Breast cancer neoadjuvant chemotherapy efficacy prediction method and system based on spatial heterogeneity of treg cells

Abstract

The present application relates to a breast cancer NAC efficacy prediction method and system based on Treg cell spatial heterogeneity, wherein the method performs multi-dimensional spatial distribution analysis on Treg cells in tumor parenchymal regions and Treg cells in tumor interstitial regions according to cell recognition results, obtains corresponding parenchymal region spatial distribution analysis results and interstitial region spatial distribution analysis results, performs multi-dimensional spatial organization analysis on Treg cells in tumor parenchymal regions and Treg cells in tumor interstitial regions based on the parenchymal region spatial distribution analysis results and the interstitial region spatial distribution analysis results, and then correlates the generated parenchymal region functional subtype classification results and interstitial region functional subtype classification results with the corresponding RCB scores to obtain breast cancer NAC efficacy prediction results. Thus, the present application can automatically capture and identify the spatial heterogeneity of Treg cells, establish the correlation between the spatial heterogeneity of Treg cells and breast cancer NAC efficacy, and provide reliable support for clinical treatment decisions.

Description

Technical Field

[0001] This invention relates to the field of medical information analysis and health assessment technology, and in particular to a method and system for predicting the efficacy of NAC in breast cancer based on the spatial heterogeneity of Treg cells. Background Technology

[0002] Breast cancer is one of the most common malignant tumors in women, and neoadjuvant chemotherapy (NAC) has become the standard treatment strategy for locally advanced breast cancer. Treg cells (Regulatory T cells) are key immunosuppressive cells in the tumor immune environment. In recent years, research results on the role of Treg cells in breast cancer have been contradictory: some studies suggest that higher Treg cell infiltration is associated with poorer NAC response and a worse prognosis; while other studies have found that a higher proportion of Treg cells in specific tumor regions can actually lead to better NAC response and significantly improved prognosis.

[0003] The core reason for this contradiction is that current research focuses only on the overall count or infiltration abundance of Treg cells, while ignoring the spatial heterogeneity of Treg cells within the tumor microenvironment. Different spatial heterogeneities of Treg cells can have drastically different impacts on the efficacy of NAC and patient prognosis.

[0004] Therefore, there is an urgent need for a predictive method that can automatically identify the spatial heterogeneity of Treg cells and establish its association with the efficacy of NAC in breast cancer, in order to resolve the research contradictions of Treg cells and provide reliable support for clinical treatment decisions. Summary of the Invention

[0005] The technical problem to be solved by the present invention is: the present invention provides a method and system for predicting the efficacy of NAC in breast cancer based on the spatial heterogeneity of Treg cells, which can automatically capture and identify the spatial heterogeneity of Treg cells and establish its correlation with the efficacy of NAC in breast cancer, providing reliable support for clinical treatment decisions.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a method for predicting the efficacy of NAC (non-invasive breast cancer) treatment based on the spatial heterogeneity of Treg cells, comprising: Obtain sample data and the corresponding RCB score of the sample data, perform cell identification and region segmentation on the sample data, and obtain cell identification results and region segmentation results. The cell identification results include the cell type corresponding to each cell, the cell type includes Treg cells, and the region segmentation results include the tumor parenchyma region and the tumor stroma region. Based on the cell identification results, multi-dimensional spatial distribution analysis was performed on Treg cells in the tumor parenchyma region and Treg cells in the tumor stroma region, respectively, to obtain corresponding spatial distribution analysis results for the parenchyma region and spatial distribution analysis results for the stroma region. Based on the spatial distribution analysis results for the parenchyma region, multi-dimensional spatial organization analysis was performed on Treg cells in the tumor parenchyma region, and simultaneously, based on the spatial distribution analysis results for the stroma region, multi-dimensional spatial organization analysis was performed on Treg cells in the tumor stroma region, to obtain corresponding spatial organization analysis results for the parenchyma region and spatial organization analysis results for the stroma region. The multi-dimensional spatial distribution analysis included: Treg cell density dimension, Ripley's L-function dimension, Treg cell occupancy rate dimension, and Treg cell ratio dimension. The multi-dimensional spatial organization analysis included: global aggregation dimension, local hotspot dimension, and spatial correlation dimension. Based on the spatial distribution analysis results and the spatial organization analysis results of the solid region, Treg cells in the tumor solid region are sequentially subjected to spatial clustering and hierarchical clustering to generate functional subtype classification results of the solid region. At the same time, based on the spatial distribution analysis results and the spatial organization analysis results of the mesenchymal region, Treg cells in the tumor mesenchymal region are sequentially subjected to spatial clustering and hierarchical clustering to generate functional subtype classification results of the mesenchymal region. The functional subtype classification results of the solid region and the functional subtype classification results of the mesenchymal region were correlated with the corresponding RCB scores to obtain the NAC efficacy prediction results for breast cancer.

[0007] The beneficial effects of this invention are as follows: It performs multi-dimensional spatial distribution and spatial organization analysis on Treg cells in both the tumor parenchyma and tumor stroma regions. The multi-dimensional spatial distribution analysis focuses on the distribution characteristics of Treg cells using dimensions such as Treg cell density, Ripley's L-function, Treg cell occupancy rate, and Treg cell ratio, comprehensively reflecting the spatial distribution patterns of Treg cells and overcoming the limitations of traditional methods that only describe cell quantity. The multi-dimensional spatial organization analysis focuses on the spatial organization patterns of Treg cells using dimensions such as global aggregation, local hotspots, and spatial correlation, enabling the functional localization of Treg cells within the tumor immune environment. The synergistic effect of these two analyses automatically captures and identifies comprehensive, multi-level spatial heterogeneity of Treg cells. A two-level clustering strategy combining spatial and hierarchical clustering is employed. This strategy preserves the spatial distribution characteristics of Treg cells while further refining functional differences. It enables the scientific classification of Treg cell functional subtypes, making the classification results more targeted and improving the accuracy of subsequent correlation analysis between Treg cells and RCB scores. This ensures the accuracy of the obtained NAC efficacy prediction results for breast cancer and provides reliable support for clinical treatment decisions.

[0008] Optionally, the step of performing cell identification and region segmentation on the sample data to obtain cell identification results and region segmentation results includes: The sample data is sliced ​​according to a preset thickness to obtain slice data, and the slice data is then subjected to mIF staining to obtain mIF-stained slice data. Cell detection algorithms are used to identify cells in mIF-stained slide data, resulting in cell identification results including the cell type corresponding to each cell. At the same time, the RTtree algorithm is used to segment regions in the mIF-stained slide data, resulting in region segmentation results including the tumor parenchyma region and the tumor stroma region.

[0009] As described above, by slicing the sample data with a preset thickness and then staining the sliced ​​data with mIF, the morphological and phenotypic characteristics of Treg cells can be effectively preserved, avoiding the loss of cell information during the sample data processing. At the same time, cell detection algorithms and RTtree algorithms are used to achieve cell identification and region segmentation, which effectively improves processing efficiency compared with manual identification and segmentation methods, and ensures the accuracy of the obtained tumor parenchyma and tumor stroma regions.

