Liver biopsy data analysis method and system

Abstract

The invention relates to the technical field of biopsy image processing, in particular to a liver biopsy data analysis method and system.The liver biopsy data analysis method comprises the following steps that based on liver biopsy slice images, geometric center points of cell nucleuses, blood vessels and fibrosis areas are extracted, and the spatial distance and the connection angle between the geometric center points are calculated; and counting node degree distribution, side length distribution and aggregation factors of the topology network, analyzing a side weight change trend, analyzing spatial distribution and association characteristics among regions, and generating an association parameter set of the organization topology network. According to the method, geometric center points of cell nucleuses, blood vessels and fibrosis areas are extracted, the spatial gradient and time evolution relation of edge weights is combined, the modeling precision of lesion diffusion behaviors is improved, the pixel-level dynamic change rule of a lesion area is captured, and by means of sparsity and aggregation analysis of local density and nuclear spacing, the lesion diffusion behaviors of the lesion diffusion behaviors can be obtained through extraction of the geometric center points of the cell nucleuses, the blood vessels and the fibrosis areas. The spatial distribution characteristics are quantified, the overall scheme is combined from multiple dimensions of geometry, time sequence and texture characteristics, and the dynamic quantification capability of pathological characteristics is enhanced.

Description

Technical Field

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

[0002] The field of liver biopsy image processing technology includes methods for analyzing and processing liver biopsy samples using image processing technology. The core content of this technical field includes obtaining high-quality tissue sample images through digital image processing technology, identifying and classifying cells and structures in tissues through image segmentation and feature extraction, and quantitatively analyzing tissue pathological characteristics in combination with statistical analysis and medical knowledge. This technical field systematically covers the entire process from image acquisition, preprocessing, analysis to information extraction, providing data support for medical diagnosis, disease research and treatment effect evaluation.

[0003] Among them, the liver biopsy data analysis method refers to the method of extracting and analyzing key information in tissue images using specific image analysis algorithms and data processing methods for the digital data of liver biopsy samples. This method mainly covers technical matters such as image preprocessing, tissue region segmentation, cell structure recognition and statistical feature extraction. It removes noise through specific image filtering methods, uses segmentation methods to mark target areas, and uses structure recognition methods to extract cells and their spatial distribution characteristics, ultimately forming a data basis for analysis. This method provides a systematic means for the quantitative and structured analysis of liver biopsy images.

[0004] In the analysis of liver biopsy images, the existing technology lacks quantitative analysis of the geometric center points between tissues, making it difficult to fully describe the structural characteristics between different regions from the perspective of spatial relationships, resulting in the ability to reflect tissue characteristics only through simple calculations of local features, and insufficient spatial correlation. In addition, the existing methods mainly focus on static feature extraction, lack extended analysis of the time dimension, and cannot provide multidimensional modeling support for the diffusion dynamics and evolution laws of the lesion area, especially in pathological processes involving dynamic changes such as fibrosis. It is difficult to accurately capture the key trends of lesion expansion. For texture characteristics, the existing technology is mostly limited to the single extraction of grayscale values, and fails to perform in-depth analysis from the perspective of neighborhood gradients and frequency distribution, resulting in significant limitations in capturing the texture change laws of the lesion area. For spatial distribution characteristics, the existing technology lacks multidimensional quantification of the sparsity and clustering patterns of internuclear spacing, and cannot carefully analyze the heterogeneity of the lesion area, affecting the accuracy and comprehensiveness of the overall pathological analysis. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a liver biopsy data analysis method and system.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme: a method for analyzing liver biopsy data, comprising the following steps: S1: Based on liver biopsy slice images, the geometric center points of cell nuclei, blood vessels and fibrosis areas are extracted, the spatial distance and connection angle between the geometric center points are calculated, the node degree distribution, edge length distribution and aggregation factor of the topological network are statistically analyzed, the edge weight change trend is analyzed, and the spatial distribution and correlation characteristics between regions are analyzed to generate the tissue topological network correlation parameter set; S2: Based on the tissue topology network association parameter set, compare the edge weight distribution changes in the fibrosis area, analyze the spatial characteristics of the edge weight gradient, calculate the expansion intensity of the boundary points, extract the expansion dynamics in combination with the time evolution law of the propagation direction, and generate the fibrosis area expansion trend parameters; S3: Based on the fibrosis area expansion trend parameter, extract the texture grayscale characteristics from the lesion area, calculate the neighborhood gradient and distribution frequency of the pixel grayscale value, analyze the differential change trend of the grayscale direction and extract the dynamic law, and generate the texture dynamic change law parameter of the lesion area; S4: combining the texture dynamic change law parameters of the lesion area and the expansion trend parameters of the fibrosis area, calculating the local area density and spatial distribution gradient between cell nuclei, analyzing the sparseness and aggregation distribution pattern of the internuclear distance, and quantifying the spatial deviation characteristics to generate the spatial distribution characteristic value of cells in the lesion area; S5: Combined with the spatial distribution characteristic values ​​of cells in the lesion area, the difference between the spatial distribution gradient and the local density is compared, the correlation analysis of the cell distribution non-uniformity and the texture dynamic change characteristics is performed, the cell distribution characteristics and expansion laws are analyzed, and the regional characteristics are divided to generate the heterogeneity characteristic partition parameters of the lesion area.

[0007] As a further scheme of the present invention, the tissue topology network association parameter set includes node degree distribution, edge length distribution, aggregation factor, edge weight change, and spatial association parameters between regions. The fibrosis area expansion trend parameters include expansion intensity parameters, fibrosis spatial gradient parameters, propagation direction parameters, and time series dynamic characteristics. The lesion area texture dynamic change law parameters include grayscale gradient parameters, distribution frequency parameters, texture change trend parameters, and dynamic evolution characteristic parameters. The lesion area cell spatial distribution characteristic values ​​include local density parameters, lesion spatial gradient parameters, sparse distribution pattern parameters, and clustered distribution pattern parameters. The lesion area heterogeneity characteristic partition parameters include spatial distribution characteristic parameters, local density difference parameters, non-uniformity characteristic parameters, and texture dynamic characteristic parameters.

