Neurological Network Spatiotemporal Metric Method, Device, Electronic Device and Storage Medium

Abstract

Embodiments of the present disclosure disclose a method, apparatus, electronic device, and storage medium for spatio-temporal measurement of a neurological network. Among them, a method for spatio-temporal measurement of a neurological network includes: a step of selecting a region of interest, wherein, based on the original relaxation time weighted imaging data of magnetic resonance imaging, a first registration is performed to obtain the relaxation time weighted imaging data after the first registration, and an accurate atlas and a region of interest are obtained based on the relaxation time weighted imaging data after the first registration; a fiber bundle analysis step, wherein, based on the diffusion tensor imaging data of the magnetic resonance imaging, a first preprocessing and a second registration are performed to obtain the diffusion tensor imaging data after the second registration, and fiber bundle construction and processing are performed on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result, the fiber bundle analysis result including a structural network; a network topology analysis step; and a spatio-temporal measurement step.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technologies, and particularly to a spatio-temporal metric method, apparatus, electronic device, and storage medium for a neurological network. Background Art

[0002] Magnetic Resonance Imaging (MRI) technology is one of the main ways to obtain neuroimaging data. Informatics methods and neuroimaging technologies have promoted research on exploring the neurostructural and functional characteristics of, for example, the human brain. Various brain projects and public datasets in the past two decades have given rise to numerous research tools. Classic methods such as radiomics, connectomics, deep learning, and probabilistic reasoning in the field of informatics have achieved research results such as the brain default mode, human brain atlas, and brain simulation models. Currently, in the field of neurological research directly related to life and health, there is a lack of intelligent system tools for mining personalized and accurate information on neurostructural and functional characteristics of, for example, the human brain after nerve tissue damage, and it cannot meet the measurement requirements for personalized and long-term evolution function changes caused by, for example, the growth of brain nerve cells.

[0003] Those of ordinary skill in the art can understand that similar problems also exist in the research on other highly nerve-intensive parts such as the spinal cord besides the brain. Similar problems also exist in the research on highly nerve-intensive parts of other animals such as mice and monkeys besides humans.

[0004] For example, in the following non-patent literature published in IEEE Trans Med Imaging 2018.37(7):1537-1550, Topology Data Analysis (TDI) was used to study the topological differences among different populations and the influence of perturbations on topological features in an undirected brain network manner, but the analysis results were not accurate and comprehensive enough, and the interpretability was insufficient.

[0005] <Prior Art Documents>

[0006] <Non-Patent Document> "Connectivity in fMRI: Blind Spots and Breakthroughs" (Chinese translation: "Connectivity in Functional Magnetic Resonance Imaging: Blind Spots and Breakthroughs") Summary of the Invention

[0007] To solve the problems in the related art, embodiments of the present disclosure provide a spatio-temporal metric method, apparatus, electronic device, and storage medium for a neurological network.

[0008] In a first aspect, embodiments of the present disclosure provide a spatio-temporal metric method for a neurological network, including:

[0009] Region of interest selection step, in which first registration is performed based on the original relaxation time weighted imaging data of magnetic resonance imaging to obtain the relaxation time weighted imaging data after the first registration, and an accurate atlas and a region of interest are obtained based on the relaxation time weighted imaging data after the first registration;

[0010] Fiber bundle analysis step, in which first preprocessing and second registration are performed based on the diffusion tensor imaging data of the magnetic resonance imaging to obtain the diffusion tensor imaging data after the second registration, and fiber bundle construction and processing are performed on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result, and the fiber bundle analysis result includes a structural network;

[0011] Network topology analysis step, in which data processing and analysis are performed based on specific functional data to obtain the specific functional data after data processing and analysis, a functional network is constructed from the specific functional data after data processing and analysis, and a topological descriptor is obtained by performing network topology analysis on at least one of the functional network and the structural network.

[0012] Combined with the first aspect, in the first implementation manner of the first aspect, the present disclosure further includes:

[0013] Space-time metric step, in which a space-time metric result is calculated based on the fiber bundle analysis result of the fiber bundle analysis step and the topological descriptor of the network topology analysis step.

[0014] Combined with the first aspect, in the second implementation manner of the first aspect, the magnetic resonance imaging includes: brain magnetic resonance imaging and / or spinal cord magnetic resonance imaging; and / or

[0015] The original relaxation time weighted imaging data includes: original longitudinal relaxation time weighted imaging data or original transverse relaxation time weighted imaging data; and / or

[0016] The relaxation time weighted imaging data after the first registration includes: the longitudinal relaxation time weighted imaging data after the first registration or the transverse relaxation time weighted imaging data after the first registration.

[0017] The accurate atlas includes: a corrected atlas.

[0018] Combined with the second implementation manner of the first aspect, in the third implementation manner of the first aspect, the first registration is performed based on the original relaxation time weighted imaging data of the magnetic resonance imaging to obtain the relaxation time weighted imaging data after the first registration, including:

[0019] Bone removal registration sub-step, in which bone removal registration is performed on the original relaxation time weighted imaging data to obtain the relaxation time weighted imaging data after the first registration; or

[0020] A bone registration sub-step, in which bone registration is performed on the original relaxation time weighted imaging data to obtain the first registered relaxation time weighted imaging data.

[0021] Combined with the third implementation manner of the first aspect, in the fourth implementation manner of the first aspect of the present disclosure, the de-boning registration sub-step includes:

[0022] Use a first command to remove bone data from the original relaxation time weighted imaging data to obtain relaxation time weighted imaging data without bone;

[0023] Use a second command to register the relaxation time weighted imaging data without bone to obtain the first registered relaxation time weighted imaging data; and / or

[0024] The bone registration sub-step includes:

[0025] Use a third command to register the original relaxation time weighted imaging data to obtain registered relaxation time weighted imaging data with bone;

[0026] Use a fourth command to remove bone data from the registered relaxation time weighted imaging data with bone to obtain the first registered relaxation time weighted imaging data,

[0027] The bone includes: the skull.

[0028] Combined with the fourth implementation manner of the first aspect, in the fifth implementation manner of the first aspect of the present disclosure, the first command includes:

[0029] A combination of the bet or bet2 command in FSL, the flirt command based on the first template, the betsurf command, and the fslmaths command; or

[0030] The fslmaths command in FSL based on the second template; or

[0031] The bet2 command in FSL; or

[0032] The mri_watershed command in FreeSurfer; and / or

[0033] The second command includes:

[0034] The flirt command in FSL based on the second template; and / or

[0035] The third command includes:

[0036] The flirt command in FSL based on the first template; and / or

[0037] The fourth command includes:

[0038] The bet2 command in the FSL; or

[0039] The fslmaths command in the FSL based on the second template.

[0040] Combined with the fifth implementation manner of the first aspect, in the sixth implementation manner of the first aspect, the first template includes: the MNI152_T1_1mm template, or the MNI152_T2_1mm template; and / or

[0041] The second template includes: the MNI152_T1_brain_1mm template, or the MNI152_T2_brain_1mm template.

[0042] Combined with the first aspect, in the seventh implementation manner of the first aspect of the present disclosure,

[0043] Obtaining the accurate atlas and the region of interest based on the first registered relaxation time weighted imaging data includes:

[0044] Using the first registered relaxation time weighted imaging data and combining with the standard MNI space atlas to obtain the accurate atlas and the region of interest.

[0045] Combined with the first aspect, in the eighth implementation manner of the first aspect of the present disclosure, performing the first preprocessing and the second registration on the diffusion tensor imaging data based on the magnetic resonance imaging to obtain the second registered diffusion tensor imaging data includes:

[0046] Using the fslroi DWI nodif instruction of FSL to extract the b0 image in the diffusion tensor imaging data, removing the bones in the b0 image with bet2 to obtain the bone-removed b0 image mask (nodif_brain_mask), and using the instruction fslmaths DWI–mas nodif_brain_mask rDWI to remove the bones for the diffusion tensor imaging data by using the mask. Performing eddy current correction and the first motion correction on the bone-removed diffusion tensor imaging data to obtain the preprocessed diffusion tensor imaging data;

[0047] For the preprocessed diffusion tensor imaging data, perform linear registration using the MNI152_T1_brain_2mm template or the MNI152_T2_brain_2mm template, or perform linear registration using the MNI152_T1_brain_2mm template or the MNI152_T2_brain_2mm template and then generate a mask using the template to remove the bones again to obtain the second registered diffusion tensor imaging data, or

[0048] Use 3D Slicer to remove the skull, perform voxel modeling and registration to obtain the second registered diffusion tensor imaging data.

[0049] Combined with the first aspect, in the ninth implementation manner of the first aspect of the present disclosure, the fiber bundle construction and processing of the second registered diffusion tensor imaging data to obtain the fiber bundle analysis result includes:

[0050] A fiber bundle tracking sub-step, wherein fiber bundle tracking is performed on the second registered diffusion tensor imaging data to obtain a fiber bundle tracking result;

[0051] A fiber bundle analysis sub-step, wherein fiber bundle analysis is performed on the fiber bundle tracking result by combining the accurate atlas and the region of interest to obtain the fiber bundle analysis result.

[0052] Combined with the ninth implementation manner of the first aspect, in the tenth implementation manner of the first aspect of the present disclosure, the fiber bundle tracking sub-step includes:

[0053] Calculate the anisotropy index and the mean diffusivity from the second registered diffusion tensor imaging data;

[0054] Use a fiber bundle tracking method to perform fiber bundle tracking on the second registered diffusion tensor imaging data and calculate the tracking error;

[0055] Select the fiber bundle tracking method with the smallest tracking error to obtain the fiber bundle tracking result.

[0056] Combined with the tenth implementation manner of the first aspect, in the eleventh implementation manner of the first aspect of the present disclosure, the fiber bundle tracking method includes at least one of the following:

[0057] The EuDX method, the deterministic method, the probabilistic method, the sfm method, or

[0058] The xtract instruction of FSL, the Tractography seeding instruction of 3D slicer.

[0059] Combined with the ninth implementation manner of the first aspect, in the twelfth implementation manner of the first aspect of the present disclosure, the fiber bundle analysis sub-step includes:

[0060] Based on the accurate atlas and the region of interest, according to the fiber bundle tracking result, calculate the number, average length, length distribution of the fiber bundles passing through the region of interest, and / or the number, average length, length distribution of the fiber bundles with the region of interest as the end points as the fiber bundle analysis result; and / or

[0061] Based on the accurate atlas and the region of interest, according to the fiber bundle tracking result, calculate the fiber bundle connectivity of each voxel in the region of interest, and / or the average fiber bundle connectivity of the voxels in the region of interest as the fiber bundle analysis result; and / or

[0062] Construct a structural network with the region of interest as the vertices and the number of fiber bundles between the regions of interest as the edge weights as the fiber bundle analysis result.

[0063] Combined with the first aspect, in the thirteenth implementation manner of the first aspect of the present disclosure,

[0064] The data processing and analysis based on the specific functional data, obtaining the specific functional data after data processing and analysis, and constructing a functional network from the specific functional data after data processing and analysis includes:

[0065] Perform second preprocessing and third registration on the blood oxygenation level-dependent data based on functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after the third registration, and construct a functional network from the blood oxygenation level-dependent data after the third registration, or

[0066] Perform data processing and analysis on the electroencephalogram data to obtain the electroencephalogram data after data processing and analysis, and construct a functional network from the electroencephalogram data after data processing and analysis.

[0067] Combined with the thirteenth implementation manner of the first aspect, in the fourteenth implementation manner of the first aspect of the present disclosure, the performing second preprocessing and third registration on the blood oxygenation level-dependent data based on functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after the third registration includes:

[0068] Remove the first number of time points of the blood oxygenation level-dependent data based on functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after removing the time points;

[0069] Perform slice time correction and second motion correction on the blood oxygenation level-dependent data after removing the time points, and register it to the MNI space using the echo planar imaging template to obtain the blood oxygenation level-dependent data in the MNI space;

[0070] Resample the blood oxygenation level-dependent data of the MNI space, after using the first kernel for smoothing, perform Detrend, remove covariates, and conduct the first bandwidth filtering to obtain the third registered blood oxygenation level-dependent data.

[0071] Combined with the fourteenth implementation manner of the first aspect, in the fifteenth implementation manner of the first aspect of the present disclosure, the first quantity includes: 0 to 10; and / or

[0072] The resampling includes: resampling at a resolution of 2x2x2 mm or 3x3x3 mm; and / or

[0073] The first kernel includes: a FWHM 4 mm Gauss kernel, or a FWHM 6 mm Gauss kernel, or a FWHM 8 mm Gauss kernel; and / or

[0074] The first bandwidth includes: 0.01 to 0.08 Hz, or 0.02 to 0.09 Hz, or 0.01 to 0.10 Hz.

[0075] Combined with the thirteenth implementation manner of the first aspect, in the sixteenth implementation manner of the first aspect of the present disclosure,

[0076] The data processing and analysis based on the electroencephalogram data to obtain the electroencephalogram data after data processing and analysis includes:

[0077] Filter the electroencephalogram data, perform independent component analysis, remove artifacts, and combine with the lead field matrix obtained by modeling through a head model and a source model from the registered relaxation time weighted imaging data to conduct signal tracing and obtain the electroencephalogram data after data processing and analysis.

[0078] Combined with the thirteenth implementation manner of the first aspect, in the seventeenth implementation manner of the first aspect of the present disclosure, the construction of the functional network from the third registered blood oxygenation level-dependent data includes:

[0079] Use the third registered blood oxygenation level-dependent data, based on the accurate atlas and the region of interest, take the voxels in the region of interest as network vertices, and perform correlation estimation on the time series of the voxels as the weights of the edges to obtain a voxel-level local undirected network; and / or

[0080] Use the third registered blood oxygenation level-dependent data, based on the accurate atlas, take the sub-regions within the global region corresponding to the accurate atlas as network vertices, calculate the time series of the sub-regions within the global region, and the interaction strength between the sub-regions within the global region as the weights of the edges to obtain a region-level global undirected network; and / or

[0081] Using the third registered blood oxygenation level-dependent data, based on the accurate atlas and the region of interest, taking the voxels in the region of interest as network vertices, and using a directed functional connectivity estimation method to obtain the weights of the edges, a voxel-level local directed network is obtained; and / or

[0082] Using the third registered blood oxygenation level-dependent data, based on the accurate atlas, taking the sub-regions in the global region corresponding to the accurate atlas as network vertices, and using the directed functional connectivity estimation method to obtain the weights of the edges, a region-level global directed network is obtained.

[0083] Combined with the thirteenth implementation manner of the first aspect, in the eighteenth implementation manner of the first aspect of the present disclosure,

[0084] Constructing a functional network from the EEG data after data processing and analysis includes:

[0085] Performing band division on the EEG data after data processing and analysis to obtain multi-band EEG data;

[0086] Using a directed functional connectivity estimation method for the multi-band EEG data to construct a directed EEG functional network, or

[0087] Using an undirected functional connectivity estimation method for the multi-band EEG data to construct a directed EEG functional network.

[0088] Combined with any one of the seventeenth implementation manner or the eighteenth implementation manner of the first aspect, in the nineteenth implementation manner of the first aspect of the present disclosure, the directed functional connectivity estimation method includes at least one of the following:

[0089] Cross-correlation method, Granger causality analysis method, convergent cross mapping method.

[0090] Combined with the eighteenth implementation manner of the first aspect, in the twentieth implementation manner of the first aspect of the present disclosure, the undirected functional connectivity estimation method includes at least one of the following:

[0091] Phase locking value method, coherence method, weighted phase lag coefficient method.

