An adrenal gland metastasis prediction analysis method and system based on image recognition

Abstract

The present application relates to the field of image recognition, and more particularly to a method and system for predicting and analyzing adrenal metastasis based on image recognition. The method comprises the following steps: collecting patient adrenal PET-CT monitoring images at multiple time points, and performing visual detection of metastasis at each time point, and extracting metastasis bounding boxes at multiple time points; according to the metastasis bounding box, the dynamic fluctuation analysis of the surrounding tissue microenvironment and the boundary immune influence degree evaluation are carried out, and a multi-cycle immune response intensity graph is constructed; the metastasis change trend evolution of the metastasis bounding box is carried out, and the metastasis image change trend graph at different time points is obtained; according to the metastasis bounding box, the surrounding tissue texture feature difference is identified, and the potential tumor metastatic cells are extracted. The present application predicts the future change of metastasis by dynamically predicting the development trend of adrenal metastasis, and provides more intuitive metastasis change at different time points.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image recognition, and in particular to a method and system for predicting and analyzing adrenal metastasis based on image recognition. BACKGROUND

[0002] With the continuous development of medical imaging technology, image recognition technology as a new artificial intelligence application has made significant breakthroughs in the field of medical diagnosis. Especially in the early detection and accurate diagnosis of tumors, image recognition technology has shown great potential. Adrenal metastasis as a common type of malignant tumor often brings great challenges due to its non-obvious symptoms and difficulty in early diagnosis. With the advancement of computer vision and deep learning algorithms, image recognition-based adrenal metastasis prediction analysis has gradually become an important tool for early tumor diagnosis, greatly improving the accuracy and efficiency of tumor detection.

[0003] The diagnosis of adrenal metastasis usually relies on CT, MRI and other medical imaging data. However, traditional image analysis methods often rely on manual interpretation, which has certain subjectivity and errors. Since the manifestations of adrenal metastasis are similar to other kidney diseases or tumors, some subtle lesion characteristics may be missed during image analysis, leading to missed diagnosis or misdiagnosis. At the same time, the diagnosis of adrenal metastasis often requires multiple image examinations, increasing the burden on patients and the consumption of medical resources. Therefore, there is an urgent need for a more efficient and accurate diagnostic method to help improve the early detection of adrenal metastasis. SUMMARY

[0004] To solve the above technical problems, the present application provides a method and system for predicting and analyzing adrenal metastasis based on image recognition to solve at least one of the above technical problems.

[0005] To achieve the above-mentioned purpose, the present application provides a method for predicting and analyzing adrenal metastasis based on image recognition, comprising the following steps:

[0006] Step S1: Collecting patient adrenal PET-CT monitoring images at multiple time points and performing metastasis visual detection at each time point, and extracting metastasis bounding boxes at multiple time points;

[0007] Step S2: Performing surrounding tissue microenvironment dynamic fluctuation analysis and boundary immune influence degree evaluation according to the metastasis bounding boxes, and constructing a multi-cycle immune response intensity map;

[0008] Step S3: Performing metastasis change trend evolution on the metastasis bounding boxes to obtain metastasis image change trend graphs at different time points;

[0009] Step S4: According to the metastasis tumor boundary box, the difference of the texture features of the surrounding tissue is identified, so as to extract the potential tumor metastatic cells;

[0010] Step S5: According to the adrenal gland PET-CT monitoring image and the metastasis tumor image change trend graph, the dynamic metastasis trajectory tracking is performed, and the whole-process metastasis trajectory fitting is performed to construct a time-series metastasis trajectory graph;

[0011] Step S6: According to the multi-cycle immune response intensity graph and the potential tumor metastatic cells, the metastasis tumor situation is predicted, and a metastasis tumor prediction time axis is constructed.

[0012] Through image monitoring at multiple time points, the present application can obtain image data of adrenal gland metastasis tumors of patients at different time points, providing more comprehensive dynamic data for subsequent analysis. Through visual detection of metastasis tumors at each time point, the boundary box of metastasis tumors at each time point can be accurately extracted, thereby providing basic data for subsequent analysis. The boundary box of metastasis tumors can provide effective spatial coordinates for tumor analysis and prediction, making the subsequent prediction more reliable and operable. Through analysis of the microenvironment of the surrounding tissue of metastasis tumors, the dynamic changes of immune cells around the tumor can be revealed, thereby providing deep insights into the evolution of the tumor microenvironment. By evaluating the immune response intensity, the changes in the immune status in the tumor microenvironment can be identified, helping to evaluate the potential response of metastasis tumors to immunotherapy. By analyzing the change trend of the metastasis tumor boundary box, the evolution process of the tumor at different time points can be determined, and the trends of tumor growth, shrinkage or stability can be revealed. The tumor image change trend graph can help identify whether the tumor has the potential to metastasize at an early stage, and the change trend graph can support early warning of metastasis tumors, helping to identify potential metastasis risks and take preventive measures. By analyzing the texture feature difference of the surrounding tissue, the potential existence of tumor metastatic cells can be identified, providing more accurate metastasis warning. The texture feature difference analysis makes the image recognition process more refined, which helps to distinguish normal tissue from metastatic tumor cells and improves the detection accuracy of tumor metastasis. By analyzing the texture changes of the surrounding tissue, the changes in the tumor microenvironment and the interaction between tumor cells and the surrounding tissue can be understood in depth, thereby providing support for subsequent clinical decision-making. Through dynamic tracking of the metastasis trajectory, the path and speed of tumor metastasis can be effectively monitored, thereby providing a more intuitive trend of disease development. The whole-process trajectory fitting can help track the evolution of the tumor at different time points and reveal the overall picture of tumor metastasis. The metastasis trajectory graph can provide key nodes on the metastasis path, and through comprehensive analysis of the immune response intensity graph and the metastasis trajectory graph, the dynamic prediction of tumor development trend can be realized, and the future changes of metastasis tumors can be predicted.

[0013] In the present specification, a kind of adrenal metastasis prediction analysis system based on image recognition is provided, for carrying out the adrenal metastasis prediction analysis method based on image recognition as described above, including:

[0014] Image processing module, for collecting the adrenal gland PET-CT monitoring image of multiple time points of patient, and carries out metastasis tumor visual detection one by one time point, extracts the metastasis tumor boundary box of multiple time points;

[0015] Immune response evaluation module, for carrying out surrounding tissue microenvironment dynamic fluctuation analysis and boundary immune influence degree evaluation according to the metastasis tumor boundary box, constructs multi-cycle immune response intensity chart;

[0016] Image change module, for carrying out metastasis tumor change trend evolution to metastasis tumor boundary box, obtains the metastasis tumor image change trend chart of different time points;

[0017] Difference identification module, for carrying out surrounding tissue texture feature difference identification according to metastasis tumor boundary box, to extract potential tumor metastatic cells;

[0018] Metastasis trajectory tracking module, for carrying out dynamic metastasis trajectory tracking according to the adrenal gland PET-CT monitoring image and the metastasis tumor image change trend chart, and carries out whole-process metastasis trajectory fitting, constructs time series metastasis trajectory chart;

[0019] Situation prediction module, for carrying out metastasis tumor situation rolling prediction to time series metastasis trajectory chart according to multi-cycle immune response intensity chart and potential tumor metastatic cells, constructs metastasis tumor prediction time axis.