[0008] As a further solution of the present invention, based on liver biopsy slice images, the geometric center points of cell nuclei, blood vessels and fibrosis areas are extracted, the spatial distances and connection angles between the geometric center points are calculated, the node degree distribution, edge length distribution and aggregation factor of the topological network are statistically analyzed, the edge weight change trend is analyzed, and the spatial distribution and correlation characteristics between regions are analyzed. The specific steps of generating a tissue topological network association parameter set are: S101: Based on the liver biopsy slice image, image denoising is performed to enhance the clarity of tissue boundaries by adjusting the image contrast, separate cell nuclei, blood vessels and fibrosis areas, extract and mark differential tissue types, and obtain a preprocessed tissue marking image; S102: extracting the geometric center points of the tissue types based on the preprocessed tissue marker image, calculating the spatial distance between the center points, calculating the connection angle between the center points through geometric relationships, and combining the spatial distance and the connection angle for analysis and processing to obtain spatial distance and connection angle data; S103: Based on the spatial distance and connection angle data, the degree distribution value of the network nodes is calculated, the length distribution of the connection edges is measured, and the aggregation factor is counted. The spatial distribution relationship between regions is analyzed through the data change characteristics of the associated edge weights, and a set of associated parameters of the organizational topology network is generated.

[0009] As a further solution of the present invention, based on the tissue topology network association parameter set, the edge weight distribution changes in the fibrosis area are compared, the spatial characteristics of the edge weight gradient are analyzed, the expansion intensity of the boundary point is calculated, and the expansion dynamics are extracted in combination with the time evolution law of the propagation direction. The specific steps of generating the fibrosis area expansion trend parameter are: S201: Based on the tissue topology network association parameter set, edge data in the fibrosis area is screened, the distribution characteristics of edge weights are compared, data of edge weight values ​​are counted and compared, change trends are extracted, spatial distribution characteristics of edge weight gradients are analyzed, and gradient distribution information in the area is calibrated to generate edge weight gradient distribution in the fibrosis area; S202: based on the edge weight gradient distribution of the fibrosis region, analyzing the geometric characteristics of the fibrosis boundary points, determining the weight change direction of the boundary points, calculating the expansion strength in combination with the gradient distribution, analyzing the boundary expansion strength data associated with the gradient distribution according to the strength calculation result, and generating the boundary expansion strength characteristics of the fibrosis region; S203: Based on the boundary expansion strength characteristics of the fiberized area, the directional characteristics of the edge weight change are analyzed, the time evolution trend during the fiberization expansion process is analyzed, the expansion dynamic data is calculated, the expansion direction and boundary propagation dynamic characteristics of the fiberized area are extracted, and the fiberized area expansion trend parameters are generated.

[0010] As a further solution of the present invention, based on the fibrosis area expansion trend parameter, the texture grayscale characteristics are extracted from the lesion area, the neighborhood gradient and distribution frequency of the pixel grayscale value are calculated, the differential change trend of the grayscale direction is analyzed and the dynamic law is extracted, and the specific steps of generating the dynamic change law parameters of the texture of the lesion area are as follows: S301: based on the fibrosis area expansion trend parameter, extracting the texture grayscale value of the pixel in the lesion area, calculating the difference between the grayscale value and the grayscale value of the neighboring pixel, analyzing the grayscale gradient change, counting the distribution frequency of the grayscale value, and collating the spatial distribution characteristic data to generate the texture grayscale distribution of the lesion area; S302: Based on the grayscale distribution of the texture of the lesion area, analyzing the change law of the grayscale value in the differentiation direction, calculating the difference between the grayscale values, comparing the dynamic characteristics of the grayscale value in the differentiation direction in the neighborhood space, extracting the grayscale change law data associated with the direction, and generating the grayscale direction dynamic characteristics; S303: Based on the grayscale direction dynamic characteristics, analyze the correlation characteristics of the grayscale change trend in spatial propagation, calculate the dynamic change intensity during texture expansion, extract texture expansion dynamic characteristic data in combination with the time evolution trend, and generate texture dynamic change law parameters in the lesion area.

[0011] As a further solution of the present invention, the difference distribution value calculation formula is specifically: ; in, Represents the grayscale difference distribution value, Representative The gray value of a pixel, Representative The gray value of a pixel, Represents the number of pixels in the neighborhood. Represents the absolute difference of grayscale values. Represents the average gray value.

[0012] As a further solution of the present invention, the specific steps of combining the dynamic change law parameters of the texture of the lesion area and the expansion trend parameters of the fibrosis area, calculating the local area density and spatial distribution gradient between cell nuclei, analyzing the sparseness and clustering distribution pattern of the internuclear distance, and quantifying the spatial deviation characteristics to generate the spatial distribution characteristic value of the cells in the lesion area are as follows: S401: based on the texture dynamic change law parameter of the lesion area and the expansion trend parameter of the fibrosis area, extract the geometric center point of the cell nucleus, count the number of cell nuclei in the local area, analyze the ratio of the number of nuclei to the area of ​​the area, calculate the cell nucleus density in the local area, and generate cell nucleus local area density data; S402: Based on the local area density data of the cell nucleus, analyzing the distribution characteristics of the geometric center point of the cell nucleus in space, calculating the gradient value of the center point, comparing the distribution changes in the local area, quantifying the difference of the gradient in space, and combining the cell nucleus distribution state to generate cell nucleus spatial distribution gradient data; S403: Based on the cell nucleus spatial distribution gradient data, measure the geometric distance between cell nuclei, calculate the discreteness and concentration characteristics of the distance data, analyze the distribution pattern and spatial deviation of the internuclear distance, count the deviation amplitude and direction, and generate the cell spatial distribution characteristic value of the lesion area.