[0092] Combined with the first aspect, in the twenty-first implementation manner of the first aspect of the present disclosure, the obtaining of topological descriptors by performing network topology analysis on at least one of the functional network and the structural network includes:

[0093] Constructing a specific simplicial complex according to at least one of the functional network and the structural network;

[0094] Constructing a homology group according to the specific simplicial complex;

[0095] Analyze the topological characteristics of the homology group to obtain the topological descriptor;

[0096] Obtain the topological descriptor according to the specific simplicial complex.

[0097] Combined with the twenty-first implementation manner of the first aspect, in the twenty-second implementation manner of the first aspect of the present disclosure, the constructing a specific simplicial complex according to at least one of the functional network and the structural network includes:

[0098] Construct a specific simplicial complex according to at least one of the functional network and the structural network by means of screening.

[0099] Combined with the twenty-second implementation manner of the first aspect, in the twenty-third implementation manner of the first aspect of the present disclosure, the constructing a specific simplicial complex according to at least one of the functional network and the structural network further includes:

[0100] Construct a comparison simplicial complex using a null model;

[0101] Compare the specific simplicial complex and the comparison simplicial complex.

[0102] Combined with the twenty-first implementation manner of the first aspect, in the twenty-fourth implementation manner of the first aspect of the present disclosure, the topological descriptor includes at least one of the following:

[0103] Topological descriptor below the threshold, persistent homology topological descriptor.

[0104] Combined with the first implementation manner of the first aspect, in the twenty-fifth implementation manner of the first aspect of the present disclosure, the space-time metric step includes: calculating the simplex change characteristics based on the topological descriptor of the network topology analysis step, and combining the fiber bundle analysis result of the fiber bundle analysis step to obtain the space-time metric result.

[0105] Combined with the twenty-fifth implementation manner of the first aspect, in the twenty-sixth implementation manner of the first aspect of the present disclosure, the simplex change characteristics include at least one of the following:

[0106] Simplex overlap rate, simplex evolution coefficient.

[0107] In a second aspect, an apparatus for space-time metric of a neurological network provided in an embodiment of the present disclosure includes:

[0108] A region of interest selection module, configured to perform a first registration based on the original relaxation time weighted imaging data of magnetic resonance imaging to obtain the relaxation time weighted imaging data after the first registration, and obtain an accurate atlas and a region of interest based on the relaxation time weighted imaging data after the first registration;

[0109] A fiber bundle analysis module, configured to perform first preprocessing and second registration based on the diffusion tensor imaging data of the magnetic resonance imaging, obtain the diffusion tensor imaging data after the second registration, perform fiber bundle construction and processing on the diffusion tensor imaging data after the second registration, and obtain a fiber bundle analysis result;

[0110] A network topology analysis module, configured to perform data processing and analysis based on specific function data, obtain the specific function data after data processing and analysis, construct a functional network from the specific function data after data processing and analysis, and obtain a topological descriptor through network topology analysis of at least one of the functional network and the structural network.

[0111] Combined with the second aspect, in the first implementation manner of the second aspect, the present disclosure further includes:

[0112] A spatio-temporal metric module, configured to calculate a spatio-temporal metric result based on the fiber bundle analysis result of the fiber bundle analysis step and the topological descriptor of the network topology analysis step.

[0113] In a third aspect, an electronic device is provided in an embodiment of the present disclosure, including a memory and a processor; wherein,

[0114] 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 according to any one of the first aspect, the first to twenty-sixth implementation manners of the first aspect.

[0115] In a fourth aspect, a readable storage medium is provided in an embodiment of the present disclosure, on which computer instructions are stored, and when the computer instructions are executed by a processor, the method according to any one of the first aspect, the first to twenty-sixth implementation manners of the first aspect is implemented.

[0116] The technical solution provided by the embodiment of the present disclosure may include the following beneficial effects:

[0117] According to the technical solution provided by the embodiments of the present disclosure, through the region of interest selection step, wherein, based on the original relaxation time weighted imaging data of magnetic resonance imaging, a first registration is performed to obtain the relaxation time weighted imaging data after the first registration, and an accurate atlas and a region of interest are obtained based on the relaxation time weighted imaging data after the first registration; a fiber bundle analysis step, wherein, based on the diffusion tensor imaging data of the magnetic resonance imaging, a first preprocessing and a second registration are performed to obtain the diffusion tensor imaging data after the second registration, and fiber bundle processing is performed on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result, and the fiber bundle analysis result includes a structural network; a network topology analysis step, wherein, based on specific functional data, data processing and analysis are performed to obtain the specific functional data after the data processing and analysis, a functional network is constructed from the specific functional data after the data processing and analysis, and a topological descriptor is obtained by performing network topology analysis on at least one of the functional network and the structural network, so as to realize accurate dynamic monitoring of the state of highly dense neural tissues such as the brain and accurate measurement of functional changes at the levels of neural regions, circuits, and systems such as brain regions, and can be extended to research fields such as non-invasive brain-computer interfaces.

[0118] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. Brief Description of the Drawings

[0119] In combination with the drawings, through the following detailed description of non-limiting embodiments, other features, objects, and advantages of the present disclosure will become more obvious. In the drawings:

[0120] Figure 1a An exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0121] Figure 1b An exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0122] Figure 1c An exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0123] Figure 1d An exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0124] Figure 1e An exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0125] Figure 1fExemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0126] Figure 1g Exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0127] Figure 1h Exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0128] Figure 1i Exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0129] Figure 1j Exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0130] Figure 1k Exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0131] Figure 2 Flowchart showing a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0132] Figure 3 Flowchart showing a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0133] Figure 4 Flowchart showing a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0134] Figure 5 Flowchart showing a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0135] Figure 6 Flowchart showing a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0136] Figure 7 Flowchart showing a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure;

[0137] Figure 8 Block diagram showing the structure of a spatio-temporal metric device for a neurological network according to an embodiment of the present disclosure;

[0138] Figure 9 Block diagram showing the structure of a spatio-temporal metric device for a neurological network according to an embodiment of the present disclosure;

[0139] Figure 10Shows a structural block diagram of an electronic device according to an embodiment of the present disclosure;

[0140] Figure 11 Is a schematic structural diagram of a computer system suitable for implementing a spatio-temporal metric method of a neurological network according to an embodiment of the present disclosure. Detailed implementation manners

[0141] Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement them. In addition, for clarity, parts irrelevant to the description of the exemplary embodiments are omitted in the drawings.

[0142] In the present disclosure, it should be understood that terms such as "including" or "having" are intended to indicate the existence of the labels, numbers, steps, actions, components, parts, or combinations thereof disclosed in this specification, and are not intended to exclude the possibility of the existence or addition of one or more other labels, numbers, steps, actions, components, parts, or combinations thereof.

[0143] In addition, it should be noted that, without conflict, the embodiments in the present disclosure and the labels in the embodiments can be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0144] Magnetic resonance imaging technology is one of the main ways to obtain neuroimaging data. Informatics methods and neuroimaging technologies have promoted the research on exploring the neural structure and functional characteristics of, for example, the human brain. Various brain projects and public data sets in the past 20 years have given rise to numerous research tools. Classic methods such as radiomics, connectomics, as well as deep learning and probabilistic inference in the field of informatics have achieved research results such as the brain default mode, human brain atlas, and brain simulation models. Currently, in the field of neurological research directly related to life and health, there is a lack of intelligent system tools for mining personalized and accurate information on neural structure and functional characteristics of, for example, the human brain after nerve tissue damage, and it cannot meet the measurement requirements for personalized and long-term evolving functional changes caused by, for example, the growth of brain nerve cells.

[0145] Those of ordinary skill in the art can understand that similar problems also exist in the research on other highly nerve-intensive parts such as the spinal cord in addition to the brain. Similar problems also exist in the research on highly nerve-intensive parts of other animals such as mice and monkeys in addition to humans.

[0146] To solve the above problems, the present disclosure proposes a spatio-temporal metric method, device, electronic device, and storage medium for a neurological network.

[0147] Figure 1a Shows an exemplary schematic diagram of an implementation scenario of a spatio-temporal metric method of a neurological network according to an embodiment of the present disclosure.

[0148] Those of ordinary skill in the art can understand that Figure 1a Exemplarily shows an implementation scenario of the neurological network spatio-temporal metric method, without constituting a limitation to the present disclosure.

[0149] As Figure 1a Shown, in implementation scenario 100 of the neurological network spatio-temporal metric method, the original relaxation time weighted imaging data 1011 based on MRI is registered to the standard MNI (Montreal Neurological Institute) space through step S1011, obtaining the registered relaxation time weighted imaging data 1012. The registered relaxation time weighted imaging data 1012 is combined with the MNI standard space atlas 1000. After atlas correction and region of interest selection in step S1012, the accurate atlas and region of interest 1013 are obtained.

[0150] In an embodiment of the present disclosure, the original relaxation time weighted imaging data 1011 can be the original longitudinal relaxation time weighted imaging data T1 of MRI, or the original transverse relaxation time weighted imaging data T2. The present disclosure does not make a limitation thereto. Correspondingly, the registered relaxation time weighted imaging data 1012 can be the registered longitudinal relaxation time weighted imaging data, or the registered transverse relaxation time weighted imaging data. The present disclosure does not make a limitation thereto. Due to local deformations caused by reasons such as brain atrophy and lesions of the subject, the brain region positions of the subject can be different from those of the standard human body. In atlas correction and region of interest selection in step S1012, the MNI standard space atlas 1000 can be combined to correct the deformed brain, obtaining an accurate atlas such as the corrected atlas. If the brain is not deformed, the MNI standard space atlas 1000 can be used as the accurate atlas. Meanwhile, according to the research requirements, a region of interest (ROI) is selected from all brain regions.

[0151] Those of ordinary skill in the art can understand that the above registration and atlas correction methods can also be used for highly nerve-dense regions such as the spinal cord, or for other animals such as mice and monkeys other than humans. The present disclosure does not make a limitation thereto.

[0152] Figure 1b An exemplary schematic diagram showing an implementation scenario of the neurological network spatio-temporal metric method according to an embodiment of the present disclosure.

[0153] Those of ordinary skill in the art can understand that Figure 1b Exemplarily shows an implementation scenario of the neurological network spatio-temporal metric method, without constituting a limitation to the present disclosure.

[0154] Figure 1b Specifically shows Figure 1aAn exemplary specific process of obtaining an accurate map and a region of interest 1013 from the original relaxation time weighted imaging data 1011 through steps S1011 and S1012.

[0155] In Figure 1b T1 data 1111 can be obtained from a brain MRI. Figure 1a The original longitudinal relaxation time weighted imaging data of the original relaxation time weighted imaging data 1011 in

[0156] As Figure 1b shown, any one of the three steps S1111, S1112, and S1113 can be used to remove the skull from the T1 data 1111 to obtain the relaxation time weighted imaging data with the bones removed, i.e., the T1_brain data 1114.

[0157] Step S1111 uses a combination of the Bet or bet2, flirt, betsurf, and fslmaths commands of the FMRIB Software Library (FSL) platform to remove the skull; step S1112 uses the fslmaths command to remove the skull based on the MNI152_T1_brain_1mm template 1112; step S1113 uses the bet2 command to remove the skull.

[0158] The T1_brain data 1114 undergoes step S1114 and is registered using the flirt command based on the MNI152_T1_brain_1mm template 1115 to obtain the first registered relaxation time weighted imaging data, i.e., the T1_brain_w data 1116. The T1_brain_w data 1116 is combined with the first MNI space map 1117 and undergoes map correction in step S1115 and the first region of interest acquisition process to obtain the accurate map and the region of interest 1013.

[0159] In the embodiments of the present disclosure, the above steps S1111, S1112, S1113, S1114, and S1115 constitute a skull removal and registration step of removing the skull first and then registering.

[0160] In the above skull removal and registration step of removing the skull first and then registering, the mri_watershed command in FreeSurfer can also be used to remove the skull.

[0161] In the embodiments of the present disclosure, it is also possible to adopt a method of registering with the skull first and then removing the skull to obtain the first registered relaxation time weighted imaging data and combine it with the MNI space map to obtain the accurate map and the region of interest.

[0162] The T1 data 1111 can first be registered using the flirt command S1211 based on the MNI152_T1_1mm template 1121 to obtain the registered relaxation time weighted imaging data with bones, i.e., the T1_w data 1112. The T1_w data 1112 is processed by the bet2 command S1222 or the fslmaths command S1123 based on the MNI152_T1_brain_1mm template 1123 to remove the skull, obtaining the first registered relaxation time weighted imaging data, i.e., the T1_w_brain data 1124. The T1_w_brain data 1124 is combined with the second MNI spatial atlas 1125 and, through the atlas correction in step S1124 and the second processing of obtaining the region of interest, the accurate atlas and the region of interest 1013 are obtained. Those of ordinary skill in the art can understand that the first MNI spatial atlas 1117 and the second MNI spatial atlas 1125 can be the same standard MNI spatial atlas or different, and the present disclosure does not limit this.

[0163] Those of ordinary skill in the art can understand that in addition to using the above MNI152_T1_1mm template and MNI152_T1_brain_1mm template with a resolution of 1 mm, templates with other resolutions or templates for other parts such as the spine can also be used to register the relaxation time weighted imaging data of other parts except the head, and the present disclosure does not limit this.

[0164] Those of ordinary skill in the art can understand that in addition to using the MNI152_T1_1mm template and MNI152_T1_brain_1mm template for the longitudinal relaxation time weighted imaging data, i.e., the T1 data, the MNI152_T2_1mm template, MNI152_T2_brain_1mm template, or T2 templates with other resolutions and for other parts can also be used for the transverse relaxation time weighted imaging data, i.e., the T2 data, for deboning and registration, and the present disclosure does not limit this.

[0165] In Figure 1a , the diffusion tensor imaging (DTI) data of MRI can also be used. The diffusion tensor imaging data 1021 undergoes the first preprocessing and the second registration in S1021 to obtain the second registered diffusion tensor imaging data 1022. The second registered diffusion tensor imaging data 1022 undergoes fiber tractography in S1022 and fiber bundle analysis in S1023 to obtain the fiber bundle analysis result 1023.

[0166] Figure 1c An exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure.

[0167] Those of ordinary skill in the art can understand that Figure 1c Exemplarily shows an implementation scenario of the spatio-temporal metric method for a neural network, without constituting a limitation to the present disclosure.

[0168] Figure 1c Specifically shows Figure 1a The exemplary specific process in [reference] from diffusion tensor imaging data 1021 through steps S1021, S1022, and S1023 to obtain the fiber bundle analysis result 1023.

[0169] In Figure 1c In [reference], the diffusion tensor imaging data 1021 undergoes a first preprocessing in S1201 to obtain the first preprocessed diffusion tensor imaging data 1201. In the first preprocessing of S1201, using the FSL tool, the b0 image is extracted, bones such as the skull are removed to obtain a mask, the bones of the DWI image of the diffusion tensor imaging data are removed, and eddy current correction and a first motion correction such as head motion correction are performed on the diffusion tensor data after bone removal, and the diffusion tensor and related parameters are calculated.

[0170] Those of ordinary skill in the art can understand that, according to actual needs, the spine can also be removed, and a first motion correction for spine motion can be performed, etc. The present disclosure does not limit this.