[0020] The present application provides comprehensive dynamic data for subsequent analysis by collecting adrenal gland PET-CT images at multiple time points, which can effectively track the changes of the tumor. Through image visual detection, metastatic tumors and their boundaries can be accurately identified, avoiding subjective differences in manual judgment and improving detection efficiency and accuracy. The extracted metastatic tumor boundary box at multiple time points provides core data support for subsequent analysis and is the basis for analysis of other modules. By analyzing the immune response around the tumor, the trend of the tumor microenvironment can be evaluated, and the sensitivity of the tumor to the immune response can be inferred. Through the image change trend chart, the changes of metastatic tumors at different time points can be tracked in real time, revealing the trend of tumor growth, shrinkage or stability. The change trend chart can help identify the development trend of metastatic tumors, providing early warning and taking preventive measures in advance. Texture feature difference recognition can reveal the subtle differences between tumor metastatic cells and surrounding normal tissues, improving the detection accuracy of metastatic cells. Through texture difference analysis, potential diffusion areas of metastatic tumors can be identified early, enhancing the early diagnosis ability of metastatic tumors. The path and speed of tumor metastasis can be accurately tracked, providing a clearer tumor development pattern to help evaluate the risk and severity of metastasis. Through the fitting of the metastasis trajectory chart, from the initial metastasis to the possible diffusion area. Through the analysis of immune response and tumor metastatic cells, combined with the metastasis trajectory chart, the future trend of tumor metastasis can be predicted. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 A step flowchart of the adrenal gland metastatic tumor prediction analysis method based on image recognition of the present application is shown in the figure.

[0022] Figure 2 A detailed implementation step flowchart of step S1 is shown in the figure.

[0023] Figure 3 A detailed implementation step flowchart of step S2 is shown in the figure.

[0024] Figure 4 A detailed implementation step flowchart of step S3 is shown in the figure. DETAILED DESCRIPTION

[0025] It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0026] The present application provides a kind of adrenal gland metastatic tumor prediction analysis method and system based on image recognition. The execution body of the adrenal gland metastatic tumor prediction analysis method and system based on image recognition includes but is not limited to the system carried by mechanical equipment, data processing platform, cloud server node, network upload equipment and the like, which can be regarded as the general computing node of the present application, and the data processing platform includes but is not limited to at least one of audio image management system, information management system and cloud data management system.

[0027] Referring to Figures 1 to 4 The application provides an adrenal metastasis tumor prediction analysis method based on image recognition, which comprises the following steps:

[0028] Step S1: Collecting patient adrenal PET-CT monitoring images at multiple time points, and performing metastasis tumor visual detection one by one, and extracting metastasis tumor bounding boxes at multiple time points;

[0029] Step S2: According to the metastasis tumor bounding box, the surrounding tissue microenvironment dynamic fluctuation analysis and the boundary immune influence degree evaluation are performed, and a multi-cycle immune reaction intensity diagram is constructed;

[0030] Step S3: The metastasis tumor change trend evolution is performed on the metastasis tumor bounding box, and a metastasis tumor image change trend diagram at different time points is obtained;

[0031] Step S4: According to the metastasis tumor bounding box, the surrounding tissue texture feature difference recognition is performed, and the potential tumor metastasis cells are extracted;

[0032] Step S5: According to the adrenal PET-CT monitoring image and the metastasis tumor image change trend diagram, the dynamic metastasis trajectory tracking is performed, and the whole process metastasis trajectory fitting is performed, and a time sequence metastasis trajectory diagram is constructed;

[0033] Step S6: According to the multi-cycle immune reaction intensity diagram and the potential tumor metastasis cells, the metastasis tumor situation rolling prediction is performed on the time sequence metastasis trajectory diagram, and a metastasis tumor prediction time axis is constructed.

[0034] The present application can obtain image data of adrenal metastasis of patients at different time points through multi-time point image monitoring, providing more comprehensive dynamic data for subsequent analysis. Through visual detection of metastasis at each time point, the boundary box of metastasis at each time point can be accurately extracted, thereby providing basic data for subsequent analysis. The boundary box of metastasis can provide effective spatial coordinates for tumor analysis and prediction, making the subsequent prediction more reliable and operable. Through analysis of the microenvironment of the tissue around the metastasis, the dynamic changes of immune cells around the tumor can be revealed, thereby providing deep insights into the evolution of the tumor microenvironment. By evaluating the intensity of the immune response, changes in the immune status in the tumor microenvironment can be identified, helping to evaluate the potential response of metastasis to immunotherapy. By analyzing the trend of changes in the boundary box of metastasis, the evolution process of the tumor at different time points can be determined, revealing trends such as tumor growth, shrinkage or stability. The tumor image change trend chart can help identify whether the tumor has the potential to metastasize at an early stage, and the change trend chart can provide support for early warning of metastasis, helping to identify potential metastasis risks and take preventive measures. By analyzing the differences in texture features of the surrounding tissue, the potential presence of metastatic cells can be identified, providing more accurate metastasis warnings. The difference analysis of texture features makes the image recognition process more refined, helping to distinguish between normal tissue and metastatic cells and improving the accuracy of tumor metastasis detection. By analyzing the texture changes of the surrounding tissue, the changes in the tumor microenvironment and the interaction between tumor cells and surrounding tissue can be understood in depth, providing support for subsequent clinical decision-making. By dynamically tracking the metastasis trajectory, the path and speed of tumor metastasis can be effectively monitored, providing a more intuitive trend of disease development. The whole-process trajectory fitting can help track the evolution of the tumor at different time points, revealing the overall picture of tumor metastasis. The metastasis trajectory chart can provide key nodes on the metastasis path, and through comprehensive analysis of the immune response intensity chart and the metastasis trajectory chart, dynamic prediction of tumor development trends can be achieved, predicting future changes in metastasis.