[0013] As a further solution of the present invention, the gradient value calculation formula is specifically: ; in, Gradient value, Represents the local area The density of cell nuclei, and Represents the density values ​​in and The rate of change of the gradient in the direction, represents the number of cell nuclei in the neighborhood, Indicates the neighborhood The cell nucleus and The absolute difference of the nuclear density values ​​of the cells.

[0014] As a further solution of the present invention, the specific steps of combining the spatial distribution characteristic value of the cells in the lesion area, comparing the difference between the spatial distribution gradient and the local density, correlating and analyzing the non-uniformity of cell distribution and the dynamic change characteristics of texture, analyzing the cell distribution characteristics and expansion law, and dividing the regional characteristics to generate the heterogeneous characteristic partition parameters of the lesion area are as follows: S501: Based on the spatial distribution characteristic value of the cells in the lesion area, calculate the spatial distribution gradient of the local area, analyze the gradient change characteristics, compare the difference between the gradient value and the local density value, count the local difference distribution data, calibrate the distribution characteristics of the non-uniform area, and generate cell distribution gradient and density difference data; S502: Based on the cell distribution gradient and density difference data, combined with the texture dynamic change characteristics, analyzing the cell distribution characteristics of the non-uniform area, analyzing the correlation between the distribution characteristics and the spatial expansion, extracting the cell distribution characteristics and the spatial law of expansion in the non-uniform area, and generating cell distribution characteristics and expansion law data; S503: Based on the cell distribution characteristics and expansion law data, the heterogeneous spatial characteristics of the lesion area are divided, the spatial differences of the characteristic parameters in the differentiated partitions are measured, the boundary information of the heterogeneous area is extracted, and a partition characteristic parameter set is constructed to generate the heterogeneous characteristic partition parameters of the lesion area.

[0015] A liver biopsy data analysis system, comprising: The tissue network module extracts the geometric center points of cell nuclei, blood vessels and fibrosis areas based on liver biopsy slice images, calculates the spatial distance and connection angle between the geometric center points, analyzes the edge weight change trend and analyzes the spatial distribution and correlation characteristics between regions, and generates a tissue topology network correlation parameter set; The fibrosis expansion module compares the edge weight distribution changes in the fibrosis area based on the tissue topology network association parameter set, calculates the expansion strength of the boundary points, extracts the expansion dynamics in combination with the time evolution law of the propagation direction, and generates the fibrosis area expansion trend parameters; The texture change module extracts texture grayscale characteristics from the lesion area based on the fibrosis area expansion trend parameter, analyzes the differential change trend in the grayscale direction and extracts the dynamic law, and generates the texture dynamic change law parameter of the lesion area; The cell distribution module combines the texture dynamic change law parameters of the lesion area and the expansion trend parameters of the fibrosis area to calculate the local area density and spatial distribution gradient between cell nuclei, and quantifies the spatial deviation characteristics to generate the spatial distribution characteristic value of cells in the lesion area; The heterogeneity partitioning module combines the spatial distribution characteristic values ​​of the cells in the lesion area, compares the difference between the spatial distribution gradient and the local density to analyze the cell distribution characteristics and expansion rules, divides the regional characteristics, and generates the heterogeneity characteristic partitioning parameters of the lesion area.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by extracting the geometric center points of cell nuclei, blood vessels and fibrosis areas and combining the spatial gradient and time evolution relationship of edge weights, the modeling accuracy of lesion diffusion behavior is improved, and the pixel-level dynamic change law of the lesion area is captured. The spatial distribution characteristics are quantified through the sparsity and clustering analysis of local density and internuclear distance. The overall solution combines geometric, temporal and texture characteristics in multiple dimensions to enhance the dynamic quantification capability of pathological characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 It is a schematic diagram of the steps of the present invention; Figure 2 is a flow chart of the steps of S1 of the present invention; Figure 3 is a flow chart of the steps of S2 of the present invention; Figure 4 is a flow chart of the steps of S3 of the present invention; Figure 5 is a flow chart of the steps of S4 of the present invention; Figure 6 is a flow chart of the steps of S5 of the present invention; Figure 7 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0019] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0020] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0021] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0022] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0023] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0024] See also Figure 1 , a liver biopsy data analysis method, comprising the following steps: S1: Based on liver biopsy slice images, the geometric center points of cell nuclei, blood vessels and fibrosis areas are extracted, the spatial distance and connection angle between the geometric center points are calculated, the node degree distribution, edge length distribution and aggregation factor of the topological network are statistically analyzed, the edge weight change trend is analyzed, and the spatial distribution and correlation characteristics between regions are analyzed to generate the tissue topological network correlation parameter set; S2: Based on the tissue topology network association parameter set, compare the edge weight distribution changes in the fibrosis area, analyze the spatial characteristics of the edge weight gradient, calculate the expansion intensity of the boundary points, extract the expansion dynamics based on the time evolution law of the propagation direction, and generate the fibrosis area expansion trend parameters; S3: Based on the fibrosis area expansion trend parameters, the texture grayscale characteristics are extracted from the lesion area, the neighborhood gradient and distribution frequency of the pixel grayscale value are calculated, the differential change trend of the grayscale direction is analyzed and the dynamic law is extracted, and the parameters of the dynamic change law of the texture of the lesion area are generated; S4: Combine the parameters of the dynamic change law of the texture of the lesion area and the expansion trend parameters of the fibrosis area to calculate the local area density and spatial distribution gradient between cell nuclei, analyze the sparseness and clustering distribution pattern of the internuclear distance, quantify the spatial deviation characteristics, and generate the spatial distribution characteristic value of cells in the lesion area; S5: Combine the spatial distribution characteristic values ​​of cells in the lesion area, compare the difference between the spatial distribution gradient and the local density, associate and analyze the non-uniformity of cell distribution and the dynamic change characteristics of texture, analyze the cell distribution characteristics and expansion rules, divide the regional characteristics, and generate the heterogeneity characteristic partition parameters of the lesion area; The tissue topology network association parameter set includes node degree distribution, edge length distribution, aggregation factor, edge weight change, and spatial association parameters between regions. The fibrosis area expansion trend parameters include expansion intensity parameters, fibrosis spatial gradient parameters, propagation direction parameters, and time series dynamic characteristics. The dynamic change law parameters of the texture in the lesion area include grayscale gradient parameters, distribution frequency parameters, texture change trend parameters, and dynamic evolution characteristic parameters. The spatial distribution characteristic values ​​of cells in the lesion area include local density parameters, lesion spatial gradient parameters, sparse distribution pattern parameters, and aggregated distribution pattern parameters. The heterogeneity characteristic partitioning parameters of the lesion area include spatial distribution characteristic parameters, local density difference parameters, non-uniformity characteristic parameters, and texture dynamic characteristic parameters.