[0171] The first preprocessed diffusion tensor imaging data 1201 undergoes a second registration in S1202 to obtain the second registered diffusion tensor imaging data 1202. The b0 is registered to the 2mm resolution template in the MNI space (the resolution closest to the acquired DTI signal) using the linear registration algorithm (flirt command) of FSL, and the coordinate space is 91×109×91 voxels. For the remaining DWI (Diffusion Weighting Imaging) images in each gradient direction, using the transformation matrix for registering the b0 to the MNI space as a reference, they are registered to the MNI space using the flirt command, or after registration using the flirt command, a mask is generated using the template to remove bones again to obtain the DTI data in the MNI space, that is, the second registered diffusion tensor imaging data 1202. The second registration can perform linear registration using the MNI152_T1_brain_2mm template or the MNI152_T2_brain_2mm template, or templates with other resolutions. The present disclosure does not limit this.

[0172] In the embodiments of the present disclosure, 3D slicer can also be used to remove the skull from the diffusion tensor imaging data 1021, and perform voxel modeling and registration using Diffusion Tensor Estimation to obtain the second registered diffusion tensor imaging data 1202.

[0173] After the second registration, the diffusion tensor imaging data 1202 is used by S1203 to calculate the FA and MD values using the Python toolkit Dipy. FA is Fractional Anisotropy, the fractional anisotropy; MD is Mean Diffusion, the mean diffusivity.

[0174] S1204 uses the threshold of FA as the termination condition, selects the fiber tractography method with the minimum tracking error, and obtains the fiber tractography result 1203.

[0175] In the embodiments of the present disclosure, at least one of the six methods of EuDX method, deterministic method, probabilistic method, sfm method, xtract instruction in FSL, and Tractography seeding instruction in 3D slicer can be used as the tracking algorithm, and the FA threshold is used as the termination condition to perform whole-brain fiber tractography on the DTI data in the MNI space, and the LiFE algorithm is used to calculate the tracking errors of the whole brain, the damaged area, the healthy side and the corresponding area of the damaged area, and the fiber tractography method with the minimum tracking error is selected. The error calculation principle of the LiFE algorithm is as follows: According to the obtained tracking result, the DTI signal is reconstructed, the reconstructed DTI signal is subtracted from the DTI signal obtained by real scanning, and the mean square value of the error of each voxel is calculated as the error of the tracking result.

[0176] According to the voxel reconstruction model of DTI, assuming that the fiber bundle tangent direction v1 of each voxel passed by the fiber bundle is the main diffusion direction, for the diffusion matrix D: λ1 = 0.5, λ2 = λ3 = 0,

[0177] v1 is the tangent direction of the fiber bundle at this voxel (known), then the diffusion matrix D of each voxel can be obtained

[0178]

[0179] After calculating the diffusion matrix of each voxel, the DTI signal can be predicted according to S1 = S0exp(x j T Dx j (b1 - b0)). Where x j is the magnetic field gradient direction at b1 (known), and S0 is the signal value of the b0 image.

[0180] The finally calculated tracking error is:

[0181]

[0182] Among them, n is the number of gradient directions of the DWI image in DTI, and x, y, and z are voxel coordinates.

[0183] In the embodiments of the present disclosure, according to the fiber tractography result 1203, combined with the accurate atlas and the region of interest 1013, in S1206, a sub-region in the region of interest can be selected, and the number and length distribution of fiber tracts passing through or having one end in the sub-region are calculated to obtain the fiber tract analysis result 1023.

[0184] In the embodiments of the present disclosure, in combination with the accurate atlas, two methods are used to define the vertices of the structural brain network, namely the brain region-based method and the voxel-based method. The brain region-based method regards the entire brain region in the human brain atlas as a vertex, and all fiber tracts connected to the voxels in the brain region are considered to be connected to this brain region vertex. The voxel-based method regards each voxel as a vertex of the network, and generally focuses on the network formed by the voxels in the brain region of the human brain atlas as vertices.

[0185] The definition of the edges of the structural brain network is as follows: If there is a fiber tract connection between two vertices, there is an edge connection between these two vertices, and the number of fiber tracts connecting the vertices is used to estimate the weight of the edge. After the brain network is established, the regions in the atlas that can contain the damaged area and the brain regions corresponding to the transplantation target are selected as the regions of interest (ROI), and the number of fiber tracts passing through the ROI, the average length, the length distribution, and the number of fiber tracts with one end point in the ROI, the average length, and the length distribution are respectively counted.

[0186] Taking each voxel in the ROI as a vertex and the fiber tract as the edge connecting the voxel vertices, since the fiber tract directions are different, in three-dimensional space, each voxel may have fiber tracts passing through in multiple directions. The number of fiber tracts passing through each voxel vertex is recorded as the fiber tract connectivity of the voxel vertex. The sum of the connectivities of all voxels in the ROI is statistically calculated, and then divided by the total number of voxels in the ROI to calculate the average fiber tract connectivity of each voxel in the ROI, or "average connectivity", which characterizes the number and length of fiber tracts in the ROI. Taking the "average connectivity" of the corresponding brain region on the healthy side as a reference, the relative average connectivity in the ROI on the affected side is calculated to reflect the characteristic differences of the nerve fibers in the bilateral hemispheres of the patient. Among the DTI data collected during multiple follow-ups, the relative average connectivity can be used to measure the fiber tract structure changes within each time interval.

[0187] In the embodiments of the present disclosure, according to the fiber tractography result 1203, combined with the accurate atlas and the region of interest 1013, in S1205, a sub-region in the global region can be used as an end point, and the number of fiber tracts between the sub-regions is used as the edge weight to construct a structural network to obtain the fiber tract analysis result 1023.

[0188] In an embodiment of the present disclosure, the fiber bundle analysis result 1023 may include at least one of the following: the number of fiber bundles passing through the ROI, the average length, the length distribution, the number of fiber bundles with one end point within the ROI, the average length, the length distribution, the fiber bundle connectivity of each voxel within the ROI, the average fiber bundle connectivity of the voxels within the ROI, and the structural network.

[0189] In an embodiment of the present disclosure, the structural network may characterize the structural connection relationship between the axons of nerve cells in the white matter of the brain.

[0190] In an embodiment of the present disclosure, in Figure 1a it is also possible to analyze the blood oxygenation level-dependent (BOLD) data of functional magnetic resonance imaging (fMRI).

[0191] As Figure 1a shown, for example, after the specific functional data 1031 of resting blood oxygenation level-dependent (rBOLD) data undergoes step S1031 data processing and analysis and step S1032 network construction, a functional network 1032 is obtained. Step S1033 network topology analysis is based on an accurate atlas and the region of interest 1013, processes the functional network 1032 and the structural network in the fiber bundle analysis result, and obtains a topological descriptor 1033.

[0192] Those of ordinary skill in the art can understand that using resting blood oxygenation level-dependent data, the activities of the human body in a resting state, such as the brain, can be analyzed, and the environment is relatively simple; while using motion blood oxygenation level-dependent data, the activities of the brain, for example, during activities such as conversations, fist clenching, and passive movements, can be analyzed. The environment is more complex, the data is more flexible, and the adaptability of the analysis is stronger. The present disclosure does not make any limitations in this regard.

[0193] Figure 1d FIG. shows an exemplary schematic diagram of an implementation scenario of a neurological network spatio-temporal metric method according to an embodiment of the present disclosure.

[0194] Those of ordinary skill in the art can understand that Figure 1d exemplarily shows an implementation scenario of a neurological network spatio-temporal metric method, without constituting a limitation to the present disclosure.

[0195] Figure 1d FIG. shows Figure 1a an exemplary specific process of constructing the functional network 1032 from the blood oxygenation level-dependent data 1031 through steps S1031 and S1032.

[0196] In an embodiment of the present disclosure, the S1031 data processing and analysis may include:

[0197] Using the DPARSFA tool for BOLD data, first removing the first 0 to 10 time points, then performing slice timing correction and a second motion correction such as head motion correction, registering to the MNI space using an echo planar imaging (EPI) template, and then resampling the data to a resolution of 2x2x2 mm or 3x3x3 mm (the resolution closest to the acquired rBOLD signal). After smoothing using a FWHM 4 mm Gauss kernel, or a FWHM 6 mm Gauss kernel, or a FWHM 8 mm Gauss kernel, detrending and removing covariates are performed, including white matter signal, whole brain signal regression, and head motion parameters, and finally filtering is performed at 0.01 - 0.08 Hz, or 0.02 - 0.09 Hz, or 0.01 - 0.10 Hz.

[0198] Those of ordinary skill in the art can understand that, according to actual requirements, other numbers of time points can be removed, resampling can be performed using other resolutions, smoothing can be performed using other kernel functions, and filtering can be performed using other bandwidths. The present disclosure does not limit this.

[0199] In an embodiment of the present disclosure, based on an accurate atlas and the region of interest 1013, voxel - level local undirected network 1301 and voxel - level local directed network 1302 are obtained by using the voxel connections within the sub - regions in step S1301. Based on the accurate atlas and the region of interest 1013, the inter - sub - region connections in step S1302 are used to obtain the region - level global undirected network 1303 and the region - level global directed network 1304. Steps S1301 and S1302 together constitute S1032 network construction. Figure 1d The S1032 network construction in Figure 1a is the same as

[0200] In an embodiment of the present disclosure, the voxel - level local undirected network and the region - level global undirected network are collectively referred to as the undirected functional network, and the voxel - level local directed network and the region - level global directed network are collectively referred to as the directed functional network. The undirected functional network and the directed functional network constitute the functional network 1032.

[0201] In an embodiment of the present disclosure, the sub - region can be a brain region, and the functional network 1032 can characterize the functional connection relationship of the brain.

[0202] In an embodiment of the present disclosure, after preprocessing the BOLD signal, brain region annotation is performed according to a selected accurate atlas, and an undirected brain network is calculated. Taking the voxels in each brain region as vertices, the time-domain correlation of the time series of each voxel is estimated, and the Pearson correlation coefficient between voxels can be calculated as the weight of the edge to describe the synchrony of the BOLD signals between voxels, thereby obtaining a voxel-level local undirected network. On the other hand, taking the average value of the time series of all voxels in a brain region as the time series of the brain region, and calculating the interaction strength of the signals between each brain region as the weight of the edge, a regional-level global undirected network such as a regional-level whole-brain undirected network is obtained.

[0203] In an embodiment of the present disclosure, the BOLD directed network can also be divided into two types: voxel-level and regional-level, namely, a voxel-level local directed network and a regional-level global directed network such as a regional-level whole-brain directed network. The vertex definition method is the same as that in the undirected network. For the weights of the edges in the network, three methods are used to estimate the directed functional connection, including the cross-correlation method, the Granger causality analysis (GCA) method, and the convergent cross-mapping (CCM) method. Among them, the cross-correlation method assigns independent sliding windows to different brain regions, allows the sliding windows to move freely on the time series, calculates the correlation values of the sliding window time series at different positions in two brain regions, and takes the maximum of these correlation values as the strength of the directed connection, which is used as the weight of the edge. The positive or negative delay between time windows is defined as the direction of the directed connection. The Granger causality method and the convergent cross-mapping method use the estimation of causal relationships to determine whether a directed edge exists, and the direction of the directed edge is defined as the brain region node as the cause pointing to the brain region node as the effect.

[0204] In an embodiment of the present disclosure, for the Granger causality analysis method, the node time series in the network can be represented by X(t) = [x1(t), x2(t),..., x n (t)], and the multivariate vector autoregressive (MVAR) model can be expressed as

[0205]

[0206] where p is the model order, A(k) is the model parameter, and E(t) is the residual error. Expanding the above formula, we can get

[0207]

[0208] After obtaining A(k) by least squares regression, the Granger causality connection strength can be calculated as

[0209] GC(i, j) = ∑aji (k),

[0210] and used as the weight of the edge.

[0211] In the embodiments of the present disclosure, the convergent cross mapping (CCM) method measures the causal effect of X on Y by calculating to what extent Y can predict X. In a non-linear deterministic dynamic system, if X has a causal effect on Y, then the information of X will be redundantly contained in Y, and thus X can be cross-mapped by Y. This cross-mapping is achieved through the neighbors of points on the system embedding manifold, and the manifolds of two variables are constructed by time series delay coordinates.

[0212] Suppose there are n nodes in the network, and each node signal is a time series. The two time series X and Y correspond to two embedding manifolds M X and M Y . Let = [x(1), x(2),..., x(T)], and the embedding manifold M x is a point set, M X = {mx(1), mx(2),..., mx(L)}, where each point is constructed by the delay coordinates of X, mx(i) = [x(i), x(i + 1),..., x(i + E)], and E is the dimension of the manifold. M Y is constructed for Y in the same way. According to the assumption of CCM, if Y has a causal effect on X, points that are close to each other on M X should also be neighbors on M Y . Based on the neighboring points on M X , their corresponding points on M Y can be found, and then Y can be predicted based on these points on M Y as follows:

[0213] For any point mx(i) on M X , find E + 1 nearest neighbors mx(i1), max(i2),..., max(i E+1 ), where the distance is measured by the L2 norm of the coordinate points, that is, d x (i, j) = ||mx(i) - mx(j)||2. Find the corresponding points my(i1), my(i2),..., my(i Y ) on M E+1 according to the subscripts of these points. Each my(i) is transformed into the time domain, corresponding to Y(i). The value of this point is predicted by the weighted sum of E + 1 time domain points, and the weight is the distance of the nearest neighbor. The specific method is as follows:

[0214]

[0215]

[0216]

[0217] in is the predicted time series, For the i k The weight of the elements, through M X The distance between the points is calculated. X A single point on The value at a point in time. X After all the points on The Pearson correlation coefficient between the true value Y and the causal effect strength of Y on X can be obtained, that is,

[0218]

[0219] The above formula can be used to calculate the strength of the mutual causal effect between two nodes. However, for strong unidirectional effects, because the causal effect strength is relatively high, it may be possible to identify a bidirectional causal effect. In this case, the delayed CCM can be used. Let the delay be τ, M X The number of upper points becomes L-τ, and based on mx(i) we can predict For the brain network established by BOLD signal, τ∈[0,5] can be taken to reflect the mutual causal effect of neural activity in the short term. For each value of τ, the above process is used to estimate The causal interaction strength CCM(i, j, τ) is calculated. The maximum CCM value calculated for all delays τ is then taken to obtain the causal interaction strength between the two brain regions.

[0220]

[0221] and serves as the weight of the edge.

[0222] In an embodiment of the present disclosure, electroencephalogram (EEG) data may also be used to construct a functional network.

[0223] Figure 1e An exemplary schematic diagram showing an implementation scenario of a neural network spatiotemporal measurement method according to an embodiment of the present disclosure.

[0224] It can be understood by those skilled in the art that Figure 1e The implementation scenario of the neural network space-time measurement method is shown by way of example only and does not constitute a limitation to the present disclosure.

[0225] Specifically, Figure 1e An exemplary schematic diagram showing the construction of a functional network using EEG data is shown.

[0226] As shown Figure 1e in FIG., for the EEG data 1401, after filtering and independent component analysis in step S1401 to remove artifacts, the preprocessed EEG data is obtained.

[0227] The preprocessing of the EEG data obtained from the scalp electrodes of the subject can use the EEGLAB toolbox and its plugins. First, a zero-phase shift non-causal FIR notch filter (for example: the upper and lower cut-off frequencies are 49 Hz and 51 Hz respectively) is used to filter out the 50 Hz mains interference, and then an FIR high-pass filter with a cut-off frequency of 0.5 Hz and a low-pass filter with a cut-off frequency of 45 Hz are used to filter the EEG data respectively to obtain the filtered EEG data. The filtered EEG data is resampled to 250 Hz. The clean_artifacts() function is used to automatically detect the electrode channels with poor signal quality, and the bad channels and noisy data segments to be removed are determined according to the electrode impedance and manual inspection, and the pop_interp() function is used to spherically interpolate the signals at the bad electrodes. Then, independent component analysis (Independent Component Analysis, ICA) is performed on the signals, and the artifact components are removed by combining the iclabel tool and visual inspection to obtain the preprocessed EEG data.