[0035] In the embodiments of the present application, reference is made to Figure 1 The present application is a step flowchart of an adrenal metastasis prediction analysis method based on image recognition. In this example, the steps of the adrenal metastasis prediction analysis method based on image recognition include:

[0036] Step S1: Collect multiple time point patient adrenal PET-CT monitoring images, and perform visual detection of metastasis at each time point to extract metastasis boundary boxes at multiple time points;

[0037] In this embodiment, after obtaining patient authorization, multiple PET-CT scans are performed, usually before treatment, after treatment and during follow-up, with intervals of several weeks to several months between each time point, and the specific time is determined according to clinical requirements. After each scan, save the image data, ensure that it is stored in DICOM standard format, and facilitate subsequent analysis and processing. Before visual inspection, the collected PET-CT images are preprocessed, including denoising, standardization and image enhancement. Gaussian filtering and other methods are usually used to remove noise and improve image quality. The images are standardized to ensure that images at different time points can be compared. This can be achieved by histogram equalization of the images. Select an appropriate visual inspection method, usually combined with the experience of medical imaging experts and computer-aided diagnosis (CAD) systems. To ensure the accuracy of metastasis identification. During visual inspection, suspicious metastasis areas are marked one by one. For each identified metastasis, draw a bounding box to ensure accurate tumor area. The bounding box should tightly surround the metastasis so that it can be analyzed later. Record the coordinates, size and shape features of each bounding box for subsequent data analysis and comparison. This information is usually stored in CSV or database format for subsequent processing. Integrate metastasis bounding box data at each time point into a database to ensure easy access and analysis at any time. This database should include patient basic information, image data at each time point and metastasis features. Ensure data integrity, back up the database regularly to prevent data loss. Quality control of collected images and labeled bounding boxes to ensure data accuracy and reliability. The labeled results can be reviewed by random sampling, and regular team discussions and training are conducted to ensure that all participants can master image analysis methods and techniques. Preliminary analysis of extracted metastasis bounding box data to assess changes in metastasis at different time points, including volume changes, morphological changes, etc. This will provide important basic data for subsequent metastasis prediction analysis. If the metastasis volume changes at different time points show significant growth, it may indicate that the tumor is progressing.

[0038] Step S2: According to the metastasis bounding box, the surrounding tissue microenvironment dynamic fluctuation analysis and the boundary immune influence degree evaluation are carried out, and a multi-cycle immune reaction intensity diagram is constructed;

[0039] In this example, first, we need to extract the tissue regions around metastatic tumors from each time point's PET-CT images. The coordinate information of the bounding box is used to crop the specific region, usually set to a range of 5-10mm beyond the bounding box to capture the microenvironment changes of the surrounding tissue. Record the specific location, size, shape, and corresponding image data of each extracted region for subsequent analysis. Image processing is performed on the extracted tissue regions, including denoising, enhancement, and standardization to improve the accuracy of analysis. Common image processing methods include Gaussian filtering, histogram equalization, etc. From the processed images, extract microenvironment features such as tissue density, blood flow, cell density, etc. These features are usually achieved through image analysis software (such as ImageJ or MATLAB), using threshold segmentation, morphological operations, and other techniques for analysis. Using the extracted microenvironment feature data, calculate the feature values at each time point and perform time series analysis to evaluate the dynamic changes of the surrounding tissue microenvironment. Moving average or difference method can be used to analyze the fluctuation of feature values. If the cell density of the surrounding tissue at a certain time point is 1000 cells / mm², and the next time point is 1200 cells / mm², record this change and calculate the change rate. According to the features of the surrounding tissue and the boundary information of the metastatic tumor, construct an immune response intensity map. This map reflects the influence of the metastatic tumor on the surrounding immune environment, usually using immunohistochemical staining or immunofluorescence technology to obtain immune cell distribution information. By staining the tissue region (such as CD3, CD8, CD68 markers), and using a microscope to take pictures, the immune cell infiltration at different time points can be evaluated. Image analysis is performed on the stained tissue region to count the number and distribution of different immune cells (such as T cells, macrophages, etc.). Image analysis software can be used to automatically count the number of cells and calculate their proportion in the surrounding tissue region. If the number of CD8+ T cells at a certain time point is 200, and the number of macrophages is 150, the proportion of immune cells and their trend can be calculated. According to the distribution and number of immune cells, the influence of the metastatic tumor on the surrounding tissue immune environment is evaluated. At each time point, the density and activity of immune cells are calculated, and the surrounding tissue microenvironment features are combined for comprehensive analysis. If significant CD8+ T cell infiltration is found near the metastatic tumor boundary, while the region far from the boundary has less infiltration, it indicates that the tumor has a significant impact on the immune environment. Integrate the immune response intensity data at each time point to construct a multi-cycle immune response intensity map. Each time point image should indicate the corresponding immune cell features, microenvironment features, and their changes. Generate a heat map containing multiple time points, with color depth representing the intensity of immune response, for intuitive analysis. Record the immune response intensity map of each time window and generate charts or images to show the immune environment changes at different time points. Visualization tools such as GraphPad Prism or Tableau can be used for data display.The generated heat map can show the dynamic change of immune response intensity on the time axis, providing support for subsequent analysis and clinical decision-making. The generated multi-cycle immune response intensity map is analyzed to explore its relationship with metastatic tumor progression. The dynamic change of immune environment is analyzed to understand how it affects tumor growth and metastasis. If it is found that the immune response intensity significantly decreases during the period of rapid tumor growth, it suggests that immune escape mechanisms may be at work.

[0040] Step S3: Perform metastatic tumor change trend evolution on the metastatic tumor boundary box to obtain a metastatic tumor image change trend graph at different time points;

[0041] In this example, before analyzing metastatic tumor trends, key imaging indicators must be determined. These indicators typically include metastatic volume, area, shape factors (such as perimeter and circularity), and image grayscale values. Selecting these indicators can help comprehensively assess tumor changes. Volume can be estimated using geometric models based on bounding box coordinate information and image resolution. At each time point, the values ​​of these indicators are recorded for each metastasis's bounding box. This data will serve as an important foundation for subsequent analysis. At time point T1, if the metastasis' volume is 15 cm³ and its perimeter is 25 cm, this information will be recorded in the database. Image data from different time points requires image registration to ensure comparisons are in the same coordinate system. This typically uses feature point-based registration methods such as SIFT (Scale-Invariant Feature Transform) or SURF (Speeded Robust Features). Registration can eliminate image deviations caused by patient position changes or device differences, thereby improving the accuracy of change analysis. Analyze metastases at each time point using automated image analysis tools (such as ImageJ, MATLAB, or specialized medical imaging analysis software), extract the metastasis boundaries, and calculate key metrics. Use threshold segmentation to extract the metastasis outline and calculate its volume and area. For each metastasis metric at time point, calculate the rate of change (e.g., percentage change) between adjacent time points. This can help identify the rate of tumor growth or shrinkage. If the volume at time point T1 is 15 cm³ and at time point T2 is 20 cm³, the rate of change is [(20 - 15) / 15] × 100% = 33.33%. Integrate the metastasis metric change data at different time points to generate a trend graph. This graph can be a line graph or a bar graph to visually display tumor changes. Draw a line graph with time points on the X-axis and metastasis volume on the Y-axis, marking the volume value and rate of change at each time point. Analyze the generated trend graph to observe tumor growth patterns and rates. By comparing changes between different time points, tumor progression can be assessed. If the rate of tumor growth accelerates significantly over a period of time, this may indicate a change in the tumor's biological behavior. Based on the trend graph and combined with clinical information, the prognosis of metastatic tumors can be assessed. The relationship between tumor changes and treatment efficacy or patient survival is analyzed. If the tumor significantly decreases in size after treatment, it can be inferred that the treatment is effective. However, if the tumor continues to grow, the treatment strategy may need to be adjusted.

