Three-dimensional reconstruction method and system for tumor cell metabolite distribution based on artificial intelligence

Abstract

The invention relates to the field of bioelectricity data processing, and discloses a tumor cell metabolite distribution three-dimensional reconstruction method and system based on artificial intelligence, and the method comprises the steps: collecting the multi-modal data of tumor cells, recognizing the entity elements of the tumor cells and the corresponding entity relationship from the multi-modal data, and obtaining the entity relationship of the entity elements of the tumor cells; constructing a metabolite knowledge map of the tumor cells; according to the space coordinates of the metabolites and the microenvironment characteristic data of the tumor cells, analyzing a space aggregation mode of the metabolites in the tumor cells; in combination with the spatial distribution mode and the biological metabolic pathway of the metabolites, key features of the metabolites in the tumor cells are extracted, and a metabolite distribution discrimination network of the tumor cells is established; and identifying an abnormal distribution mode of the metabolite in the tumor cells, constructing a metabolite distribution three-dimensional model of the tumor cells, and outputting a metabolite three-dimensional distribution result of the tumor cells. According to the method, the accuracy and comprehensiveness of the three-dimensional distribution result of the tumor cell metabolites can be improved.

Description

Technical Field

[0001] The present invention relates to a three-dimensional reconstruction method and system for tumor cell metabolite distribution based on artificial intelligence, belonging to the field of bioelectrical data processing. Background Art

[0002] Tumor cell metabolite distribution refers to small molecule substances produced or consumed by tumor cells during growth and proliferation, including carbohydrates, amino acids, lipids, nucleotides, etc. The distribution and concentration changes of these metabolites can reflect key biological processes such as energy metabolism, biosynthesis, and signal transduction of tumors, thereby revealing the molecular mechanism of tumor occurrence and development, and providing key basis for early tumor diagnosis, personalized treatment, and prognosis evaluation. Therefore, it is particularly important to obtain accurate information on tumor cell metabolite distribution.

[0003] Currently, traditional three-dimensional reconstruction methods mainly rely on tissue section analysis techniques. By using image registration and interpolation techniques, two-dimensional images of multiple tissue sections are stacked into a three-dimensional model. Although this method can obtain certain metabolite information, it may be difficult to restore the true distribution of tumor metabolites, resulting in the lack of accuracy and comprehensiveness of the three-dimensional reconstruction results of tumor cell metabolite distribution.

[0004] Therefore, there is an urgent need for a solution to improve the accuracy and comprehensiveness of the three-dimensional distribution results of tumor cell metabolites. Summary of the Invention

[0005] The present invention provides a three-dimensional reconstruction method and system for tumor cell metabolite distribution based on artificial intelligence, and its main purpose is to improve the accuracy and comprehensiveness of the three-dimensional distribution results of tumor cell metabolites.

[0006] To achieve the above object, a three-dimensional reconstruction method for tumor cell metabolite distribution based on artificial intelligence provided by the present invention includes: Obtain tumor cells to be detected, collect imaging data, metabolomics data, genomics data, and proteomics data of the tumor cells, and perform alignment and fusion processing on the imaging data, the metabolomics data, the genomics data, and the proteomics data to obtain multi-modal data; Identify the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data, construct a metabolite knowledge graph of the tumor cells according to the entity elements and the entity relationships, and use the multi-modal data and the metabolite knowledge graph to identify the metabolites of the tumor cells; Based on the multi-modal data, determine the spatial coordinates of the metabolite, and extract the microenvironment characteristic data of the tumor cells. According to the spatial coordinates and the microenvironment characteristic data, analyze the spatial aggregation pattern of the metabolite in the tumor cells, and based on the metabolite knowledge graph, identify the biological metabolic pathway of the metabolite; According to the spatial aggregation pattern and the biological metabolic pathway, extract the key features of the metabolite in the tumor cells, where the key features include spatial distribution features, pathway correlation features, and concentration gradient features. Based on the key features, create a metabolite distribution discrimination network for the tumor cells; Based on the metabolite distribution discrimination network, identify the abnormal distribution pattern of the metabolite in the tumor cells. According to the abnormal distribution pattern and the multi-modal data, construct a three-dimensional model of the metabolite distribution in the tumor cells, and based on the three-dimensional model of the metabolite distribution, output the three-dimensional distribution result of the metabolite in the tumor cells.

[0007] Optionally, the aligning and fusing the imaging data, the metabolomics data, the genomics data, and the proteomics data to obtain multi-modal data includes: Perform data standardization processing on the imaging data, the metabolomics data, the genomics data, and the proteomics data to obtain first-modal standardized data, second-modal standardized data, third-modal standardized data, and fourth-modal standardized data; Based on the first-modal standardized data, extract the voxel coordinates in the imaging data, and perform spatial transformation processing on the voxel coordinates to obtain first spatial transformation coordinates; According to the second-modal standardized data, extract the second spatial coordinates corresponding to the metabolomics data; Based on the third-modal standardized data and the fourth-modal standardized data, generate third spatial coordinates and fourth spatial coordinates corresponding to the genomics data and the proteomics data; Perform coordinate alignment processing on the first spatial transformation coordinates with the second spatial coordinates, the third spatial coordinates, and the fourth spatial coordinates to obtain aligned spatial coordinates; According to the aligned spatial coordinates, perform data fusion processing on the first-modal standardized data, the second-modal standardized data, the third-modal standardized data, and the fourth-modal standardized data to obtain multi-modal data.

[0008] Optionally, the identifying the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data includes: Based on the multi-modal data, extract the molecular characteristics and pathological features of the tumor cells; Determine the cancer type of the tumor cells according to the molecular characteristics and the pathological characteristics; Identify the metabolic characteristics of the tumor cells based on the cancer type; Extract the metabolites of the tumor cells according to the metabolic characteristics, and analyze the metabolic pathways of the metabolites; Extract the regulatory factors of the metabolic pathway based on the metabolic characteristics and the metabolites; Combining the metabolic characteristics, the metabolites, the metabolic pathway and the regulatory factors, identify the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data.

[0009] Optionally, constructing a metabolite knowledge graph of the tumor cells according to the entity elements and the entity relationships includes: Extract the cancer cell type, metabolites, metabolic pathways and metabolic regulatory factors corresponding to the tumor cells according to the entity elements; Define the top-level entity and sub-entities of the tumor cells based on the entity relationships; Identify the relationship attributes among the cancer cell types, the metabolites, the metabolic pathways and the metabolic regulatory factors; Define the semantic rules of the top-level entity and the sub-entities according to the relationship attributes; Generate a metabolite knowledge graph of the tumor cells based on the top-level entity, the sub-entities and the semantic rules; Perform knowledge storage processing on the metabolite knowledge graph to obtain a metabolite knowledge map.

[0010] Optionally, analyzing the spatial aggregation pattern of the metabolites in the tumor cells according to the spatial coordinates and the microenvironment characteristic data includes: Perform spatial alignment processing on the spatial coordinates and the microenvironment characteristic data to obtain a spatial alignment result; Identify the hot spots of the metabolites in the tumor cells based on the spatial alignment result; Measure the metabolite concentration in the hot spots, and analyze the gradient change law of the metabolite concentration in the hot spots by using the following formula:

[0011] where, represents the gradient change law of the metabolite concentration in the hot spots, represents the change rate of the metabolite concentration in the x direction of the hot spots, represents the change rate of the metabolite concentration in the y direction of the hot spots, Indicates the rate of change of metabolite concentration in the z - direction of the hot spot region, which represents the partial derivative of metabolite concentration; According to the hot spot region and the gradient change rule, analyze the spatial aggregation pattern of the metabolite in the tumor cells.