[0025] See also Figure 2 , the specific steps of S1 are: S101: Based on the liver biopsy slice image, image denoising is performed to enhance the clarity of tissue boundaries by adjusting the image contrast, separate cell nuclei, blood vessels and fibrosis areas, extract and mark differential tissue types, and obtain a preprocessed tissue marking image; A high-resolution microscope is used to obtain digital image data of the slices, and the noise in the image is removed by a multi-scale filtering method. The filters used include Gaussian filtering and bilateral filtering. The information of the high-frequency area of ​​the image is extracted while retaining the tissue details. The image contrast is then enhanced by histogram equalization to reveal the tissue boundaries. The cell nuclei, blood vessels, and fibrosis areas in the image are then segmented by morphological processing technology. The cell nuclei are marked with blue, blood vessels with red, and fibrosis areas with green using specific colors to form a preprocessed tissue marking image.

[0026] S102: extracting the geometric center points of the tissue types based on the preprocessed tissue marker image, calculating the spatial distance between the center points, calculating the connection angle between the center points through geometric relationships, combining the spatial distance and the connection angle for analysis and processing, and obtaining spatial distance and connection angle data; Through the extraction of geometric center points and spatial analysis, according to the formula ; Calculate the connection angle and spatial distance between center points.

[0027] In the formula, represents the connection angle, and are the direction vectors of the two center points, and is the modulus of the vector, is the spatial distance between center point i and center point j, and are the coordinates of the two center points.

[0028] First, calculate the spatial distance between the two center points. Assume that the coordinates of center point 1 are , the coordinates of center point 2 are , the distance formula is ; If the collected data is , ,but ; Next, the connection angle is calculated and the direction vector can be set to and Assume another point ,but ; The dot product of the direction vector is ; The vector modulus is ; The angle formula is ; The result shows that the spatial distance between the two center points is 28.28, the connection angle is 0 degrees, and the topological relationship can be further analyzed by combining multiple center points.

[0029] S103: Based on the spatial distance and connection angle data, the degree distribution value of the network nodes is calculated, the length distribution of the connection edges is measured, and the aggregation factor thereof is counted. The spatial distribution relationship between regions is analyzed through the data variation characteristics of the associated edge weights, and an organizational topology network associated parameter set is generated. Firstly, the preprocessed center point coordinates are extracted in batches to build the node adjacency matrix. The number of connections of each node is calculated by statistical methods. The length distribution of the connecting edges is calculated and the distribution histogram is drawn. The changing trend of the aggregation factor is further compared. Combined with the changing characteristics of the node weight values ​​between different regions, the correlation between the node edge weight and the regional distribution is analyzed. Finally, the association parameter set of the organizational topology network is generated.

[0030] See also Figure 3 , the specific steps of S2 are: S201: based on the tissue topology network association parameter set, edge data in the fibrosis area are screened, the distribution characteristics of edge weights are compared, data of edge weight values ​​are counted and compared, change trends are extracted, spatial distribution characteristics of edge weight gradients are analyzed, and gradient distribution information in the area is calibrated to generate edge weight gradient distribution in the fibrosis area; The edge data are sorted and classified according to the weight value. The mean and standard deviation of the edge weight are calculated by statistical methods to obtain the weight distribution characteristics. The kernel density estimation method is used to extract the changing trend of the edge weight. The gradient characteristics of the distribution are combined to analyze the gradient characteristics of the edge weight changes in different regions. The gradient distribution information is calibrated by the spatial interpolation method, and the gradient distribution results are expressed in the form of a two-dimensional heat map. Finally, the edge weight gradient distribution of the fibrosis area is generated.

[0031] S202: based on the edge weight gradient distribution of the fibrosis region, analyzing the geometric characteristics of the fibrosis boundary points, determining the weight change direction of the boundary points, calculating the expansion strength in combination with the gradient distribution, analyzing the boundary expansion strength data associated with the gradient distribution according to the strength calculation result, and generating the boundary expansion strength characteristics of the fibrosis region; Based on the gradient distribution, the geometric characteristics of the boundary points are analyzed according to the formula ; The boundary extension strength data associated with the extension strength and the gradient distribution are calculated.

[0032] In the formula, Represents the expansion strength, is the gradient distribution, is the normal vector of the boundary point, is the boundary extension strength associated with the gradient distribution, and are the weighting factor of the gradient distribution and the local gradient characteristic value, respectively. are the boundary point coordinates, and is the integration interval.

[0033] Extension Strength By integrating the dot product of the gradient distribution of the boundary point and the normal vector, assuming that the gradient distribution of a boundary point , the normal vector , integration interval , and the dot product is calculated as ; The integral result is ; Extended Strength Correlation Value Calculate, assuming the weighting factor of the gradient distribution , local gradient characteristic value ,but ; The results show that the boundary extension intensity values ​​associated with the gradient distribution are and , by applying it to the boundary points of the fibrosis area, its expansion characteristics and direction can be further analyzed.

