A radiology image data analysis system

Abstract

The present invention relates to the field of medical data processing technology, specifically to a radiology image data analysis system, the system comprising: a self-matching texture recognition module to collect image data, analyze the texture type in the image, adjust the size and shape of the search window, calculate the structural fit of the area in the window according to the texture features, and aggregate the data of similar areas. In the present invention, by adjusting the size and shape of the search window, the structural fit of the area is accurately calculated according to the texture features, and a more refined data analysis is achieved. By aggregating similar area data and updating the fit weight, the target of data analysis is optimized, and the iterative soft threshold algorithm is applied to effectively process texture features and edges, reduce the complexity of data, and improve the processing speed. In addition, the topological structure analysis strengthens the intuitive understanding of complex pathological structures, and the global pattern features are analyzed through the graph network, which enhances the accuracy of diagnosis and provides data support for disease prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a radiology image data analysis system. Background Art

[0002] The field of medical data processing technology covers a range of methods and systems for analyzing, processing and interpreting medical data, including clinical information, medical images and laboratory test results. In this technical field, advanced computer algorithms and software tools are used to process data to support clinical decision-making, pathological diagnosis, patient monitoring and medical research. This field pays special attention to improving the efficiency and accuracy of data processing, improving medical quality and treatment effects. Technological development includes but is not limited to the application of automated data entry, image segmentation, feature extraction, pattern recognition and machine learning technologies, which help to extract useful information from huge amounts of medical data, thereby providing more accurate diagnostic support and efficacy evaluation.

[0003] Among them, the radiology image data analysis system refers to a system specifically used to process and analyze image data generated by the radiology department. The purpose of this type of system is mainly to assist doctors in diagnosing diseases, monitoring disease progression, and evaluating treatment effects by analyzing medical images such as X-rays, CT, and MRI. Radiology image data analysis improves the accuracy and efficiency of diagnosis through advanced image processing technologies such as image enhancement, three-dimensional reconstruction, and automatic identification of abnormal structures. Such systems are extremely important in medical practice, especially in diagnosing complex diseases and making precise treatment plans.

[0004] Existing technologies often face challenges in efficiency and accuracy when processing large-scale or complex data, especially in the recognition of subtle texture differences and complex structures, which affects the accurate diagnosis of complex diseases. Existing systems usually use fixed algorithm parameters and lack the flexibility to process non-standardized data, which may lead to delayed or incorrect diagnostic information in clinical applications, thus affecting treatment plans and patient prognosis. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a radiology image data analysis system.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: A radiology image data analysis system comprises:

[0007] The self-matching texture recognition module collects image data, analyzes the texture type in the image, adjusts the size and shape of the search window, calculates the structural fit of the area within the window based on the texture features, gathers data from similar areas, updates the fit weights, and establishes a regional fit matrix;

[0008] The dynamic threshold adjustment module receives the regional fit matrix, performs sparseness on the matrix data, adjusts the iteration threshold according to the data sparsity, modifies the regularization parameter, applies the iterative soft threshold algorithm to optimize the matrix structure, processes the texture features and edges of the matrix, and generates an optimized sparse matrix;

[0009] The topological structure analysis module processes the optimized sparse matrix, identifies the ring and hollow structures in the image as nodes, constructs a spatial and functional relationship graph between the nodes, adjusts the nodes and edges of the graph structure according to the image features, and generates a structural relationship network;

[0010] The pattern feature analysis module uses the node and edge information in the structural relationship network to analyze the data between nodes, extract global pattern features, learn nodes through the graph network, analyze image data, and generate feature analysis results.

[0011] The regional fitting matrix specifically includes structural fitting, fitting weight, and similar regional data; the optimized sparse matrix specifically refers to texture features and edges; the structural relationship network includes nodes, spatial relationship graphs, and functional relationship graphs; and the feature analysis results specifically include global pattern features and graph network learning results.

[0012] As a further solution of the present invention, the step of obtaining the structural fit is specifically:

[0013] Collect image data, analyze the texture type in the image, extract the texture features of each area, and construct the corresponding preliminary texture feature set;

[0014] According to the preliminary texture feature set, the size and shape of the search window are adjusted to optimize the window configuration to match the requirements of the differentiated texture structure, and an optimized search window configuration is obtained;

[0015] The optimized search window configuration is applied to scan the image region by region, and the local mean, standard deviation and grayscale value of the pixels in the window are combined to use the formula:

[0016] ;

[0017] Calculate the regional structure fit , generating the structural fit, where Represents the window Line The gray value of the pixel. Represents the mean grayscale value of all pixels in the window. Represents the standard deviation of the grayscale values ​​of all pixels in the window. and Represent the number of rows and columns of the pixel matrix in the window respectively.