[0228] The MRI data 1402 of the subject is registered to the MNI space in step S1402 and tissue segmentation is performed in step S1403 to obtain the head model 1403. The head model can be a volume conduction model, which describes the characteristics of current passing through tissues, including the description of the head geometry and tissue conductivity. The more accurate the description of the head geometry, the better the effect of forward modeling, and a real head model (finite element model (Finite Element Model, FEM), boundary element model (Boundary Element Model, BEM)) can be used. The MRI data 1402 is subjected to source modeling in step S1404 to obtain the source model 1404. A distributed source model can be used to model with a large number of current dipoles distributed on a three-dimensional brain volume or a two-dimensional cortical surface, and the positions of the dipoles are evenly distributed in the volume. After obtaining the head model 1403 and the source model 1404, forward modeling can be performed in step S1405, considering the head model and the channel electrode positions, to calculate the lead field matrix 1405 at each grid point.

[0229] The preprocessed EEG data and the lead field matrix 1405 can be subjected to a source tracing algorithm in step S1406 to obtain the source distribution 1406.

[0230] EEG data from scalp electrodes cannot directly reflect the actual location where nerve electrical signals are generated. To reconstruct the location of the brain electrical signal source and the time series of the source signals, source tracing algorithms such as the minimum norm method, beamforming method, sloreta method, eloreta method, and LAURA method can be used to process the preprocessed EEG data, combined with the lead field matrix obtained from forward modeling, so as to perform signal source tracing and obtain the source distribution 1406.

[0231] The source distribution 1406 is the current density intensity that varies with time and is uniformly distributed in the brain. The source distribution 1406 can be registered S1407 through a brain atlas 1407 (such as: Brodmann atlas, Brodmann_ce atlas, AAL, AAL2, AAL3, AICHA, Brainnetome atlas, etc.) to obtain the sources in each brain region in the atlas. After calculations such as averaging, extreme values, median, variance, eigenvectors, etc., the source signals 1408 in the brain region are obtained.

[0232] In an embodiment of the present disclosure, the source signals 1408 in the brain region are subjected to functional connectivity calculation in step S1408 to obtain the functional network 1409.

[0233] After obtaining the source signals 1408 in the brain region through the above process, the full-band signal or the signal can be decomposed into frequency bands such as [1-4]Hz, [4-8]Hz, [8-13]Hz, [8-10]Hz, [10-13]Hz, [13-30]Hz, [13-17]Hz, [17-21]Hz, [21-30]Hz, [30-45]Hz, etc. For the signals in each frequency band, the aforementioned cross-correlation method, GCA method, and CCM method are respectively used to construct a directed functional network. In addition to using the cross-correlation method, GCA method, and CCM method to construct a directed functional network, methods such as the phase value locking method, coherence method, and weighted phase lag index method can also be used to construct an undirected functional network. When constructing a functional network, the data can also be divided into segments of equal length such as 2 seconds or in the form of a sliding window to calculate the directed or undirected functional connection, and the average functional connection can also be obtained by averaging all the segmented functional connections.

[0234] Those of ordinary skill in the art can understand that in addition to the cross-correlation method, Granger causality analysis method, and convergent cross-mapping method, other methods can also be used to estimate the directed functional connection, and the present disclosure does not limit this.

[0235] Figure 1f An exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure.

[0236] Those of ordinary skill in the art can understand that Figure 1f exemplarily shows an implementation scenario of the spatio-temporal metric method of the neural network, without constituting a limitation to the present disclosure.

[0237] Figure 1f shows Figure 1a the exemplary specific process of obtaining the topological descriptor 1033 by performing topological analysis on the structural network in the functional network 1032 and the fiber bundle analysis result 1023 from the network topology analysis S1033.

[0238] Such as Figure 1f shown, in step S1501, by the formula re-determine the weights of the edges of the structural network, and by the formula d ij = 1 - |w ij | re-determine the weights of the edges of the undirected functional network 1502 and the directed functional network 1503.

[0239] In step S1502, construct a simplicial complex by means of filtration. In S1502, as the threshold ∈ of the connection strength between vertices increases, one-dimensional simplices are gradually connected to each other to generate two-dimensional simplices, three-dimensional simplices, and so on.

[0240] The simplicial complex obtained by the null model 1504 method is used for comparison with the simplicial complexes obtained by the filtration method for the structural network, the undirected functional network, and the directed functional network in the present application. The null model method randomly rearranges the weights of the edges, or rearranges both the edges and the weights of the edges. Compared with the random null model 1504, the simplicial complexes obtained by using the structural network, the undirected functional network, and the directed functional network in the present application through the filtration method better reflect the structural connection relationship in, for example, white matter in the brain and the functional connection relationship in gray matter in the brain, can achieve accurate dynamic monitoring, and have better effectiveness and interpretability.

[0241] The topological descriptor 1505 at each threshold can be obtained from the simplicial complex at each threshold. The topological descriptor 1505 at each threshold includes at least one of the following: the number of simplices of each dimension, the vertex numbers of the simplices of each dimension, the Betti numbers of each order, the vertex numbers of the holes of each order, the Euler characteristic numbers of each order, and the spectral entropies of each order. In addition to the topological descriptors at the above thresholds, other topological descriptors at different thresholds can also be used, and the present disclosure does not limit this.

[0242] In step S1503, the calculation and display of persistent homology features can be performed based on the simplicial complex.

[0243] In an embodiment of the present disclosure, homology is an algebraic topology concept that can study the structure of simplicial complexes in different dimensions and is an algebraic invariant of topological spaces. To perform calculations consistently, a direction is added to each simplex in the simplicial complex, that is, the order of vertices is specified. The nodes in the structure network and the function network are numbered, and this serves as the order of vertices.

[0244] A k - dimensional chain is defined as , where σ i is a k - dimensional simplex, is the coefficient in the field . The k - dimensional chain group or the k - dimensional chain vector space is all possible combinations of k - dimensional chains:

[0245]

[0246] is the set of k - dimensional simplices in . Here, a vector space is used. Of course, any field F or the integer field Z can also be used as the coefficient for a more general definition.

[0247] In an embodiment of the present disclosure, the boundary operator can be used to relate the k - dimensional chain space and the (k - 1) - dimensional chain space There is:

[0248]

[0249] To find the cycle in the k - chain group, the algebraic equation needs to be solved. And there is always The kernel of is denoted as which is the group of boundaries and the image of is the set of boundaries. The only difference is that is in polynomial form, is in set form. Since means not contained in the elements in can be represented by the quotient group, and the k - th homology group is defined as:

[0250]

[0251] k is the dimension of the generator of the homology group, and the elements in H k are the k - dimensional holes in the complex. Hk is defined as the k-th Betti number β k , specifically, β0 is the number of connected components in the simplicial complex, β1 is the number of cycles, β2 is the number of holes surrounded by 2-dimensional simplices, and so on. Another topological invariant besides the Betti numbers is the Euler characteristic,

[0252]

[0253] where denotes the number of n-dimensional simplices in

[0254]

[0255] is the eigenvalue of the k-th combinatorial Laplacian L k of

[0256] In the embodiments of the present disclosure, persistent homology is a method of computing the homology of a changing simplicial complex as a threshold is continuously changed and continuously tracking how homology features (holes of each order, etc.) evolve along the filtration. So persistent homology does not analyze the network at a fixed threshold and is a threshold-independent framework. As the threshold increases, homology features are born and then disappear at ∈ b and ∈ d -∈ b which can represent the persistence of homology features. Although the persistence of homology features is related to how the filtration is constructed, it can be considered that the more persistent the homology features are, the more important they are. The main manifestation of topological features is the Persistence Diagram (PD). Each point (∈ b , ∈ d ) on the PD represents the birth and disappearance thresholds of a topological feature. Features with a long "persistent time" are far from the diagonal, while features with a short "persistent time" are near the diagonal and can be considered noise. Persistence Landscape (PL) and Persistence Images (PI) are two transformations of the PD. PL transforms the PD into a series of piecewise linear functions, and PI transforms the PD into a "heat map" Gaussian weighted by each point in the PD. PI and PL or the parameters extracted from them can be used as features describing topology, that is, topological descriptors.

[0257] In an embodiment of the present disclosure, persistent entropy (PE) can also be calculated from PD to measure the "information" encoded in persistent homology (specific dimension).

[0258] PD = {(∈ b1 , ∈ d1 ),..., (∈ bI , ∈ dI )}

[0259] p i = ∈ di - ∈ bi , i ∈ I

[0260]

[0261] PE = -Σ i∈I s i log(s i )

[0262] As Figure 1f shown, through the calculation and display of persistent homology features in S1503, a persistent homology descriptor 1506 can be obtained. The persistent homology descriptor 1506 includes at least one of the following: PD and its extractable features, PL and its extractable features, PI and its extractable features, BV and its extractable features. In addition to the above persistent homology descriptors, other persistent homology descriptors can also be used, and the present disclosure does not limit this.

[0263] In an embodiment of the present disclosure, the topological descriptor 1505 and the persistent homology descriptor 1506 at each threshold can form a topological descriptor 1033. Figure 1e The topological descriptor 1033 in Figure 1a is the same as the topological descriptor 1033 in

[0264] In Figure 1a , the spatio-temporal metric S1041 can obtain a spatio-temporal metric result 1041 through the fiber bundle analysis result 1023 and the topological descriptor 1033.

[0265] In an embodiment of the present disclosure, the spatio-temporal metric evaluates the comparability of the neurological network in terms of spatial scale change and time line. Both the "relative mean connectivity" and the "persistent homology topological descriptor" are comparable factors of the spatio-temporal metric.

[0266] In an embodiment of the present disclosure, for the spatio-temporal metric of the structural network of the fiber bundle, the aforementioned relative mean connectivity can be used to calculate the change in the relative mean connectivity of the fiber bundle connections at different time intervals, measure the change in the fiber bundle structure within the ROI, or the change in the relative mean connectivity between different ROIs at the same time point.

[0267] In the embodiments of the present disclosure, for the persistent homology topological descriptor, the above-mentioned Persistent Entropy (PE), Betti number, Euler characteristic number, etc. are still comparable in the time dimension. For the sub-threshold topological descriptor, a relative comparability factor for counting the n-dimensional simplices of the connected network within the ROI is defined: the nth power of the number of voxels within the ROI. That is, the analysis count result of the n-dimensional simplices within the ROI is divided by the nth power of the total number of voxels within the ROI to obtain the number of simplices corresponding to each voxel in the n-dimensional space, which is used to measure the network connection scale or complexity within the ROI. It can be called the "number of simplices per vertex" of the n-dimensional space network. It can be used to compare the changes in the network scale of the same brain region at different time intervals, or the network scale comparison between different brain regions at the same time.

[0268] Define the simplex overlap rate to measure the evolution relationship or inheritance relationship of the simplices of each dimension within each time interval. There is a one-to-one mapping relationship between the order on the vertex set and the simplex, so the equality relationship of the simplex can be defined therefrom. It can be considered that two simplices are the same if and only if the order on the corresponding vertex set is exactly the same. For the simplicial complex on the functional network and the structural network at each time point, all the simplices of dimension k form a simplex set, which is called the k-dimensional simplex set. The simplicial complex overlap ratio can be used to measure the proportion of the number of overlapping simplices in the k-dimensional simplex set at different time points. Specifically, given the k-dimensional simplex sets S k，t1 , S k，t2 , the simplex overlap rate is

[0269]

[0270] where |S| is the cardinality of the set (the number of elements in the set). The simplex overlap rate quantifies the similarity degree of the two simplex sets and can measure the similarity degree of the topological structures of the structural and functional networks in the time dimension.

[0271] Define the inclusion relationship of simplices of different dimensions, that is, for s m,i and s n,j which are the directed vertex sequences of the ith m-dimensional simplex and the jth n-dimensional simplex respectively, m < n. When and only when s m,i is a continuous subsequence of s n,j , it can be considered that the simplex corresponding to s n,j contains the simplex corresponding to s m,i , denoted as Define the maximal simplex, that is, for a certain m-dimensional simplex s m,i , satisfy Combining the inclusion relationship of simplices in different dimensions and time information, the evolution and degradation relationships of maximal simplices can be defined. That is, for a certain low-dimensional maximal simplex at a past time point, if there exists a certain high-dimensional maximal simplex at a future time point that contains this low-dimensional simplex, then it is considered that the high-dimensional simplex evolves from the low-dimensional simplex at the past time point; conversely, it is considered that the low-dimensional simplex degrades from the high-dimensional simplex at the past time point. Furthermore, given time points t1 and t2, the numbers of maximal simplices that evolve and degrade are N evol ,N deg , define the simplex evolution coefficient R c,(t1,t2) :

[0272]

[0273] The simplex evolution coefficient R c describes the intensity and trend of simplex evolution on the time scale. Specifically, when R c is greater than 1, the simplex tends to evolve; when R c is less than 1, it tends to degrade.

[0274] In the embodiments of the present disclosure, the simplex overlap rate and the simplex evolution coefficient can form the simplex change characteristics. Through the simplex change characteristics, combined with the fiber bundle analysis result 1023, the spatio-temporal metric result 1041 can be obtained.

[0275] Figure 1g Exemplary schematic diagram showing an implementation scenario of a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure.

[0276] Those of ordinary skill in the art can understand that Figure 1g exemplarily shows an implementation scenario of a spatio-temporal metric method for a neurological network, without constituting a limitation to the present disclosure.

[0277] Such as Figure 1gAs shown, 1601 is the brain MRI image of human P1 in the stem cell transplantation group, and 1603 is the MRI image of human P2 in the sham transplantation control reference group. The three-dimensional T1 data with labeled targets was registered into the MNI space, and the Brainnetome atlas (bnatlas) was used for brain region division to obtain the regions of interest in the MNI space. The nine brain regions in the right hemisphere corresponding to the transplantation target and the stroke injury area of P1 are numbered bnatlas (164, 166, 168, 170, 172, 174, 222, 226, 230), that is, the area shown in 1602. The corresponding brain regions on the healthy side of the left hemisphere are numbered bnatlas (163, 165, 167, 169, 171, 173, 221, 225, 229). The affected brain regions of P2 in the control reference group are almost symmetrical to those of P1 in the transplantation group. Therefore, the regions of interest for P2 are its affected brain regions bnatlas (163, 165, 167, 169, 171, 173, 221, 225, 229), that is, the area shown in 1604.

[0278] In the embodiments of the present disclosure, the MRI data collected three times for P1 and P2 is used, including the pre-transplantation baseline pre, the 6-month follow-up 6mopo after transplantation, and the 12-month follow-up 12mopo.

[0279] In the embodiments of the present disclosure, the tractography results at the voxel level within the brain region can be calculated according to the brain region numbers of the regions of interest, such as the relative mean connectivity of fiber bundles.

[0280] In the embodiments of the present disclosure, the voxel-level local directed network within the brain region corresponding to the region of interest can also be estimated according to the CCM method described above in this application, the simplicial complex can be calculated, the topological descriptors can be calculated, the number of simplices in different dimensions can be calculated, and the simplex overlap rate can be calculated.