[0034] S203: based on the boundary expansion strength characteristics of the fibrosis region, analyzing the direction characteristics of the edge weight change, analyzing the time evolution trend of the fibrosis expansion process, calculating the expansion dynamic data, extracting the expansion direction and boundary propagation dynamic characteristics of the fibrosis region, and generating the fibrosis region expansion trend parameters; The dynamic change characteristics of fibrosis extension are extracted by collecting time series data, the data in the time series are discretized into equally spaced time points, and the expansion rate corresponding to each time point is calculated. The trend characteristics of the fibrosis extension direction are extracted by using a polynomial fitting method. The dynamic changes of the direction vector and boundary propagation are combined, and the expansion trend information is analyzed by establishing a dynamic characteristic parameter matrix, and finally the fibrosis area expansion trend parameters are generated.

[0035] See also Figure 4 , the specific steps of S3 are: S301: based on the fibrosis area expansion trend parameter, extract the texture grayscale value of the pixel in the lesion area, calculate the difference between the grayscale value and the grayscale value of the neighboring pixel, analyze the grayscale gradient change, count the distribution frequency of the grayscale value, and organize the spatial distribution characteristic data to generate the texture grayscale distribution of the lesion area; By quantifying the grayscale value, the sliding window method is used to calculate the grayscale difference between each pixel and its neighboring pixels, the distribution characteristics of the grayscale differences of all pixels are counted, and the local change trend of the grayscale gradient is analyzed in combination with the change of the gradient direction. The distribution frequency of the grayscale value is calculated using the frequency statistics method, and the spatial distribution map of the grayscale value is generated through the two-dimensional spatial interpolation method. Finally, the texture grayscale distribution characteristic data of the lesion area are sorted out.

[0036] S302: Based on the grayscale distribution of the texture in the lesion area, analyzing the change law of the grayscale value in the differentiation direction, calculating the difference distribution value between the grayscale values, comparing the dynamic characteristics of the grayscale value in the differentiation direction in the neighborhood space, extracting the grayscale change law data associated with the direction, and generating the grayscale direction dynamic characteristics; The specific calculation formula for the difference distribution value is: ; in, Represents the grayscale difference distribution value, Representative The gray value of a pixel, Representative The gray value of a pixel, Represents the number of pixels in the neighborhood. Represents the absolute difference of grayscale values. Represents the average gray value.

[0037] Indicates the grayscale difference distribution value. The following parameters need to be obtained in sequence during the calculation process: Gray value and : Grayscale value is the grayscale intensity of the pixel in the image, ranging from 0 to 255. Grayscale value is collected through the image grayscale histogram. The unit of grayscale value is pixel intensity. In the example, five pixels in a local area of ​​an image are selected, and the grayscale values ​​are , the gray value of the neighborhood pixel is .

[0038] Number of neighboring points : The number of neighborhood points is determined by the size of the sliding window or convolution kernel. , the number of neighboring points is 9. In the example, Pixels are calculated.

[0039] Grayscale Difference : This parameter is the absolute difference of grayscale values, and the calculation steps are: ; Grayscale average : The steps to calculate the grayscale average are: ; Sum of Squares and Square Roots: The sum of squares of grayscale differences is: ; The calculation results are: ; Take the square root: ; Grayscale difference distribution value : Calculate the mean: ; Substitute the values ​​to calculate: ; The result shows that the grayscale difference distribution value is 72.71, which characterizes the mean square error characteristics of the grayscale value changes in the local area of ​​the image, and is closely related to the dynamic changes in the grayscale direction of the neighborhood, and can be further used to analyze the dynamic characteristics and change trends of the grayscale direction.

[0040] S303: Based on the dynamic characteristics of grayscale direction, the correlation characteristics of grayscale change trend in spatial propagation are analyzed, the dynamic change intensity in the texture expansion process is calculated, and the texture expansion dynamic characteristic data is extracted in combination with the time evolution trend to generate the parameters of the dynamic change law of the texture in the lesion area; A dynamic change matrix of grayscale value and time is constructed, and the change rate of grayscale value in the time series is extracted. The linear regression model is used to fit the dynamic characteristic data of texture extension. The dynamic change intensity of texture extension is calculated by solving the time derivative of the change intensity. The texture dynamic change association parameters are established by combining the gradient direction and the time evolution trend, and the texture dynamic change law parameters of the lesion area are generated.

[0041] See also Figure 5 , the specific steps of S4 are: S401: based on the parameters of the dynamic change law of the texture of the lesion area and the expansion trend parameters of the fibrosis area, the geometric center point of the cell nucleus is extracted, the number of cell nuclei in the local area is counted, the ratio of the number of nuclei to the area of ​​the area is analyzed, the cell nucleus density in the local area is calculated, and the cell nucleus local area density data is generated; The geometric center point of the cell nucleus in the local area is extracted, and the position of each cell nucleus is located using an image processing algorithm. The number of cell nuclei in the area is counted by a counting method, and the area of ​​the local area is calculated by the number of pixels. The local cell nucleus density is calculated using the ratio of the area to the number of cell nuclei. The density data is organized into a matrix form and associated with the local area position to generate local area density data of cell nuclei.

[0042] S402: Based on the local area density data of the cell nucleus, the distribution characteristics of the geometric center point of the cell nucleus in space are analyzed, the gradient value of the center point is calculated, the distribution changes in the local area are compared, the difference of the gradient in space is quantified, and the spatial distribution gradient data of the cell nucleus is generated in combination with the distribution state of the cell nucleus; The specific formula for calculating the gradient value is: ; in, Gradient value, Represents the local area The density of cell nuclei, and Represents the density values ​​in and The rate of change of the gradient in the direction, represents the number of cell nuclei in the neighborhood, Indicates the neighborhood The cell nucleus and The absolute difference of the nuclear density values ​​of the cells.