[0042] Step S4: identifying differences in surrounding tissue texture features based on the metastatic tumor bounding box to extract potential tumor metastatic cells;

[0043] In this embodiment, based on the metastasis tumor boundary box determined in step S1, the boundary box is expanded to define the surrounding tissue area. The expansion range is usually 5-10 mm to ensure that enough peripheral tissue features are captured. The coordinates and size of these expanded areas are recorded for subsequent analysis. The extracted peripheral tissue images are preprocessed, including denoising, standardization and enhancement. Common denoising methods such as Gaussian filtering, mean filtering, etc. can improve image quality and reduce interference. A Gaussian filter is applied with a standard deviation of 1 to smooth the image and remove small noise points. The gray level co-occurrence matrix (GLCM) method is used to extract the texture features of the surrounding tissue. For each extracted area, GLCM is calculated in multiple directions (such as 0°, 45°, 90°, 135°), and feature values such as contrast, correlation, energy and entropy are extracted from it. If the GLCM energy value calculated in a certain direction is 0.5, the texture uniformity of that direction can be recorded and compared with other directions. Compare the texture features of the metastatic tumor surrounding tissue with the texture features of the normal tissue area, and use statistical analysis methods (such as t-test or ANOVA) to evaluate the significant differences of the features. If the average energy of the metastatic tumor surrounding tissue is 0.45 and the average energy of the normal tissue is 0.65, perform a t-test analysis to determine the statistical significance of the difference. Set the significance level (usually p<0.05) to determine whether the extracted texture features have significant differences between the metastatic tumor surrounding tissue and the normal tissue. If the p-value is less than the set threshold, it is considered that there is a significant difference in texture features between the two. If the p-value is found to be 0.03 in the statistical analysis, it can be considered that there is a significant difference between the metastatic tumor surrounding tissue and the normal tissue. Potential metastatic cell identification: According to the difference in texture features, identify potential tumor metastatic cells. Generally, the presence of tumor metastatic cells will cause significant changes in the texture features of the surrounding tissue (such as increased cell density, disordered tissue structure, etc.). If the entropy value of the texture features in a certain area is significantly higher than that of the normal tissue, it may indicate that there are potential tumor metastatic cells in that area. Choose appropriate cell recognition methods, usually including image segmentation techniques (such as threshold segmentation, region growing or deep learning-based segmentation methods). Combine the texture feature difference information to implement cell recognition and extraction. Use threshold-based segmentation method to set appropriate gray threshold to separate tumor metastatic cells from surrounding tissue. Count the identified potential tumor metastatic cells and record their location, morphological features (such as size, shape, etc.). These data will provide a basis for subsequent biological analysis. If 50 potential metastatic cells are counted in a certain area, record the cell density and distribution characteristics of that area. Integrate the data of the extracted potential tumor metastatic cells with the analysis results of the surrounding tissue texture features to form a complete analysis report. This report should include cell number, location, texture features and their clinical significance.The generated report can include a cell density of 200 cells / mm2 in the tissue surrounding the metastasis, and the texture features of the region are significantly different from normal tissue, thereby suggesting a risk of potential metastasis.

[0044] Step S5: Dynamic metastasis trajectory tracking is performed according to the adrenal gland PET-CT monitoring image and the metastasis image change trend graph, and full-process metastasis trajectory fitting is performed to construct a time-series metastasis trajectory graph.

[0045] In this example, we collect multiple time-point adrenal PET-CT monitoring images and metastasis trend graphs. These images should have been pre-processed and analyzed to ensure reliable data quality. Integrate the metastasis location, volume, bounding box, and other information at each time point to facilitate subsequent tracking. Typically, a database is used to record this information to ensure data traceability. Review the metastasis bounding box extracted in step S1 and confirm whether the metastasis location at each time point is consistent. This can be achieved through image registration techniques to ensure that images at different time points are aligned in the same coordinate system. Use feature point matching methods such as SIFT or SURF algorithms for image registration to ensure accurate identification of metastasis dynamics. Track the location of metastasis at each time point and record the center coordinates of the bounding box at each time point. By calculating the coordinate changes between consecutive time points, we can understand the tumor's movement trajectory. If the center coordinates of the metastasis at time point T1 are (30, 40) and at time point T2 are (32, 42), record the displacement between these two time points. Organize the coordinate data at each time point into a sequence to form the dynamic trajectory data of the metastasis. These data can include coordinates, volume, and change rate at each time point. Organize the coordinate data at each time point into a list to facilitate subsequent trajectory analysis. Use data visualization tools such as Matplotlib or Tableau to plot the dynamic trajectory of the metastasis into a graph. The X-axis usually represents time, and the Y-axis represents the position or volume change of the metastasis. Plot the tumor's movement trajectory at different time points, mark the position and volume change at each time point, and form an intuitive dynamic trajectory graph. Select an appropriate fitting model to analyze the dynamic metastasis trajectory, commonly used models include linear regression, polynomial regression, or more complex machine learning models such as support vector regression (SVR). Determine the model parameters and fitting method to accurately describe the tumor growth and migration trend. Fit the trajectory data, calculate the fitting curve. Optimize the model parameters through least squares method and other methods to obtain the best fitting effect. If a polynomial regression model is used, you may need to choose an appropriate polynomial order (such as quadratic or cubic polynomial) to ensure the accuracy of the fitting. Evaluate the fitting results and use R² (coefficient of determination) and other indicators to judge the goodness of the fitting. An R² value close to 1 indicates a good fitting effect. If the R² value of the model is 0.95, it indicates that the fitting effect is good and can accurately reflect the dynamic changes of the metastasis. Use the fitted results to construct a time-series metastasis trajectory graph. This graph should show the changes of the metastasis at each time point, including position, volume, and growth rate. Generate a time-series graph containing the fitting curve to visually display the tumor's change trend over the entire observation period. Record the constructed time-series metastasis trajectory graph and related data in the database to ensure data integrity and traceability. These records will provide a basis for subsequent research and clinical decision-making.Generate a report to record the metastasis tumor location, volume change and fitting model parameters at each time point. Based on the analysis results of the time series metastasis trajectory map, discuss its application value in clinical practice. Analyze how the dynamic changes of metastasis tumor affect the treatment strategy and prognosis.

