficient; X i , Y i Respectively represent the value of the i-th point of the blood oxygen level dependent time series signal X, Y, Represent the mean of blood oxygen level dependent time series signals X and Y respectively; 9. The modeling method of the whole-brain spatiotemporal multi-layer functional connection network according to claim 8 is characterized in that: A three-layer functional connection network is constructed at the spatial scale. The first layer of the functional connection network consists of the association matrix H of the hypergraph ij Representation to describe high-order correlations between large-scale resting-state networks in gray matter; The second layer of the functional connection network is composed of the adjacency matrix Representations are made to describe low-order connections between regions of interest within each large-scale resting-state network; The third layer functional connection network is composed of the adjacency matrix Represented to describe the low-order association between gray matter regions of interest and white matter regions of interest; the adjacency matrix A used ij It is expressed as: r ij is the Pearson correlation coefficient between the blood oxygen level-dependent time series signals of node i and node j, which is regarded as the functional connection strength here; the threshold of each brain FCN is determined by the average value of all functional connection strengths of the current layer plus one standard deviation.

Images