[0010] Optionally, the cell identification result includes the cell coordinates of each cell. The step of performing multi-dimensional spatial distribution analysis on Treg cells in the tumor parenchyma region and Treg cells in the tumor stroma region based on the cell identification result yields corresponding spatial distribution analysis results for the parenchyma region and spatial distribution analysis results for the stroma region. Based on the spatial distribution analysis results for the parenchyma region, a multi-dimensional spatial organization analysis is performed on the Treg cells in the tumor parenchyma region. Simultaneously, based on the spatial distribution analysis results for the stroma region, a multi-dimensional spatial organization analysis is performed on the Treg cells in the tumor stroma region. The resulting spatial organization analysis results for the parenchyma region and spatial organization analysis results for the stroma region include: The tumor parenchyma region and the tumor stroma region are divided into grids according to preset sizes to obtain the gridded tumor parenchyma region and the gridded tumor stroma region. Based on the cell type and cell coordinates of each cell, the number of Treg cells in each grid within the tumor parenchyma region after grid division is statistically analyzed to obtain the first Treg cell count for each grid. Simultaneously, based on the cell type and cell coordinates of each cell, the number of Treg cells in each grid within the tumor stroma region after grid division is statistically analyzed to obtain the second Treg cell count for each grid. The first and second Treg cell counts for each grid are then input into the Treg cell density formula for calculation, yielding the corresponding first and second Treg cell density results. The Treg cell density formula is as follows: ; ; in, This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. T represents the i-th grid in the meshed region R. quantity, This represents the area of ​​the i-th grid in the region R after gridding; The number of first Treg cells corresponding to all grids within the preset range is summarized to obtain the total number of first Treg cells. Simultaneously, the number of second Treg cells corresponding to all grids within the preset range is summarized to obtain the total number of second Treg cells. The total numbers of the first and second Treg cells are then input into Ripley's L function formula for calculation to obtain the corresponding first and second Ripley's L function results. The Ripley's L function formula is as follows: ; ; in, This represents the Ripley's L function result for the meshed region R within a preset range r, where r represents the preset range. This represents the total area of ​​the region R after meshing. This represents the total number of regions R after grid division. Represents the region R after meshing, the first... One Treg cell, Represents the region R after meshing, the first... One Treg cell, This indicates that the a-th Treg cell in the region R after gridding is related to the a-th Treg cell. Euclidean distance between Treg cells This indicates that the a-th Treg cell in the region R after gridding is related to the a-th Treg cell. Edge correction weights between Treg cells Indicates an indicator function; The number of grid cells containing Treg cells in the tumor parenchyma region and the number of grid cells containing Treg cells in the tumor stroma region after grid division are statistically analyzed to obtain the corresponding first grid number and second grid number. These first and second grid numbers are then input into the Treg cell occupancy rate formula for calculation to obtain the corresponding first and second Treg cell occupancy rate results. The Treg cell occupancy rate formula is as follows: ; ; in, This represents the Treg cell density result corresponding to region R after meshing. This indicates the number of grid cells contained in the region R after gridding. This represents the total number of grid cells corresponding to the region R after meshing. The number of cells of each cell type in each grid within the tumor parenchyma region after grid division is counted according to the cell type and cell coordinates of each cell, resulting in a first statistical result. At the same time, the number of cells of each cell type in each grid within the tumor stroma region after grid division is counted according to the cell type and cell coordinates of each cell, resulting in a second statistical result. The first statistical result and the second statistical result are then input into the Treg cell ratio formula for calculation to obtain the corresponding first Treg cell ratio result and second Treg cell ratio result. The formula for the Treg cell ratio is: ; ; in, This represents the Treg cell ratio in region R after gridding. T represents the i-th grid in the meshed region R. quantity, This represents the number of cells of type m in the i-th grid within the statistical results of the region R after grid division; The results of the first Treg cell density, the first Ripley's L function, the first Treg cell occupancy rate, and the first Treg cell ratio are used as the spatial distribution analysis results of the parenchymal region, and the results of the second Treg cell density, the second Ripley's L function, the second Treg cell occupancy rate, and the second Treg cell ratio are used as the spatial distribution analysis results of the interstitial region. Based on the spatial distribution analysis results of the solid region, a multi-dimensional spatial organization analysis was performed on the Treg cells in the tumor solid region. At the same time, based on the spatial distribution analysis results of the mesenchymal region, a multi-dimensional spatial organization analysis was performed on the Treg cells in the tumor mesenchymal region, resulting in the corresponding spatial organization analysis results of the solid region and the mesenchymal region.

[0011] As described above, grid division and standardized formula calculations ensure the scientific validity and rationality of the obtained spatial distribution analysis results for the tumor parenchyma and stromal regions. The Treg cell density result obtained based on the Treg cell density formula directly reflects the sparseness of Treg cell distribution in the corresponding tumor parenchyma and stromal regions. The Ripley's L-function result obtained based on the Ripley's L-function formula accurately reflects the spatial aggregation degree of Treg cells in the corresponding tumor parenchyma and stromal regions. The Treg cell occupancy rate result obtained based on the Treg cell occupancy rate formula comprehensively reflects the breadth of Treg cell distribution in the corresponding tumor parenchyma and stromal regions. The Treg cell ratio result obtained based on the Treg cell proportion formula clearly reflects the relative proportion of Treg cells in the corresponding tumor parenchyma and stromal regions. This allows for a comprehensive capture of the spatial distribution characteristics of Treg cells in the corresponding tumor parenchyma and stromal regions, providing a high-quality analytical foundation for subsequent multi-dimensional spatial tissue analysis.

[0012] Optionally, the multi-dimensional spatial distribution analysis further includes: Shannon entropy quantification dimension, Treg cell radial frequency dimension, and nearest neighbor distance dimension. The step of using the first Treg cell density result, the first Ripley's L function result, the first Treg cell occupancy rate result, and the first Treg cell ratio result as the spatial distribution analysis result of the parenchymal region, and using the second Treg cell density result, the second Ripley's L function result, the second Treg cell occupancy rate result, and the second Treg cell ratio result as the spatial distribution analysis result of the mesenchymal region includes: The first statistical result and the second statistical result are respectively input into the Shannon entropy quantization formula for calculation to obtain the corresponding first Shannon entropy quantization result and second Shannon entropy quantization result. The Shannon entropy quantization formula is as follows: ; ; in, This represents the Shannon entropy quantization result corresponding to the region R after gridding. This represents the total number of all cell types in the region R after meshing. This represents the number of cells of type m in the i-th grid within the corresponding statistical results of the gridded region R. This represents the total number of cells of all cell types in the i-th grid within the gridded region R. Based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor parenchyma region as the center, the number of Treg cells within a preset radius is counted to obtain a fifth number for each Treg cell. Simultaneously, based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor stroma region as the center, the number of Treg cells within a preset radius is counted to obtain a sixth number for each Treg cell. The fifth and sixth numbers are then input into the Treg cell radial frequency formula for calculation to obtain the corresponding first and second Treg cell radial frequency results. The Treg cell radial frequency formula is: ; ; in, This represents the Treg cell ratio in region R after gridding. This represents the number of Treg cells within a preset radius L, centered on the a-th Treg cell, in the region R after gridding. Indicates the preset radius range. This represents the total number of all cell types within a preset radius centered on the a-th Treg cell in the region R after grid division. Based on the cell type and cell coordinates of each cell, and using each Treg cell in the gridded tumor stroma region as the center, the KNN algorithm is used to select the first nearest neighbor Treg cell for each Treg cell. The first nearest neighbor Treg cell and its corresponding Treg cell are then input into the nearest neighbor distance formula for calculation, yielding the first nearest neighbor distance result for each Treg cell. Simultaneously, based on the cell type and cell coordinates of each cell, and using each Treg cell in the gridded tumor stroma region as the center, the KNN algorithm is used to select the second nearest neighbor Treg cell for each Treg cell. The second nearest neighbor Treg cell and its corresponding Treg cell are then input into the nearest neighbor distance formula for calculation, yielding the second nearest neighbor distance result for each Treg cell. The nearest neighbor distance formula is as follows: ; ; in, This represents the nearest neighbor distance result for the a-th Treg cell in the gridded region R. This represents the x-coordinate of the a-th Treg cell in the meshed region R. Let x represent the x-coordinate of the nearest neighbor Treg cell c to the a-th Treg cell. This represents the ordinate of the a-th Treg cell in the meshed region R. The ordinate represents the nearest neighbor Treg cell c to the a-th Treg cell. This represents the c-th Treg cell in the gridded region R. The results of the first Treg cell density, the first Ripley's L function, the first Treg cell occupancy rate, the first Treg cell ratio, the first Shannon entropy quantification, the first Treg cell radial frequency, and the first nearest neighbor distance of each Treg cell are used as the spatial distribution analysis results of the parenchymal region. The results of the second Treg cell density, the second Ripley's L function, the second Treg cell occupancy rate, the second Treg cell ratio, the second Shannon entropy quantification, the second Treg cell radial frequency, and the second nearest neighbor distance of each Treg cell are used as the spatial distribution analysis results of the mesenchymal region.

[0013] As described above, incorporating Shannon entropy quantification, Treg cell radial frequency, and nearest neighbor distance dimensions into multi-dimensional spatial distribution analysis allows for several key improvements. First, the Shannon entropy quantification result, based on the formula, reflects the mixed distribution characteristics of Treg cells with other cell types in the corresponding tumor parenchyma and stromal regions. Second, the Treg cell radial frequency result reflects the local aggregation characteristics of Treg cells in these regions. Third, the nearest neighbor distance result reflects the spatial proximity relationships between individual Treg cells within the tumor parenchyma and stromal regions. This enables a more comprehensive and refined capture of the spatial heterogeneity of Treg cells.