[0046] Step S6: According to the multi-cycle immune response intensity map and the potential tumor metastasis cell, the metastasis tumor situation rolling prediction is carried out on the time series metastasis trajectory map, and the metastasis tumor prediction time axis is constructed.

[0047] In this example, the correlation between the multi-cycle immune response intensity map and the data of potential tumor metastatic cells is collected. The immune response intensity map should reflect the immune cell infiltration status at each time point and its changes, while the data of potential tumor metastatic cells should include the number, location and characteristics of the identified cells. Record the immune response intensity values and the count of potential metastatic cells at different time points for subsequent analysis. Review the previously generated time series metastasis trajectory map to ensure that the dynamic change data of metastatic tumors (such as volume, location, etc.) are accurate. This graph will serve as the basis data for the prediction model. Confirm the metastatic tumor volume and location information at each time point and ensure consistency with the data of the immune response intensity map. Integrate the dynamic change data of metastatic tumors with the data of immune response intensity and potential metastatic cells to form a comprehensive data set. The data at each time point should include the volume of metastatic tumors, immune response intensity, and the number of potential metastatic cells. Construct a table containing time points, metastatic tumor volume, immune intensity, and metastatic cell count for subsequent analysis. Select an appropriate prediction model for rolling prediction of metastatic tumor trends. Common models include time series analysis models (such as ARIMA) and machine learning models (such as regression models, random forests, etc.). Model selection should consider the characteristics of the data and the purpose of prediction. If the changes of metastatic tumors show obvious seasonality or periodicity, consider using a seasonal ARIMA model for prediction. Use the integrated data set to train the selected prediction model. Divide the historical data into training set and test set to ensure that the model can accurately capture the relationship between metastatic tumor volume, immune response and potential metastatic cells. Use the past 6 months of data to train the model and evaluate its performance on the validation set to ensure good prediction ability. Perform rolling prediction on future time windows to generate the expected volume of metastatic tumors, immune response intensity and changes of potential metastatic cells at future time points. The prediction results of each time window will provide the basis for subsequent analysis. Predict the volume change of metastatic tumors and immune response intensity in each of the next three months to identify potential tumor progression risks. According to the results of rolling prediction, construct a metastatic tumor prediction timeline. The timeline should show the changes in metastatic tumor volume, immune response intensity and potential metastatic cells at each time window to visually display the dynamic change trend of the tumor. Generate a timeline containing different time points, labeled with the predicted volume and immune response intensity at each time point, to facilitate observation of tumor development trends. Record the generated prediction timeline and related data in the database to ensure data integrity and traceability. At the same time, use visualization tools (such as Matplotlib, Tableau, etc.) to display the prediction results in graphical form to enhance readability. Create charts to show the volume changes of metastatic tumors and immune intensity at different time points.

[0048] In this example, refer to Figure 2For the detailed implementation step flowchart of step S1, in the embodiment, the detailed implementation step of step S1 includes:

[0049] Collecting PET-CT monitoring images of adrenal glands of the patient at multiple time points;

[0050] Sharpening and histogram equalization are performed on the PET-CT monitoring images to construct globally optimized monitoring images;

[0051] Adaptive filtering and noise reduction are performed on the globally optimized monitoring images to obtain filtered and noise-reduced optimized images;

[0052] Visual detection of metastatic tumors is performed on the filtered and noise-reduced optimized images at each time point to mark the positions of metastatic tumors in the images;

[0053] Boundary box segmentation is performed according to the positions of metastatic tumors in the images to extract boundary boxes of metastatic tumors at multiple time points.

[0054] In this embodiment, after obtaining patient authorization, ensure that the device is calibrated and in good condition when performing PET-CT scanning, select the appropriate radioactive tracer (such as [18F]FDG), and calculate the appropriate dose according to the patient's weight. Generally, PET-CT scanning needs to be performed in multiple follow-ups, such as once every three months, and image data at multiple time points is recorded to monitor changes in adrenal metastases. Adopt standardized scanning protocols to ensure consistent image quality for each scan. After each scan, use professional software (such as PACS system) to store and manage image data, ensuring its traceability. Record the scan parameters at each time point, such as scan time (60 minutes), image resolution (512x512 pixels), etc., for subsequent processing and analysis. Before image processing, first read the PET-CT monitoring images at each time point to ensure that all images are processed under the same reference framework. Perform a preliminary check on the images to confirm that there are no artifacts or obvious scanning errors. Check the signal-to-noise ratio (SNR) and contrast of the images to ensure they are suitable for subsequent sharpening and equalization processing. Use image sharpening techniques (such as Laplacian filtering or high-pass filtering) to sharpen the images at each time point to enhance the boundaries of the adrenal glands and metastases in the images. This step helps improve the accuracy of subsequent detection. By applying the Laplacian operator, calculate the second derivative of the image to enhance edge information and record the changes in the sharpened image. Perform histogram equalization on the processed image to improve the contrast of the image, making the metastases more visible. This step improves image visibility by expanding the grayscale range. Use the CLAHE (Contrast Limited Adaptive Histogram Equalization) method to ensure equalization in local regions, thereby avoiding excessive noise enhancement. Combine the sharpened and equalized images to form a globally optimized monitoring image. This image will serve as the basis for subsequent analysis and detection. Record the optimized image parameters at each time point, such as mean grayscale value and contrast enhancement factor, for subsequent analysis and comparison. Choose an appropriate adaptive filtering method, such as Wiener filtering or adaptive median filtering, with the goal of reducing random noise in the image while preserving important edge information. Determine the filtering parameters, such as window size and filter type, to facilitate subsequent processing. Apply the adaptive filtering algorithm to the globally optimized monitoring image at each time point, dynamically adjusting the filtering strength for each pixel based on the statistical properties of its neighborhood pixels. Use Wiener filtering to analyze local regions of the image, calculate local mean and variance, and adaptively adjust the filter based on noise levels to ensure that the details of the image are preserved. Record the image parameters after filtering, such as peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM), to evaluate the filtering effect. Ensure that the recognizability of the adrenal glands and metastases is improved during the noise reduction process. Evaluate the change in PSNR value before and after filtering. If the PSNR value increases from 20 to 30, it indicates that the noise reduction effect is significant.Collect multiple time-point filtered denoising optimization images and ensure that the images are under the same coordinate system for visual inspection. Prepare detection tools and marking software to facilitate subsequent metastasis marking work. Develop a standardized visual inspection process to ensure consistency in the definition and marking of metastases among inspectors. Visually inspect each image one by one to identify the location and size of metastases. Use professional software tools for marking to ensure accuracy. Record the coordinate position (x, y) and diameter of each metastasis to ensure that the characteristic information of the metastasis is fully recorded. Mark the metastasis location in each time-point image to form a labeled image library for subsequent analysis and comparison. Ensure that each mark has corresponding image number and time-point information. Record the marking information, including the location, size, shape, etc. of the metastasis in the image, and generate a marking report for subsequent data analysis. Select appropriate bounding box segmentation algorithms, such as threshold-based segmentation, contour detection (such as Canny edge detection), or region growing methods, with the goal of extracting the bounding box from the marked metastasis. Determine segmentation parameters such as threshold range and minimum contour area to ensure the accuracy and completeness of the bounding box. Apply the selected bounding box segmentation algorithm to each time-point labeled image to extract the boundary information of the metastasis. Generate the corresponding bounding box based on the algorithm output. Identify the edges of the metastasis through Canny edge detection and generate a bounding box using the minimum enclosing rectangle method. Record the metastasis bounding box information for each time point, including the coordinates, width, and height of the bounding box. Form a dataset for subsequent comparison and analysis.