[0281] The test and calculation results show that there is no obvious trend change in the relative mean connectivity of fiber bundles, the number of high-dimensional simplices (2D, 3D), and the simplex overlap ratio of the control reference group P2 at the pre-transplantation baseline pre, the 6-month follow-up 6mopo after transplantation, and the 12-month follow-up 12mopo. Correspondingly, in the regions of ROI166 and ROI168 of human P1 in the stem cell transplantation group, that is, the transplantation target regions, as Figure 1h shown, compared with pre, the relative mean connectivity of fiber bundles at 6mopo and 12mopo increased significantly; as Figure 1i 、 1j shown, the number of 2D and 3D simplices at 6mopo and 12mopo increased significantly; as Figure 1k shown, the simplex overlap ratio at 6mopo and 12mopo also changed significantly. The method of this application accurately detected the changes in nerve cell growth and made accurate measurements.

[0282] Those of ordinary skill in the art can understand that Figure 1h , Figure 1i , Figure 1j , Figure 1k exemplarily shows an implementation scenario of a neurological network spatio-temporal metric method, without constituting a limitation to the present disclosure.

[0283] Figure 2 Shows a flowchart of a neurological network spatio-temporal metric method according to an embodiment of the present disclosure.

[0284] As Figure 2 shown, the neurological network spatio-temporal metric method includes: steps S201, S202, and S203.

[0285] In step S201, based on the original relaxation time weighted imaging data of magnetic resonance imaging, a first registration is performed to obtain the relaxation time weighted imaging data after the first registration, and an accurate atlas and a region of interest are obtained based on the relaxation time weighted imaging data after the first registration.

[0286] The step S201 is a region of interest selection step.

[0287] In step S202, based on the diffusion tensor imaging data of magnetic resonance imaging, a first preprocessing and a second registration are performed to obtain the diffusion tensor imaging data after the second registration, and fiber bundle construction and processing are performed on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result, and the fiber bundle analysis result includes a structural network.

[0288] The step S202 is a fiber bundle analysis step.

[0289] In step S203, based on specific functional data, data processing and analysis are performed to obtain the specific functional data after data processing and analysis, a functional network is constructed from the specific functional data after data processing and analysis, and a topological descriptor is obtained by performing network topology analysis on at least one of the functional network and the structural network.

[0290] The step S203 is a network topology analysis step.

[0291] According to the technical solution provided by the embodiments of the present disclosure, through the region of interest selection step, wherein, based on the original relaxation time weighted imaging data of magnetic resonance imaging, the first registration is performed to obtain the relaxation time weighted imaging data after the first registration, and based on the relaxation time weighted imaging data after the first registration, an accurate atlas and a region of interest are obtained; the fiber bundle analysis step, wherein, based on the diffusion tensor imaging data of magnetic resonance imaging, the first preprocessing and the second registration are performed to obtain the diffusion tensor imaging data after the second registration, and the fiber bundle processing is performed on the diffusion tensor imaging data after the second registration to obtain the fiber bundle analysis result, and the fiber bundle analysis result includes a structural network; the network topology analysis step, wherein, based on the specific functional data, data processing and analysis are performed to obtain the specific functional data after the data processing and analysis, a functional network is constructed from the specific functional data after the data processing and analysis, and a topological descriptor is obtained by performing network topology analysis on at least one of the functional network and the structural network, so as to realize the accurate dynamic monitoring of the state of the highly dense neural tissue such as the brain, and the accurate measurement of the functional changes at the levels of neural regions, circuits and systems such as the brain, and can be extended to research fields such as non-invasive brain-computer interfaces.

[0292] Figure 3 The flowchart showing a spatio-temporal metric method for a neurological network according to an embodiment of the present disclosure.

[0293] As Figure 3 shown, the spatio-temporal metric method for a neurological network includes the same steps S201, S202, S203 and further includes step S301. Figure 2

[0294] In step S301, based on the fiber bundle analysis result of the fiber bundle analysis step and the topological descriptor of the network topology analysis step, the spatio-temporal metric result is calculated.

[0295]

[0296]

[0297] According to the technical solution provided by the embodiments of the present disclosure, it further includes: the spatio-temporal metric step, wherein, based on the fiber bundle analysis result of the fiber bundle analysis step and the topological descriptor of the network topology analysis step, the spatio-temporal metric result is calculated, and the spatio-temporal characteristics of the structural network and the functional network are accurately quantitatively analyzed, so as to realize the accurate dynamic monitoring of the state of the highly dense neural tissue such as the brain, and the accurate measurement of the functional changes.

[0297] In an embodiment of the present disclosure, magnetic resonance imaging includes: brain magnetic resonance imaging, and / or spinal cord magnetic resonance imaging, or imaging of other highly neural-dense regions, which can be performed on the human body or on animal bodies such as mice and monkeys. The original relaxation time weighted imaging data includes: original longitudinal relaxation time weighted imaging data T1, or original transverse relaxation time weighted imaging data T2. Correspondingly, the first registered relaxation time weighted imaging data includes: the first registered longitudinal relaxation time weighted imaging data, or the first registered transverse relaxation time weighted imaging data, and the accurate atlas includes: the corrected atlas.

[0298] As Figure 1a shown, in step S1012 of atlas correction and region of interest selection, the MNI standard space atlas 1000 can be combined to correct the deformed brain to obtain an accurate atlas such as the corrected atlas. Those of ordinary skill in the art can understand that if the brain is not deformed, the MNI standard space atlas 1000 can be used as the accurate atlas, and the present disclosure does not limit this.

[0299] According to the technical solution provided by the embodiment of the present disclosure, through magnetic resonance imaging including: brain magnetic resonance imaging, and / or spinal cord magnetic resonance imaging; and / or the original relaxation time weighted imaging data including: original longitudinal relaxation time weighted imaging data, or original transverse relaxation time weighted imaging data; and / or the first registered relaxation time weighted imaging data including: the first registered longitudinal relaxation time weighted imaging data, or the first registered transverse relaxation time weighted imaging data, and the accurate atlas includes: the corrected atlas, the flexibility and wide adaptability of the technical solution of the present application are realized.

[0300] Figure 4 The flowchart of the spatio-temporal metric method of the neurological network according to an embodiment of the present disclosure is shown.

[0301] Specifically, Figure 4 shows Figure 2 the specific process of "performing the first registration based on the original relaxation time weighted imaging data of magnetic resonance imaging to obtain the first registered relaxation time weighted imaging data" in step S201 in

[0302] In step S401, the original relaxation time weighted imaging data is subjected to de-skeleton registration to obtain the first registered relaxation time weighted imaging data.

[0303] The step S401 is a de-skeleton registration sub-step.

[0304] In step S402, the original relaxation time weighted imaging data is subjected to skeletonized registration to obtain the first registered relaxation time weighted imaging data.

[0305] The step S401 is a bone registration sub-step.

[0306] Those of ordinary skill in the art can understand that depending on the different parts scanned by MRI, the bone can be the skull, the spine, or the bone of other parts, and the present disclosure does not limit this.

[0307] According to the technical solution provided by the embodiments of the present disclosure, the first registration is performed on the original relaxation time weighted imaging data based on magnetic resonance imaging to obtain the first registered relaxation time weighted imaging data, including: a bone removal registration sub-step, in which the bone removal registration is performed on the original relaxation time weighted imaging data to obtain the first registered relaxation time weighted imaging data; or a bone registration sub-step, in which the bone registration is performed on the original relaxation time weighted imaging data to obtain the first registered relaxation time weighted imaging data, thereby realizing the flexibility and accuracy of the relaxation time weighted imaging data registration.

[0308] In the embodiments of the present disclosure, as Figure 1b shown, as shown in the upper branch, the steps S1111, S1112, and S1113 can be used to remove the bone, and then the step S1114 is used for registration to obtain the T1_brain_w1 data 1116; or as shown in the lower branch, the step S1121 can be used for bone registration, and then the steps S1122 and S1123 are used to remove the bone to obtain the T1_w_brain data 1124. The T1 data 1111 can be the head T1 signal, and the bone can be the skull.

[0309] Those of ordinary skill in the art can understand that registration can also be performed on other parts with dense nerves such as the spinal cord, and the present disclosure does not limit this.

[0310] According to the technical solution provided by the embodiments of the present disclosure, the bone removal registration sub-step includes: using a first command to remove the bone data in the original relaxation time weighted imaging data to obtain the relaxation time weighted imaging data after bone removal; using a second command to register the relaxation time weighted imaging data after bone removal to obtain the first registered relaxation time weighted imaging data; and / or the bone registration sub-step includes: using a third command to register the original relaxation time weighted imaging data to obtain the registered relaxation time weighted imaging data with bone; using a fourth command to remove the bone data in the registered relaxation time weighted imaging data with bone to obtain the first registered relaxation time weighted imaging data, and the bone includes: the skull, thereby realizing the flexibility and accuracy of the relaxation time weighted imaging data registration.

[0311] In the embodiments of the present disclosure, as Figure 1bAs shown, the first command includes: a combination of the bet or bet2 command in FSL, the flirt command based on the first template, the betsurf command, and the fslmaths command; or the fslmaths command in FSL based on the second template; or the bet2 command in FSL; or the mri_watershed command in FreeSurfer;

[0312] The second command includes: the flirt command in FSL based on the second template.

[0313] The third command includes: the flirt command in FSL based on the first template.

[0314] The fourth command includes: the bet2 command in FSL; or the fslmaths command in FSL based on the second template.

[0315] According to the technical solution provided by the embodiments of the present disclosure, through the first command including: the bet command in FSL, the flirt command based on the first template, the betsurf command, and the fslmaths command; or the fslmaths command in FSL based on the second template; or the bet2 command in FSL; or the mri_watershed command in FreeSurfer; and / or the second command including: the flirt command in FSL based on the second template; and / or the third command including: the flirt command in FSL based on the first template; and / or the fourth command including: the bet2 command in FSL; or the fslmaths command in FSL based on the second template, the flexible and accurate registration of relaxation time weighted imaging data is achieved.

[0316] In the embodiments of the present disclosure, the first template may be the MNI152_T1_1mm template, or the MNI152_T2_1mm template, and the second template may be the MNI152_T1_brain_1mm template, or the MNI152_T2_brain_1mm template. The first template and the second template may also be templates for other resolutions outside 1 mm, or templates for other parts, and the present disclosure does not limit this.

[0317] According to the technical solution provided by the embodiments of the present disclosure, through the first template including: the MNI152_T1_1mm template, or the MNI152_T2_1mm template; and / or the second template including: the MNI152_T1_brain_1mm template, or the MNI152_T2_brain_1mm template, the flexible and accurate registration of relaxation time weighted imaging data is achieved.

[0318] In the embodiments of the present disclosure, as Figure 1bAs shown, the T1_brain_w data 1116 is combined with the first MNI spatial atlas 1117. Through the atlas correction in step S1115 and the first processing of obtaining the region of interest, an accurate atlas and the region of interest 1013 are obtained. The T1_w_brain data 1124 is combined with the second MNI spatial atlas 1125. Through the atlas correction in step S1124 and the second processing of obtaining the region of interest, an accurate atlas and the region of interest 1013 are obtained.

[0319] Those of ordinary skill in the art can understand that the first MNI spatial atlas 1117 and the second MNI spatial atlas 1125 can be the same standard MNI spatial atlas or different, and the present disclosure does not limit this.

[0320] According to the technical solution provided by the embodiments of the present disclosure, obtaining an accurate atlas and the region of interest based on the first registered relaxation time weighted imaging data includes: using the first registered relaxation time weighted imaging data and combining it with the standard MNI spatial atlas to obtain an accurate atlas and the region of interest, so that atlas correction can be performed on local deformations of tissue regions such as brain regions, and the region of interest can be selected for key analysis.

[0321] In the embodiments of the present disclosure, as Figure 1c shown, the diffusion tensor imaging data 1021 undergoes the first preprocessing in S1201 to obtain the first preprocessed diffusion tensor imaging data 1201. In the first preprocessing in S1201, using the FSL tool, the b0 image is extracted, bones such as the skull are removed, eddy current correction is performed, and the first motion correction such as head motion correction is performed, and the diffusion tensor and related parameters are calculated.

[0322] Those of ordinary skill in the art can understand that according to actual needs, the spine can also be removed, and the first motion correction for spine motion can be performed, and the present disclosure does not limit this.

[0323] The first preprocessed diffusion tensor imaging data 1201 undergoes the second registration in S1202, and the second registered diffusion tensor imaging data 1202 is obtained using the flirt command in FSL.

[0324] In the embodiments of the present disclosure, 3D slicer can also be used to remove the skull from the diffusion tensor imaging data 1021 using Diffusion Brain Mask, and voxel modeling and registration are performed using Diffusion Tensor Estimation to obtain the second registered diffusion tensor imaging data 1202.

[0325] According to the technical solution provided by the embodiments of the present disclosure, by performing first preprocessing and second registration on diffusion tensor imaging data based on magnetic resonance imaging, obtaining the diffusion tensor imaging data after the second registration includes: removing bones from the b0 image in the diffusion tensor imaging data and generating a bone-removed mask, using the generated mask to remove bones from the DWI image in the diffusion tensor data, performing eddy current correction and first motion correction on the diffusion tensor data after bone removal to obtain the preprocessed diffusion tensor imaging data; for the preprocessed diffusion tensor imaging data, performing linear registration using the MNI152_T1_brain_2mm template or the MNI152_T2_brain_2mm template, or performing linear registration using the MNI152_T1_brain_2mm template or the MNI152_T2_brain_2mm template and then using the template to generate a mask to remove bones again to obtain the diffusion tensor imaging data after the second registration, or using 3D Slicer to remove the skull, performing voxel modeling and registration to obtain the diffusion tensor imaging data after the second registration, thereby eliminating the error of the diffusion tensor imaging data and facilitating subsequent analysis.

[0326] Figure 5 The flowchart showing a spatio-temporal metric method of a neurological network according to an embodiment of the present disclosure.

[0327] Specifically, Figure 5 Showing Figure 2 The specific process of "performing fiber bundle construction and processing on the diffusion tensor imaging data after the second registration to obtain fiber bundle analysis results" in step S202.

[0328] Figure 5 The process in includes steps S501 and S502.

[0329] In step S501, fiber bundle tracking is performed on the diffusion tensor imaging data after the second registration to obtain fiber bundle tracking results.

[0330] The step S501 is a fiber bundle tracking sub-step.

[0331] In step S502, fiber bundle analysis is performed on the fiber bundle tracking results by combining an accurate atlas and a region of interest to obtain fiber bundle analysis results.

[0332] The step S502 is a fiber bundle analysis sub-step.

[0333] According to the technical solution provided by the embodiments of the present disclosure, by performing fiber bundle processing on the second-registered diffusion tensor imaging data, the obtained fiber bundle analysis results include: a fiber bundle tracking sub-step, wherein fiber bundle tracking is performed on the second-registered diffusion tensor imaging data to obtain a fiber bundle tracking result; a fiber bundle analysis sub-step, wherein fiber bundle analysis is performed on the fiber bundle tracking result by combining an accurate atlas and a region of interest to obtain a fiber bundle analysis result, thereby realizing precise dynamic monitoring of the fiber bundle state of highly nerve-dense tissues such as the brain and accurate measurement of functional changes.

[0334] As Figure 1c shown, the second-registered diffusion tensor imaging data 1202 is used in S1203 to calculate FA and MD values using the Python toolkit Dipy. In S1204, the threshold of FA is used as the termination condition, and the fiber bundle tracking method with the smallest tracking error is selected to obtain the fiber bundle tracking result 1203.

[0335] According to the technical solution provided by the embodiments of the present disclosure, the fiber bundle tracking sub-step includes: calculating the anisotropy index and the mean diffusivity from the second-registered diffusion tensor imaging data; performing fiber bundle tracking on the second-registered diffusion tensor imaging data using a fiber bundle tracking method and calculating the tracking error; selecting the fiber bundle method with the smallest tracking error to obtain a fiber bundle tracking result, thereby realizing precise tracking and analysis of the fiber bundle state of highly nerve-dense tissues such as the brain.