[0014] Optionally, the step of performing multi-dimensional spatial organization analysis on Treg cells in the tumor parenchyma region based on the spatial distribution analysis results of the parenchyma region, and simultaneously performing multi-dimensional spatial organization analysis on Treg cells in the tumor stroma region based on the spatial distribution analysis results of the stroma region, to obtain the corresponding spatial organization analysis results of the parenchyma region and the spatial organization analysis results of the stroma region, includes: The first Treg cell density result from the spatial distribution analysis of the mesenchymal region is input into the Moran's I index formula for calculation to obtain the first global aggregation result. Simultaneously, the second Treg cell density result from the spatial distribution analysis of the mesenchymal region is input into the Moran's I index formula for calculation to obtain the second global aggregation result. The Moran's I index formula is: ; ; in, This represents the first global aggregation result corresponding to region R after meshing. This represents the total number of grid cells corresponding to the region R after meshing. This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. This represents the average Treg cell density result across all grids in the meshed region R. This represents the i-th grid cell in the meshed region R. This represents the j-th grid cell in the meshed region R. This represents the spatial weight matrix between the i-th and j-th grids in the meshed region R. The first Treg cell density result from the spatial distribution analysis of the interstitial region is input into the local hotspot formula for calculation to obtain the first local hotspot result. Simultaneously, the second Treg cell density result from the spatial distribution analysis of the interstitial region is input into the local hotspot formula for calculation to obtain the second local hotspot result. The local hotspot formula is as follows: ; ; in, This represents the first local hotspot result corresponding to region R after meshing, where h represents the preset distance threshold. This represents the total number of grid cells corresponding to the region R after meshing. This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. This represents the average Treg cell density result across all grids in the meshed region R. This represents the i-th grid in the region R after gridding. This represents the j-th grid cell in the meshed region R. This represents the spatial weight matrix between the i-th and j-th grids in the gridded region R, based on a preset distance threshold h. The first Treg cell ratio from the spatial distribution analysis of the solid region is input into a spatial correlation formula for calculation to obtain the first spatial correlation result. Simultaneously, the second Treg cell ratio from the spatial distribution analysis of the mesenchymal region is input into a spatial correlation formula for calculation to obtain the second spatial correlation result. The spatial correlation formula is as follows: ; ; in, This represents the first spatial association result corresponding to the region R after meshing. This indicates that within the same grid cell in the region R after meshing, there are simultaneously... The ratio of the number of grid cells of cell type m to the total number of grid cells. This represents the ratio of the number of grid cells containing Treg cells in the meshed region R to the total number of grid cells. This represents the ratio of the number of grid cells containing cell type m in the meshed region R to the total number of grid cells. The first global aggregation result, the first local hotspot result, and the first spatial correlation result are used as the spatial organization analysis result of the substance region, and the second global aggregation result, the second local hotspot result, and the second spatial correlation result are used as the spatial organization analysis result of the interstitial region.

[0015] As described above, the global aggregation results obtained based on Moran's I index formula can reflect the overall aggregation trend of Treg cells in the corresponding tumor parenchyma and tumor stroma regions. The local hotspot results obtained based on the local hotspot formula can reflect the local aggregation trend of Treg cells in the corresponding tumor parenchyma and tumor stroma regions. The spatial correlation results obtained based on the spatial correlation formula can reflect the spatial interaction relationship between Treg cells and other cell types in the corresponding tumor parenchyma and tumor stroma regions. The synergistic effect of the three can comprehensively analyze the spatial organization pattern of Treg cells in the corresponding tumor parenchyma and tumor stroma regions, and improve the accuracy and comprehensiveness of the corresponding spatial organization analysis results.

[0016] Optionally, the cell identification result includes the cell coordinates of each cell. The step of sequentially performing spatial clustering and hierarchical clustering on the Treg cells in the tumor parenchyma region based on the spatial distribution analysis results and the spatial organization analysis results of the parenchyma region to generate a functional subtype classification result for the parenchyma region, and simultaneously performing spatial clustering and hierarchical clustering on the Treg cells in the tumor stroma region based on the spatial distribution analysis results and the spatial organization analysis results of the stroma region to generate a functional subtype classification result for the stroma region, includes: The K-means unsupervised clustering algorithm was used to spatially cluster the cell coordinates of Treg cells in the tumor parenchyma region to obtain K first clusters. At the same time, the K-means unsupervised clustering algorithm was used to spatially cluster the cell coordinates of Treg cells in the tumor stroma region to obtain K second clusters. Based on the spatial distribution analysis results of the solid region and the spatial organization analysis results of the solid region, a first spatial feature vector is constructed for each first cluster to obtain a set of all first spatial feature vectors. At the same time, based on the spatial distribution analysis results of the interstitial region and the spatial organization analysis results of the interstitial region, a second spatial feature vector is constructed for each second cluster to obtain a set of all second spatial feature vectors. Hierarchical clustering algorithm is used to perform hierarchical clustering on all first spatial feature vector sets to generate functional subtype classification results for the solid region. At the same time, hierarchical clustering algorithm is used to perform hierarchical clustering on all second spatial feature vector sets to generate functional subtype classification results for the interstitial region.

[0017] As described above, spatial clustering of Treg cells is first achieved using the K-means unsupervised clustering algorithm based on the cell coordinates of Treg cells. Then, spatial feature vectors are constructed by combining the spatial distribution analysis results and spatial organization analysis results. Hierarchical clustering is then performed using a hierarchical clustering algorithm, which not only achieves accurate classification of functional subtypes in the parenchymal region and functional subtypes in the mesenchymal region, but also ensures that the classification results conform to the spatial distribution pattern.

[0018] Optionally, the step of performing correlation analysis between the functional subtype classification results of the solid region and the functional subtype classification results of the stromal region and the corresponding RCB scores to obtain the NAC efficacy prediction results for breast cancer includes: The composition ratio of functional subtypes in the solid region is obtained based on the classification results of functional subtypes in the solid region, and the composition ratio of functional subtypes in the interstitial region is obtained based on the classification results of functional subtypes in the interstitial region. Based on the RCB scores of the sample data, the sample data is divided into a pathological complete remission group and a pathological incomplete remission group. The pathological complete remission group is correlated with the corresponding composition ratio of functional subtypes in the parenchyma area and the corresponding composition ratio of functional subtypes in the mesenchyma area to obtain a first complete parenchyma correlation result and a first complete mesenchyma correlation result. At the same time, the pathological incomplete remission group is correlated with the corresponding composition ratio of functional subtypes in the parenchyma area and the corresponding composition ratio of functional subtypes in the mesenchyma area to obtain a first incomplete parenchyma correlation result and a first incomplete mesenchyma correlation result. A pre-defined testing method is used to perform a difference analysis between the first substantially complete association result and the first substantially incomplete association result to obtain a first difference analysis result. At the same time, a difference analysis is performed between the first interstitial completely association result and the first interstitial incomplete association result to obtain a second difference analysis result. Based on the results of the first and second difference analyses, the composition ratio of key functional subtypes is obtained. The key functional subtypes are then input into a breast cancer NAC efficacy prediction model constructed based on the Logistic regression algorithm to obtain the breast cancer NAC efficacy prediction results.

[0019] As described above, by correlating the pathological complete remission group and the pathological incomplete remission group, which are divided based on the RCB score, with the corresponding proportions of functional subtypes in the parenchymal region and the stromal region, and performing differential analysis using pre-set test methods, the proportions of key functional subtypes can be accurately identified, interference from irrelevant functional subtype proportions can be eliminated, and the accuracy of predicting the efficacy of NAC in breast cancer can be improved.

[0020] In a second aspect, the present invention provides a breast cancer NAC efficacy prediction system based on Treg cell spatial heterogeneity, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the breast cancer NAC efficacy prediction method based on Treg cell spatial heterogeneity described in the first aspect.

[0021] The technical effects of the breast cancer NAC efficacy prediction system based on Treg cell spatial heterogeneity provided in the second aspect are the same as those of the breast cancer NAC efficacy prediction method based on Treg cell spatial heterogeneity provided in the first aspect. Attached Figure Description

[0022] Figure 1 This is a flowchart of the breast cancer NAC efficacy prediction method based on Treg cell spatial heterogeneity provided in this embodiment; Figure 2 This is a schematic diagram of the overall process of the breast cancer NAC efficacy prediction method based on Treg cell spatial heterogeneity provided in this embodiment; Figure 3 This embodiment provides a breast cancer NAC efficacy prediction system based on Treg cell spatial heterogeneity.

[0023] [Explanation of Labels in the Attached Image] 1. A breast cancer NAC efficacy prediction system based on Treg cell spatial heterogeneity; 2. Processor; 3. Memory. Detailed Implementation

[0024] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0025] Example 1 Please refer to Figures 1 to 2This invention provides a method for reconstructing a 3D scene based on 3D Gaussian splashing, comprising the following steps: S1. Obtain sample data and the RCB score corresponding to the sample data, perform cell identification and region segmentation on the sample data, and obtain cell identification results and region segmentation results. The cell identification results include the cell type corresponding to each cell, the cell type includes Treg cells, and the region segmentation results include the tumor parenchyma region and the tumor stroma region. In this embodiment, 153 patients who were diagnosed with non-specific invasive breast cancer by histopathology and received neoadjuvant chemotherapy between 2014 and 2022 were included in the study. Their clinicopathological data and postoperative follow-up data were retrospectively analyzed to obtain the sample data. The selection and exclusion criteria for the sample data are as follows: Screening criteria: (1) At the initial diagnosis, the patient was confirmed to have non-specific invasive breast cancer by core needle biopsy; (2) Received a standard neoadjuvant chemotherapy regimen before surgery; (3) After neoadjuvant chemotherapy, the patient underwent modified radical mastectomy for breast cancer at our hospital; (4) Complete clinical and follow-up data.