[0012] Optionally, extract the key features of the metabolite in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathway, including: Extract the regional aggregation characteristics of the metabolite in the tumor cells according to the spatial aggregation pattern; Based on the regional aggregation characteristics, analyze the concentration change of the metabolite in the tumor cells; According to the concentration change, identify the concentration gradient characteristics of the metabolite; Determine the specific position of the metabolite in the biological metabolic pathway, and based on the specific position, analyze the functional role of the metabolite in the biological metabolic pathway; According to the functional role, identify the pathway - associated characteristics of the metabolite; Combining the regional aggregation characteristics, the concentration gradient characteristics and the pathway - associated characteristics, extract the key features of the metabolite in the tumor cells.

[0013] Optionally, create a metabolite distribution discrimination network for the tumor cells based on the key features, including: Based on the key features, analyze the metabolite distribution pattern of the tumor cells; Identify the disease stage of the tumor cells, and according to the disease stage, extract the key metabolites of the tumor cells; Calculate the correlation coefficient between the key metabolites and the disease stage; Based on the correlation coefficient, construct an association network between the key metabolites and the disease stage; According to the association network, identify the classification performance of the key metabolites in the disease stage; Based on the metabolite distribution pattern and the classification performance, create a metabolite distribution discrimination network for the tumor cells.

[0014] Optionally, identify the abnormal distribution pattern of the metabolite in the tumor cells based on the metabolite distribution discrimination network, including: Based on the metabolite distribution discrimination network, identify the abnormal distribution regions of the metabolite in the tumor cells; Analyze the metabolic pathways of the metabolite in the abnormal distribution regions and identify the activity levels of the metabolic pathways; Precipitate the tumor microenvironment of the metabolite according to the activity level; Identify the spatial heterogeneous characteristics of the metabolite based on the metabolic pathway and the tumor microenvironment; Analyze the disease progression of the tumor cells according to the spatial heterogeneous characteristics; Identify the abnormal distribution pattern of the metabolite in the tumor cells based on the metabolic pathway, the tumor microenvironment, the spatial heterogeneous characteristics and the disease progression;

[0015] Optionally, constructing a three-dimensional model of the metabolite distribution of the tumor cells according to the abnormal distribution pattern and the multimodal data includes: Extract the abnormal metabolites of the tumor cells according to the abnormal distribution pattern; Identify the key regulatory nodes of the abnormal metabolites and collect the expression data of the key regulatory nodes; Determine the three-dimensional coordinates of the tumor cells based on the multimodal data; Collect the concentration data of the abnormal metabolites and map the concentration data onto the three-dimensional coordinates to generate a spatial distribution map of the abnormal metabolites; According to the expression data and the three-dimensional coordinates, use the following formula to identify the spatial co-localization relationship between the abnormal metabolites and the key regulatory nodes:

[0016] where r represents the spatial co-localization relationship between the abnormal metabolite and the key regulatory node, represents the metabolite concentration value of the abnormal metabolite at the e-th observation point in the three-dimensional coordinates, represents the gene-protein expression level of the key regulatory node at the e-th observation point in the three-dimensional coordinates, represents the average concentration of the abnormal metabolite, represents the average value of the gene-protein expression level of the key regulatory node, x represents the metabolite concentration corresponding to the abnormal metabolite in the expression data, y represents the gene-protein expression level corresponding to the key regulatory node in the expression data, m represents the number of observation points in the three-dimensional coordinates, and e represents the index of the observation point; Construct a three-dimensional model of the metabolite distribution of the tumor cells based on the spatial distribution map and the spatial co-localization relationship.

[0017] To solve the above problems, the present invention also provides a three-dimensional reconstruction system for the metabolite distribution of tumor cells based on artificial intelligence, and the system includes: A data collection module, which is used to obtain tumor cells to be detected, collect imaging data, metabolomics data, genomics data and proteomics data of the tumor cells, and perform alignment and fusion processing on the imaging data, the metabolomics data, the genomics data and the proteomics data to obtain multi-modal data; A knowledge graph construction module, which is used to identify entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data, construct a metabolite knowledge graph of the tumor cells according to the entity elements and the entity relationships, and use the multi-modal data and the metabolite knowledge graph to identify metabolites of the tumor cells; An aggregation pattern analysis module, which is used to determine spatial coordinates of the metabolites based on the multi-modal data, extract microenvironment feature data of the tumor cells, analyze a spatial aggregation pattern of the metabolites in the tumor cells according to the spatial coordinates and the microenvironment feature data, and identify a biological metabolic pathway of the metabolites based on the metabolite knowledge graph; A distribution prediction module, which is used to extract key features of the metabolites in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathway, where the key features include spatial distribution features, pathway correlation features and concentration gradient features, and create a metabolite distribution discrimination network of the tumor cells based on the key features; A three-dimensional reconstruction module, which is used to identify an abnormal distribution pattern of the metabolites in the tumor cells based on the metabolite distribution discrimination network, construct a three-dimensional model of the metabolite distribution of the tumor cells according to the abnormal distribution pattern and the multi-modal data, and output a three-dimensional distribution result of the metabolites of the tumor cells based on the three-dimensional model of the metabolite distribution.

[0018] Compared with the problems described in the background art, in the embodiments of the present invention, by performing alignment and fusion processing on the imaging data, the metabolomics data, the genomics data, and the proteomics data, multi-modal data is obtained, which can ensure that data of different modalities are in the same spatial coordinate system to improve the comprehensiveness and accuracy of metabolite distribution detection; further, in the embodiments of the present invention, by identifying the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data, a metabolite knowledge graph of the tumor cells is constructed, which can infer the causal relationships between genes, proteins, metabolites, and tissue structures, helping to understand the network structure of metabolites in tumor cells, thereby providing a basis for determining the relative positions and connection modes of metabolites in three-dimensional reconstruction and ensuring the rationality and accuracy of the three-dimensional reconstruction results; in the embodiments of the present invention, by analyzing the spatial aggregation patterns of metabolites in tumor cells according to the spatial coordinates of metabolites and the microenvironment characteristic data of tumor cells, the interactions between different metabolites and the regulatory mechanisms at specific spatial positions can be discovered; further, in the embodiments of the present invention, by extracting the key features of the metabolites in the tumor cells according to the spatial aggregation patterns and the biological metabolic pathways, potential metabolic hotspots or abnormal metabolic pathways can be discovered, so as to more comprehensively predict the dynamic changes of metabolite distribution to create a discriminant network for metabolite distribution in tumor cells, real-time monitor the growth, metastasis, and response to treatment of tumors, and improve the accuracy of diagnosis; in the embodiments of the present invention, by identifying the abnormal distribution patterns of the metabolites in the tumor cells, it can ensure that the metabolite distribution in the key regions can be restored with high precision, helping doctors to diagnose tumors more accurately; further, in the embodiments of the present invention, by constructing a three-dimensional model of metabolite distribution in tumor cells according to the abnormal distribution patterns and the multi-modal data, the spatial distribution of metabolites in tumor cells and the interaction relationships between different metabolites can be presented intuitively and comprehensively, which helps doctors to detect tumors in the early stage of the disease and formulate more appropriate treatment plans; finally, in the embodiments of the present invention, by outputting the three-dimensional distribution results of metabolites in tumor cells based on the three-dimensional model of metabolite distribution, the spatial distribution of metabolites in tumor tissues can be intuitively displayed, the accuracy of three-dimensional reconstruction of metabolite distribution in tumor cells can be improved, and the spatial associations between metabolites and genes and proteins can be revealed, helping to understand the metabolic reprogramming mechanisms of tumor cells in different regions and improve the accuracy and specificity of diagnosis. Therefore, the method and system for three-dimensional reconstruction of metabolite distribution in tumor cells based on artificial intelligence provided by the embodiments of the present invention can improve the accuracy and comprehensiveness of the three-dimensional distribution results of metabolites in tumor cells. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic flowchart of a method for three-dimensional reconstruction of metabolite distribution in tumor cells based on artificial intelligence provided by an embodiment of the present invention; Figure 2 This is a schematic diagram of a module for implementing the three-dimensional reconstruction method of tumor cell metabolite distribution based on artificial intelligence provided by an embodiment of the present invention.