[0055] In this embodiment, the specific steps for sharpening enhancement and histogram equalization of the PET-CT monitoring image to construct a globally optimized monitoring image are as follows:

[0056] Gaussian blur filtering is performed on the PET-CT monitoring image to obtain a blurred filtered image;

[0057] The pixel difference between the blurred filtered image and the PET-CT monitoring image is calculated to obtain a sharpening mask;

[0058] The PET-CT monitoring image is sharpened and enhanced according to the sharpening mask to construct a sharpened and enhanced monitoring image;

[0059] The gray level histogram of the sharpened and enhanced monitoring image is calculated;

[0060] The occurrence frequency of each gray level of the gray level histogram is counted;

[0061] The cumulative frequency of pixels is calculated according to the occurrence frequency of each gray level to obtain a cumulative distribution function;

[0062] The cumulative distribution function is normalized to obtain a gray level mapping range parameter;

[0063] The sharpening enhancement monitoring image is globally and pixel- uniformly converted by the gray scale mapping range parameter, so as to construct a globally optimized monitoring image.

[0064] In this embodiment, Gaussian blur is a common image processing technique aimed at reducing image noise and details. This process smooths the image by applying a Gaussian kernel to convolve the image. The size of the Gaussian kernel and the standard deviation (σ) are determined, typically a 3x3 or 5x5 kernel is chosen, and the σ value can be set to 1.0 or 1.5, depending on the noise level of the image. The selected Gaussian kernel is applied to each pixel, calculating the weighted average of its surrounding pixels to generate a blurred filter image. If a 3x3 Gaussian kernel is used, the value of the center pixel is calculated considering the weighted values of its surrounding 8 pixels, resulting in a smoother image with reduced details. The parameters of the blurred filter image are recorded, including the peak signal-to-noise ratio (PSNR) and the structural similarity index (SSIM), to evaluate the effect of the blur processing. If the PSNR value of the blurred image is 30, it indicates that the noise reduction effect is significant, making it suitable for subsequent sharpening enhancement processing. The blurred filter image is compared pixel by pixel with the original PET-CT monitoring image to calculate the difference between them. This process aims to identify the details hidden in the original image by the blur processing. For each pixel, the difference formula is calculated as: D(x, y) = |I(x, y) - F(x, y)|, where I is the original image, F is the blurred image, and D is the difference image. According to the calculated difference, a sharpening mask is generated. Typically, a difference threshold is set to a certain critical value (e.g., 10) to filter out significant edge information. If the difference of a certain pixel is greater than 10, it is marked as 1 (significant edge) on the sharpening mask, otherwise it is marked as 0 (non-significant edge). The statistical properties of the sharpening mask are recorded, including the number and distribution of edge pixels, to serve as a reference for subsequent sharpening enhancement processing. If the significant edge pixels in the sharpening mask account for 5% of the total pixels, it indicates that there is certain detail information in the image that needs to be enhanced. An appropriate sharpening enhancement method is selected, typically using weighted average filtering or Laplacian filtering. By combining the sharpening mask with the original image, the edges and details are enhanced. The parameters of the sharpening enhancement are determined, such as the enhancement factor (e.g., 1.5), to control the intensity of the sharpening and avoid excessive sharpening that may cause artifacts. The sharpening mask is applied to the original PET-CT monitoring image. The specific steps are: for each pixel, apply the formula E(x, y) = I(x, y) + α⋅D(x, y)E(x, y) = I(x, y) + \alpha \cdot D(x, y)E(x, y)=I(x,y)+α⋅D(x,y), where E is the sharpening enhanced image, and α\alphaα is the enhancement factor. If the original value of a certain pixel is 100 and the difference value of the sharpening mask is 15, the sharpened value is 100 + 1.5 × 15 = 122.5100 + 1.5 \times 15 = 122.5100+1.5×15=122.5, and this pixel becomes more prominent in the enhanced image.Perform gray scale histogram calculation on the sharpened enhanced image, record the frequency of each gray level (usually 0 to 255) pixel. This process helps to evaluate the brightness distribution of the image. The specific steps are: traverse each pixel of the sharpened enhanced image, count the number of each gray level, and record it in the histogram. Visualize the calculated gray scale histogram to facilitate the analysis of the brightness and contrast information of the image. The generated histogram shows that most pixels are concentrated in the middle gray level, indicating the uniformity of the image brightness. Perform statistics on the generated gray scale histogram to record the frequency of each gray level, forming a frequency distribution table. This step provides basic data for the subsequent cumulative distribution function calculation. If the frequency of gray level 255 is 50 times, it is recorded as part of the frequency data. Record the frequency of each gray level and analyze the characteristics of the frequency distribution, such as whether there is a significant deviation or peak value. If the frequency of gray level 128 is much higher than that of other gray levels, it means that the image has more details at this gray level. According to the frequency of each gray level, calculate the cumulative frequency of each gray level. Cumulative frequency: where f(j) is the frequency of gray level j. If the frequency of gray level 0 is 10 and the frequency of gray level 1 is 20, then the cumulative frequency of gray level 2 is 30. Normalize the cumulative distribution function to adjust its value range to 0 to 1 to facilitate subsequent gray level mapping. Record the normalized gray level mapping parameters to facilitate subsequent image processing and conversion. According to the normalized gray level mapping range parameters, perform global pixel equalization conversion on the sharpened enhanced monitoring image. The specific steps are: map each pixel value using the mapping formula M(x,y)=N(I(x,y)), where M is the equalized converted pixel value and I is the sharpened enhanced image. If a pixel value is 120, it may become 200 after mapping, ensuring that the brightness and contrast of the image are improved. Record the main parameters of the global optimization monitoring image, such as contrast, brightness and image quality, to evaluate the effect of equalization conversion.