[0336] In the embodiments of the present disclosure, the fiber bundle tracking method includes at least one of the following: EuDX method, deterministic method, probabilistic method, sfm method, xtract instruction, Tractography seeding instruction.

[0337] According to the technical solution provided by the embodiments of the present disclosure, the fiber bundle tracking method includes at least one of the following: EuDX method, deterministic method, probabilistic method, sfm method, xtract instruction, Tractography seeding instruction, thereby using multiple fiber bundle tracking methods for comparison to obtain accurate fiber bundle tracking results.

[0338] In the embodiments of the present disclosure, as Figure 1cAs shown, according to the fiber tractography result 1203, in combination with the accurate atlas and the region of interest 1013, in S1206, a sub-region in the region of interest can be selected, and the number and length distribution of the fiber tracts passing through or having one end in the sub-region can be calculated to obtain, for example, the number of fiber tracts passing through the ROI, the average length, the length distribution, and the number of fiber tracts having one end point in the ROI, the average length, the length distribution of the fiber tracts, the fiber tract connectivity of each voxel in the region of interest, and the average fiber tract connectivity of the voxels in the region of interest as the fiber tract analysis result 1023.

[0339] In an embodiment of the present disclosure, according to the fiber tractography result 1203, in combination with the accurate atlas and the region of interest 1013, in S1205, a sub-region in the global region of the whole brain area, for example, can be used as an end point, and the number of fiber tracts between the sub-regions can be used as the edge weight to construct, for example, the fiber tract analysis result 1023 of a structural network.

[0340] According to the technical solution provided by the embodiment of the present disclosure, the fiber tract analysis sub-steps include: based on the accurate atlas and the region of interest, according to the fiber tractography result, calculating the number, average length, length distribution of the fiber tracts passing through the region of interest, and / or the number, average length, length distribution of the fiber tracts having the region of interest as an end point as the fiber tract analysis result; and / or based on the accurate atlas and the region of interest, according to the fiber tractography result, calculating the fiber tract connectivity of each voxel in the region of interest, and / or the average fiber tract connectivity of the voxels in the region of interest as the fiber tract analysis result; and / or using the region of interest as vertices and the number of fiber tracts between the regions of interest as the weight of the edges to construct a structural network as the fiber tract analysis result, so as to comprehensively and fully reflect the statistical characteristics and topological characteristics of the nerve fiber bundles.

[0341] In an embodiment of the present disclosure, based on specific functional data for data processing and analysis, after obtaining the specific functional data after data processing and analysis, constructing a functional network from the specific functional data after data processing and analysis includes: performing a second preprocessing and a third registration on the blood oxygenation level-dependent data based on functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after the third registration, and constructing a functional network from the blood oxygenation level-dependent data after the third registration, or performing data processing and analysis on electroencephalogram data to obtain the electroencephalogram data after data processing and analysis, and constructing a functional network from the electroencephalogram data after data processing and analysis.

[0342] According to the technical solution provided by the embodiments of the present disclosure, by performing data processing and analysis based on specific functional data, and obtaining the specific functional data after data processing and analysis, constructing a functional network from the specific functional data after data processing and analysis includes: performing second preprocessing and third registration on the blood oxygenation level-dependent (BOLD) data based on functional magnetic resonance imaging (fMRI), obtaining the BOLD data after the third registration, and constructing a functional network from the BOLD data after the third registration, or performing data processing and analysis on electroencephalogram (EEG) data, obtaining the EEG data after data processing and analysis, and constructing a functional network from the EEG data after data processing and analysis, so as to obtain a functional network in multiple ways, which has great flexibility.

[0343] In the embodiments of the present disclosure, Figure 1a and Figure 1d the data processing and analysis in S1031 may include: using the DPARSFA tool for BOLD data, first removing the first 0 to 10 time points, then performing slice timing correction and a second motion correction such as head motion correction, registering to the MNI space using an EPI template, and then resampling the data to a resolution of 2x2x2 mm or 3x3x3 mm (the resolution closest to the acquired rBOLD signal). After smoothing with a FWHM 4 mm Gauss kernel, detrending and removing covariates, including white matter signal, whole brain signal regression, and head motion parameters, and finally performing filtering at 0.01 - 0.08 Hz.

[0344] According to the technical solution provided by the embodiments of the present disclosure, obtaining the BOLD data after the third registration by performing second preprocessing and third registration on the BOLD data based on fMRI includes: removing a first number of time points from the BOLD data based on fMRI, obtaining the BOLD data after removing the time points; performing slice time correction and a second motion correction on the BOLD data after removing the time points, registering to the MNI space using an echo planar imaging template, obtaining the BOLD data in the MNI space; resampling the BOLD data in the MNI space, performing detrending and removing covariates after smoothing with a first kernel, and performing a first bandwidth filtering to obtain the BOLD data after the third registration, thereby eliminating interference and errors in the fMRI data.

[0345] In an embodiment of the present disclosure, the first quantity includes: 0 to 10; and / or resampling includes: resampling using a resolution of 2x2x2 mm or 3x3x3 mm; and / or the first kernel includes: a FWHM 4 mm Gauss kernel, or a FWHM 6 mm Gauss kernel, or a FWHM 8 mm Gauss kernel; and / or the first bandwidth includes: 0.01 to 0.08 Hz, or 0.02 to 0.09 Hz, or 0.01 to 0.10 Hz.

[0346] Those of ordinary skill in the art can understand that, according to actual needs, other time points can be removed, resampling can be performed using other resolutions, smoothing can be performed using other kernel functions, and filtering can be performed using other bandwidths. The present disclosure does not limit this.

[0347] According to the technical solution provided by the embodiment of the present disclosure, by the first quantity including: 0 to 10; and / or resampling including: resampling using a resolution of 2x2x2 mm or 3x3x3 mm; and / or the first kernel includes: a FWHM 4 mm Gauss kernel, or a FWHM 6 mm Gauss kernel, or a FWHM 8 mm Gauss kernel; and / or the first bandwidth includes: 0.01 to 0.08 Hz, or 0.02 to 0.09 Hz, or 0.01 to 0.10 Hz, noise and interference in the BOLD signal are removed, and clear and accurate BOLD imaging data is obtained.

[0348] According to the technical solution provided by the embodiment of the present disclosure, by performing data processing and analysis based on electroencephalogram data, the electroencephalogram data after data processing and analysis is obtained, including: filtering the electroencephalogram data, independent component analysis, removing artifacts, and combining the lead field matrix obtained by head model and source model modeling from the registered relaxation time weighted imaging data to perform signal tracing, and obtaining the electroencephalogram data after data processing and analysis, and obtaining the signal in the registered brain region.

[0349] Figure 6 A flowchart showing a spatio-temporal metric method of a neurological network according to an embodiment of the present disclosure.

[0350] Figure 6 Show Figure 2 The specific process of "constructing a functional network from the third registered blood oxygenation level dependent data" in step S203.

[0351] Figure 6 Includes: steps S601, S602, S603, S604.

[0352] In step S601, using the third-registered blood oxygenation level-dependent data, based on the accurate atlas and the region of interest, with the voxels in the region of interest as network vertices, perform correlation estimation on the time series of the voxels as the weights of the edges to obtain a voxel-level local undirected network.

[0353] In step S602, using the third-registered blood oxygenation level-dependent data, based on the accurate atlas, with the sub-regions within the global region corresponding to the accurate atlas as network vertices, calculate the time series of the sub-regions within the global region and the interaction strength between the sub-regions within the global region as the weights of the edges to obtain a region-level global undirected network.

[0354] In step S603, using the third-registered blood oxygenation level-dependent data, based on the accurate atlas and the region of interest, with the voxels in the region of interest as network vertices, use a directed functional connectivity estimation method to obtain the weights of the edges to obtain a voxel-level local directed network.

[0355] In step S604, using the third-registered blood oxygenation level-dependent data, based on the accurate atlas, with the sub-regions within the global region corresponding to the accurate atlas as network vertices, use a directed functional connectivity estimation method to obtain the weights of the edges to obtain a region-level global directed network.

[0356] In the embodiments of the present disclosure, the voxel-level local undirected network and the region-level global undirected network are collectively referred to as the undirected functional network, and the voxel-level local directed network and the region-level global directed network are collectively referred to as the directed functional network. The undirected functional network and the directed functional network form Figure 1d the functional network 1032 therein.

[0357] According to the technical solution provided by the embodiments of the present disclosure, constructing a functional network from the third registered blood oxygenation level-dependent data includes: using the third registered blood oxygenation level-dependent data, based on an accurate atlas and the region of interest, taking the voxels in the region of interest as network vertices, performing correlation estimation on the time series of the voxels as the weights of the edges to obtain a voxel-level local undirected network; and / or using the third registered blood oxygenation level-dependent data, based on the accurate atlas, taking the sub-regions within the global region corresponding to the accurate atlas as network vertices, calculating the time series of the sub-regions within the global region and the interaction strength between the sub-regions within the global region as the weights of the edges to obtain a region-level global undirected network; and / or using the third registered blood oxygenation level-dependent data, based on the accurate atlas and the region of interest, taking the voxels in the region of interest as network vertices, using a directed functional connectivity estimation method to obtain the weights of the edges to obtain a voxel-level local directed network; and / or using the third registered blood oxygenation level-dependent data, based on the accurate atlas, taking the sub-regions within the global region corresponding to the accurate atlas as network vertices, using a directed functional connectivity estimation method to obtain the weights of the edges to obtain a region-level global directed network, thereby accurately estimating the functional network in, for example, the gray matter of the brain.

[0358] According to the technical solution provided by the embodiments of the present disclosure, constructing a functional network from the electroencephalogram data after data processing and analysis includes: dividing the electroencephalogram data after data processing and analysis into frequency bands to obtain multi-band electroencephalogram data; using a directed functional connectivity estimation method for the multi-band electroencephalogram data to construct a directed electroencephalogram functional network, or using an undirected functional connectivity estimation method for the multi-band electroencephalogram data to construct an undirected electroencephalogram functional network, thereby accurately estimating a directed functional network or an undirected functional network based on the electroencephalogram.

[0359] In the embodiments of the present disclosure, the directed functional connectivity estimation method includes at least one of the following: cross-correlation method, Granger causality analysis method, convergence cross mapping method.

[0360] Those of ordinary skill in the art can understand that in addition to the cross-correlation method, Granger causality analysis method, and convergence cross mapping method, other methods can also be used to estimate directed functional connectivity, and the present disclosure does not limit this.

[0361] According to the technical solution provided by the embodiments of the present disclosure, the directed functional connectivity estimation method includes at least one of the following: cross-correlation method, Granger causality analysis method, convergence cross mapping method, so as to be applied to different scenarios to achieve accurate estimation of the directed functional network.

[0362] According to the technical solution provided by the embodiments of the present disclosure, the undirected functional connectivity estimation method includes at least one of the following: the phase locking value method, the coherence method, and the weighted phase lag index method, so as to be applied to different scenarios and achieve accurate estimation of the undirected functional network.

[0363] Figure 7 The flowchart showing the spatio-temporal metric method of the neurological network according to an embodiment of the present disclosure.

[0364] Specifically, Figure 7 shows Figure 2 the specific implementation process of "obtaining topological descriptors by performing network topology analysis on at least one of the functional network and the structural network" in step S203.

[0365] Figure 7 including: steps S701, S702, S703, S704.

[0366] In step S701, a specific simplicial complex is constructed according to at least one of the functional network and the structural network.

[0367] In step S702, a homology group is constructed according to the specific simplicial complex.

[0368] In step S703, the topological features of the homology group are analyzed to obtain topological descriptors.

[0369] In step S704, topological descriptors are obtained according to the specific simplicial complex

[0370] Such as Figure 1e shown, a simplicial complex can be constructed by means of screening, for example, and a homology group can be constructed to calculate persistent homology topological descriptors, or threshold-based topological descriptors can be directly calculated according to the simplicial complex.

[0371] According to the technical solution provided by the embodiments of the present disclosure, obtaining topological descriptors by performing network topology analysis on at least one of the functional network and the structural network includes: constructing a specific simplicial complex according to at least one of the functional network and the structural network; constructing a homology group according to the specific simplicial complex; analyzing the topological features of the homology group to obtain topological descriptors; obtaining topological descriptors according to the specific simplicial complex, so as to accurately analyze the topological structure features of the functional network and the structural network, and achieve precise dynamic monitoring of the state of highly nerve-dense tissues such as the brain and accurate measurement of functional changes.

[0372] In the embodiments of the present disclosure, constructing a specific simplicial complex according to at least one of the functional network and the structural network includes: constructing a specific simplicial complex by means of screening according to at least one of the functional network and the structural network.

[0373] In an embodiment of the present disclosure, constructing a specific simplicial complex based on at least one of a functional network and a structural network further includes: constructing a comparison simplicial complex using a null model; comparing the specific simplicial complex and the comparison simplicial complex.

[0374] In an embodiment of the present disclosure, compared with the simplicial complex constructed by the null model method, the simplicial complex constructed by the screening method better reflects the structural connection relationship of nerve cell axons in, for example, white matter of the brain and the functional connection relationship in gray matter of the brain, and can achieve accurate dynamic monitoring.

[0375] According to the technical solution provided by the embodiment of the present disclosure, constructing a specific simplicial complex based on at least one of a functional network and a structural network includes: by means of screening, constructing a specific simplicial complex based on at least one of a functional network and a structural network, so as to better reflect the structural connection relationship in, for example, white matter of the brain and the functional connection relationship in gray matter of the brain, and can achieve accurate dynamic monitoring.

[0376] According to the technical solution provided by the embodiment of the present disclosure, constructing a specific simplicial complex based on at least one of a functional network and a structural network further includes: constructing a comparison simplicial complex using a null model; comparing the specific simplicial complex and the comparison simplicial complex, so as to make a comparison with the null model and verify that the screening method better reflects the structural connection relationship in, for example, white matter of the brain and the functional connection relationship in gray matter of the brain, and can achieve accurate dynamic monitoring.

[0377] In an embodiment of the present disclosure, the topological descriptor includes at least one of the following: sub-threshold topological descriptor, persistent homology topological descriptor.

[0378] According to the technical solution provided by the embodiment of the present disclosure, the topological descriptor includes at least one of the following: sub-threshold topological descriptor, persistent homology topological descriptor, so as to comprehensively describe the topological characteristics of the neurological network and facilitate accurate detection and analysis.

[0379] In an embodiment of the present disclosure, the spatio-temporal metric evaluates the comparability of the neurological network at the spatial scale change and on the time line. Both the "relative average connectivity" and the "persistent homology topological descriptor" are comparable factors of the spatio-temporal metric. The simplex change feature quantitatively describes the spatio-temporal change feature of the simplex set and can measure the similarity degree of the topological structures of the structural network and the functional network in the time dimension.

[0380] According to the technical solution provided by the embodiment of the present disclosure, the spatio-temporal metric step includes: calculating the simplex change feature based on the topological descriptor of the network topology analysis step, and combining the fiber bundle analysis result of the fiber bundle analysis step to obtain the spatio-temporal metric result, so as to accurately detect and analyze the spatio-temporal characteristics of the neurological network.

[0381] According to the technical solution provided by the embodiments of the present disclosure, the simplex variation features include at least one of the following: simplex overlap rate, simplex evolution coefficient, so as to accurately describe the spatio-temporal variation of the simplex in multiple dimensions.

[0382] Figure 8 The structural block diagram of a neurological network spatio-temporal metric device according to an embodiment of the present disclosure is shown.