[0026] Exclusion criteria: (1) Clinical pathology data is missing or lost to follow-up; (2) Diagnosed with other types of malignant tumors, or the biopsy specimen only shows carcinoma in situ; (3) No tumor residue was found in the paraffin-embedded tissue specimen; (4) Cases with pathological diagnosis of Luminal type A breast cancer.

[0027] like Figure 2As shown, sample data and corresponding RCB scores were obtained, where the RCB score is actually an evaluation of the efficacy after NAC (Non-Anaerobic Cerebral Angiogenesis). The RCB score is calculated based on five pathological features: the extent of residual tumor in the primary breast lesion, the density of residual tumor cells, the proportion of carcinoma in situ, the number of positive lymph nodes, and the maximum diameter of residual metastatic lymph nodes. Different efficacy grades exist for the RCB score: RCB-0: indicates complete pathological remission, i.e., no residual tumor cells after NAC treatment; RCB-1: indicates a small amount of residual tumor, indicating a good NAC treatment effect; RCB-2: indicates moderate residual tumor, indicating a moderate treatment effect; RCB-3: indicates extensive residual tumor, indicating a poor treatment effect. Cell identification and region segmentation were performed on the sample data to obtain cell identification results and region segmentation results. The cell identification results include the cell type and cell coordinates for each cell. The cell types include seven immune / tumor cell subpopulations: Treg cells, circulating tumor cells (CTLs), exhausted CD8+ T cells, PD-L1+ tumor cells, tumor cells, PD-L1+ macrophages, macrophages, and other PD-L1+ cells. The region segmentation results include the tumor parenchyma and the tumor stroma.

[0028] At this point, the cell identification and region segmentation of the sample data described in step S1, to obtain the cell identification results and region segmentation results, includes: S11. The sample data is sliced ​​according to a preset thickness to obtain slice data. The slice data is then subjected to mIF staining to obtain mIF-stained slice data. S12. Cell identification is performed on the mIF-stained slide data using a cell detection algorithm to obtain cell identification results including the cell type corresponding to each cell. At the same time, the RTtree algorithm is used to perform region segmentation on the mIF-stained slide data to obtain region segmentation results including the tumor parenchyma region and the tumor stroma region.

[0029] In this embodiment, as Figure 2 As shown, the sample data was sliced ​​according to a preset thickness of 4 μm, which can be adjusted according to actual conditions. After the slices were adhered to a glass slide and baked in a 65°C oven for 4 hours, they were then subjected to mlF staining to obtain the mIF-stained slice data. The mlF staining process is as follows: After dewaxing with xylene for 3 × 10 minutes and hydration with graded ethanol, the slide data were subjected to high-temperature heat repair in pH 9.0 EDTA repair solution for 20 minutes; endogenous peroxidase was blocked with 3% H2O2 (room temperature for 10 minutes), washed with TBS, and blocked with BSA or normal goat serum for 30 minutes. A panel containing 5-color antibodies including CKpan, CD8, FOXP3, CD68, and PD-L1 was used for staining in successive rounds using TSA multiplex fluorescence synchronous staining technology.

[0030] Cell detection algorithms are used to identify cells in mIF-stained slide data, resulting in cell identification results including the cell type and cell coordinates of each cell. After pathologists manually annotate typical tumor and stroma regions, the RTtree algorithm performs region segmentation on the mIF-stained slide data, obtaining region segmentation results including tumor parenchyma and tumor stroma regions.

[0031] S2. Based on the cell identification results, perform multi-dimensional spatial distribution analysis on the Treg cells in the tumor parenchyma region and the Treg cells in the tumor stroma region respectively to obtain corresponding spatial distribution analysis results for the parenchyma region and spatial distribution analysis results for the stroma region. Based on the spatial distribution analysis results for the parenchyma region, perform multi-dimensional spatial organization analysis on the Treg cells in the tumor parenchyma region, and simultaneously perform multi-dimensional spatial organization analysis on the Treg cells in the tumor stroma region based on the spatial distribution analysis results for the stroma region, to obtain corresponding spatial organization analysis results for the parenchyma region and spatial organization analysis results for the stroma region. The multi-dimensional spatial distribution analysis includes: Treg cell density dimension, Ripley's L-function dimension, Treg cell occupancy rate dimension, and Treg cell ratio dimension. The multi-dimensional spatial organization analysis includes: global aggregation dimension, local hotspot dimension, and spatial correlation dimension. At this point, step S2 includes: S21. Divide the tumor parenchyma region and the tumor stroma region into grids according to preset sizes to obtain the tumor parenchyma region and the tumor stroma region after grid division. S22. Based on the cell type and cell coordinates of each cell, the number of Treg cells in each grid of the tumor parenchyma region after grid division is statistically analyzed to obtain the first Treg cell count for each grid. Simultaneously, based on the cell type and cell coordinates of each cell, the number of Treg cells in each grid of the tumor stroma region after grid division is statistically analyzed to obtain the second Treg cell count for each grid. The first and second Treg cell counts for each grid are then input into the Treg cell density formula for calculation to obtain the corresponding first and second Treg cell density results. The Treg cell density formula is: ; ; in, This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. T represents the i-th grid in the meshed region R. quantity, This represents the area of ​​the i-th grid in the region R after gridding; S23. Summarize the number of first Treg cells corresponding to all grids within the preset range to obtain the total number of first Treg cells. Simultaneously, summarize the number of second Treg cells for all grids within the preset range to obtain the total number of second Treg cells. Input the total number of first Treg cells and the total number of second Treg cells into Ripley's L function formula for calculation to obtain the corresponding first Ripley's L function result and second Ripley's L function result. The Ripley's L function formula is: ; ; in, This represents the Ripley's L function result for the meshed region R within a preset range r, where r represents the preset range. This represents the total area of ​​the region R after meshing. This represents the total number of regions R after grid division. Represents the region R after meshing, the first... One Treg cell, Represents the region R after meshing, the first... One Treg cell, This indicates that the a-th Treg cell in the region R after gridding is related to the a-th Treg cell. Euclidean distance between Treg cells This indicates that the a-th Treg cell in the region R after gridding is related to the a-th Treg cell. Edge correction weights between Treg cells Indicates an indicator function; S24. Count the number of grid cells containing Treg cells in the tumor parenchyma region and the number of grid cells containing Treg cells in the tumor stroma region after grid division to obtain the corresponding first grid number and second grid number. Input the first grid number and second grid number into the Treg cell occupancy rate formula to calculate the corresponding first Treg cell occupancy rate and second Treg cell occupancy rate results. The Treg cell occupancy rate formula is: ; ; in, This represents the Treg cell density result corresponding to region R after meshing. This indicates the number of grid cells contained in the region R after gridding. This represents the total number of grid cells corresponding to the region R after meshing. S25. Based on the cell type and cell coordinates of each cell, the number of cells corresponding to each cell type in each grid in the tumor parenchyma region after grid division is counted to obtain the first statistical result. At the same time, based on the cell type and cell coordinates of each cell, the number of cells corresponding to each cell type in each grid in the tumor stroma region after grid division is counted to obtain the second statistical result. The first statistical result and the second statistical result are respectively input into the Treg cell ratio formula for calculation to obtain the corresponding first Treg cell ratio result and second Treg cell ratio result. The formula for the Treg cell ratio is: ; ; in, This represents the Treg cell ratio in region R after gridding. T represents the i-th grid in the meshed region R. quantity, This represents the number of cells of type m in the i-th grid within the statistical results of the region R after grid division; S26. The results of the first Treg cell density, the first Ripley's L function, the first Treg cell occupancy rate, and the first Treg cell ratio are used as the spatial distribution analysis results of the parenchymal region, and the results of the second Treg cell density, the second Ripley's L function, the second Treg cell occupancy rate, and the second Treg cell ratio are used as the spatial distribution analysis results of the interstitial region. In this embodiment, the tumor parenchyma and tumor stroma regions are divided into grids according to preset sizes, including 25µm, 50µm, 75µm, and 100µm square grids, resulting in gridded tumor parenchyma and tumor stroma regions. The number of Treg cells in each grid within the divided tumor parenchyma region is counted based on the cell type and cell coordinates of each cell, yielding the first Treg cell count for each grid. Similarly, the number of Treg cells in each grid within the divided tumor stroma region is counted based on the cell type and cell coordinates of each cell, yielding the second Treg cell count for each grid. The first and second Treg cell counts for each grid are then input into the Treg cell density formula for calculation, resulting in first and second Treg cell density results that reflect the sparse distribution of Treg cells in the corresponding tumor parenchyma and tumor stroma regions.