[0018] As a further solution of the present invention, the step of obtaining the regional fitness matrix is ​​specifically as follows:

[0019] Based on the calculation result of the structural fit, regional data with similar texture features are aggregated, and regional classification standards are constructed by comparing the fits of multiple regions to obtain a classified regional data set;

[0020] Using the classified regional data set, the relationship between multiple regions is calculated, and the regional fit weight is updated using the formula:

[0021] ;

[0022] Re-adjust the weight of each region according to the matrix value to generate the regional fitness matrix;

[0023] in, Indicates area and Region The degree of fit between them reflects the degree of fit between the two regions in terms of structure and texture features. Respectively expressed in Area in window and Region The structural fit of is calculated based on the texture data in multiple regions. It is The confidence of each window is used to adjust the influence of the window data in the overall fit calculation. A high confidence means that the window data has a greater weight in the overall calculation. Represents the number of windows involved in the calculation and is used for normalization.

[0024] As a further solution of the present invention, the step of obtaining the optimized sparse matrix is ​​specifically as follows:

[0025] Receiving the regional fit matrix, analyzing the data distribution in the matrix, determining the need for data sparsification, and obtaining a preliminary plan for data sparsification;

[0026] According to the preliminary plan of data sparsification, the iteration threshold is adjusted and the regularization parameter is modified to match the requirements of the sparsification process. The formula is adopted:

[0027] ;

[0028] Calculate and generate updated iteration threshold;

[0029] in, represents the updated iteration threshold, Represents the original iteration threshold, which is the parameter used to control the iteration process before optimization. Represents the sparse parameters in the matrix, which are used to adjust the iteration threshold to match the data sparse requirements. is a regularization parameter used to adjust the influence of the sparse parameter on the iterative threshold adjustment. is the number of sparse parameters, indicating the total number of parameters involved in the calculation;

[0030] Using the updated iteration threshold, an iterative soft threshold algorithm is applied to process the matrix, optimize the matrix structure, and generate an optimized sparse matrix.

[0031] As a further solution of the present invention, the steps of obtaining the spatial and functional relationship graph between the nodes are specifically as follows:

[0032] Extract image features from the optimized sparse matrix, identify ring and hollow structures through image processing technology, determine the structures as nodes of the graph, and generate node identification data;

[0033] Using the node identification data, construct the spatial relationship between nodes, calculate the physical distance and structural fit between nodes, ensure the accurate spatial positioning of each node, and generate spatial relationship data;

[0034] Integrate the spatial relationship data, iteratively analyze the functional attributes of the nodes, and use the formula:

[0035] ;

[0036] Optimize the connection strategy between nodes and generate spatial and functional relationship diagrams between nodes;

[0037] in, Representative Node and nodes The strength of the relationship between and is the weight coefficient, which is used to adjust the influence of distance and functional fit in node relationships. Representative Node and nodes The physical distance between Representative Node and nodes The fitness of the functional characteristics.

[0038] As a further solution of the present invention, the step of acquiring the structural relationship network is specifically:

[0039] Analyze the spatial and functional relationship diagram between the nodes, determine the key nodes in the diagram and their connection patterns, build a preliminary structural framework of the diagram based on the structural positions and functional attributes of the nodes, and obtain structural framework data;

[0040] Using the structural framework data, perform deep optimization of nodes and edges, adjust the connections in the graph according to actual functional requirements and interactions between nodes, ensure that the structure of the graph meets the predetermined functional requirements, and generate adjusted graph structure data;

[0041] Apply a graph structure optimization algorithm to optimize the adjusted graph structure data, using the formula:

[0042] ;

[0043] Computationally generate structural relationship networks;

[0044] in, An adjustment factor representing the criticality of a node, used to enhance the criticality score of a node function The impact of represents the adjustment coefficient of spatial interaction, which is used to adjust the effect of the distance between nodes on the connection strength. Representation Node The functional criticality score highlights the centrality and influence of the node. Representation Node and The strength of the edge between Representation Node and The distance between.