[0043] Represents the local distribution gradient value, which is used to quantify the spatial variation characteristics of cell nucleus density in the local area. The following parameters are involved in the formula calculation: Density value gradient change rate and : The local density data of the cell nucleus were extracted by image processing and calculated. and Gradient in the direction. Taking a 3×3 local area as an example, assume that the density value matrix is: ; The density value of the center point is .exist The formula for calculating the gradient change rate in a direction is: ; Substitute the data: ; exist The formula for calculating the gradient change rate in a direction is: ; Substitute the data: ; Gradient modulus : The gradient modulus calculation formula is: ; Density difference within the neighborhood : Calculate the absolute difference between the density value of each cell nucleus in the neighborhood and the center point. The neighborhood density value is , the center point density is The difference is calculated as follows: ; The calculation results are: ; Neighborhood Difference Mean : The number of cell nuclei in the neighborhood is , the mean calculation formula is: ; Local distribution gradient value : Substitute the gradient modulus and the mean of the neighborhood differences into the formula: ; ; The result shows that the gradient value is 9.475, which characterizes the density variation characteristics of the cell nucleus in the local area, is closely related to the gradient characteristics of the spatial distribution of the cell nucleus, and can be further used to generate the spatial distribution gradient data of the cell nucleus.

[0044] S403: Based on the spatial distribution gradient data of cell nuclei, measure the geometric distance between cell nuclei, calculate the discreteness and concentration characteristics of the distance data, analyze the distribution pattern and spatial deviation of the internuclear distance, calculate the deviation amplitude and direction, and generate the spatial distribution characteristic value of cells in the lesion area; The distance characteristics between cell nuclei were calculated by measuring the Euclidean distance between geometric center points. The distribution range of all distance values ​​was analyzed by statistical methods, and the standard deviation and mean of the distance values ​​were calculated to represent the discreteness and centrality. The statistical matrix of the deviation amplitude was established through the distribution pattern of the internuclear distance and the spatial deviation. Combined with the directional quantitative characteristics of the spatial deviation, the spatial distribution characteristic values ​​of the cells in the lesion area were generated.

[0045] See also Figure 6 , the specific steps of S5 are: S501: Based on the spatial distribution characteristic value of cells in the lesion area, calculate the spatial distribution gradient of the local area, analyze the gradient change characteristics, compare the difference between the gradient value and the local density value, count the local difference distribution data, calibrate the distribution characteristics of the non-uniform area, and generate cell distribution gradient and density difference data; The center point of each local area was selected, and the spatial gradient value of each area was calculated using the distribution gradient formula. The deviation between the gradient and the density was analyzed by comparing the local gradient value with the cell density value. The statistical model was used to calculate the differential distribution value within the local area, and the distribution characteristics of the non-uniform area were marked in the spatial map to generate cell distribution gradient and density difference data.

[0046] S502: Based on the cell distribution gradient and density difference data, combined with the texture dynamic change characteristics, analyze the cell distribution characteristics of the non-uniform area, analyze the correlation between the distribution characteristics and the spatial expansion, extract the cell distribution characteristics and the spatial law of expansion in the non-uniform area, and generate cell distribution characteristics and expansion law data; Based on the cell distribution gradient and density difference data, according to the formula ; Calculate the cell distribution characteristics and expansion laws in non-uniform areas.

[0047] In the formula, Represents the total difference value within the non-uniform area, and Respectively represent the distribution gradient value and density value in the area, is the difference vector modulus in the non-uniform region, and is the rate of change of the gradient value in the x and y directions, and is the rate of change of density value in the x and y directions.

[0048] First, calculate the total difference value of the non-uniform area, assuming that the distribution gradient value of a certain area is , the density value is ,but ; Then calculate the difference vector modulus of the non-uniform area, assuming that the rates of change of the gradient values ​​in the x and y directions are and The rates of change of density values ​​in the x and y directions are and ,but ; ; The results show that the total difference value of the non-uniform area is 0.7, and the difference vector modulus is 0.49. By analyzing these values, the non-uniform characteristics and expansion laws of cell distribution can be further extracted.

[0049] S503: Based on the cell distribution characteristics and expansion law data, the heterogeneous spatial characteristics of the lesion area are divided, the spatial differences of the characteristic parameters in the differentiated partitions are measured, the boundary information of the heterogeneous area is extracted, and a partition characteristic parameter set is constructed to generate the heterogeneous characteristic partition parameters of the lesion area; By calculating the spatial difference values ​​of the characteristic parameters within the partition, the characteristic parameter set of each partition is extracted. The boundary information of the heterogeneous area and the characteristic parameter distribution characteristics are combined to count the boundary characteristic parameter values ​​of each partition, and a characteristic parameter matrix is ​​established. The heterogeneous spatial characteristics of the partitions are marked as independent areas, and the heterogeneous characteristic partition parameters of the lesion area are generated.

[0050] See also Figure 7 , a liver biopsy data analysis system, comprising: The tissue network module extracts the geometric center points of cell nuclei, blood vessels and fibrosis areas based on liver biopsy slice images, calculates the spatial distance and connection angle between the geometric center points, analyzes the edge weight change trend and analyzes the spatial distribution and correlation characteristics between regions, and generates a tissue topology network correlation parameter set; The fibrosis expansion module compares the edge weight distribution changes in the fibrosis area based on the tissue topology network association parameter set, calculates the expansion strength of the boundary points, extracts the expansion dynamics based on the time evolution law of the propagation direction, and generates the fibrosis area expansion trend parameters; The texture change module extracts the texture grayscale characteristics from the lesion area based on the fibrosis area expansion trend parameters, analyzes the differential change trend in the grayscale direction, extracts the dynamic law, and generates the parameters of the dynamic change law of the texture in the lesion area; The cell distribution module combines the parameters of the dynamic change law of the texture of the lesion area and the expansion trend parameters of the fibrosis area to calculate the local area density and spatial distribution gradient between cell nuclei, quantify the spatial deviation characteristics, and generate the spatial distribution characteristic value of cells in the lesion area; The heterogeneity partitioning module combines the spatial distribution characteristic values ​​of cells in the lesion area, compares the difference between the spatial distribution gradient and the local density to analyze the cell distribution characteristics and expansion rules, divides the regional characteristics, and generates the heterogeneity characteristic partitioning parameters of the lesion area.