[0020] The implementation, functional characteristics, and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. Specific embodiments

[0021] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0022] An embodiment of the present application provides a three-dimensional reconstruction method for tumor cell metabolite distribution based on artificial intelligence. The execution subject of the three-dimensional reconstruction method for tumor cell metabolite distribution based on artificial intelligence includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the three-dimensional reconstruction method for tumor cell metabolite distribution based on artificial intelligence can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster, etc. Embodiment 1:

[0023] Referring to Figure 1 As shown, it is a flowchart of a three-dimensional reconstruction method for tumor cell metabolite distribution based on artificial intelligence provided by an embodiment of the present invention. In this embodiment, the three-dimensional reconstruction method for tumor cell metabolite distribution based on artificial intelligence includes: S1. Obtain the tumor cells to be detected, collect the imaging data, metabolomics data, genomics data, and proteomics data of the tumor cells, and perform alignment and fusion processing on the imaging data, the metabolomics data, the genomics data, and the proteomics data to obtain multimodal data.

[0024] By obtaining the tumor cells to be detected in the embodiment of the present invention, a data source can be provided for the subsequent three-dimensional reconstruction of metabolite distribution. The tumor cells refer to cells in an organism that have lost normal growth and differentiation regulation due to gene mutations or other factors, and thus have the ability of abnormal proliferation, such as sarcoma cells, lymphoma cells, leukemia cells, etc.

[0025] Furthermore, the embodiments of the present invention can comprehensively reveal the biological characteristics of tumor cells by collecting the imaging data, metabolomics data, genomics data and proteomics data of the tumor cells, and provide multi-dimensional information support for constructing a three-dimensional model of metabolite distribution. The imaging data refers to information about the internal structure and function of tumor cells obtained through various medical imaging technologies, such as the spatial structure of tumor cells, the shape and size of tumor cells, etc. The metabolomics data refers to information on all metabolites in tumor cells obtained through metabolomics technology, such as metabolite types, metabolite concentrations, etc. The genomics data refers to all information on the tumor cell genome obtained through genomics technology, including genome sequences, gene mutations, gene expression levels, etc. The proteomics data refers to information on all proteins in tumor cells obtained through proteomics technology, such as protein types, protein structures, etc.

[0026] Optionally, the imaging data collection of the tumor cells can be obtained by magnetic resonance imaging technology, the metabolomics data collection of the tumor cells can be achieved using mass spectrometry imaging technology, the genomic data collection of the tumor cells can be obtained by gene sequencing technology, such as third-generation sequencing technology, and the proteomics data collection of the tumor cells can be achieved using liquid chromatography-mass spectrometry technology.

[0027] The embodiment of the present invention obtains multimodal data by aligning and fusing the imaging data, the metabolomics data, the genomics data and the proteomics data, thereby ensuring that the data of different modalities are in the same spatial coordinate system to improve the comprehensiveness and accuracy of metabolite distribution detection. The alignment and fusion processing refers to adjusting the data from different sources to the same spatial coordinate system and obtaining fused data after standardization processing.

[0028] As an embodiment of the present invention, the alignment and fusion processing of the imaging data, the metabolomics data, the genomics data, and the proteomics data to obtain multimodal data includes: performing data standardization processing on the imaging data, the metabolomics data, the genomics data, and the proteomics data to obtain first-modal standardized data, second-modal standardized data, third-modal standardized data, and fourth-modal standardized data; based on the first-modal standardized data, extracting voxel coordinates in the imaging data and performing spatial transformation processing on the voxel coordinates to obtain first spatial transformation coordinates; according to the second-modal standardized data, extracting second spatial coordinates corresponding to the metabolomics data; based on the third-modal standardized data and the fourth-modal standardized data, generating third spatial coordinates and fourth spatial coordinates corresponding to the genomics data and the proteomics data; performing coordinate alignment processing on the first spatial transformation coordinates, the second spatial coordinates, the third spatial coordinates, and the fourth spatial coordinates to obtain aligned spatial coordinates; and according to the aligned spatial coordinates, performing data fusion processing on the first-modal standardized data, the second-modal standardized data, the third-modal standardized data, and the fourth-modal standardized data to obtain multimodal data.

[0029] Among them, the first-modal standardized data refers to the imaging data after standardization processing, the second-modal standardized data refers to the metabolomics data after standardization processing, the third-modal standardized data refers to the genomics data after standardization processing, the fourth-modal standardized data refers to the proteomics data after standardization processing. The voxel coordinates refer to the specific positions of the three-dimensional grids divided inside tumor cells in the imaging data, usually corresponding to the coordinate system of the imaging device. For example, the x coordinate represents the position of the voxel in the horizontal direction, the y coordinate represents the position of the voxel in the vertical direction, and the z coordinate represents the position of the voxel in the depth direction. The first spatial transformation coordinates refer to the voxel coordinates after spatial transformation processing, the second spatial coordinates refer to the spatial coordinates corresponding to the metabolomics data, the third spatial coordinates refer to the spatial coordinates corresponding to the genomics data, the fourth spatial coordinates refer to the spatial coordinates corresponding to the proteomics data. The coordinate alignment processing refers to the process of aligning the spatial coordinates of different-modal data, and the aligned spatial coordinates refer to the unified spatial coordinates obtained after coordinate alignment processing.

[0030] It should be explained that in imaging data (such as CT, MRI, PET, etc.), the interior of an organism is divided into many small three-dimensional grids, each grid unit is called a voxel, and each voxel contains a specific value, which usually represents a certain physical property of the voxel, such as density, signal intensity, or radioactive concentration, etc.

[0031] Optionally, the data standardization processing of the imaging data, the metabolomics data, the genomics data, and the proteomics data can be achieved by using the Transformer architecture to learn the complex relationships between the data. The spatial transformation processing of the voxel coordinates can be obtained through a Spatial Transformer Network (STN). The coordinate alignment processing of the first spatial transformation coordinate with the second, third, and fourth spatial coordinates can be achieved by using the affine transformation method. The generation of the third and fourth spatial coordinates corresponding to the genomics data and the proteomics data based on the third-modal standardized data and the fourth-modal standardized data can be determined by tissue section localization.

[0032] S2. Identify the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data. According to the entity elements and the entity relationships, construct a metabolite knowledge graph of the tumor cells, and use the multi-modal data and the metabolite knowledge graph to identify the metabolites of the tumor cells.

[0033] In an embodiment of the present invention, by identifying the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data, the causal relationships between genes, proteins, metabolites, and tissue structures can be inferred, which helps to construct a knowledge graph of tumor cell metabolites. The entity elements refer to specific substances or structural units that can be identified and measured in tumor cells, including proteins, genes, metabolites, and metabolic pathways. The entity relationships refer to the logical associations or interactions existing between entity elements, such as the regulation of proteins by genes, the reactions between metabolites and proteins, and the signal transduction between cell types.