[0045] As a further solution of the present invention, the step of obtaining the characteristic analysis result is specifically:

[0046] Combined with the node and edge information in the structural relationship network, the data flow between nodes is analyzed, and according to the interaction and data exchange characteristics between nodes, the key data indicators of each node are identified and recorded to obtain an overview of node characteristics;

[0047] Using graph network learning technology, deep learning processing is performed on the node feature overview to extract and identify global pattern features, and global pattern feature data is generated by adjusting network parameters to match differentiated node characteristics;

[0048] The global pattern feature data is applied to image data analysis using the formula:

[0049] ;

[0050] Calculate and generate characteristic analysis results;

[0051] in, Represents the feature analysis result, which is used to measure the output effect after integrating image data and network model features. Is a node The weight adjustment coefficient is used to balance the influence of each node in the overall model. is a global weighting factor that adjusts the emphasis on the overall model analysis. is a data-weighted nonlinear response index used to adjust the intensity of feature data processing. Represents nodes learned from the structural relationship network The global pattern characteristics of For the corresponding node Image feature data is used to extract features from visual information and enhance the model's ability to parse image data.

[0052] Compared with the prior art, the advantages and positive effects of the present invention are:

[0053] In the present invention, by adjusting the search window size and shape, the structural fit of the region is accurately calculated based on the texture features, thereby achieving more refined data analysis. By aggregating similar regional data and updating the fit weight, the target of data analysis is optimized. The iterative soft threshold algorithm is applied to effectively process texture features and edges, reduce the complexity of data, and improve the processing speed. In addition, topological structure analysis strengthens the intuitive understanding of complex pathological structures, analyzes global pattern features through graph networks, enhances the accuracy of diagnosis, and provides data support for disease prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a system flow chart of the present invention;

[0055] Figure 2 The figure is a flow chart of the steps for obtaining the structural fit of the present invention;

[0056] Figure 3 The flowchart of the steps of obtaining the regional fit matrix of the present invention;

[0057] Figure 4 A flow chart of the steps for obtaining the optimized sparse matrix of the present invention;

[0058] Figure 5 A flow chart of the steps for obtaining the spatial and functional relationship diagram between nodes of the present invention;

[0059] Figure 6 A flowchart of the steps for obtaining the structural relationship network of the present invention;

[0060] Figure 7 The figure is a flow chart of the steps for obtaining the characteristic analysis results of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0062] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0063] Example 1: Please refer to Figure 1 , a radiology image data analysis system comprises:

[0064] The self-matching texture recognition module collects image data, analyzes the texture type in the image, adjusts the size and shape of the search window, calculates the structural fit of the area within the window based on the texture features, gathers data from similar areas, updates the fit weights, and establishes a regional fit matrix;

[0065] The dynamic threshold adjustment module receives the regional fit matrix, sparses the matrix data, adjusts the iteration threshold according to the data sparsity, modifies the regularization parameter, applies the iterative soft threshold algorithm to optimize the matrix structure, processes the texture features and edges of the matrix, and generates an optimized sparse matrix;

[0066] The topological structure analysis module processes the optimized sparse matrix, identifies the ring and hollow structures in the image as nodes, constructs a spatial and functional relationship graph between nodes, adjusts the nodes and edges of the graph structure according to the image features, and generates a structural relationship network;

[0067] The pattern feature analysis module uses the node and edge information in the structural relationship network to analyze the data between nodes, extract global pattern features, learn nodes through the graph network, analyze image data, and generate feature analysis results.

[0068] The regional fitting matrix specifically includes structural fitting, fitting weight, and similar regional data. The optimized sparse matrix specifically refers to texture features and edges. The structural relationship network includes nodes, spatial relationship graphs, and functional relationship graphs. The feature analysis results specifically refer to global pattern features and graph network learning results.

[0069] See also Figure 2 , the steps to obtain the structural fit are as follows:

[0070] Collect image data, analyze the texture type in the image, extract the texture features of each area, and construct the corresponding preliminary texture feature set;

[0071] After collecting image data, the texture characteristics of differentiated areas in the image are analyzed. By classifying the texture of each area, its feature vector is extracted and a feature database is constructed to provide a basis for subsequent steps. Based on the collected feature information and combined with image processing technology, the texture density and distribution pattern of each area are calculated. These data support the adjustment of the size and shape of the search window to ensure that the search window in subsequent steps can match the differentiated texture structure, thereby improving the accuracy and efficiency of the overall image analysis. This process involves batch data comparison and pattern recognition calculations, and the results obtained will directly affect the optimization decision of the search window configuration.