[0032] The number of first Treg cells corresponding to all grids within a preset range is summarized, that is, the number of all Treg cells within the preset range in the tumor parenchyma region after grid division is summarized to obtain the total number of first Treg cells. Similarly, the number of second Treg cells corresponding to all grids within a preset range is summarized, that is, the number of all Treg cells within the preset range in the tumor stroma region after grid division is summarized to obtain the total number of second Treg cells. The total number of first Treg cells and the total number of second Treg cells are respectively input into Ripley's L function formula for calculation to obtain the first Ripley's L function result and the second Ripley's L function result, which can reflect the degree of spatial aggregation of Treg cells in the corresponding tumor parenchyma region and tumor stroma region. The preset range in Ripley's L function formula is [0,200]um, which can be adjusted according to the actual situation. When the first Ripley's L function result or the second Ripley's L function result is 0, it means that the Treg cells in the corresponding segmented region conform to complete spatial randomness, with no aggregated state and no discrete state, and present a random state. When the result of the first Ripley's L function or the result of the second Ripley's L function is greater than 0, it indicates that the Treg cells are aggregated; when the result of the first Ripley's L function or the result of the second Ripley's L function is less than 0, it indicates that the Treg cells are discretely distributed.

[0033] The number of Treg cells in the tumor parenchyma region and the number of Treg cells in the tumor stroma region after gridding are statistically analyzed. These first and second grid counts are then input into the Treg cell occupancy rate formula for calculation, yielding the first and second Treg cell occupancy rates, which reflect the distribution breadth of Treg cells in the corresponding tumor parenchyma and tumor stroma regions. The first statistical result is obtained by counting the number of cells of each cell type within each grid in the tumor parenchyma region after gridding, based on the cell type and cell coordinates of each cell. Similarly, the second statistical result is obtained by counting the number of cells of each cell type within each grid in the tumor stroma region after gridding. These first and second statistical results are then input into the Treg cell ratio formula for calculation, yielding the first and second Treg cell ratio results, which reflect the relative proportion of Treg cells in the corresponding tumor parenchyma and tumor stroma regions.

[0034] At this point, step S26 includes: S261. Input the first statistical result and the second statistical result into the Shannon entropy quantization formula respectively for calculation to obtain the corresponding first Shannon entropy quantization result and second Shannon entropy quantization result. The Shannon entropy quantization formula is as follows: ; ; in, This represents the Shannon entropy quantization result corresponding to the region R after gridding. This represents the total number of all cell types in the region R after meshing. This represents the number of cells of type m in the i-th grid within the corresponding statistical results of the gridded region R. This represents the total number of cells of all cell types in the i-th grid within the gridded region R. S262. Based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor stroma region as the center, count the number of Treg cells within a preset radius to obtain a fifth number for each Treg cell. Simultaneously, based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor stroma region as the center, count the number of Treg cells within a preset radius to obtain a sixth number for each Treg cell. Input the fifth and sixth numbers into the Treg cell radial frequency formula for calculation to obtain the corresponding first and second Treg cell radial frequency results. The Treg cell radial frequency formula is: ; ; in, This represents the Treg cell ratio in region R after gridding. This represents the number of Treg cells within a preset radius L, centered on the a-th Treg cell, in the region R after gridding. Indicates the preset radius range. This represents the total number of all cell types within a preset radius centered on the a-th Treg cell in the region R after grid division. S263. Based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor stroma region as the center, the KNN algorithm is used to select the first nearest neighbor Treg cell to each Treg cell. The first nearest neighbor Treg cell and its corresponding Treg cell are then input into the nearest neighbor distance formula for calculation to obtain the first nearest neighbor distance result for each Treg cell. Simultaneously, based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor stroma region as the center, the KNN algorithm is used to select the second nearest neighbor Treg cell to each Treg cell. The second nearest neighbor Treg cell and its corresponding Treg cell are then input into the nearest neighbor distance formula for calculation to obtain the second nearest neighbor distance result for each Treg cell. The nearest neighbor distance formula is: ; ; in, This represents the nearest neighbor distance result for the a-th Treg cell in the gridded region R. This represents the x-coordinate of the a-th Treg cell in the meshed region R. Let x represent the x-coordinate of the nearest neighbor Treg cell c to the a-th Treg cell. This represents the ordinate of the a-th Treg cell in the meshed region R. The ordinate represents the nearest neighbor Treg cell c to the a-th Treg cell. This represents the c-th Treg cell in the gridded region R. S264. The results of the first Treg cell density, the first Ripley's L function, the first Treg cell occupancy rate, the first Treg cell ratio, the first Shannon entropy quantification, the first Treg cell radial frequency, and the first nearest neighbor distance of each Treg cell are taken as the spatial distribution analysis results of the parenchymal region. The results of the second Treg cell density, the second Ripley's L function, the second Treg cell occupancy rate, the second Treg cell ratio, the second Shannon entropy quantification, the second Treg cell radial frequency, and the second nearest neighbor distance of each Treg cell are taken as the spatial distribution analysis results of the mesenchymal region.

[0035] In this embodiment, the multi-dimensional spatial distribution analysis also includes: Shannon entropy quantification dimension, Treg cell radial frequency dimension, and nearest neighbor distance dimension. The first and second statistical results are respectively input into the Shannon entropy quantification formula for calculation to obtain the first and second Shannon entropy quantification results, which can reflect the mixed distribution characteristics of Treg cells with other cell types in the corresponding tumor parenchyma and tumor stroma regions.

[0036] Based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor parenchyma region as the center, the number of Treg cells within a preset radius is counted, resulting in a fifth number for each Treg cell. Similarly, taking each Treg cell in the gridded tumor stroma region as the center, the number of Treg cells within a preset radius is counted, resulting in a sixth number for each Treg cell. The fifth and sixth numbers are then input into the Treg cell radial frequency formula for calculation, yielding the first and second Treg cell radial frequency results, respectively reflecting the local aggregation characteristics of Treg cells in the corresponding tumor parenchyma and tumor stroma regions.

[0037] Based on the cell type and cell coordinates of each cell, and using each Treg cell in the gridded tumor parenchyma region as the center, the KNN algorithm is used to select the first nearest neighbor Treg cell. Similarly, using each Treg cell in the gridded tumor stroma region as the center, the KNN algorithm is used to select the second nearest neighbor Treg cell. The first nearest neighbor Treg cell and its corresponding Treg cell are input into the nearest neighbor distance formula for calculation. Similarly, the second nearest neighbor Treg cell and its corresponding Treg cell are input into the nearest neighbor distance formula for calculation. This yields the first and second nearest neighbor distance results, which reflect the spatial proximity relationship of each Treg cell in the corresponding tumor parenchyma and tumor stroma regions.

[0038] The results of the spatial distribution analysis of the parenchymal region were obtained by using the following parameters: first Treg cell density, first Ripley's L-function result, first Treg cell occupancy rate, first Treg cell ratio, first Shannon entropy quantification result, first Treg cell radial frequency result, and the first nearest neighbor distance result for each Treg cell. The results of the spatial distribution analysis of the interstitial region were obtained by using the following parameters: second Treg cell density, second Ripley's L-function result, second Treg cell occupancy rate, second Treg cell ratio, second Shannon entropy quantification result, second Treg cell radial frequency result, and the second nearest neighbor distance result for each Treg cell.

[0039] S27. Based on the spatial distribution analysis results of the solid region, perform multi-dimensional spatial organization analysis on the Treg cells in the tumor solid region, and simultaneously perform multi-dimensional spatial organization analysis on the Treg cells in the tumor stroma region based on the spatial distribution analysis results of the stroma region, to obtain the corresponding spatial organization analysis results of the solid region and the stroma region.

[0040] At this point, step S27 includes: S271. The first Treg cell density result from the spatial distribution analysis of the mesenchymal region is input into the Moran's I index formula for calculation to obtain the first global aggregation result. Simultaneously, the second Treg cell density result from the spatial distribution analysis of the mesenchymal region is input into the Moran's I index formula for calculation to obtain the second global aggregation result. The Moran's I index formula is: ; ; in, This represents the first global aggregation result corresponding to region R after meshing. This represents the total number of grid cells corresponding to the region R after meshing. This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. This represents the average Treg cell density result across all grids in the meshed region R. This represents the i-th grid in the region R after gridding. This represents the j-th grid cell in the meshed region R. This represents the spatial weight matrix between the i-th and j-th grids in the meshed region R. S272. The first Treg cell density result from the spatial distribution analysis of the interstitial region is input into the local hotspot formula for calculation to obtain the first local hotspot result. Simultaneously, the second Treg cell density result from the spatial distribution analysis of the interstitial region is input into the local hotspot formula for calculation to obtain the second local hotspot result. The local hotspot formula is: ; ; in, This represents the first local hotspot result corresponding to region R after meshing, where h represents the preset distance threshold. This represents the total number of grid cells corresponding to the region R after meshing. This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. This represents the average Treg cell density result across all grids in the meshed region R. This represents the i-th grid in the region R after gridding. This represents the j-th grid cell in the meshed region R. This represents the spatial weight matrix between the i-th and j-th grids in the gridded region R, based on a preset distance threshold h. S273. The first Treg cell ratio result from the spatial distribution analysis of the parenchymal region is input into the spatial correlation formula for calculation to obtain the first spatial correlation result. Simultaneously, the second Treg cell ratio result from the spatial distribution analysis of the mesenchymal region is input into the spatial correlation formula for calculation to obtain the second spatial correlation result. The spatial correlation formula is: ; ; in, This represents the first spatial association result corresponding to the region R after meshing. This indicates that within the same grid cell in the region R after meshing, there are simultaneously... The ratio of the number of grid cells of cell type m to the total number of grid cells. This represents the ratio of the number of grid cells containing Treg cells in the meshed region R to the total number of grid cells. This represents the ratio of the number of grid cells containing cell type m in the meshed region R to the total number of grid cells. S274. The first global aggregation result, the first local hotspot result, and the first spatial correlation result are taken as the spatial organization analysis result of the substance region, and the second global aggregation result, the second local hotspot result, and the second spatial correlation result are taken as the spatial organization analysis result of the interstitial region.