[0065] In this embodiment, refer to Figure 3 The detailed implementation steps of step S2 include:

[0066] According to the metastatic tumor boundary box, the surrounding tissue blood vessel density change quantitative analysis is performed, and the blood vessel density change feature is extracted;

[0067] The infiltration at the boundary of the metastatic tumor boundary box is identified;

[0068] The metastatic tumor boundary box is subjected to boundary tissue metabolic level change calculation to generate metabolic level change data;

[0069] According to the blood vessel density change characteristics, the infiltration situation at the boundary, and the metabolic level change data, the dynamic fluctuation of the microenvironment of the surrounding tissue is analyzed, so as to generate the change rule of the microenvironment of the surrounding tissue;

[0070] According to the change rule of the microenvironment of the surrounding tissue, the influence degree of the boundary immunity is evaluated, so as to generate the evaluation value of the influence degree of the boundary immunity;

[0071] According to the evaluation value of the influence degree of the boundary immunity, the time sequence immune interaction intensity evolution of the metastatic tumor boundary box is performed, and a multi-cycle immune reaction intensity map is constructed.

[0072] In this embodiment, a suitable blood vessel density calculation method is selected, typically using image processing techniques such as thresholding and morphological operations to extract the vascular structures from the image. The goal of the analysis is to quantify the vascular density in the tissue within and outside the metastatic tumor boundary box. A thresholding method is determined, for example using the Otsu algorithm, to automatically determine the optimal segmentation threshold, enhancing the visualization of the vascular structures. The extraction of the vascular structures is performed separately within and outside the metastatic tumor boundary box, and the ratio of the total area of the blood vessels to the total area of the tissue in each region is calculated to obtain the vascular density (expressed as a percentage of the area of the blood vessels). If the area of the blood vessels within the metastatic tumor boundary box is 500 square millimeters and the total area of the tissue is 2000 square millimeters, then the vascular density is 25%. The characteristics of the change in vascular density are recorded, including the values of the vascular density within and outside the boundary box, and the changes in vascular density at different time points are compared. If the vascular density of the tissue surrounding the metastatic tumor at different time points is 25%, 30%, and 35% respectively, it indicates that the vascular density has a rising trend over time. A suitable method for identifying the infiltration is selected, typically using image analysis techniques (such as edge detection and region growing) to determine the tissue infiltration at the boundary of the metastatic tumor. The degree of infiltration is confirmed by observing the degree of edge blurring and changes in tissue structure. An edge detection algorithm is determined, for example using Canny edge detection or Sobel operator, to enhance the boundary features. The edges of the metastatic tumor boundary box are analyzed to identify whether there is significant tissue infiltration. By comparing the grayscale levels and texture features of normal tissue and infiltrated tissue, the severity of the infiltration is determined. If the tissue grayscale at the edge is significantly reduced and the texture becomes blurred, it can be determined that there is infiltration. The evaluation results of the infiltration are recorded, including the range, depth, and impact on the surrounding tissue of the infiltration, for subsequent analysis. If the depth of the infiltration at the boundary of the metastatic tumor is 2 millimeters and it affects the surrounding normal tissue, special attention should be paid to the pathological changes in this area. A suitable method for calculating the metabolic level is selected, typically using image processing and analysis techniques, combined with the metabolic information of the PET-CT image, to evaluate the metabolic activity of the metastatic tumor and its surrounding tissue. An evaluation index for metabolic level is determined, for example the standard uptake value (SUV), as a quantitative analysis of metabolic activity. The metabolic level is calculated within and outside the metastatic tumor boundary box, and the SUV value of each region is recorded. This process includes extracting the distribution information of the radioactive tracer in the image and calculating the average value in the metastatic tumor and the surrounding tissue. If the SUV value in the metastatic tumor is 5.0 and the SUV value in the surrounding tissue is 2.0, it indicates that the metabolic activity of the metastatic tumor is significantly higher than that of the normal tissue. The metabolic level change data is recorded, and the metabolic activity changes at different time points are analyzed to identify the biological behavior of the metastatic tumor. If the SUV value of the metastatic tumor gradually increases over time, it may indicate that the tumor is progressing. Combining the data of blood vessel density, infiltration, and metabolic level, a suitable statistical analysis method is selected, such as regression analysis and analysis of variance, to explore the relationship between these factors.Determine the key parameters of the analysis, such as correlation coefficient, P value, to assess the impact of different factors on the surrounding tissue microenvironment. Analyze the changes in vascular density, infiltration, and metabolic level comprehensively to identify the dynamic fluctuation rules of the surrounding tissue microenvironment. Compare the data at different time points to assess the trend of the microenvironment changes. If an increase in vascular density is positively correlated with an increase in metabolic level, it may indicate a link between angiogenesis and tumor metabolism. Record the change rules of the surrounding tissue microenvironment, including the change amplitude, trend, and clinical significance, and generate charts for subsequent analysis. Generate a microenvironment feature change curve graph to show the changes in vascular density, infiltration, and metabolic level at different time points. Select an appropriate immune influence evaluation method, usually combined with clinical data, histological examination, and image analysis, to evaluate the infiltration and distribution of immune cells at the metastatic tumor boundary. Determine evaluation indicators such as the density and distribution characteristics of tumor-infiltrating lymphocytes (TILs). According to the microenvironment change rules, evaluate the degree of immune influence at the metastatic tumor boundary. Observe the distribution and activity of immune cells to determine their inhibitory or promotional effects on tumor growth. If a large number of TILs are observed at the metastatic tumor boundary, it indicates that the immune system has a certain response to the tumor. Record the immune influence evaluation values, including the density, distribution of immune cells, and their correlation with tumor growth, to support subsequent analysis. If the evaluation values show high-density infiltrating lymphocytes, it may indicate the immune escape mechanism of the tumor. Select an appropriate immune interaction intensity evaluation method, usually using time series analysis technology, to evaluate the interaction between immune cells and tumor cells. Determine evaluation parameters such as immune cell activity index and cytokine level to quantify the immune response. Perform time-series immune interaction intensity evolution analysis on the metastatic tumor boundary box, analyze the intensity changes of the immune response based on immune evaluation values at different time points. Draw a curve of immune response intensity over time to observe the relationship between immune cell activity and tumor growth. Record the change data of multi-cycle immune response intensity and generate a multi-cycle immune response intensity graph to visually display the dynamic evolution of the immune response. The generated intensity graph shows the changes in immune response at different time points, providing a basis for further clinical decision-making.