[0072] According to the preliminary texture feature set, the size and shape of the search window are adjusted to optimize the window configuration to match the requirements of the differentiated texture structure and obtain the optimized search window configuration;

[0073] After obtaining the preliminary texture feature set, the size and shape of the search window are adjusted. Based on the data obtained above, including the statistical analysis results of texture types and feature distribution, geometric and mathematical models are used to optimize the window shape to ensure that the key texture areas can be covered to the greatest extent, improve the scanning efficiency and coverage accuracy, and verify the effectiveness of the window configuration through actual image data. The optimization operation determines the shape and size of the window by calculating the frequency of occurrence and distribution pattern of each texture type.

[0074] Apply the optimized search window configuration to scan the image region by region, combining the local mean, standard deviation and grayscale value of the pixels in the window, using the formula:

[0075] ;

[0076] Calculate the regional structure fit , generating the structural fit, where Represents the window Line The gray value of the pixel. Represents the mean grayscale value of all pixels in the window. Represents the standard deviation of the grayscale values ​​of all pixels in the window. and Respectively represent the number of rows and columns of the pixel matrix in the window;

[0077] The optimized search window configuration is applied to scan the image region by region. First, the pixel data of the input image is acquired, the grayscale information of the image is extracted and stored in a matrix form, and the grayscale value of each pixel is set to , then according to the preset search window size Divide the image area and obtain the area containing The local image area data of pixels is used to calculate the mean value of all pixels in the area. , the mean value is calculated using the formula:

[0078] ;

[0079] Then calculate the standard deviation of the pixels in the window area , the standard deviation is calculated using the formula:

[0080] ;

[0081] After completing the regional data calculation, it is necessary to further calculate the degree of deviation between each pixel in the window and the mean. The absolute value of the difference between the grayscale value of all pixels and the mean is calculated and summed. The calculation formula is:

[0082] ;

[0083] This value can be used for subsequent structural fit calculations. Next, the structural fit of the window area is calculated. , the calculation process is to normalize the deviation and standard deviation and local total variation obtained by the above formula, assuming that the total variation is:

[0084] ;

[0085] The structural fit of the final calculation window for:

[0086] ;

[0087] For example, suppose the window size is , the grayscale values ​​of the regional pixels are as follows:

[0088] ;

[0089] Then calculate the mean: ;

[0090] Calculate the standard deviation: ;

[0091] Calculate the total variation: ;

[0092] Calculate the structural fit: ;

[0093] Finally, the structural fitting degree of the area within the window is .

[0094] See also Figure 3 , the steps to obtain the regional fitness matrix are as follows:

[0095] Based on the calculation results of structural fit, regional data with similar texture features are aggregated, and by comparing the fit of multiple regions, a regional classification standard is constructed to obtain a classified regional data set;

[0096] Based on image data set collection and texture feature analysis, the image data is processed in a fine-grained manner, the images are classified and labeled according to their texture types, and a machine learning algorithm is used to analyze each pixel in the image to determine which type of texture it belongs to, and then the regions with similar features are clustered by category. This process involves data processing, including image preprocessing, feature extraction and clustering analysis. All steps are aimed at improving the accuracy and reliability of classification, ensuring that each classified regional data set can accurately reflect the texture information in the original image and obtain a preliminary texture feature set.

[0097] Use the classified regional dataset to calculate the relationship between multiple regions and update the regional fit weights using the formula:

[0098] ;

[0099] Re-adjust the weight of each region according to the matrix value to generate the regional fitness matrix;

[0100] in, Indicates area and Region The degree of fit between them reflects the degree of fit between the two regions in terms of structure and texture features. Respectively expressed in Area in window and Region The structural fit of is calculated based on the texture data in multiple regions. It is The confidence of each window is used to adjust the influence of the window data in the overall fit calculation. A high confidence means that the window data has a greater weight in the overall calculation. Represents the number of windows involved in the calculation and is used for normalization.

[0101] formula:

[0102] ;

[0103] The benefit of the formula is that it also introduces confidence as a weight with reference to the fit between the two regions in terms of texture features, thus enhancing the credibility and accuracy of the results.

[0104] Detailed explanation of the formula and the process of formula calculation and derivation:

[0105] According to an example, three areas are set, where , , , , , , and the confidence , , ,So The calculation process is:

[0106] ;

[0107] The results show that the texture fitting degree of region 1 and region 2 is 0.5395, which shows that these two regions have a high fitting degree in texture, which is very useful in adjusting regional weights and image processing.