[0041] In this embodiment, the first Treg cell density result from the spatial distribution analysis of the solid region and the second Treg cell density result from the spatial distribution analysis of the mesenchymal region are respectively input into Moran's I index formula for calculation, obtaining a first global clustering result and a second global clustering result that can reflect the overall aggregation trend of Treg cells in the corresponding tumor solid region and tumor mesenchymal region. Specifically, when the first global clustering result > E(first global clustering result) or the second global clustering result > E(second global clustering result), it indicates that Treg cells exhibit an overall aggregated distribution; when the first global clustering result < E(first global clustering result) or the second global clustering result < E(second global clustering result), it indicates that Treg cells exhibit an overall discrete distribution. The first Treg cell density result from the spatial distribution analysis of the solid region and the second Treg cell density result from the spatial distribution analysis of the mesenchymal region are respectively input into local hotspot formula for calculation, obtaining a first local hotspot result and a second local hotspot result that can reflect the local aggregation trend of Treg cells in the corresponding tumor solid region and tumor mesenchymal region. Specifically, when the first local hotspot result is >1.96 and the first Treg cell density result is <0.05, or the second local hotspot result is >1.96 and the second Treg cell density result is <0.05, it indicates that Treg cells exhibit local aggregation, i.e., significant hotspots; when the first local hotspot result is <-1.96, or the second local hotspot result is <-1.96, it indicates that Treg cells exhibit local discrete distribution, i.e., significant cold spots. The first Treg cell ratio result from the spatial distribution analysis of the solid region and the second Treg cell ratio result from the spatial distribution analysis of the mesenchymal region are respectively input into the spatial correlation formula for calculation to obtain the first spatial correlation result and the second spatial correlation result, which can reflect the spatial organization pattern of Treg cells in the corresponding tumor solid region and tumor mesenchymal region. Specifically, when the first spatial association result > 0 or the second spatial association result > 0, it indicates that Treg cells are positively correlated with other corresponding cell types and colocalize; when the first spatial association result = 0 or the second spatial association result = 0, it indicates that Treg cells are independent; when the first spatial association result < 0 or the second spatial association result < 0, it indicates that Treg cells are mutually exclusive with other corresponding cell types.

[0042] S3. Based on the spatial distribution analysis results and the spatial organization analysis results of the solid region, the Treg cells in the tumor solid region are sequentially subjected to spatial clustering and hierarchical clustering to generate functional subtype classification results of the solid region. At the same time, based on the spatial distribution analysis results and the spatial organization analysis results of the mesenchymal region, the Treg cells in the tumor mesenchymal region are sequentially subjected to spatial clustering and hierarchical clustering to generate functional subtype classification results of the mesenchymal region. At this point, step S3 includes: S31. The K-means unsupervised clustering algorithm is used to spatially cluster the cell coordinates of Treg cells in the tumor parenchyma region to obtain K first clusters. At the same time, the K-means unsupervised clustering algorithm is used to spatially cluster the cell coordinates of Treg cells in the tumor stroma region to obtain K second clusters. S32. Based on the spatial distribution analysis results of the solid region and the spatial organization analysis results of the solid region, construct a first spatial feature vector for each first cluster to obtain a set of all first spatial feature vectors. At the same time, based on the spatial distribution analysis results of the interstitial region and the spatial organization analysis results of the interstitial region, construct a second spatial feature vector for each second cluster to obtain a set of all second spatial feature vectors. S33. Hierarchical clustering algorithm is used to perform hierarchical clustering on all first space feature vector sets to generate functional subtype classification results for the solid region. At the same time, hierarchical clustering algorithm is used to perform hierarchical clustering on all second space feature vector sets to generate functional subtype classification results for the interstitial region.

[0043] In this embodiment, as Figure 2 As shown, the K-means unsupervised clustering algorithm is used to spatially cluster the cell coordinates of Treg cells in the tumor parenchyma and the tumor stroma, respectively, resulting in K first clusters and K second clusters. Based on the spatial distribution and organization analysis results of the parenchyma, a first spatial feature vector is constructed for each first cluster, and a second spatial feature vector is constructed for each second cluster, resulting in a set of all first and second spatial feature vectors. A hierarchical clustering algorithm is then used to perform hierarchical clustering on all first and second spatial feature vector sets, generating corresponding functional subtype classification results for the parenchyma and stroma.

[0044] S4. The functional subtype classification results of the solid region and the functional subtype classification results of the mesenchymal region are respectively correlated with the corresponding RCB scores to obtain the NAC efficacy prediction results for breast cancer.

[0045] At this point, step S4 includes: S41. Obtain the composition ratio of functional subtypes in the solid region based on the classification results of functional subtypes in the solid region, and simultaneously obtain the composition ratio of functional subtypes in the interstitial region based on the classification results of functional subtypes in the interstitial region. S42. Based on the RCB score of the sample data, the sample data is divided into a pathological complete remission group and a pathological incomplete remission group. The pathological complete remission group is correlated with the corresponding functional subtype composition ratio of the parenchyma area and the corresponding functional subtype composition ratio of the mesenchymal area to obtain a first parenchyma complete correlation result and a first mesenchymal complete correlation result. At the same time, the pathological incomplete remission group is correlated with the corresponding functional subtype composition ratio of the parenchyma area and the corresponding functional subtype composition ratio of the mesenchymal area to obtain a first parenchyma incomplete correlation result and a first mesenchymal incomplete correlation result. S43. Using a preset testing method, perform a difference analysis on the first substantially fully associated result and the first substantially incompletely associated result to obtain a first difference analysis result. At the same time, perform a difference analysis on the first interstitial fully associated result and the first interstitial incompletely associated result to obtain a second difference analysis result. S44. Based on the results of the first and second difference analyses, the composition ratio of key functional subtypes is obtained. The key functional subtypes are then input into a breast cancer NAC efficacy prediction model constructed based on the Logistic regression algorithm to obtain the breast cancer NAC efficacy prediction results.

[0046] In this embodiment, as Figure 2 As shown, based on the classification results of functional subtypes of the parenchyma region, the proportion of each functional subtype of the parenchyma region in the corresponding sample data is calculated to obtain the composition ratio of functional subtypes of the parenchyma region; simultaneously, based on the classification results of functional subtypes of the mesenchymal region, the proportion of each functional subtype of the mesenchymal region in the corresponding sample data is calculated to obtain the composition ratio of functional subtypes of the mesenchymal region. Based on the RCB scores corresponding to the sample data, the sample data is divided into a pathological complete remission group and a pathological incomplete remission group. The pathological complete remission group is correlated with the composition ratio of functional subtypes of the parenchyma region and the corresponding composition ratio of functional subtypes of the mesenchymal region; simultaneously, the pathological incomplete remission group is correlated with the composition ratio of functional subtypes of the parenchyma region and the corresponding composition ratio of functional subtypes of the mesenchymal region, resulting in the first complete parenchyma correlation result, the first complete mesenchymal correlation result, the first incomplete parenchyma correlation result, and the first incomplete mesenchymal correlation result.

[0047] Pre-defined testing methods, such as the Wilcoxon rank-sum test and chi-square test, were used to perform difference analysis on the results of the first complete association of the first parenchymal tissue and the first imperfect association of the first parenchymal tissue, respectively, to obtain the first difference analysis result. Simultaneously, difference analysis was performed on the results of the first complete association of the first mesenchymal tissue and the first imperfect association of the first mesenchymal tissue, to obtain the second difference analysis result. Multiple tests were then performed using the Benjamini-Hochberg method to obtain the final first and second difference analysis results. Based on the first and second difference analysis results, the composition ratio of key functional subtypes was obtained. Specifically, the composition ratio of parenchymal functional subtypes with Treg cell density <0.05 in the final first and second difference analysis results was used as the key functional subtype composition ratio. That is, the key functional subtype composition ratio is the composition ratio of functional subtypes significantly associated with the efficacy of NAC in breast cancer. The key functional subtype composition ratio was then input into a breast cancer NAC efficacy prediction model constructed based on a logistic regression algorithm to obtain the breast cancer NAC efficacy prediction result.