[0383] As Figure shown, the neurological network spatio-temporal metric device 800 includes: a region of interest selection module 801, a fiber bundle analysis module 802, and a network topology analysis module 803.

[0384] The region of interest selection module 801 is configured to perform a first registration based on the original relaxation time weighted imaging data of magnetic resonance imaging, obtain the relaxation time weighted imaging data after the first registration, and obtain an accurate atlas and a region of interest based on the relaxation time weighted imaging data after the first registration.

[0385] The fiber bundle analysis module 802 is configured to perform a first preprocessing and a second registration based on the diffusion tensor imaging data of the magnetic resonance imaging, obtain the diffusion tensor imaging data after the second registration, perform fiber bundle construction and processing on the diffusion tensor imaging data after the second registration, and obtain a fiber bundle analysis result.

[0386] The network topology analysis module 803 is configured to perform data processing and analysis based on specific functional data, obtain the specific functional data after data processing and analysis, construct a functional network from the specific functional data after data processing and analysis, and obtain a topological descriptor through network topology analysis of at least one of the functional network and the structural network.

[0387] According to the technical solution provided by the embodiments of the present disclosure, through a region of interest selection module, which is configured to perform a first registration based on the original relaxation time weighted imaging data of magnetic resonance imaging to obtain the first registered relaxation time weighted imaging data, and obtain an accurate atlas and a region of interest based on the first registered relaxation time weighted imaging data; a fiber bundle analysis module, which is configured to perform a first preprocessing and a second registration based on the diffusion tensor imaging data of the magnetic resonance imaging to obtain the second registered diffusion tensor imaging data, and perform fiber bundle construction and processing on the second registered diffusion tensor imaging data to obtain a fiber bundle analysis result; a network topology analysis module, which is configured to perform data processing and analysis based on specific functional data to obtain the specific functional data after data processing and analysis, construct a functional network from the specific functional data after data processing and analysis, and obtain a topological descriptor through network topology analysis of at least one of the functional network and the structural network, so as to realize accurate dynamic monitoring of the state of highly dense neural tissues such as the brain and accurate measurement of functional changes at the levels of neural regions, circuits, and systems such as brain regions, and can be extended to research fields such as non-invasive brain-computer interfaces.

[0388] ​ The structural block diagram of a neurological network spatio-temporal metric device according to an embodiment of the present disclosure is shown.

[0389] As ​ shown, in addition to including the same region of interest selection module 801, fiber bundle analysis module 802, and network topology analysis module 803 as ​ , the neurological network spatio-temporal metric device 900 further includes: a spatio-temporal metric module 901.

[0390] The spatio-temporal metric module 901 is configured to calculate a spatio-temporal metric result based on the fiber bundle analysis result of the fiber bundle analysis step and the topological descriptor of the network topology analysis step.

[0391] According to the technical solution provided by the embodiments of the present disclosure, through the spatio-temporal metric module being configured to calculate a spatio-temporal metric result based on the fiber bundle analysis result of the fiber bundle analysis step and the topological descriptor of the network topology analysis step, so as to realize accurate dynamic monitoring of the state of highly dense neural tissues such as the brain and accurate measurement of functional changes at the levels of neural regions, circuits, and systems such as brain regions, and can be extended to research fields such as non-invasive brain-computer interfaces.

[0392] ​ The structural block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0393] The embodiments of the present disclosure also provide an electronic device, as ​As shown, the electronic device 1000 includes a processor 1001 and a memory 1002; wherein, the memory 1002 stores instructions executable by at least one processor 1001, and the instructions are executed by at least one processor 1001 to implement the following steps:

[0394] Step of selecting region of interest, wherein, based on the original relaxation time weighted imaging data of magnetic resonance imaging, a first registration is performed to obtain the relaxation time weighted imaging data after the first registration, and an accurate atlas and a region of interest are obtained based on the relaxation time weighted imaging data after the first registration;

[0395] Step of fiber bundle analysis, wherein, based on the diffusion tensor imaging data of the magnetic resonance imaging, a first preprocessing and a second registration are performed to obtain the diffusion tensor imaging data after the second registration, and fiber bundle construction and processing are performed on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result, and the fiber bundle analysis result includes a structural network;

[0396] Step of network topology analysis, wherein, based on specific functional data, data processing and analysis are performed to obtain the specific functional data after data processing and analysis, a functional network is constructed from the specific functional data after data processing and analysis, and a topological descriptor is obtained by performing network topology analysis on at least one of the functional network and the structural network.

[0397] In an embodiment of the present disclosure, the instructions are further executed by at least one processor 1001 to implement the following steps:

[0398] Step of spatio-temporal metric, wherein, based on the fiber bundle analysis result of the fiber bundle analysis step and the topological descriptor of the network topology analysis step, a spatio-temporal metric result is calculated.

[0399] In an embodiment of the present disclosure, the magnetic resonance imaging includes: brain magnetic resonance imaging, and / or spinal cord magnetic resonance imaging; and / or

[0400] The original relaxation time weighted imaging data includes: original longitudinal relaxation time weighted imaging data, or original transverse relaxation time weighted imaging data; and / or

[0401] The relaxation time weighted imaging data after the first registration includes: the longitudinal relaxation time weighted imaging data after the first registration, or the transverse relaxation time weighted imaging data after the first registration; and / or

[0402] The accurate atlas includes: a corrected atlas.

[0403] In an embodiment of the present disclosure, the performing a first registration based on the original relaxation time weighted imaging data of magnetic resonance imaging to obtain the relaxation time weighted imaging data after the first registration includes:

[0404] A de - boning registration sub - step, in which the original relaxation time - weighted imaging data is de - boned and registered to obtain the first registered relaxation time - weighted imaging data; or

[0405] A bone - included registration sub - step, in which the original relaxation time - weighted imaging data is registered with bones to obtain the first registered relaxation time - weighted imaging data.

[0406] In an embodiment of the present disclosure, the de - boning registration sub - step includes:

[0407] Using a first command to remove bone data from the original relaxation time - weighted imaging data to obtain relaxation time - weighted imaging data without bones;

[0408] Using a second command to register the relaxation time - weighted imaging data without bones to obtain the first registered relaxation time - weighted imaging data; and / or

[0409] The bone - included registration sub - step includes:

[0410] Using a third command to register the original relaxation time - weighted imaging data to obtain registered relaxation time - weighted imaging data with bones;

[0411] Using a fourth command to remove bone data from the registered relaxation time - weighted imaging data with bones to obtain the first registered relaxation time - weighted imaging data,

[0412] The bone includes: the skull.

[0413] In an embodiment of the present disclosure, the first command includes:

[0414] A combination of the bet or bet2 command in FSL, the flirt command based on the first template, the betsurf command, and the fslmaths command; or

[0415] The fslmaths command in FSL based on the second template; or

[0416] The bet2 command in FSL; or

[0417] The mri_watershed command in FreeSurfer; and / or

[0418] The second command includes:

[0419] The flirt command in FSL based on the second template; and / or

[0420] The third command includes:

[0421] The flirt command in the FSL based on the first template; and / or

[0422] The fourth command includes:

[0423] The bet2 command in the FSL; or

[0424] The fslmaths command in the FSL based on the second template.

[0425] In an embodiment of the present disclosure, the first template includes: the MNI_{152}_T1_1mm template, or the MNI_{152}_T2_1mm template; and / or

[0426] The second template includes: the MNI_{152}_T1_brain_1mm template, or the MNI_{152}_T2_brain_1mm template.

[0427] In an embodiment of the present disclosure, obtaining the accurate atlas and the region of interest based on the first registered relaxation time weighted imaging data includes:

[0428] Using the first registered relaxation time weighted imaging data and combining with the standard MNI spatial atlas to obtain the accurate atlas and the region of interest.

[0429] In an embodiment of the present disclosure, performing the first preprocessing and the second registration on the diffusion tensor imaging data based on the magnetic resonance imaging to obtain the second registered diffusion tensor imaging data includes:

[0430] Deboning the b0 image in the diffusion tensor imaging data and generating a deboned mask, using the generated mask to remove the bone from the DWI image in the diffusion tensor imaging data, and performing eddy current correction and the first motion correction on the diffusion tensor data after bone removal to obtain the preprocessed diffusion tensor imaging data;

[0431] For the preprocessed diffusion tensor imaging data, performing linear registration using the MNI_{152}_T1_brain_2mm template or the MNI_{152}_T2_brain_2mm template, or performing linear registration using the MNI_{152}_T1_brain_2mm template or the MNI_{152}_T2_brain_2mm template and then using the template to generate a mask to remove the bone again to obtain the second registered diffusion tensor imaging data, or

[0432] Using 3D Slicer to remove the skull, performing voxel modeling and registration to obtain the second registered diffusion tensor imaging data.

[0433] In an embodiment of the present disclosure, the fiber bundle construction and processing of the second registered diffusion tensor imaging data to obtain the fiber bundle analysis result includes:

[0434] A fiber bundle tracking sub-step, wherein fiber bundle tracking is performed on the second registered diffusion tensor imaging data to obtain a fiber bundle tracking result;

[0435] A fiber bundle analysis sub-step, wherein fiber bundle analysis is performed on the fiber bundle tracking result by combining the accurate atlas and the region of interest to obtain the fiber bundle analysis result.

[0436] In an embodiment of the present disclosure, the fiber bundle tracking sub-step includes:

[0437] Calculating the anisotropy index and the mean diffusivity from the second registered diffusion tensor imaging data;

[0438] Using a fiber bundle tracking method to perform fiber bundle tracking on the second registered diffusion tensor imaging data and calculating the tracking error;

[0439] Selecting the fiber bundle tracking method with the minimum tracking error to obtain the fiber bundle tracking result.

[0440] In an embodiment of the present disclosure, the fiber bundle tracking method includes at least one of the following:

[0441] EuDX method, deterministic method, probabilistic method, sfm method, xtract instruction, Tractography seeding instruction.

[0442] In an embodiment of the present disclosure, the fiber bundle analysis sub-step includes:

[0443] Based on the accurate atlas and the region of interest, according to the fiber bundle tracking result, calculating the number, average length, length distribution of the fiber bundles passing through the region of interest, and / or the number, average length, length distribution of the fiber bundles with the region of interest as endpoints as the fiber bundle analysis result; and / or

[0444] Based on the accurate atlas and the region of interest, according to the fiber bundle tracking result, calculating the fiber bundle connectivity of each voxel in the region of interest, and / or the average fiber bundle connectivity of the voxels in the region of interest as the fiber bundle analysis result; and / or

[0445] Constructing a structural network with the region of interest as vertices and the number of fiber bundles between the regions of interest as the edge weights as the fiber bundle analysis result.

[0446] In an embodiment of the present disclosure, the data processing and analysis based on specific function data, obtaining the specific function data after data processing and analysis, and constructing a function network from the specific function data after data processing and analysis include:

[0447] Performing second preprocessing and third registration on the blood oxygenation level-dependent data based on functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after the third registration, and constructing a function network from the blood oxygenation level-dependent data after the third registration, or

[0448] Performing data processing and analysis on electroencephalogram data to obtain the electroencephalogram data after data processing and analysis, and constructing a function network from the electroencephalogram data after data processing and analysis.

[0449] In an embodiment of the present disclosure, the performing second preprocessing and third registration on the blood oxygenation level-dependent data based on functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after the third registration includes:

[0450] Removing a first number of time points from the blood oxygenation level-dependent data based on functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after removing the time points;

[0451] Performing slice time correction and second motion correction on the blood oxygenation level-dependent data after removing the time points, and registering it to the MNI space using an echo planar imaging template to obtain the blood oxygenation level-dependent data in the MNI space;

[0452] Resampling the blood oxygenation level-dependent data in the MNI space, performing Detrend, removing covariates, and performing first bandwidth filtering after smoothing with a first kernel to obtain the blood oxygenation level-dependent data after the third registration.

[0453] In an embodiment of the present disclosure, the first number includes: 0 to 10; and / or

[0454] The resampling includes: resampling using a resolution of 2×2×2 mm or 3×3×3 mm; and / or

[0455] The first kernel includes: a FWHM 4 mm Gauss kernel, or a FWHM 6 mm Gauss kernel, or a FWHM 8 mm Gauss kernel; and / or

[0456] The first bandwidth includes: 0.01 to 0.08 Hz, or 0.02 to 0.09 Hz, or 0.01 to 0.10 Hz.

[0457] In an embodiment of the present disclosure, the performing data processing and analysis on electroencephalogram data to obtain the electroencephalogram data after data processing and analysis includes:

[0458] Filter the electroencephalogram (EEG) data, perform independent component analysis, remove artifacts, and combine the lead field matrix obtained by head model and source model modeling from the post-registration relaxation time weighted imaging data to perform signal source tracing and obtain the EEG data after data processing and analysis.

[0459] In an embodiment of the present disclosure, constructing a functional network from the third post-registration blood oxygenation level-dependent (BOLD) data includes:

[0460] Using the third post-registration BOLD data, based on the accurate atlas and the region of interest, taking the voxels in the region of interest as network vertices, and performing correlation estimation on the time series of the voxels as the weights of the edges to obtain a voxel-level local undirected network; and / or

[0461] Using the third post-registration BOLD data, based on the accurate atlas, taking the sub-regions within the global region corresponding to the accurate atlas as network vertices, calculating the time series of the sub-regions within the global region, and the interaction strength between the sub-regions within the global region as the weights of the edges to obtain a region-level global undirected network; and / or

[0462] Using the third post-registration BOLD data, based on the accurate atlas and the region of interest, taking the voxels in the region of interest as network vertices, and using a directed functional connectivity estimation method to obtain the weights of the edges to obtain a voxel-level local directed network; and / or

[0463] Using the third post-registration BOLD data, based on the accurate atlas, taking the sub-regions within the global region corresponding to the accurate atlas as network vertices, and using the directed functional connectivity estimation method to obtain the weights of the edges to obtain a region-level global directed network.

[0464] In an embodiment of the present disclosure, constructing a functional network from the EEG data after data processing and analysis includes:

[0465] Perform band division on the EEG data after data processing and analysis to obtain multi-band EEG data;

[0466] Use a directed functional connectivity estimation method for the multi-band EEG data to construct a directed EEG functional network, or

[0467] Use an undirected functional connectivity estimation method for the multi-band EEG data to construct a directed EEG functional network.

[0468] In an embodiment of the present disclosure, the directed functional connectivity estimation method includes at least one of the following:

[0469] Cross-correlation method, Granger causality analysis method, convergent cross mapping method.

[0470] In an embodiment of the present disclosure, obtaining a topological descriptor by performing network topology analysis on at least one of the functional network and the structural network includes:

[0471] Constructing a specific simplicial complex based on at least one of the functional network and the structural network;

[0472] Constructing a homology group based on the specific simplicial complex;

[0473] Analyzing topological features of the homology group to obtain the topological descriptor;

[0474] Obtaining the topological descriptor based on the specific simplicial complex.

[0475] In an embodiment of the present disclosure, constructing a specific simplicial complex based on at least one of the functional network and the structural network includes:

[0476] Constructing a specific simplicial complex based on at least one of the functional network and the structural network by means of screening.

[0477] In an embodiment of the present disclosure, constructing a specific simplicial complex based on at least one of the functional network and the structural network further includes:

[0478] Constructing a comparison simplicial complex using a null model;

[0479] Comparing the specific simplicial complex and the comparison simplicial complex.

[0480] In an embodiment of the present disclosure, the topological descriptor includes at least one of the following:

[0481] A topological descriptor below a threshold, a persistent homology topological descriptor.