[0108] See also Figure 4 , the steps to obtain the optimized sparse matrix are as follows:

[0109] Receive the regional fit matrix, analyze the data distribution in the matrix, determine the need for data sparsification, and obtain a preliminary plan for data sparsification;

[0110] After receiving the regional fit matrix, we first organize the data, screen representative data points, extract the elements whose values ​​in the matrix are greater than the average value, and use them as the key analysis objects. We use statistical analysis methods to analyze the data distribution and determine the need for data sparsification. By calculating the standard deviation and variance of the data, we evaluate the trend and degree of dispersion of the data. Based on this, we build a preliminary data sparsification scheme, which aims to reduce the proportion of zero elements in the matrix, optimize storage and computing efficiency, and determine the preliminary plan for data sparsification, providing a basis for the next iterative threshold adjustment.

[0111] According to the preliminary plan of data sparsification, adjust the iteration threshold and modify the regularization parameter to match the requirements of the sparsification process. Use the formula:

[0112] ;

[0113] Calculate and generate updated iteration threshold;

[0114] in, represents the updated iteration threshold, Represents the original iteration threshold, which is the parameter used to control the iteration process before optimization. Represents the sparse parameters in the matrix, which are used to adjust the iteration threshold to match the data sparse requirements. is a regularization parameter used to adjust the influence of the sparse parameter on the iterative threshold adjustment. is the number of sparse parameters, indicating the total number of parameters involved in the calculation;

[0115] formula:

[0116] ;

[0117] The benefit of the formula is that, by dynamically adjusting the iteration threshold, it can flexibly respond to the actual situation of data sparsity, increase the flexibility and accuracy of adjusting the iteration threshold, and match the processing needs of differentiated data sets.

[0118] Detailed explanation of the formula and the process of formula calculation and derivation:

[0119] Set the original iteration threshold , refer to the data sparsification needs, adjust the iteration threshold to match the data sparsity, and set , each sparse parameter They are , the regularization parameter , substitute into the formula and calculate:

[0120] ;

[0121] The results show that the updated iteration threshold is 0.6, which is significantly higher than the original threshold, indicating that a larger scale of sparse processing is needed to optimize the matrix structure.

[0122] Using the updated iteration threshold, an iterative soft threshold algorithm is applied to process the matrix, optimize the matrix structure, and generate an optimized sparse matrix.

[0123] When using the updated iteration threshold for matrix processing, the iterative soft threshold algorithm is first used to judge and adjust each element of the matrix according to the updated iteration threshold. If the absolute value of the element is less than the iteration threshold, the element is set to 0, otherwise the sign times of the threshold are subtracted. This method can effectively compress the number of non-zero elements in the matrix, thereby achieving the purpose of data sparsification. In this way, the matrix structure is optimized, the processing efficiency is improved, and an optimized sparse matrix is ​​generated. The matrix is ​​more in line with the needs of subsequent processing and storage.

[0124] See also Figure 5 , the steps for obtaining the spatial and functional relationship graph between nodes are as follows:

[0125] Extract image features from the optimized sparse matrix, identify ring and hollow structures through image processing technology, identify the structures as nodes of the graph, and generate node identification data;

[0126] Image features are extracted from the optimized sparse matrix. The feature extraction uses image processing algorithms to identify ring and hollow structures. The structures are identified as highly connected or low-connected areas in mathematical processing. Applications include Sobel or Canny edge detection algorithms, which iteratively extract tiny structures in the image and use these ring and hollow structures as nodes of the graph. The identification process involves not only pixel-level operations, but also image segmentation techniques such as region growing methods to distinguish the boundaries of differentiated nodes, laying the foundation for the next step of building spatial relationships between nodes. Through the image processing steps, information about each independent structure in the image is obtained, which will be used to construct a spatial relationship graph.

[0127] Using node identification data, we construct the spatial relationship between nodes, calculate the physical distance and structural fit between nodes, ensure the accurate spatial positioning of each node, and generate spatial relationship data;

[0128] Using the preliminary node identification data, the spatial relationship diagram between nodes is constructed through geometric distance calculation and structural fit analysis. The spatial location of each node is determined by calculating its center coordinates in the image, and the physical distance between nodes is calculated using the Euclidean distance formula, that is, ,in and are the coordinates of the two nodes. The structural fit is analyzed by comparing the pixel intensity distribution of the nodes and applying the histogram cross-correlation method to determine whether there is a functional relationship or whether the two nodes should be connected. These calculations not only enhance the accuracy of the relationship between nodes, but also provide a quantitative basis for the construction of functional relationship diagrams.

[0129] Integrate spatial relationship data and iteratively analyze the functional attributes of nodes using the formula:

[0130] ;

[0131] Optimize the connection strategy between nodes and generate spatial and functional relationship diagrams between nodes;

[0132] in, Representative Node and nodes The strength of the relationship between and is the weight coefficient, which is used to adjust the influence of distance and functional fit in node relationships. Representative Node and nodes The physical distance between Representative Node and nodes The fitness of the functional characteristics.