[0048] Example 2 Please refer to Figure 3 The present invention provides a breast cancer NAC efficacy prediction system 1 based on Treg cell spatial heterogeneity, including a memory 3, a processor 2, and a computer program stored on the memory 3 and executable on the processor 2. When the processor 2 executes the computer program, it implements the steps in Embodiment 1.

[0049] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0050] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0051] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0052] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0053] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0054] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0055] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A method for predicting the efficacy of NAC (Non-Acute Breast Cancer) based on the spatial heterogeneity of Treg cells, characterized in that, include: Obtain sample data and the corresponding RCB score of the sample data, perform cell identification and region segmentation on the sample data, and obtain cell identification results and region segmentation results. The cell identification results include the cell type corresponding to each cell, the cell type includes Treg cells, and the region segmentation results include the tumor parenchyma region and the tumor stroma region. Based on the cell identification results, multi-dimensional spatial distribution analysis was performed on Treg cells in the tumor parenchyma region and Treg cells in the tumor stroma region, respectively, to obtain corresponding spatial distribution analysis results for the parenchyma region and spatial distribution analysis results for the stroma region. Based on the spatial distribution analysis results for the parenchyma region, multi-dimensional spatial organization analysis was performed on Treg cells in the tumor parenchyma region, and simultaneously, based on the spatial distribution analysis results for the stroma region, multi-dimensional spatial organization analysis was performed on Treg cells in the tumor stroma region, to obtain corresponding spatial organization analysis results for the parenchyma region and spatial organization analysis results for the stroma region. The multi-dimensional spatial distribution analysis included: Treg cell density dimension, Ripley's L-function dimension, Treg cell occupancy rate dimension, and Treg cell ratio dimension. The multi-dimensional spatial organization analysis included: global aggregation dimension, local hotspot dimension, and spatial correlation dimension. Based on the spatial distribution analysis results and the spatial organization analysis results of the solid region, Treg cells in the tumor solid region are sequentially subjected to spatial clustering and hierarchical clustering to generate functional subtype classification results of the solid region. At the same time, based on the spatial distribution analysis results and the spatial organization analysis results of the mesenchymal region, Treg cells in the tumor mesenchymal region are sequentially subjected to spatial clustering and hierarchical clustering to generate functional subtype classification results of the mesenchymal region. The functional subtype classification results of the solid region and the functional subtype classification results of the mesenchymal region were correlated with the corresponding RCB scores to obtain the NAC efficacy prediction results for breast cancer.

2. The method for predicting the efficacy of NAC (Non-Acute Breast Cancer) based on the spatial heterogeneity of Treg cells as described in claim 1, characterized in that, The process of performing cell identification and region segmentation on the sample data to obtain cell identification results and region segmentation results includes: The sample data is sliced ​​according to a preset thickness to obtain slice data, and the slice data is then subjected to mIF staining to obtain mIF-stained slice data. Cell detection algorithms are used to identify cells in mIF-stained slide data, resulting in cell identification results including the cell type corresponding to each cell. At the same time, the RTtree algorithm is used to segment regions in the mIF-stained slide data, resulting in region segmentation results including the tumor parenchyma region and the tumor stroma region.

3. The method for predicting the efficacy of NAC (Non-Acute Breast Cancer) based on the spatial heterogeneity of Treg cells as described in claim 1, characterized in that, The cell identification results include the cell coordinates of each cell. Based on the cell identification results, multi-dimensional spatial distribution analysis is performed on the Treg cells in the tumor parenchyma region and the Treg cells in the tumor stroma region, respectively, to obtain corresponding spatial distribution analysis results for the parenchyma region and the stroma region. Based on the spatial distribution analysis results for the parenchyma region, multi-dimensional spatial organization analysis is performed on the Treg cells in the tumor parenchyma region. Simultaneously, based on the spatial distribution analysis results for the stroma region, multi-dimensional spatial organization analysis is performed on the Treg cells in the tumor stroma region, to obtain corresponding spatial organization analysis results for the parenchyma region and the stroma region, including: The tumor parenchyma region and the tumor stroma region are divided into grids according to preset sizes to obtain the gridded tumor parenchyma region and the gridded tumor stroma region. Based on the cell type and cell coordinates of each cell, the number of Treg cells in each grid within the tumor parenchyma region after grid division is statistically analyzed to obtain the first Treg cell count for each grid. Simultaneously, based on the cell type and cell coordinates of each cell, the number of Treg cells in each grid within the tumor stroma region after grid division is statistically analyzed to obtain the second Treg cell count for each grid. The first and second Treg cell counts for each grid are then input into the Treg cell density formula for calculation, yielding the corresponding first and second Treg cell density results. The Treg cell density formula is as follows: ； ； in, This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. T represents the i-th grid in the meshed region R. quantity, This represents the area of ​​the i-th grid in the region R after gridding; The number of first Treg cells corresponding to all grids within the preset range is summarized to obtain the total number of first Treg cells. Simultaneously, the number of second Treg cells corresponding to all grids within the preset range is summarized to obtain the total number of second Treg cells. The total numbers of the first and second Treg cells are then input into Ripley's L function formula for calculation to obtain the corresponding first and second Ripley's L function results. The Ripley's L function formula is as follows: ； ； in, This represents the Ripley's L function result for the meshed region R within a preset range r, where r represents the preset range. This represents the total area of ​​the region R after meshing. This represents the total number of regions R after grid division. Represents the region R after meshing, the first... One Treg cell, Represents the region R after meshing, the first... One Treg cell, This indicates that the a-th Treg cell in the region R after gridding is related to the a-th Treg cell. Euclidean distance between Treg cells This indicates that the a-th Treg cell in the region R after gridding is related to the a-th Treg cell. Edge correction weights between Treg cells Indicates an indicator function; The number of grid cells containing Treg cells in the tumor parenchyma region and the number of grid cells containing Treg cells in the tumor stroma region after grid division are statistically analyzed to obtain the corresponding first grid number and second grid number. These first and second grid numbers are then input into the Treg cell occupancy rate formula for calculation to obtain the corresponding first and second Treg cell occupancy rate results. The Treg cell occupancy rate formula is as follows: ； ； in, This represents the Treg cell density result corresponding to region R after meshing. This indicates the number of grid cells contained in the region R after gridding. This represents the total number of grid cells corresponding to the region R after meshing. The number of cells of each cell type in each grid within the tumor parenchyma region after grid division is counted according to the cell type and cell coordinates of each cell, resulting in a first statistical result. At the same time, the number of cells of each cell type in each grid within the tumor stroma region after grid division is counted according to the cell type and cell coordinates of each cell, resulting in a second statistical result. The first statistical result and the second statistical result are then input into the Treg cell ratio formula for calculation to obtain the corresponding first Treg cell ratio result and second Treg cell ratio result. The formula for the Treg cell ratio is: ； ； in, This represents the Treg cell ratio in region R after gridding. T represents the i-th grid in the meshed region R. quantity, This represents the number of cells of type m in the i-th grid within the statistical results of the region R after grid division; The results of the first Treg cell density, the first Ripley's L function, the first Treg cell occupancy rate, and the first Treg cell ratio are used as the spatial distribution analysis results of the parenchymal region, and the results of the second Treg cell density, the second Ripley's L function, the second Treg cell occupancy rate, and the second Treg cell ratio are used as the spatial distribution analysis results of the interstitial region. Based on the spatial distribution analysis results of the solid region, a multi-dimensional spatial organization analysis was performed on the Treg cells in the tumor solid region. At the same time, based on the spatial distribution analysis results of the mesenchymal region, a multi-dimensional spatial organization analysis was performed on the Treg cells in the tumor mesenchymal region, resulting in the corresponding spatial organization analysis results of the solid region and the mesenchymal region.