[0051] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for analyzing liver biopsy data, characterized in that: The following steps are involved: S1: Based on liver biopsy slice images, the geometric center points of cell nuclei, blood vessels and fibrosis areas are extracted, the spatial distance and connection angle between the geometric center points are calculated, the node degree distribution, edge length distribution and aggregation factor of the topological network are statistically analyzed, the edge weight change trend is analyzed, and the spatial distribution and correlation characteristics between regions are analyzed to generate the tissue topological network correlation parameter set; S2: Based on the tissue topology network association parameter set, compare the edge weight distribution changes in the fibrosis area, analyze the spatial characteristics of the edge weight gradient, calculate the expansion intensity of the boundary points, extract the expansion dynamics in combination with the time evolution law of the propagation direction, and generate the fibrosis area expansion trend parameters; S3: Based on the fibrosis area expansion trend parameter, extract the texture grayscale characteristics from the lesion area, calculate the neighborhood gradient and distribution frequency of the pixel grayscale value, analyze the differential change trend of the grayscale direction and extract the dynamic law, and generate the texture dynamic change law parameter of the lesion area; S4: combining the texture dynamic change law parameters of the lesion area and the expansion trend parameters of the fibrosis area, calculating the local area density and spatial distribution gradient between cell nuclei, analyzing the sparseness and aggregation distribution pattern of the internuclear distance, and quantifying the spatial deviation characteristics to generate the spatial distribution characteristic value of cells in the lesion area; S5: Combined with the spatial distribution characteristic values ​​of cells in the lesion area, the difference between the spatial distribution gradient and the local density is compared, the correlation analysis of the cell distribution non-uniformity and the texture dynamic change characteristics is performed, the cell distribution characteristics and expansion laws are analyzed, and the regional characteristics are divided to generate the heterogeneity characteristic partition parameters of the lesion area.

2. The method for analyzing liver biopsy data according to claim 1, characterized in that: The tissue topology network association parameter set includes node degree distribution, edge length distribution, aggregation factor, edge weight change, and spatial association parameters between regions. The fibrosis area expansion trend parameters include expansion intensity parameters, fibrosis spatial gradient parameters, propagation direction parameters, and time series dynamic characteristics. The lesion area texture dynamic change law parameters include grayscale gradient parameters, distribution frequency parameters, texture change trend parameters, and dynamic evolution characteristic parameters. The lesion area cell spatial distribution characteristic values ​​include local density parameters, lesion spatial gradient parameters, sparse distribution pattern parameters, and clustered distribution pattern parameters. The lesion area heterogeneity characteristic partition parameters include spatial distribution characteristic parameters, local density difference parameters, non-uniformity characteristic parameters, and texture dynamic characteristic parameters.

3. The method for analyzing liver biopsy data according to claim 1, characterized in that: Based on liver biopsy slice images, the geometric center points of cell nuclei, blood vessels and fibrosis areas were extracted, the spatial distances and connection angles between the geometric center points were calculated, the node degree distribution, edge length distribution and aggregation factor of the topological network were statistically analyzed, the edge weight change trend was analyzed, and the spatial distribution and association characteristics between regions were analyzed. The specific steps for generating the tissue topological network association parameter set are as follows: S101: Based on the liver biopsy slice image, image denoising is performed to enhance the clarity of tissue boundaries by adjusting the image contrast, separate cell nuclei, blood vessels and fibrosis areas, extract and mark differential tissue types, and obtain a preprocessed tissue marking image; S102: extracting the geometric center points of the tissue types based on the preprocessed tissue marker image, calculating the spatial distance between the center points, calculating the connection angle between the center points through geometric relationships, and combining the spatial distance and the connection angle for analysis and processing to obtain spatial distance and connection angle data; S103: Based on the spatial distance and connection angle data, the degree distribution value of the network nodes is calculated, the length distribution of the connection edges is measured, and the aggregation factor is counted. The spatial distribution relationship between regions is analyzed through the data change characteristics of the associated edge weights, and a set of associated parameters of the organizational topology network is generated.

4. The method for analyzing liver biopsy data according to claim 1, characterized in that: Based on the tissue topology network association parameter set, the edge weight distribution changes in the fibrosis area are compared, the spatial characteristics of the edge weight gradient are analyzed, the expansion intensity of the boundary points is calculated, and the expansion dynamics are extracted in combination with the time evolution law of the propagation direction. The specific steps of generating the fibrosis area expansion trend parameters are as follows: S201: Based on the tissue topology network association parameter set, edge data in the fibrosis area is screened, the distribution characteristics of edge weights are compared, data of edge weight values ​​are counted and compared, change trends are extracted, spatial distribution characteristics of edge weight gradients are analyzed, and gradient distribution information in the area is calibrated to generate edge weight gradient distribution in the fibrosis area; S202: based on the edge weight gradient distribution of the fibrosis region, analyzing the geometric characteristics of the fibrosis boundary points, determining the weight change direction of the boundary points, calculating the expansion strength in combination with the gradient distribution, analyzing the boundary expansion strength data associated with the gradient distribution according to the strength calculation result, and generating the boundary expansion strength characteristics of the fibrosis region; S203: Based on the boundary expansion strength characteristics of the fiberized area, the directional characteristics of the edge weight change are analyzed, the time evolution trend during the fiberization expansion process is analyzed, the expansion dynamic data is calculated, the expansion direction and boundary propagation dynamic characteristics of the fiberized area are extracted, and the fiberized area expansion trend parameters are generated.