[0133] formula:

[0134] ;

[0135] The benefit of the formula is that it comprehensively considers the physical distance between reference nodes and the functional fit, and adjusts the weight coefficient and , which can flexibly reflect the priority of node connections in differentiated scenarios, and provides an adjustable method for constructing spatial and functional relationship diagrams that better meet actual needs.

[0136] Detailed explanation of the formula and the process of formula calculation and derivation:

[0137] Set the actual physical distance between the two nodes to Unit length, functional fit score , weight coefficient , The calculation process is:

[0138] ;

[0139] The results show that under the given weights and parameters, the overall relationship strength between nodes i and j is 1.6, which indicates that there is a strong sense of connection between the nodes.

[0140] See also Figure 6 , the specific steps for obtaining the structural relationship network are:

[0141] Analyze the spatial and functional relationship diagram between nodes, determine the key nodes and their connection patterns in the diagram, build a preliminary structural framework of the diagram based on the structural position and functional attributes of the nodes, and obtain the structural framework data;

[0142] Based on the preliminary structural framework data of the graph, the connection mode and structural layout of the nodes are analyzed, and the spatial position and functional attributes between the nodes are determined. This process includes evaluating the functional requirements and structural position of each node, and identifying key nodes by calculating the distance and connection type between nodes. These key nodes have centralized connection characteristics or play a key role in data flow in the graph. After refining the functional attributes of the nodes, the connection strength and distance between the nodes are adjusted to ensure that the structure of the graph can effectively support the predetermined functional requirements, and the adjusted graph structure data is generated. This data is the direct basis for graph structure optimization and is the key input for iterative analysis and optimization.

[0143] Using the structural framework data, perform deep optimization of nodes and edges, adjust the connections in the graph according to the actual functional requirements and the interactions between nodes, ensure that the structure of the graph meets the predetermined functional requirements, and generate the adjusted graph structure data;

[0144] The adjusted graph structure data is used to optimize the overall graph structure. Each node is adjusted according to its actual performance and functional requirements, including recalculating the optimal path and network flow between nodes based on the existing graph structure data using graph algorithms, analyzing the centrality and betweenness centrality of the nodes, and thus determining the relative criticality of each node in the graph. Based on the analysis results, the configuration of the edges in the graph is adjusted, including adding or reducing connections and modifying the weights of connections, in order to improve the information flow and functional efficiency of the network, and generate optimized graph structure data, which reflects the optimal connection configuration of each node and the overall optimization status of the graph structure.

[0145] Apply the graph structure optimization algorithm to optimize the adjusted graph structure data, using the formula:

[0146] ;

[0147] Computationally generate structural relationship networks;

[0148] in, An adjustment factor representing the criticality of a node, used to enhance the criticality score of a node function The impact of represents the adjustment coefficient of spatial interaction, which is used to adjust the effect of the distance between nodes on the connection strength. Representation Node The functional criticality score highlights the centrality and influence of the node. Representation Node and The strength of the edge between Representation Node and The distance between.

[0149] formula:

[0150] ;

[0151] The benefit of the formula is that it optimizes the structure of the graph by adjusting the criticality of nodes and the strength of connections, thereby enhancing the functionality and matching of the network.

[0152] Detailed explanation of the formula and the process of formula calculation and derivation:

[0153] There is a network consisting of three nodes, the criticality of which are , , , the connection strength and distance are , , , , the adjustment factor is and , the calculation process is:

[0154] ;

[0155] The results show that by adjusting node criticality and connection strength, the network structure is optimized to enhance its functionality and matching, allowing the network to more effectively handle information flow and functional requirements.

[0156] See also Figure 7 , the specific steps for obtaining the feature analysis results are:

[0157] Combine the node and edge information in the structural relationship network to analyze the data flow between nodes. According to the interaction and data exchange characteristics between nodes, identify and record the key data indicators of each node to obtain an overview of node characteristics.

[0158] Data flow analysis is performed in combination with node and edge information in the structural relationship network to analyze the data interaction and transmission dynamics between multiple nodes. The amount, speed and direction of data flow are recorded and analyzed. The data flow dynamics are used to reveal the activity levels and interdependencies of multiple nodes in the network. This data flow analysis helps to identify key nodes in the network and potential data transmission bottlenecks. Through this process, the optimization points and potential risks of the network can be determined, and the network structure can be designed and adjusted more reasonably to improve the efficiency and stability of the entire network.