[0034] As an embodiment of the present invention, the identification of the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data includes: based on the multi-modal data, extract the molecular characteristics and pathological features of the tumor cells; according to the molecular characteristics and the pathological features, determine the cancer type of the tumor cells; based on the cancer type, identify the metabolic characteristics of the tumor cells; according to the metabolic characteristics, analyze the metabolic products of the tumor cells and the metabolic pathways of the metabolic products; based on the metabolic characteristics and the metabolic products, extract the regulatory factors of the metabolic pathways; and combine the metabolic characteristics, the metabolic products, the metabolic pathways, and the regulatory factors to identify the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data.

[0035] Among them, the molecular characteristics refer to the characteristics of tumor cells at the molecular level, including the expression and interactions of genes, proteins, metabolites, etc. within the cells, such as mutations or overexpressions of specific genes in tumor cells. The pathological characteristics refer to the morphological and structural characteristics of tumor cells in histology and cytology, such as tumor cells may exhibit characteristics such as enlarged nuclei and increased mitotic figures. The cancer types refer to the tumor types classified according to the origin tissues, cell types, and pathological characteristics of tumor cells. For example, lung cancer and breast cancer originate from epithelial tissues. The metabolic characteristics refer to the unique manifestations of tumor cells during the metabolic process, including changes in metabolic pathways and the generation of metabolites. The metabolites refer to the small molecule compounds generated by tumor cells during the metabolic process, such as lactic acid, glutamic acid, etc. The metabolic pathways refer to the generation and conversion pathways of metabolites, such as the glycolysis pathway, the tricarboxylic acid cycle, etc. The regulatory factors refer to the molecules or signals that can regulate the reaction rate and direction of metabolic reactions in the metabolic pathway, such as hexokinase 2, lactate dehydrogenase A, etc.

[0036] Optionally, based on the multi-modal data, the extraction of the molecular characteristics of the tumor cells can be obtained through genomics data and proteomics data. The pathological characteristics of the tumor cells can be observed through a microscope. Based on the cancer types, the identification of the metabolic characteristics of the tumor cells can be determined through metabolomics data. According to the metabolic characteristics, the analysis of the metabolic pathways of the metabolites can be achieved by using metabolic flux analysis methods, such as stable isotope labeling methods. Combining the metabolic characteristics, the metabolites, the metabolic pathways, and the regulatory factors, the identification of the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data can be realized through a rule engine, such as the Jena rule engine.

[0037] Furthermore, by constructing a metabolite knowledge graph of the tumor cells according to the entity elements and the entity relationships in the embodiments of the present invention, it can help understand the network structure of metabolites within tumor cells, thereby providing a basis for determining the relative positions and connection methods of metabolites in 3D reconstruction, and ensuring the rationality and accuracy of the 3D reconstruction results. The metabolite knowledge graph refers to a tool for describing tumor cell metabolites and their metabolic processes using a graph model.

[0038] As an embodiment of the present invention, constructing the metabolite knowledge graph of the tumor cells according to the entity elements and the entity relationships includes: extracting the cancer cell type, metabolite, metabolic pathway, and metabolic regulatory factor corresponding to the tumor cells according to the entity elements; defining the top-level entity and sub-entities of the tumor cells based on the entity relationships; identifying the relationship attributes among the cancer cell types, the metabolites, the metabolic pathways, and the metabolic regulatory factors; defining the semantic rules of the top-level entity and the sub-entities according to the relationship attributes; generating the metabolite knowledge graph of the tumor cells based on the top-level entity, the sub-entities, and the semantic rules; and performing knowledge storage processing on the metabolite knowledge graph to obtain the metabolite knowledge map.

[0039] Among them, the metabolic regulatory factor refers to a gene, enzyme, or protein that regulates the metabolic pathway. The top-level entity refers to the highest-level entity in the metabolite knowledge graph, such as a tumor cell. The sub-entity refers to a specific instance or subdivision of the top-level entity. For example, if the top-level entity is a tumor cell, the sub-entity can be a breast cancer cell or a lung cancer cell. The relationship attribute refers to the association method between entities, such as production (breast cancer cells produce lactic acid), through (lactic acid through the glycolysis pathway), influence (HK2 affects lactic acid), etc. The semantic rule refers to the logical and semantic constraint rules that define the relationship between entities, such as cancer cell type A produces metabolite C through metabolic pathway B. The metabolite knowledge graph refers to a graph structure used to store and display entities related to tumor cell metabolism and their relationships, including nodes and edges. The knowledge storage processing refers to the process of storing the metabolite knowledge graph in a database or knowledge base.

[0040] Optionally, the definition of the top-level entity and sub-entities of the tumor cells based on the entity relationships can be achieved by classifying the top-level entity using domain knowledge or data sources. The identification of the relationship attributes among the cancer cell types, the metabolites, the metabolic pathways, and the metabolic regulatory factors can be determined by analyzing the relevance of experimental data, such as the correlation between metabolite concentration and gene expression. The knowledge storage processing of the metabolite knowledge graph can be implemented using the Neo4j tool.

[0041] By using the multi-modal data and the metabolite knowledge graph, the embodiments of the present invention can identify the metabolites of the tumor cells, which can improve the detection accuracy of the metabolite distribution of the tumor cells. The metabolite refers to an intermediate or end product in the cell metabolism process, such as fatty acid, adenosine acid, etc.

[0042] S3. Based on the multi-modal data, determine the spatial coordinates of the metabolite, extract the microenvironment feature data of the tumor cells, analyze the spatial aggregation pattern of the metabolite in the tumor cells according to the spatial coordinates and the microenvironment feature data, and identify the biological metabolic pathway of the metabolite based on the metabolite knowledge graph.

[0043] In the embodiment of the present invention, by determining the spatial coordinates of the metabolite based on the multi-modal data and extracting the microenvironment feature data of the tumor cells, the spatial relationship between the metabolite and different cell types can be revealed, which helps to understand the mechanism of action of the metabolite. The spatial coordinates refer to the specific coordinates of the metabolite in the tumor cells, and the microenvironment feature data refer to data such as the oxygen level, blood vessel distribution, pH value, extracellular matrix composition, and immune cell infiltration in the local area where the tumor cells are located.

[0044] Optionally, the extraction of the microenvironment feature data of the tumor cells based on the multi-modal data can be realized by immunofluorescence imaging technology. For example, the spatial distribution of the hypoxic region and blood vessels can be obtained by using immunofluorescence imaging technology. The determination of the spatial coordinates of the metabolite based on the multi-modal data can be realized by secondary ion mass spectrometry imaging.

[0045] Furthermore, in the embodiment of the present invention, by analyzing the spatial aggregation pattern of the metabolite in the tumor cells according to the spatial coordinates and the microenvironment feature data, the interaction between different metabolites and the regulation mechanism at specific spatial positions can be discovered. The spatial aggregation pattern refers to the specific rules and characteristics presented by the spatial distribution of metabolites in the tumor, mainly including the distribution state, aggregation form, and relationship with the surrounding environment, etc.

[0046] As an embodiment of the present invention, the analysis of the spatial aggregation pattern of the metabolite in the tumor cells according to the spatial coordinates and the microenvironment feature data includes: performing spatial alignment processing on the spatial coordinates and the microenvironment feature data to obtain a spatial alignment result; identifying the hot spots of the metabolite in the tumor cells based on the spatial alignment result; measuring the metabolite concentration in the hot spots and analyzing the gradient change law of the metabolite concentration in the hot spots; and analyzing the spatial aggregation pattern of the metabolite in the tumor cells according to the hot spots and the gradient change law.