4. The method for predicting the efficacy of NAC in breast cancer based on the spatial heterogeneity of Treg cells as described in claim 3, characterized in that, The multi-dimensional spatial distribution analysis further includes: Shannon entropy quantification dimension, Treg cell radial frequency dimension, and nearest neighbor distance dimension. The step of using the first Treg cell density result, the first Ripley's L function result, the first Treg cell occupancy rate result, and the first Treg cell ratio result as the spatial distribution analysis result of the parenchymal region, and using the second Treg cell density result, the second Ripley's L function result, the second Treg cell occupancy rate result, and the second Treg cell ratio result as the spatial distribution analysis result of the mesenchymal region includes: The first statistical result and the second statistical result are respectively input into the Shannon entropy quantization formula for calculation to obtain the corresponding first Shannon entropy quantization result and second Shannon entropy quantization result. The Shannon entropy quantization formula is as follows: ； ； in, This represents the Shannon entropy quantization result corresponding to the region R after gridding. This represents the total number of all cell types in the region R after gridding. This represents the number of cells of type m in the i-th grid within the corresponding statistical results of the gridded region R. This represents the total number of cells of all cell types in the i-th grid within the gridded region R. Based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor parenchyma region as the center, the number of Treg cells within a preset radius is counted to obtain a fifth number for each Treg cell. Simultaneously, based on the cell type and cell coordinates of each cell, and taking each Treg cell in the gridded tumor stroma region as the center, the number of Treg cells within a preset radius is counted to obtain a sixth number for each Treg cell. The fifth and sixth numbers are then input into the Treg cell radial frequency formula for calculation to obtain the corresponding first and second Treg cell radial frequency results. The Treg cell radial frequency formula is: ； ； in, This represents the Treg cell ratio in region R after gridding. This represents the number of Treg cells within a preset radius L, centered on the a-th Treg cell, in the region R after gridding. Indicates the preset radius range. This represents the total number of all cell types within a preset radius centered on the a-th Treg cell in the region R after grid division. Based on the cell type and cell coordinates of each cell, and using each Treg cell in the gridded tumor stroma region as the center, the KNN algorithm is used to select the first nearest neighbor Treg cell for each Treg cell. The first nearest neighbor Treg cell and its corresponding Treg cell are then input into the nearest neighbor distance formula for calculation, yielding the first nearest neighbor distance result for each Treg cell. Simultaneously, based on the cell type and cell coordinates of each cell, and using each Treg cell in the gridded tumor stroma region as the center, the KNN algorithm is used to select the second nearest neighbor Treg cell for each Treg cell. The second nearest neighbor Treg cell and its corresponding Treg cell are then input into the nearest neighbor distance formula for calculation, yielding the second nearest neighbor distance result for each Treg cell. The nearest neighbor distance formula is as follows: ； ； in, This represents the nearest neighbor distance result for the a-th Treg cell in the gridded region R. This represents the x-coordinate of the a-th Treg cell in the meshed region R. Let x represent the x-coordinate of the nearest neighbor Treg cell c to the a-th Treg cell. This represents the ordinate of the a-th Treg cell in the meshed region R. The ordinate represents the nearest neighbor Treg cell c to the a-th Treg cell. This represents the c-th Treg cell in the gridded region R. The results of the first Treg cell density, the first Ripley's L function, the first Treg cell occupancy rate, the first Treg cell ratio, the first Shannon entropy quantification, the first Treg cell radial frequency, and the first nearest neighbor distance of each Treg cell are used as the spatial distribution analysis results of the parenchymal region. The results of the second Treg cell density, the second Ripley's L function, the second Treg cell occupancy rate, the second Treg cell ratio, the second Shannon entropy quantification, the second Treg cell radial frequency, and the second nearest neighbor distance of each Treg cell are used as the spatial distribution analysis results of the mesenchymal region.

5. The method for predicting the efficacy of NAC in breast cancer based on the spatial heterogeneity of Treg cells as described in claim 3, characterized in that, The step involves performing multi-dimensional spatial organization analysis on Treg cells in the tumor parenchyma region based on the spatial distribution analysis results of the parenchyma region, and simultaneously performing multi-dimensional spatial organization analysis on Treg cells in the tumor stroma region based on the spatial distribution analysis results of the stroma region. The resulting spatial organization analysis results for the parenchyma region and the stroma region include: The first Treg cell density result from the spatial distribution analysis of the mesenchymal region is input into the Moran's I index formula for calculation to obtain the first global aggregation result. Simultaneously, the second Treg cell density result from the spatial distribution analysis of the mesenchymal region is input into the Moran's I index formula for calculation to obtain the second global aggregation result. The Moran's I index formula is as follows: ； ； in, This represents the first global aggregation result corresponding to region R after meshing. This represents the total number of grid cells corresponding to the region R after meshing. This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. This represents the average Treg cell density result across all grids in the meshed region R. This represents the i-th grid in the region R after gridding. This represents the j-th grid cell in the meshed region R. This represents the spatial weight matrix between the i-th and j-th grids in the meshed region R. The first Treg cell density result from the spatial distribution analysis of the interstitial region is input into the local hotspot formula for calculation to obtain the first local hotspot result. Simultaneously, the second Treg cell density result from the spatial distribution analysis of the interstitial region is input into the local hotspot formula for calculation to obtain the second local hotspot result. The local hotspot formula is as follows: ； ； in, This represents the first local hotspot result corresponding to region R after meshing, where h represents the preset distance threshold. This represents the total number of grid cells corresponding to the region R after meshing. This represents the Treg cell density result corresponding to the i-th grid in the gridded region R. This represents the average Treg cell density result across all grids in the meshed region R. This represents the i-th grid in the region R after gridding. This represents the j-th grid cell in the meshed region R. This represents the spatial weight matrix between the i-th and j-th grids in the gridded region R, based on a preset distance threshold h. The first Treg cell ratio from the spatial distribution analysis of the parenchymal region is input into a spatial correlation formula for calculation to obtain the first spatial correlation result. Simultaneously, the second Treg cell ratio from the spatial distribution analysis of the mesenchymal region is input into the spatial correlation formula for calculation to obtain the second spatial correlation result. The spatial correlation formula is as follows: ； ； in, This represents the first spatial association result corresponding to the region R after meshing. This indicates that within the same grid cell in the region R after meshing, there are simultaneously... The ratio of the number of grid cells of cell type m to the total number of grid cells. This represents the ratio of the number of grid cells containing Treg cells in the meshed region R to the total number of grid cells. This represents the ratio of the number of grid cells containing cell type m in the meshed region R to the total number of grid cells. The first global aggregation result, the first local hotspot result, and the first spatial correlation result are used as the spatial organization analysis result of the substance region, and the second global aggregation result, the second local hotspot result, and the second spatial correlation result are used as the spatial organization analysis result of the interstitial region.

6. The method for predicting the efficacy of NAC in breast cancer based on the spatial heterogeneity of Treg cells as described in claim 1, characterized in that, The cell identification results include the cell coordinates of each cell. The process of performing spatial clustering and hierarchical clustering on Treg cells in the tumor parenchyma region based on the spatial distribution analysis results and the spatial organization analysis results of the parenchyma region to generate functional subtype classification results for the parenchyma region, and simultaneously performing spatial clustering and hierarchical clustering on Treg cells in the tumor stroma region based on the spatial distribution analysis results and the spatial organization analysis results of the stroma region to generate functional subtype classification results for the stroma region, includes: The K-means unsupervised clustering algorithm was used to spatially cluster the cell coordinates of Treg cells in the tumor parenchyma region to obtain K first clusters. At the same time, the K-means unsupervised clustering algorithm was used to spatially cluster the cell coordinates of Treg cells in the tumor stroma region to obtain K second clusters. Based on the spatial distribution analysis results of the solid region and the spatial organization analysis results of the solid region, a first spatial feature vector is constructed for each first cluster to obtain a set of all first spatial feature vectors. At the same time, based on the spatial distribution analysis results of the interstitial region and the spatial organization analysis results of the interstitial region, a second spatial feature vector is constructed for each second cluster to obtain a set of all second spatial feature vectors. Hierarchical clustering algorithm is used to perform hierarchical clustering on all first spatial feature vector sets to generate functional subtype classification results for the solid region. At the same time, hierarchical clustering algorithm is used to perform hierarchical clustering on all second spatial feature vector sets to generate functional subtype classification results for the interstitial region.

7. The method for predicting the efficacy of NAC in breast cancer based on the spatial heterogeneity of Treg cells as described in claim 1, characterized in that, The step of performing correlation analysis between the functional subtype classification results of the solid region and the functional subtype classification results of the stromal region and their corresponding RCB scores to obtain the NAC efficacy prediction results for breast cancer includes: The composition ratio of functional subtypes in the solid region is obtained based on the classification results of functional subtypes in the solid region, and the composition ratio of functional subtypes in the interstitial region is obtained based on the classification results of functional subtypes in the interstitial region. Based on the RCB scores of the sample data, the sample data is divided into a pathological complete remission group and a pathological incomplete remission group. The pathological complete remission group is correlated with the corresponding composition ratio of functional subtypes in the parenchyma area and the corresponding composition ratio of functional subtypes in the mesenchyma area to obtain a first complete parenchyma correlation result and a first complete mesenchyma correlation result. At the same time, the pathological incomplete remission group is correlated with the corresponding composition ratio of functional subtypes in the parenchyma area and the corresponding composition ratio of functional subtypes in the mesenchyma area to obtain a first incomplete parenchyma correlation result and a first incomplete mesenchyma correlation result. A pre-defined testing method is used to perform a difference analysis between the first substantially complete association result and the first substantially incomplete association result to obtain a first difference analysis result. At the same time, a difference analysis is performed between the first interstitial completely association result and the first interstitial incomplete association result to obtain a second difference analysis result. Based on the results of the first and second difference analyses, the composition ratio of key functional subtypes is obtained. The key functional subtypes are then input into a breast cancer NAC efficacy prediction model constructed based on the Logistic regression algorithm to obtain the breast cancer NAC efficacy prediction results.

8. A breast cancer NAC efficacy prediction system based on Treg cell spatial heterogeneity, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 7.

Images