[0159] Using graph network learning technology, deep learning is performed on the node feature overview to extract and identify global pattern features. By adjusting network parameters to match differentiated node characteristics, global pattern feature data is generated.

[0160] Use graph network learning technology to learn and analyze node features, and analyze how to identify and extract global pattern features in the network through machine learning models, including using deep learning frameworks for feature learning and adjusting model hyperparameters including learning rate and number of layers to match differentiated data characteristics. This process not only involves algorithm selection and optimization, but also includes data preprocessing, normalization, and division into training and test sets. Through this technical means, pattern features can be accurately identified to provide support for subsequent data analysis.

[0161] The global pattern feature data is applied to image data analysis using the formula:

[0162] ;

[0163] Calculate and generate characteristic analysis results;

[0164] in, Represents the feature analysis result, which is used to measure the output effect after integrating image data and network model features. Is a node The weight adjustment coefficient is used to balance the influence of each node in the overall model. is a global weighting factor that adjusts the emphasis on the overall model analysis. It is a data-weighted nonlinear response index used to adjust the intensity of feature data processing. Represents nodes learned from the structural relationship network The global pattern characteristics of For the corresponding node Image feature data is used to extract features from visual information and enhance the model's ability to parse image data.

[0165] formula:

[0166] ;

[0167] The benefit of the formula is that, through weighted and nonlinear operations, the influence of node characteristics and image data are integrated to achieve higher analysis accuracy and pattern recognition capabilities.

[0168] Detailed explanation of the formula and the process of formula calculation and derivation:

[0169] First, set , Represents the simulated values ​​of node features and image data features, weight parameters , adjustment coefficient , nonlinear response index , calculate each term:

[0170] ;

[0171] Then sum it up, for n=10:

[0172] ;

[0173] The results show that the characteristic analysis results obtained by formula , integrates data from differentiated sources and enhances nonlinear correlations between data features through exponential operations.

[0174] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A radiology image data analysis system, characterized in that: The system comprises: The self-matching texture recognition module collects image data, analyzes the texture type in the image, adjusts the size and shape of the search window, calculates the structural fit of the area within the window based on the texture features, gathers data from similar areas, updates the fit weights, and establishes a regional fit matrix; The dynamic threshold adjustment module receives the regional fit matrix, performs sparseness on the matrix data, adjusts the iteration threshold according to the data sparsity, modifies the regularization parameter, applies the iterative soft threshold algorithm to optimize the matrix structure, processes the texture features and edges of the matrix, and generates an optimized sparse matrix; The topological structure analysis module processes the optimized sparse matrix, identifies the ring and hollow structures in the image as nodes, constructs a spatial and functional relationship graph between the nodes, adjusts the nodes and edges of the graph structure according to the image features, and generates a structural relationship network; The pattern feature analysis module uses the node and edge information in the structural relationship network to analyze the data between nodes, extract global pattern features, learn nodes through the graph network, analyze image data, and generate feature analysis results.

2. The radiology image data analysis system according to claim 1, characterized in that: The regional fitting matrix specifically includes structural fitting, fitting weight, and similar regional data; the optimized sparse matrix specifically refers to texture features and edges; the structural relationship network includes nodes, spatial relationship graphs, and functional relationship graphs; and the feature analysis results specifically include global pattern features and graph network learning results.

3. The radiology image data analysis system according to claim 2, characterized in that: The steps for obtaining the structural fit are specifically as follows: Collect image data, analyze the texture type in the image, extract the texture features of each area, and construct the corresponding preliminary texture feature set; According to the preliminary texture feature set, the size and shape of the search window are adjusted to optimize the window configuration to match the requirements of the differentiated texture structure, and an optimized search window configuration is obtained; The optimized search window configuration is applied to scan the image region by region, and the local mean, standard deviation and grayscale value of the pixels in the window are combined to use the formula: ； Calculate the regional structure fit , generating structural fit, where Represents the window Line The gray value of the pixel. Represents the mean grayscale value of all pixels in the window. Represents the standard deviation of the grayscale values ​​of all pixels in the window. and Represent the number of rows and columns of the pixel matrix in the window respectively.