[0047] Among them, the spatial alignment process refers to the process of registering the spatial coordinates of metabolites and the microenvironment feature data in space. The spatial alignment result refers to the matching result of metabolite distribution data and microenvironment feature data in the same coordinate system after the spatial alignment process. The hot spot region refers to the spatial region where metabolites are highly concentrated in tumor cells. The metabolite concentration refers to the content or abundance of metabolites in the hot spot region, such as the mass spectrometry signal intensity of lactate in the hot spot region. The gradient change rule refers to the change trend of metabolite concentration in space, such as the lactate concentration gradually decreasing from the tumor core to the edge.

[0048] Optionally, the spatial alignment process of the spatial coordinates and the microenvironment feature data can be achieved through an image registration tool, such as the ImageJ tool. Based on the spatial alignment result, the identification of the hot spot region of metabolites in the tumor cells can be achieved by using the kernel density estimation method. The measurement of the metabolite concentration in the hot spot region can be obtained through the SCiLSLab software.

[0049] In an optional embodiment of the present invention, the following formula is used to analyze the gradient change rule of the metabolite concentration in the hot spot region:

[0050] Among them, represents the gradient change rule of metabolite concentration in the hot spot region, represents the change rate of metabolite concentration in the x direction of the hot spot region, represents the change rate of metabolite concentration in the y direction of the hot spot region, represents the change rate of metabolite concentration in the z direction of the hot spot region, represents the partial derivative of metabolite concentration.

[0051] In the embodiment of the present invention, by identifying the biological metabolic pathway of the metabolite based on the metabolite knowledge graph, it can help to understand the dynamic changes of metabolites under different physiological or pathological conditions. The biological metabolic pathway refers to the specific biochemical reaction path or network participated by metabolites in tumor cells, such as glucose being converted into pyruvate through the glycolysis pathway and then generating lactate.

[0052] Optionally, based on the metabolite knowledge graph, the identification of the biological metabolic pathway of the metabolite can be achieved through a pathway visualization tool, such as the Cytoscape tool.

[0053] S4. Extract the key features of the metabolite in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathway. Among them, the key features include spatial distribution features, pathway correlation features, and concentration gradient features. Based on the key features, create a metabolite distribution discrimination network for the tumor cells.

[0054] In an embodiment of the present invention, by extracting the key features of the metabolite in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathway, potential metabolic hotspots or abnormal metabolic pathways can be discovered, so as to more comprehensively predict the dynamic changes in the metabolite distribution. The key features refer to the features that can accurately reflect the characteristics and behaviors of metabolites, including spatial distribution features, pathway correlation features, and concentration gradient features.

[0055] As an embodiment of the present invention, the extracting the key features of the metabolite in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathway includes: extracting the regional aggregation features of the metabolite in the tumor cells according to the spatial aggregation pattern; analyzing the concentration change of the metabolite in the tumor cells based on the regional aggregation features; identifying the concentration gradient features of the metabolite according to the concentration change; determining the specific position of the metabolite in the biological metabolic pathway, and analyzing the functional role of the metabolite in the biological metabolic pathway based on the specific position; identifying the pathway correlation features of the metabolite according to the functional role; and extracting the key features of the metabolite in the tumor cells by combining the regional aggregation features, the concentration gradient features, and the pathway correlation features.

[0056] Among them, the regional aggregation features refer to the distribution features of metabolites in different regions within tumor cells. For example, metabolites are mainly concentrated in specific regions of the cytoplasm. The concentration change refers to the concentration difference of metabolites in different regions or at different time points within tumor cells. The concentration gradient features refer to the concentration difference formed between the inside and outside of cells or between different regions within cells. For example, the concentration of metabolites is lower outside the cells and higher in a certain region inside the cells. The specific position refers to the specific link or step of the metabolite in the biological metabolic pathway. For example, whether the metabolite is a substrate of a certain enzyme, or an intermediate or final product of the metabolic pathway. The functional role refers to the specific physiological function of the metabolite in the biological metabolic pathway. For example, the metabolite participates in processes such as energy metabolism, substance synthesis, or signal transduction. The pathway correlation features refer to the correlation between the metabolite and other metabolic pathways or cellular processes. For example, whether the metabolite participates in multiple metabolic pathways simultaneously, or whether its function is regulated by other cellular signaling pathways.

[0057] Optionally, according to the spatial aggregation pattern, the extraction of the regional aggregation characteristics of the metabolite in the tumor cells can be obtained by analyzing metabolite imaging data through a machine learning algorithm, such as a support vector machine. Based on the regional aggregation characteristics, the analysis of the concentration change of the metabolite in the tumor cells can be achieved by simulating the dynamic process of metabolite concentration change using the metabolic network model of tumor cells. The determination of the specific position of the metabolite in the biological metabolic pathway can be realized by using the stable isotope labeling method. Based on the specific position, the analysis of the functional role of the metabolite in the biological metabolic pathway can be determined by calculating the distribution of metabolic fluxes and analyzing the functional contribution of the metabolite in the pathway.

[0058] Furthermore, in the embodiment of the present invention, by creating a metabolite distribution discrimination network of the tumor cells based on the key features, the growth, metastasis and response to treatment of the tumor can be monitored in real time, and the accuracy of diagnosis can be improved. The metabolite distribution discrimination network refers to a model for identifying, classifying and analyzing the metabolic state of tumor cells.

[0059] As an embodiment of the present invention, creating the metabolite distribution discrimination network of the tumor cells based on the key features includes: analyzing the metabolite distribution pattern of the tumor cells based on the key features; identifying the disease course stage of the tumor cells, and extracting the key metabolites of the tumor cells according to the disease course stage; calculating the correlation coefficient between the key metabolites and the disease course stage; constructing an association network between the key metabolites and the disease course stage based on the correlation coefficient; identifying the classification performance of the key metabolites in the disease course stage according to the association network; and creating the metabolite distribution discrimination network of the tumor cells based on the metabolite distribution pattern and the classification performance.

[0060] Among them, the metabolite distribution pattern refers to the spatial distribution characteristics and rules of metabolites in tumor cells or tissues. The disease course stage refers to different biological stages of tumor development from occurrence to progression, such as early-stage, mid-stage, and late-stage tumors. The key metabolites refer to important metabolites with significant biological significance or functions in tumor cells, such as lactate, glutamine, ATP, etc. The correlation coefficient refers to a statistical index used to measure the strength and direction of the linear relationship between the key metabolites and the disease course stage. The association network refers to a network model used to describe the relationship between metabolites and the disease course stage of tumors. The classification performance refers to the characteristic performance of metabolites in different disease course stages. For example, some metabolites may show low expression in early-stage tumor cells and high expression in late-stage tumor cells.

[0061] Optionally, based on the key features, the analysis of the metabolite distribution pattern of the tumor cells can be achieved through a clustering analysis algorithm, the identification of the disease course stage of the tumor cells can be determined by molecular markers, the construction of the association network between the key metabolites and the disease course stage based on the correlation coefficient can be achieved by using graph theory tools, such as the NetworkX tool, and the identification of the classification performance of the key metabolites in the disease course stage according to the association network can be obtained through a classification model, such as a random forest model.

[0062] S5. Based on the metabolite distribution discrimination network, identify the abnormal distribution pattern of the metabolite in the tumor cells, construct a three-dimensional model of the metabolite distribution of the tumor cells according to the abnormal distribution pattern and the multi-modal data, and output the three-dimensional distribution result of the metabolite of the tumor cells based on the three-dimensional model of the metabolite distribution.

[0063] In the embodiment of the present invention, by identifying the abnormal distribution pattern of the metabolite in the tumor cells based on the metabolite distribution discrimination network, it can ensure that the metabolite distribution in the key region can be restored with high precision, helping doctors diagnose tumors more accurately. The abnormal distribution pattern refers to a pattern in which the spatial distribution, concentration or type of the metabolite in the tumor cells shows significant differences compared with normal cells or tissues, such as local enrichment, local deletion, etc.