5. The method for analyzing liver biopsy data according to claim 1, characterized in that: Based on the fibrosis area expansion trend parameter, the texture grayscale characteristics are extracted from the lesion area, the neighborhood gradient and distribution frequency of the pixel grayscale value are calculated, the differential change trend of the grayscale direction is analyzed and the dynamic law is extracted, and the specific steps of generating the dynamic change law parameters of the texture of the lesion area are as follows: S301: based on the fibrosis area expansion trend parameter, extracting the texture grayscale value of the pixel in the lesion area, calculating the difference between the grayscale value and the grayscale value of the neighboring pixel, analyzing the grayscale gradient change, counting the distribution frequency of the grayscale value, and collating the spatial distribution characteristic data to generate the texture grayscale distribution of the lesion area; S302: Based on the grayscale distribution of the texture of the lesion area, analyzing the change law of the grayscale value in the differentiation direction, calculating the difference between the grayscale values, comparing the dynamic characteristics of the grayscale value in the differentiation direction in the neighborhood space, extracting the grayscale change law data associated with the direction, and generating the grayscale direction dynamic characteristics; S303: Based on the grayscale direction dynamic characteristics, analyze the correlation characteristics of the grayscale change trend in spatial propagation, calculate the dynamic change intensity during texture expansion, extract texture expansion dynamic characteristic data in combination with the time evolution trend, and generate texture dynamic change law parameters in the lesion area.

6. The method for analyzing liver biopsy data according to claim 5, characterized in that: The difference distribution value calculation formula is specifically: ； in, Represents the grayscale difference distribution value, Representative The gray value of a pixel, Representative The gray value of a pixel, Represents the number of pixels in the neighborhood. Represents the absolute difference of grayscale values. Represents the average gray value.

7. The method for analyzing liver biopsy data according to claim 1, characterized in that: The specific steps of combining the texture dynamic change law parameters of the lesion area and the expansion trend parameters of the fibrosis area, calculating the local area density and spatial distribution gradient between cell nuclei, analyzing the sparseness and aggregation distribution pattern of the internuclear distance, and quantifying the spatial deviation characteristics to generate the spatial distribution characteristic value of cells in the lesion area are as follows: S401: based on the texture dynamic change law parameter of the lesion area and the expansion trend parameter of the fibrosis area, extract the geometric center point of the cell nucleus, count the number of cell nuclei in the local area, analyze the ratio of the number of nuclei to the area of ​​the area, calculate the cell nucleus density in the local area, and generate cell nucleus local area density data; S402: Based on the local area density data of the cell nucleus, analyzing the distribution characteristics of the geometric center point of the cell nucleus in space, calculating the gradient value of the center point, comparing the distribution changes in the local area, quantifying the difference of the gradient in space, and combining the cell nucleus distribution state to generate cell nucleus spatial distribution gradient data; S403: Based on the cell nucleus spatial distribution gradient data, measure the geometric distance between cell nuclei, calculate the discreteness and concentration characteristics of the distance data, analyze the distribution pattern and spatial deviation of the internuclear distance, count the deviation amplitude and direction, and generate the cell spatial distribution characteristic value of the lesion area.

8. The method for analyzing liver biopsy data according to claim 7, characterized in that: The gradient value calculation formula is specifically: ； in, Gradient value, Represents the local area The density of cell nuclei, and Represents the density values ​​in and The rate of change of the gradient in the direction, represents the number of cell nuclei in the neighborhood, Indicates the neighborhood The cell nucleus and The absolute difference of the nuclear density values ​​of the cells.

9. The method for analyzing liver biopsy data according to claim 1, characterized in that: Combined with the spatial distribution characteristic value of cells in the lesion area, the difference between the spatial distribution gradient and the local density is compared, the non-uniformity of cell distribution and the dynamic change characteristics of texture are correlated and analyzed, the cell distribution characteristics and expansion law are analyzed, and the regional characteristics are divided. The specific steps of generating the heterogeneous characteristic partition parameters of the lesion area are as follows: S501: Based on the spatial distribution characteristic value of the cells in the lesion area, calculate the spatial distribution gradient of the local area, analyze the gradient change characteristics, compare the difference between the gradient value and the local density value, count the local difference distribution data, calibrate the distribution characteristics of the non-uniform area, and generate cell distribution gradient and density difference data; S502: Based on the cell distribution gradient and density difference data, combined with the texture dynamic change characteristics, analyzing the cell distribution characteristics of the non-uniform area, analyzing the correlation between the distribution characteristics and the spatial expansion, extracting the cell distribution characteristics and the spatial law of expansion in the non-uniform area, and generating cell distribution characteristics and expansion law data; S503: Based on the cell distribution characteristics and expansion law data, the heterogeneous spatial characteristics of the lesion area are divided, the spatial differences of the characteristic parameters in the differentiated partitions are measured, the boundary information of the heterogeneous area is extracted, and a partition characteristic parameter set is constructed to generate the heterogeneous characteristic partition parameters of the lesion area.

10. A liver biopsy data analysis system, characterized in that: A method for analyzing liver biopsy data according to any one of claims 1 to 9, wherein the system comprises: The tissue network module extracts the geometric center points of cell nuclei, blood vessels and fibrosis areas based on liver biopsy slice images, calculates the spatial distance and connection angle between the geometric center points, analyzes the edge weight change trend and analyzes the spatial distribution and correlation characteristics between regions, and generates a tissue topology network correlation parameter set; The fibrosis expansion module compares the edge weight distribution changes in the fibrosis area based on the tissue topology network association parameter set, calculates the expansion strength of the boundary points, extracts the expansion dynamics in combination with the time evolution law of the propagation direction, and generates the fibrosis area expansion trend parameters; The texture change module extracts texture grayscale characteristics from the lesion area based on the fibrosis area expansion trend parameter, analyzes the differential change trend in the grayscale direction and extracts the dynamic law, and generates the texture dynamic change law parameter of the lesion area; The cell distribution module combines the texture dynamic change law parameters of the lesion area and the expansion trend parameters of the fibrosis area to calculate the local area density and spatial distribution gradient between cell nuclei, and quantifies the spatial deviation characteristics to generate the spatial distribution characteristic value of cells in the lesion area; The heterogeneity partitioning module combines the spatial distribution characteristic values ​​of the cells in the lesion area, compares the difference between the spatial distribution gradient and the local density to analyze the cell distribution characteristics and expansion rules, divides the regional characteristics, and generates the heterogeneity characteristic partitioning parameters of the lesion area.

Images