4. The radiology image data analysis system according to claim 3, characterized in that: The steps for obtaining the regional fitness matrix are specifically as follows: Based on the calculation result of the structural fit, regional data with similar texture features are aggregated, and regional classification standards are constructed by comparing the fits of multiple regions to obtain a classified regional data set; Using the classified regional data set, the relationship between multiple regions is calculated, and the regional fit weight is updated using the formula: ； Re-adjust the weight of each region according to the matrix value to generate the regional fitness matrix; in, Indicates area and Region The degree of fit between them reflects the degree of fit between the two regions in terms of structure and texture features. Respectively expressed in Area in window and Region The structural fit of is calculated based on the texture data in multiple regions. It is The confidence of each window is used to adjust the influence of the window data in the overall fit calculation. A high confidence means that the window data has a greater weight in the overall calculation. Represents the number of windows involved in the calculation and is used for normalization.

5. The radiology image data analysis system according to claim 4, characterized in that: The steps for obtaining the optimized sparse matrix are specifically as follows: Receiving the regional fit matrix, analyzing the data distribution in the matrix, determining the need for data sparsification, and obtaining a preliminary plan for data sparsification; According to the preliminary plan of data sparsification, the iteration threshold is adjusted and the regularization parameter is modified to match the requirements of the sparsification process. The formula is adopted: ； Calculate and generate updated iteration threshold; in, represents the updated iteration threshold, Represents the original iteration threshold, which is the parameter used to control the iteration process before optimization. Represents the sparse parameters in the matrix, which are used to adjust the iteration threshold to match the data sparse requirements. is a regularization parameter used to adjust the influence of the sparse parameter on the iterative threshold adjustment. is the number of sparse parameters, indicating the total number of parameters involved in the calculation; Using the updated iteration threshold, an iterative soft threshold algorithm is applied to process the matrix, optimize the matrix structure, and generate an optimized sparse matrix.

6. The radiology image data analysis system according to claim 5, characterized in that: The steps for obtaining the spatial and functional relationship diagram between the nodes are specifically as follows: Extract image features from the optimized sparse matrix, identify ring and hollow structures through image processing technology, determine the structures as nodes of the graph, and generate node identification data; Using the node identification data, construct the spatial relationship between nodes, calculate the physical distance and structural fit between nodes, ensure the accurate spatial positioning of each node, and generate spatial relationship data; Integrate the spatial relationship data, iteratively analyze the functional attributes of the nodes, and use the formula: ； Optimize the connection strategy between nodes and generate spatial and functional relationship diagrams between nodes; in, Representative Node and nodes The strength of the relationship between and is the weight coefficient, which is used to adjust the influence of distance and functional fit in node relationships. Representative Node and nodes The physical distance between Representative Node and nodes The fitness of the functional characteristics.

7. The radiology image data analysis system according to claim 6, characterized in that: The steps for obtaining the structural relationship network are specifically as follows: Analyze the spatial and functional relationship diagram between the nodes, determine the key nodes in the diagram and their connection patterns, build a preliminary structural framework of the diagram based on the structural positions and functional attributes of the nodes, and obtain structural framework data; Using the structural framework data, perform deep optimization of nodes and edges, adjust the connections in the graph according to actual functional requirements and interactions between nodes, ensure that the structure of the graph meets the predetermined functional requirements, and generate adjusted graph structure data; Apply a graph structure optimization algorithm to optimize the adjusted graph structure data, using the formula: ； Computationally generate structural relationship networks; in, An adjustment factor representing the criticality of a node, used to enhance the criticality score of a node function The impact of Represents the adjustment coefficient of spatial interaction, which is used to adjust the effect of the distance between nodes on the connection strength. Representation Node The functional criticality score highlights the centrality and influence of the node. Representation Node and The strength of the edge between Representation Node and The distance between.

8. The radiology image data analysis system according to claim 7, characterized in that: The steps for obtaining the characteristic analysis results are specifically as follows: Combined with the node and edge information in the structural relationship network, the data flow between nodes is analyzed, and according to the interaction and data exchange characteristics between nodes, the key data indicators of each node are identified and recorded to obtain an overview of node characteristics; Using graph network learning technology, deep learning processing is performed on the node feature overview to extract and identify global pattern features, and global pattern feature data is generated by adjusting network parameters to match differentiated node characteristics; The global pattern feature data is applied to image data analysis using the formula: ； Calculate and generate characteristic analysis results; in, Represents the feature analysis result, which is used to measure the output effect after integrating image data and network model features. Is a node The weight adjustment coefficient is used to balance the influence of each node in the overall model. is a global weighting factor that adjusts the emphasis on the overall model analysis. It is a data-weighted nonlinear response index used to adjust the intensity of feature data processing. Represents nodes learned from the structural relationship network The global pattern characteristics of For the corresponding node Image feature data is used to extract features from visual information and enhance the model's ability to parse image data.

Images