[0064] As an embodiment of the present invention, the identification of the abnormal distribution pattern of the metabolite in the tumor cells based on the metabolite distribution discrimination network includes: based on the metabolite distribution discrimination network, identifying the abnormal distribution region of the metabolite in the tumor cells; analyzing the metabolic pathway of the metabolite in the abnormal distribution region and identifying the activity degree of the metabolic pathway; according to the activity degree, extracting the tumor microenvironment of the metabolite; based on the metabolic pathway and the tumor microenvironment, identifying the spatial heterogeneous characteristics of the metabolite; according to the spatial heterogeneous characteristics, analyzing the disease course progression of the tumor cells; and based on the metabolic pathway, the tumor microenvironment, the spatial heterogeneous characteristics and the disease course progression, identifying the abnormal distribution pattern of the metabolite in the tumor cells.

[0065] Among them, the abnormal distribution region refers to the region in the tumor microenvironment where the distribution of metabolites is significantly different from the normal physiological state or the expected distribution. For example, a large amount of lactic acid may accumulate in the tumor core region due to hypoxia, while the metabolite distribution in the region rich in blood vessels may be more complex. The metabolic pathway refers to the path by which metabolites are transformed through a series of enzymatic reactions within or between cells, including glycolysis, the tricarboxylic acid cycle (TCA cycle), fatty acid metabolism, amino acid metabolism, etc. The activity level refers to the level of metabolic flux or metabolite conversion rate of each reaction in the metabolic pathway. The tumor microenvironment refers to the complex environment composed of non-tumor cells (such as immune cells, fibroblasts) and physicochemical factors (such as hypoxia, acidic environment) surrounding the tumor cells corresponding to the metabolites. The spatial heterogeneity feature refers to the non-uniformity of metabolite distribution in tumor cells, manifested as enrichment or depletion of metabolites in certain regions. The disease course progression refers to the dynamic process of tumor occurrence, development, and metastasis.

[0066] Optionally, the analysis of the metabolic pathway of the metabolite in the abnormal distribution region can be achieved by using a metabolic pathway enrichment analysis tool, such as the MetaboAnalyst tool. According to the activity level, the precipitation of the metabolite in the tumor microenvironment can be simulated using an in vitro model. Based on the metabolic pathway and the tumor microenvironment, the identification of the spatial heterogeneity feature of the metabolite can be determined by fluorescence microscopy or confocal microscopy imaging.

[0067] Furthermore, in the embodiment of the present invention, by constructing a three-dimensional model of the metabolite distribution of the tumor cells according to the abnormal distribution pattern and the multi-modal data, the spatial distribution of metabolites in the tumor cells and the interaction relationship between different metabolites can be intuitively and comprehensively presented, which helps doctors detect tumors at an early stage of the disease and formulate more appropriate treatment plans. The three-dimensional model of metabolite distribution refers to a model that intuitively presents the distribution of metabolites in an organism in a three-dimensional space form.

[0068] As an embodiment of the present invention, constructing the three-dimensional model of the metabolite distribution of the tumor cells according to the abnormal distribution pattern and the multi-modal data includes: extracting the abnormal metabolites of the tumor cells according to the abnormal distribution pattern; identifying the key regulatory nodes of the abnormal metabolites and collecting the expression data of the key regulatory nodes; determining the three-dimensional coordinates of the tumor cells based on the multi-modal data; collecting the concentration data of the abnormal metabolites and mapping the concentration data onto the three-dimensional coordinates to generate a spatial distribution map of the abnormal metabolites; identifying the spatial co-localization relationship between the abnormal metabolites and the key regulatory nodes according to the expression data and the three-dimensional coordinates; and constructing the three-dimensional model of the metabolite distribution of the tumor cells based on the spatial distribution map and the spatial co-localization relationship.

[0069] Among them, the abnormal metabolite refers to a metabolite whose concentration in tumor cells deviates significantly from the normal range. The key regulatory node refers to a molecule or reaction that plays an important regulatory role in the metabolic network, such as a key enzyme, a signaling molecule, or a metabolite, etc. The expression data refers to the expression level of a key regulatory node (such as a gene or a protein) in tumor cells that can reflect its activity or functional state. The three-dimensional coordinates refer to the position information of tumor cells or metabolites in three-dimensional space. The concentration data refers to the content information of abnormal metabolites in tumor cells. The spatial distribution map refers to the visualized result of the spatial distribution of abnormal metabolites in tumor cells or tissues. The spatial co-localization relationship refers to the coexistence or proximity relationship between key regulatory nodes and abnormal metabolites in space.

[0070] Optionally, the identification of key regulatory nodes of the abnormal metabolite can be obtained by visualizing the metabolic network through topological analysis techniques. Mapping the concentration data onto the three-dimensional coordinates to generate the spatial distribution map of the abnormal metabolite can be achieved using a heat map. The acquisition of the concentration data of the abnormal metabolite can be obtained through spatial transcriptomics techniques. Based on the spatial distribution map and the spatial co-localization relationship, the construction of the three-dimensional model of the metabolite distribution in tumor cells can be achieved using three-dimensional modeling software, such as Blender software.

[0071] In an optional embodiment of the present invention, according to the expression data and the three-dimensional coordinates, the following formula is used to identify the spatial co-localization relationship between the abnormal metabolite and the key regulatory node:

[0072] Among them, r represents the spatial co-localization relationship between the abnormal metabolite and the key regulatory node. represents the metabolite concentration value of the abnormal metabolite at the e-th observation point in the three-dimensional coordinates. represents the gene-protein expression level of the key regulatory node at the e-th observation point in the three-dimensional coordinates. represents the average concentration of the abnormal metabolite. represents the average of the gene-protein expression levels of the key regulatory node. x represents the metabolite concentration corresponding to the abnormal metabolite in the expression data. y represents the gene-protein expression level corresponding to the key regulatory node in the expression data. m represents the number of observation points in the three-dimensional coordinates. e represents the index of the observation point.

[0073] In an embodiment of the present invention, based on the three-dimensional metabolite distribution model, the three-dimensional metabolite distribution result of the tumor cells is output, which can visually display the spatial distribution of metabolites in the tumor tissue, improve the accuracy of the three-dimensional reconstruction of the metabolite distribution of the tumor cells, and reveal the spatial association between metabolites and genes and proteins, helping to understand the metabolic reprogramming mechanism of tumor cells in different regions and improve the accuracy and specificity of diagnosis. The three-dimensional metabolite distribution result refers to the visualized spatial distribution information of metabolites in tumor cells presented by the three-dimensional metabolite distribution model, such as metabolite concentration information, the spatial association between metabolite distribution and the expression of key regulatory genes and proteins, etc.

[0074] Optionally, the output of the three-dimensional metabolite distribution result of the tumor cells based on the three-dimensional metabolite distribution model can be implemented using an interactive tool, such as the Plotly tool.