[0482] In an embodiment of the present disclosure, the space-time metric step includes: calculating a simplicial change feature based on the topological descriptor of the network topology analysis step, and combining the fiber bundle analysis result of the fiber bundle analysis step to obtain the space-time metric result.

[0483] In an embodiment of the present disclosure, the simplicial change feature includes at least one of the following: a simplicial overlap rate, a simplicial evolution coefficient.

[0484] ​ It is a schematic structural diagram of a computer system suitable for implementing the space-time metric method of a neurological network according to an embodiment of the present disclosure.

[0485] As ​As shown, computer system 1100 includes a processing unit 1101, which can perform various processes in the embodiments shown in the above-mentioned drawings according to the programs stored in the read-only memory (ROM) 1102 or the programs loaded from the storage section 1108 into the random access memory (RAM) 1103. In the RAM 1103, various programs and data required for the operation of the system 1100 are also stored. The processing unit 1101, ROM 1102, and RAM 1103 are connected to each other via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0486] The following components are connected to the I / O interface 1105: an input section 1106 including a keyboard, a mouse, etc.; an output section 1107 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 1108 including a hard disk, etc.; and a communication section 1109 including a network interface card such as a LAN card, a modem, etc. The communication section 1109 performs communication processing via a network such as the Internet. A drive 1110 is also connected to the I / O interface 1105 as needed. A removable medium 1111, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1110 as needed so that a computer program read from it can be installed into the storage section 1108 as needed. Among them, the processing unit 1101 can be implemented as a processing unit such as a CPU, GPU, TPU, FPGA, NPU, etc.

[0487] In particular, according to the embodiments of the present disclosure, the methods described above with reference to the drawings can be implemented as computer software programs. For example, the embodiments of the present disclosure include a computer program product, which includes a computer program tangibly contained on a computer-readable medium, and the computer program includes program codes for executing the methods in the drawings. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 1109, and / or installed from the removable medium 1111.

[0488] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system that performs the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0489] The units or modules described in the embodiments of the present disclosure may be implemented in software or in hardware. The units or modules described may also be provided in a processor, and the names of these units or modules do not, in some cases, constitute a limitation on the units or modules themselves.

[0490] As another aspect, the present disclosure also provides a computer-readable storage medium, which may be the computer-readable storage medium included in the nodes described in the above embodiments, or may be a computer-readable storage medium that exists separately and is not assembled into a device. The computer-readable storage medium stores one or more programs, and the programs are used by one or more processors to execute the methods described in the present disclosure.

[0491] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles applied. Those skilled in the art should understand that the scope of the invention involved in the present disclosure is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, technical solutions formed by mutually replacing the above features with technical features (but not limited to) having similar functions disclosed in the present disclosure.

Claims

1. A spatio-temporal measurement method for a neurological network, comprising: A step of selecting a region of interest, wherein, based on the original relaxation time weighted imaging data of magnetic resonance imaging, a first registration is performed to obtain the relaxation time weighted imaging data after the first registration, and an accurate atlas and a region of interest are obtained based on the relaxation time weighted imaging data after the first registration; A fiber bundle analysis step, wherein, based on the diffusion tensor imaging data of the magnetic resonance imaging, a first preprocessing and a second registration are performed to obtain the diffusion tensor imaging data after the second registration, and fiber bundle construction and processing are performed on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result, and the fiber bundle analysis result includes a structural network; wherein, the performing fiber bundle construction and processing on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result includes: a fiber bundle tracking sub-step, wherein fiber bundle tracking is performed on the diffusion tensor imaging data after the second registration to obtain a fiber bundle tracking result; a fiber bundle analysis sub-step, wherein fiber bundle analysis is performed on the fiber bundle tracking result by combining the accurate atlas and the region of interest to obtain the fiber bundle analysis result; A network topology analysis step, wherein data processing and analysis are performed based on specific functional data to obtain the specific functional data after data processing and analysis, a functional network is constructed from the specific functional data after data processing and analysis, and a topological descriptor is obtained by performing network topology analysis on at least one of the functional network and the structural network; A spatio-temporal measurement step, wherein a simplex change feature is calculated based on the topological descriptor of the network topology analysis step, and the spatio-temporal measurement result is obtained by combining the fiber bundle analysis result of the fiber bundle analysis step.

2. The method according to claim 1, wherein: The magnetic resonance imaging includes: brain magnetic resonance imaging, and / or spinal cord magnetic resonance imaging; and / or The original relaxation time weighted imaging data includes: original longitudinal relaxation time weighted imaging data, or original transverse relaxation time weighted imaging data; and / or The relaxation time weighted imaging data after the first registration includes: the longitudinal relaxation time weighted imaging data after the first registration, or the transverse relaxation time weighted imaging data after the first registration, and / or The accurate atlas includes: a corrected atlas.

3. The method according to claim 2, wherein, The performing a first registration based on the original relaxation time weighted imaging data of magnetic resonance imaging to obtain the relaxation time weighted imaging data after the first registration includes: A de-boning registration sub-step, wherein de-boning registration is performed on the original relaxation time weighted imaging data to obtain the relaxation time weighted imaging data after the first registration; or A bone-included registration sub-step, wherein bone-included registration is performed on the original relaxation time weighted imaging data to obtain the relaxation time weighted imaging data after the first registration.

4. The method according to claim 3, wherein, The de-boning registration sub-step includes: Use a first command to remove bone data from the original relaxation time weighted imaging data, and obtain relaxation time weighted imaging data with bone removed; Use a second command to register the relaxation time weighted imaging data with bone removed, and obtain the first registered relaxation time weighted imaging data; and / or The bone-included registration sub-step includes: Use a third command to register the original relaxation time weighted imaging data, and obtain registered relaxation time weighted imaging data with bone; Use a fourth command to remove bone data from the registered relaxation time weighted imaging data with bone, and obtain the first registered relaxation time weighted imaging data, The bone includes: the skull.

5. The method according to claim 4, wherein The first command includes: A combination of the bet or bet2 command in FSL, the flirt command based on a first template, the betsurf command, and the fslmaths command; or The fslmaths command in FSL based on a second template; or The bet2 command in FSL; or The mri_watershed command in FreeSurfer; and / or The second command includes: The flirt command in FSL based on the second template; and / or The third command includes: The flirt command in FSL based on the first template; and / or The fourth command includes: The bet2 command in FSL; or The fslmaths command in FSL based on the second template.

6. The method according to claim 5, wherein The first template includes: the MNI152_T1_1mm template, or the MNI152_T2_1mm template; and / or The second template includes: the MNI152_T1_brain_1mm template, or the MNI152_T2_brain_1mm template.

7. The method according to claim 1, wherein Obtaining an accurate atlas and regions of interest based on the first registered relaxation time weighted imaging data includes: Using the first registered relaxation time weighted imaging data, and combining with a standard MNI space atlas, to obtain the accurate atlas and the regions of interest.

8. The method according to claim 1, wherein Performing a first preprocessing and a second registration on the diffusion tensor imaging data based on the magnetic resonance imaging, and obtaining second registered diffusion tensor imaging data includes: Debone the b0 image in the diffusion tensor imaging data and generate a deboned mask, use the deboned mask to remove bone from the DWI image in the diffusion tensor imaging data, perform eddy current correction and a first motion correction on the diffusion tensor data after bone removal, and obtain preprocessed diffusion tensor imaging data; For the preprocessed diffusion tensor imaging data, perform linear registration using the MNI152_T1_brain_2mm template or the MNI152_T2_brain_2mm template, or perform linear registration using the MNI152_T1_brain_2mm template or the MNI152_T2_brain_2mm template and then generate a mask using the template to remove the bones again to obtain the second registered diffusion tensor imaging data, or Use 3D Slicer to remove the skull, perform voxel modeling and registration to obtain the second registered diffusion tensor imaging data.

9. The method according to claim 1, characterized in that, The fiber tractography sub-step includes: Calculate the anisotropy index and the mean diffusivity from the second registered diffusion tensor imaging data; Use the fiber tractography method to perform fiber tractography on the second registered diffusion tensor imaging data and calculate the tracking error; Select the fiber tractography method with the minimum tracking error to obtain the fiber tractography result.

10. The method according to claim 9, characterized in that, The fiber tractography method includes at least one of the following: EuDX method, deterministic method, probabilistic method, sfm method, xtract instruction, Tractography seeding instruction.

11. The method according to claim 1, characterized in that, The fiber bundle analysis sub-step includes: Based on the accurate atlas and the region of interest, according to the fiber tractography result, calculate the number, average length, length distribution of the fiber bundles passing through the region of interest, and / or the number, average length, length distribution of the fiber bundles with the region of interest as endpoints as the fiber bundle analysis result; and / or Based on the accurate atlas and the region of interest, according to the fiber tractography result, calculate the fiber bundle connectivity of each voxel in the region of interest, and / or the average fiber bundle connectivity of the voxels in the region of interest as the fiber bundle analysis result; and / or Construct a structural network with the region of interest as vertices and the number of fiber bundles between the regions of interest as the edge weights as the fiber bundle analysis result.

12. The method according to claim 1, wherein The data processing and analysis based on the specific functional data to obtain the specific functional data after data processing and analysis, and constructing a functional network from the specific functional data after data processing and analysis includes: Perform second preprocessing and third registration on the blood oxygenation level-dependent data of functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after the third registration, and construct a functional network from the blood oxygenation level-dependent data after the third registration, or Perform data processing and analysis on the electroencephalogram data to obtain the electroencephalogram data after data processing and analysis, and construct a functional network from the electroencephalogram data after data processing and analysis.

13. The method according to claim 12, wherein The performing second preprocessing and third registration on the blood oxygenation level-dependent data of functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after the third registration includes: Remove the first quantity of time points from the blood oxygenation level-dependent data based on functional magnetic resonance imaging to obtain the blood oxygenation level-dependent data after the removal time points; the first quantity includes: 0 to 10; Perform slice time correction and second motion correction on the blood oxygenation level-dependent data after the removal time points, and register it to the MNI space using an echo planar imaging template to obtain the blood oxygenation level-dependent data in the MNI space; Resample the blood oxygenation level-dependent data in the MNI space, smooth it using the first kernel, Detrend, remove covariates, and perform first bandwidth filtering to obtain the third registered blood oxygenation level-dependent data.

14. The method according to claim 13, wherein The resampling includes: resampling at a resolution of 2x2x2 mm or 3x3x3 mm; and / or The first kernel includes: a FWHM 4 mm Gauss kernel, or a FWHM 6 mm Gauss kernel, or a FWHM 8 mm Gauss kernel; and / or The first bandwidth includes: 0.01 to 0.08 Hz, or 0.02 to 0.09 Hz, or 0.01 to 0.10 Hz.

15. The method according to claim 12, wherein Performing data processing and analysis based on the electroencephalogram data to obtain the electroencephalogram data after data processing and analysis includes: Filtering the electroencephalogram data, performing independent component analysis, removing artifacts, and combining the lead field matrix obtained by head model and source model modeling from the registered relaxation time weighted imaging data to perform signal tracing to obtain the electroencephalogram data after data processing and analysis.

16. The method according to claim 12, wherein Constructing a functional network from the third registered blood oxygenation level-dependent data includes: Using the third registered blood oxygenation level-dependent data, based on the accurate atlas and the region of interest, using the voxels in the region of interest as network vertices, and performing correlation estimation on the time series of the voxels as the weights of the edges to obtain a voxel-level local undirected network; and / or Using the third registered blood oxygenation level-dependent data, based on the accurate atlas, using the sub-regions within the global region corresponding to the accurate atlas as network vertices, calculating the time series of the sub-regions within the global region, and the interaction strength between the sub-regions within the global region as the weights of the edges to obtain a region-level global undirected network; and / or Using the third registered blood oxygenation level-dependent data, based on the accurate atlas and the region of interest, using the voxels in the region of interest as network vertices, and using a directed functional connectivity estimation method to obtain the weights of the edges to obtain a voxel-level local directed network; and / or Using the third registered blood oxygenation level-dependent data, based on the accurate atlas, using the sub-regions within the global region corresponding to the accurate atlas as network vertices, and using the directed functional connectivity estimation method to obtain the weights of the edges to obtain a region-level global directed network.

17. The method according to claim 12, characterized in that, Constructing a functional network from the electroencephalogram data after data processing and analysis includes: Perform band division on the EEG data after data processing and analysis to obtain multi-band EEG data; Adopt a directed functional connection estimation method for the multi-band EEG data to construct a directed functional EEG network, or Adopt an undirected functional connection estimation method for the multi-band EEG data to construct a directed functional EEG network.

18. The method according to any one of claims 16 or 17, characterized in that The directed functional connection estimation method includes at least one of the following: Cross-correlation method, Granger causality analysis method, convergence cross mapping method.

19. The method according to claim 17, wherein The undirected functional connection estimation method includes at least one of the following: Phase locking value method, coherence method, weighted phase lag coefficient method.

20. The method according to claim 1, characterized in that The obtaining of topological descriptors by performing network topology analysis on at least one of the functional network and the structural network includes: Construct a specific simplicial complex according to at least one of the functional network and the structural network; Construct a homology group according to the specific simplicial complex; Analyze the topological characteristics of the homology group to obtain the topological descriptors; Obtain the topological descriptors according to the specific simplicial complex.

21. The method according to claim 20, wherein The constructing of a specific simplicial complex according to at least one of the functional network and the structural network includes: Construct a specific simplicial complex by a screening method according to at least one of the functional network and the structural network.

22. The method according to claim 21, wherein The constructing of a specific simplicial complex according to at least one of the functional network and the structural network further includes: Construct a comparison simplicial complex using a null model; Compare the specific simplicial complex and the comparison simplicial complex.

23. The method according to claim 20, characterized in that, The topological descriptors include at least one of the following: Sub-threshold topological descriptors, persistent homology topological descriptors.

24. The method according to claim 1, characterized in that The simplex change characteristics include at least one of the following: Simplex overlap rate, simplex evolution coefficient.

25. A spatio-temporal metric device for a neurological network, comprising: A region of interest selection module configured to perform a first registration based on original relaxation time weighted imaging data of magnetic resonance imaging to obtain relaxation time weighted imaging data after the first registration, and obtain an accurate atlas and a region of interest based on the relaxation time weighted imaging data after the first registration; A fiber bundle analysis module configured to perform a first preprocessing and a second registration based on the diffusion tensor imaging data of the magnetic resonance imaging to obtain diffusion tensor imaging data after the second registration, perform fiber bundle construction and processing on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result, the fiber bundle analysis result including a structural network; wherein, the performing of fiber bundle construction and processing on the diffusion tensor imaging data after the second registration to obtain a fiber bundle analysis result includes: a fiber bundle tracking sub-step, wherein fiber bundle tracking is performed on the diffusion tensor imaging data after the second registration to obtain a fiber bundle tracking result; a fiber bundle analysis sub-step, wherein fiber bundle analysis is performed on the fiber bundle tracking result in combination with the accurate atlas and the region of interest to obtain the fiber bundle analysis result; A network topology analysis module, configured to perform data processing and analysis based on specific functional data, obtain the specific functional data after data processing and analysis, construct a functional network from the specific functional data after data processing and analysis, and obtain a topology descriptor through network topology analysis of at least one of the functional network and the structural network; A spatio-temporal metric module, configured to calculate simplex change features based on the topology descriptor of the network topology analysis step, and combine with the fiber bundle analysis result of the fiber bundle analysis step to obtain the spatio-temporal metric result.

26. An electronic device, characterized in that, Comprising 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 according to any one of claims 1-24.

27. A readable storage medium storing computer instructions thereon, characterized in that, When the computer instructions are executed by the processor, the method according to any one of claims 1-24 is implemented.

Images