[0075] Compared with the problems described in the background art, in the embodiments of the present invention, by performing alignment and fusion processing on the imaging data, the metabolomics data, the genomics data, and the proteomics data, multi-modal data is obtained, which can ensure that data of different modalities are in the same spatial coordinate system to improve the comprehensiveness and accuracy of metabolite distribution detection; further, in the embodiments of the present invention, by identifying the entity elements of the tumor cells and their corresponding entity relationships from the multi-modal data, a metabolite knowledge graph of the tumor cells is constructed, which can infer the causal relationships between genes, proteins, metabolites, and tissue structures, helping to understand the network structure of metabolites in tumor cells, thereby providing a basis for determining the relative positions and connection methods of metabolites in three-dimensional reconstruction and ensuring the rationality and accuracy of the three-dimensional reconstruction results; in the embodiments of the present invention, by analyzing the spatial aggregation patterns of metabolites in tumor cells according to the spatial coordinates of metabolites and the microenvironment characteristic data of tumor cells, the interactions between different metabolites and the regulatory mechanisms at specific spatial positions can be discovered; further, in the embodiments of the present invention, by extracting the key features of metabolites in the tumor cells according to the spatial aggregation patterns and the biological metabolic pathways, potential metabolic hotspots or abnormal metabolic pathways can be discovered, so as to more comprehensively predict the dynamic changes of metabolite distribution to create a discriminant network for metabolite distribution in tumor cells, monitor the growth, metastasis, and response to treatment of tumors in real time, and improve the accuracy of diagnosis; in the embodiments of the present invention, by identifying the abnormal distribution patterns of metabolites in the tumor cells, it can ensure that the metabolite distribution in the key areas can be restored with high precision, helping doctors to diagnose tumors more accurately; further, in the embodiments of the present invention, by constructing a three-dimensional model of metabolite distribution in the tumor cells according to the abnormal distribution patterns and the multi-modal data, the spatial distribution of metabolites in tumor cells and the interaction relationships between different metabolites can be presented intuitively and comprehensively, which helps doctors to detect tumors in the early stage of the disease and formulate more appropriate treatment plans; finally, in the embodiments of the present invention, by outputting the three-dimensional metabolite distribution results of the tumor cells based on the three-dimensional model of metabolite distribution, the spatial distribution of metabolites in tumor tissues can be intuitively displayed, the accuracy of three-dimensional reconstruction of metabolite distribution in tumor cells can be improved, and the spatial associations between metabolites and genes and proteins can be revealed, helping to understand the metabolic reprogramming mechanisms of tumor cells in different regions and improve the accuracy and specificity of diagnosis. Therefore, the method and system for three-dimensional reconstruction of metabolite distribution in tumor cells based on artificial intelligence provided by the embodiments of the present invention can improve the accuracy and comprehensiveness of the three-dimensional metabolite distribution results of tumor cells.

[0076] Embodiment 2: As Figure 2 shown, it is a functional module diagram of the system for three-dimensional reconstruction of metabolite distribution in tumor cells based on artificial intelligence of the present invention.

[0077] The three-dimensional reconstruction system 200 for tumor cell metabolite distribution based on artificial intelligence according to the present invention can be installed in an electronic device. According to the functions achieved, the three-dimensional reconstruction system for tumor cell metabolite distribution based on artificial intelligence may include a data collection module 201, a knowledge graph construction module 202, an aggregation pattern analysis module 203, a distribution prediction module 204, and a three-dimensional reconstruction module 205. The modules in the present invention may also be referred to as units, which refer to a series of computer program segments that can be executed by a processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.

[0078] In the embodiments of the present invention, the functions of each module / unit are as follows: The data collection module 201 is used to obtain tumor cells to be detected, collect imaging data, metabolomics data, genomics data, and proteomics data of the tumor cells, and perform alignment and fusion processing on the imaging data, the metabolomics data, the genomics data, and the proteomics data to obtain multimodal data; The knowledge graph construction module 202 is used to identify entity elements of the tumor cells and their corresponding entity relationships from the multimodal data, construct a metabolite knowledge graph of the tumor cells according to the entity elements and the entity relationships, and identify metabolites of the tumor cells by using the multimodal data and the metabolite knowledge graph; The aggregation pattern analysis module 203 is used to determine the spatial coordinates of the metabolites based on the multimodal data, extract microenvironment feature data of the tumor cells, analyze the spatial aggregation pattern of the metabolites in the tumor cells according to the spatial coordinates and the microenvironment feature data, and identify the biological metabolic pathways of the metabolites based on the metabolite knowledge graph; The distribution prediction module 204 is used to extract key features of the metabolites in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathways, where the key features include spatial distribution features, pathway correlation features, and concentration gradient features, and create a metabolite distribution discrimination network of the tumor cells based on the key features; The three-dimensional reconstruction module 205 is used to identify abnormal distribution patterns of the metabolites in the tumor cells based on the metabolite distribution discrimination network, construct a three-dimensional model of the metabolite distribution of the tumor cells according to the abnormal distribution patterns and the multimodal data, and output a three-dimensional distribution result of the metabolites of the tumor cells based on the three-dimensional model of the metabolite distribution.

[0079] Specifically, when the modules in the three-dimensional reconstruction system 200 for tumor cell metabolite distribution based on artificial intelligence in the embodiments of the present invention are used, they adopt the same technical means as those in the Figure 1 three-dimensional reconstruction method for tumor cell metabolite distribution based on artificial intelligence described above, and can produce the same technical effects, which will not be elaborated here.

[0080] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence, characterized in that: The method comprises: Acquire tumor cells to be detected, collect imaging data, metabolomics data, genomics data and proteomics data of the tumor cells, align and fuse the imaging data, the metabolomics data, the genomics data and the proteomics data to obtain multimodal data; Identifying entity elements of the tumor cells and their corresponding entity relationships from the multimodal data, constructing a metabolite knowledge graph of the tumor cells based on the entity elements and the entity relationships, and identifying metabolites of the tumor cells using the multimodal data and the metabolite knowledge graph; Based on the multimodal data, determining the spatial coordinates of the metabolites, and extracting the microenvironmental characteristic data of the tumor cells, analyzing the spatial aggregation pattern of the metabolites in the tumor cells according to the spatial coordinates and the microenvironmental characteristic data, and identifying the biological metabolic pathway of the metabolites based on the metabolite knowledge graph; Extracting key features of the metabolites in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathway, wherein the key features include spatial distribution features, pathway correlation features, and concentration gradient features, and creating a metabolite distribution discrimination network for the tumor cells based on the key features; Based on the metabolite distribution discrimination network, the abnormal distribution pattern of the metabolites in the tumor cells is identified, and according to the abnormal distribution pattern and the multimodal data, a three-dimensional model of the metabolite distribution of the tumor cells is constructed; based on the three-dimensional metabolite distribution model, the three-dimensional distribution results of the metabolites of the tumor cells are output.

2. The method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence according to claim 1, characterized in that: The imaging data, the metabolomics data, the genomics data and the proteomics data are aligned and fused to obtain multimodal data, including: Performing data standardization processing on the imaging data, the metabolomics data, the genomics data, and the proteomics data to obtain first modality standardized data, second modality standardized data, third modality standardized data, and fourth modality standardized data; Based on the first modality standardized data, extracting voxel coordinates in the imaging data, and performing spatial transformation processing on the voxel coordinates to obtain first spatial transformation coordinates; extracting second spatial coordinates corresponding to the metabolomics data according to the second modality standardized data; Based on the third modality standardized data and the fourth modality standardized data, generating third spatial coordinates and fourth spatial coordinates corresponding to the genomics data and the proteomics data; Performing coordinate alignment processing of the first spatial transformation coordinates with the second spatial coordinates, the third spatial coordinates, and the fourth spatial coordinates to obtain aligned spatial coordinates; According to the alignment space coordinates, data fusion processing is performed on the first modality standardized data, the second modality standardized data, the third modality standardized data and the fourth modality standardized data to obtain multimodal data.

3. The method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence according to claim 1, characterized in that: The identifying the entity elements of the tumor cells and their corresponding entity relationships from the multimodal data includes: Extracting molecular characteristics and pathological features of the tumor cells based on the multimodal data; Determining the cancer type of the tumor cell based on the molecular characteristics and the pathological features; Based on the type of cancer, identifying the metabolic characteristics of the tumor cells; According to the metabolic characteristics, metabolites of the tumor cells are separated, and the metabolic pathways of the metabolites are analyzed; Extracting regulatory factors of the metabolic pathway based on the metabolic characteristics and the metabolites; In combination with the metabolic characteristics, the metabolites, the metabolic pathways and the regulatory factors, entity elements of the tumor cells and their corresponding entity relationships are identified from the multimodal data.

4. The method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence according to claim 1, characterized in that: The step of constructing the metabolite knowledge graph of the tumor cell according to the entity elements and the entity relationships includes: Extracting cancer cell types, metabolites, metabolic pathways, and metabolic regulatory factors corresponding to the tumor cells based on the entity elements; Based on the entity relationship, define the top-level entity and sub-entities of the tumor cell; Identifying the relationship attributes between the cancer cell type, the metabolites, the metabolic pathways, and the metabolic regulators; According to the relationship attributes, defining semantic rules between the top-level entity and the sub-entities; Based on the top-level entity, the sub-entities and the semantic rules, generating a metabolite knowledge graph of the tumor cell; The knowledge storage processing of the metabolite knowledge graph is performed to obtain a metabolite knowledge graph.

5. The method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence according to claim 1, characterized in that: Analyzing the spatial aggregation pattern of the metabolites in the tumor cells according to the spatial coordinates and the microenvironment characteristic data comprises: Performing spatial alignment processing on the spatial coordinates and the microenvironment feature data to obtain a spatial alignment result; Based on the spatial alignment result, identifying the hotspot region of the metabolite in the tumor cell; The metabolite concentration in the hot spot area is measured, and the gradient variation law of the metabolite concentration in the hot spot area is analyzed using the following formula: ； in, Indicates the gradient change law of metabolite concentration in the hot spot area, represents the rate of change of metabolite concentration in the x direction of the hot spot area, represents the rate of change of metabolite concentration in the y direction of the hot spot area, represents the rate of change of metabolite concentration in the z direction of the hot spot area, represents the partial derivative of the metabolite concentration; The spatial aggregation pattern of the metabolites in the tumor cells is analyzed based on the hot spot areas and the gradient variation rules.

6. The method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence according to claim 1, characterized in that: Extracting key features of the metabolites in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathway includes: extracting regional aggregation characteristics of the metabolites in the tumor cells according to the spatial aggregation pattern; Based on the regional aggregation characteristics, analyzing the concentration changes of the metabolites in the tumor cells; According to the concentration change, identifying the concentration gradient characteristics of the metabolite; Determining the specific position of the metabolite in the biological metabolic pathway, and based on the specific position, analyzing the function of the metabolite in the biological metabolic pathway; According to the function, identifying the pathway association characteristics of the metabolite; The key features of the metabolites in the tumor cells are extracted by combining the regional aggregation features, the concentration gradient features and the pathway association features.

7. The method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence according to claim 1, characterized in that: The step of creating a metabolite distribution discrimination network for the tumor cells based on the key features comprises: Based on the key features, analyzing the metabolite distribution pattern of the tumor cells; Identifying the disease stage of the tumor cells, and extracting key metabolites of the tumor cells according to the disease stage; Calculating the correlation coefficient between the key metabolites and the disease stage; Based on the correlation coefficient, constructing a correlation network between the key metabolites and the disease stages; According to the association network, identifying the classification performance of the key metabolites at the stage of the disease course; Based on the metabolite distribution pattern and the classification performance, a metabolite distribution discrimination network of the tumor cells is created.

8. The method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence according to claim 1, characterized in that: The step of identifying the abnormal distribution pattern of the metabolite in the tumor cell based on the metabolite distribution discrimination network includes: Based on the metabolite distribution discrimination network, identifying the abnormal distribution area of ​​the metabolite in the tumor cell; analyzing the metabolic pathway of the metabolite in the abnormal distribution area and identifying the activity level of the metabolic pathway; according to the degree of activity, precipitating a tumor microenvironment of the metabolite; Based on the metabolic pathway and the tumor microenvironment, identifying the spatial heterogeneity characteristics of the metabolites; analyzing the disease progression of the tumor cells according to the spatial heterogeneity characteristics; Based on the metabolic pathway, the tumor microenvironment, the spatial heterogeneity characteristics and the disease progression, the abnormal distribution pattern of the metabolite in the tumor cells is identified.

9. The method for three-dimensional reconstruction of tumor cell metabolite distribution based on artificial intelligence according to claim 1, characterized in that: The step of constructing a three-dimensional metabolite distribution model of the tumor cells according to the abnormal distribution pattern and the multimodal data includes: extracting abnormal metabolites of the tumor cells according to the abnormal distribution pattern; Identifying key regulatory nodes of the abnormal metabolites and collecting expression data of the key regulatory nodes; Determining the three-dimensional coordinates of the tumor cells based on the multimodal data; Collecting concentration data of the abnormal metabolite, and mapping the concentration data to the three-dimensional coordinates to generate a spatial distribution map of the abnormal metabolite; Based on the expression data and the three-dimensional coordinates, the spatial co-localization relationship between the abnormal metabolites and the key regulatory nodes is identified using the following formula: ； Among them, r represents the spatial co-localization relationship between abnormal metabolites and key regulatory nodes, represents the metabolite concentration value of the abnormal metabolite at the e-th observation point in the three-dimensional coordinates, represents the gene-protein expression level of the key regulatory node at the e-th observation point in three-dimensional coordinates, represents the average concentration of abnormal metabolites, represents the average value of the gene-protein expression level of the key regulatory node, x represents the metabolite concentration corresponding to the abnormal metabolite in the expression data, y represents the gene-protein expression level corresponding to the key regulatory node in the expression data, m represents the number of observation points in the three-dimensional coordinates, and e represents the index of the observation point; Based on the spatial distribution map and the spatial co-localization relationship, a three-dimensional model of metabolite distribution of the tumor cells is constructed.

10. A three-dimensional reconstruction system for tumor cell metabolite distribution based on artificial intelligence, characterized in that: The system comprises: A data collection module, used to obtain tumor cells to be detected, collect imaging data, metabolomics data, genomics data and proteomics data of the tumor cells, align and fuse the imaging data, the metabolomics data, the genomics data and the proteomics data to obtain multimodal data; A knowledge graph construction module, used to identify entity elements of the tumor cells and their corresponding entity relationships from the multimodal data, construct a metabolite knowledge graph of the tumor cells based on the entity elements and the entity relationships, and identify metabolites of the tumor cells using the multimodal data and the metabolite knowledge graph; an aggregation pattern analysis module, for determining the spatial coordinates of the metabolites based on the multimodal data, and extracting the microenvironment characteristic data of the tumor cells, analyzing the spatial aggregation patterns of the metabolites in the tumor cells according to the spatial coordinates and the microenvironment characteristic data, and identifying the biological metabolic pathways of the metabolites based on the metabolite knowledge graph; A distribution prediction module, used to extract key features of the metabolites in the tumor cells according to the spatial aggregation pattern and the biological metabolic pathway, wherein the key features include spatial distribution features, pathway correlation features, and concentration gradient features, and to create a metabolite distribution discrimination network for the tumor cells based on the key features; A three-dimensional reconstruction module is used to identify the abnormal distribution pattern of the metabolites in the tumor cells based on the metabolite distribution discrimination network, construct a three-dimensional model of the metabolite distribution of the tumor cells according to the abnormal distribution pattern and the multimodal data, and output the three-dimensional distribution results of the metabolites of the tumor cells based on the three-dimensional metabolite distribution model.

Images