[0073] In this embodiment, refer to Figure 4 For the detailed implementation step flowchart of step S3, in this embodiment, the detailed implementation steps of step S3 include:

[0074] Perform individual boundary box metastatic tumor three-dimensional morphology analysis on the metastatic tumor boundary box to generate metastatic tumor three-dimensional morphology features;

[0075] Identify the morphological change characteristics of the metastatic tumor three-dimensional morphology features at multiple time points;

[0076] Perform multi-time point metastatic tumor density calculation according to the metastatic tumor boundary box to generate metastatic tumor densities at multiple time points;

[0077] spatial density distribution mining is performed on the metastasis tumor densities of multiple time points to generate metastasis tumor spatial density distribution features;

[0078] Based on the morphological change features and the metastasis tumor spatial density distribution features, a metastasis tumor change trend evolution is performed to obtain a metastasis tumor image change trend graph at different time points.

[0079] In this example, suitable three-dimensional morphology analysis tools are selected, typically using medical image processing software (e.g., 3DSlicer or ITK-SNAP) to ensure that the image data within the metastasis bounding box can be reconstructed in three dimensions. The objectives and parameters of the analysis are determined, such as the volume, surface area, and shape characteristics (e.g., major axis, minor axis ratio) of the metastasis. For each metastasis bounding box, the three-dimensional data is extracted, segmented, and reconstructed. Threshold segmentation and region growing techniques are used to ensure accurate extraction of the three-dimensional morphology of the metastasis. If a metastasis has a bounding box size of 30mm x 20mm x 15mm, its three-dimensional model can be reconstructed, and its volume and surface area are recorded. The three-dimensional morphological characteristic data of each metastasis, including volume, surface area, shape index, etc., are recorded and compared to identify morphological changes at different time points. If a metastasis has a volume change of 5.0cm³, 6.5cm³, and 8.0cm³ at different time points, it indicates that the tumor is growing. Suitable morphological change analysis methods are selected, typically combining statistical analysis and visualization techniques, to compare the three-dimensional morphological characteristics at different time points and identify their change trends. Key parameters such as growth rate and morphological change rate are determined to quantify morphological changes. The volume change rate and shape change rate are calculated by comparing the three-dimensional characteristics of each metastasis at each time point. Growth rate = (Vt−V0) / V0×100% where Vt is the current volume, and V0 is the initial volume. If a metastasis grows from 5.0cm³ to 6.5cm³, the growth rate is 30%, and this change is recorded for subsequent analysis. For each metastasis bounding box at each time point, the volume of the metastasis and the surrounding tissue is calculated, and the density of the metastasis is calculated. If a metastasis has a volume of 6.5cm³ and the surrounding tissue has a volume of 100cm³, the density of the metastasis is 0.065cm³ / cm³. The densities of multiple metastases at different time points are recorded and compared to identify the density change trend of the metastasis. If the densities of a metastasis at different time points are 0.065, 0.070, and 0.075, respectively, it indicates that the density of the metastasis is gradually increasing. Suitable spatial density analysis methods are selected, typically using heat map generation or spatial interpolation methods (e.g., Kriging interpolation) to visualize the spatial distribution of metastasis density. The spatial range and resolution of the analysis are determined to ensure that the generated density distribution map has high precision. Spatial density data of multiple metastases at different time points are interpolated to generate the corresponding spatial density distribution map. The Kriging interpolation method is used to evaluate the density of the metastasis in different regions. The generated heat map shows the distribution of tumor density in different parts, making it easy to observe high-density areas. The generated spatial density distribution characteristic data, including the center position, distribution range, and high-density area of the density distribution, are recorded for subsequent analysis. If the density of a metastasis in a certain area is significantly higher than that in other areas, special attention should be paid to the pathological changes in that area.In combination with the morphological change characteristics and spatial density distribution characteristics, a suitable trend analysis method, such as linear regression analysis or nonlinear fitting, is selected to evaluate the change trend of the metastatic tumor. Key parameters of the analysis are determined, such as the rate of change, the direction of the trend, and the like. Statistical analysis methods are used to comprehensively analyze the morphological changes and density distribution data at different time points to identify the change trend of the metastatic tumor. If the analysis results show that the volume and density of the metastatic tumor are both increasing, it indicates that the tumor is progressing.

[0080] In this embodiment, step S4 includes the following steps:

[0081] The metastatic tumor bounding box is subjected to multi-scale decomposition processing to obtain metastatic tumor sub-images of different frequency components;

[0082] The pixel gray levels of the metastatic tumor sub-images of different frequency components are traversed to extract the gray level co-occurrence matrix of each frequency;

[0083] The metastatic tumor sub-images of different frequency components are subjected to texture analysis one by one to generate the metastatic tumor texture features of each frequency;

[0084] The gray level co-occurrence matrix of each frequency and the metastatic tumor texture features of each frequency are subjected to deep convolution learning to extract deep metastatic tumor texture features;

[0085] The deep metastatic tumor texture features are used to identify the texture feature differences of the surrounding tissues of the metastatic tumor bounding box, and the texture difference features of the metastatic tumor and the surrounding normal tissues are extracted;

[0086] Based on the texture difference features, the surrounding component micro details are identified to extract potential tumor metastatic cells.

[0087] In this embodiment, a suitable multi-scale decomposition technique is selected, typically using a wavelet transform or a Laplacian pyramid method, to decompose the image within the metastasis bounding box into different frequency components. This process helps extract image features at different scales. The wavelet basis function (such as Haar wavelet or Daubechies wavelet) is determined, and the number of decomposition layers is set, typically 3 to 5 layers, to ensure that sufficient details and contour information are captured. The original image within the metastasis bounding box is subjected to a wavelet transform to generate sub-images of multiple frequency components. The first layer of wavelet decomposition will produce low-frequency and high-frequency sub-images, representing the overall structure and details of the image, respectively. If the image size of the metastasis tumor bounding box is 256x256 pixels, a three-layer wavelet decomposition is performed to obtain low-frequency components and multiple high-frequency components (such as horizontal, vertical and diagonal high frequencies). The image characteristics of each frequency component, including resolution, composition and information volume, are recorded for subsequent texture analysis. The image analysis of the low-frequency components shows the overall structure, while the high-frequency components show the detailed features, providing a basis for subsequent texture analysis. Select a suitable gray-level co-occurrence matrix calculation method, usually calculating its gray-level co-occurrence matrix based on the different frequency components of the image. GLCM is used to describe the spatial relationship between the gray levels of pixels in the image. Determine the direction and distance parameters. Commonly used directions include 0° (horizontal), 90° (vertical) and 45° (diagonal), and the distance parameter is usually set to 1. Perform a pixel gray-level traversal on the metastasis tumor sub-image of each frequency component and construct the corresponding GLCM. For each pixel, check the gray-level values ​​of its neighboring pixels and update the co-occurrence matrix. If in the horizontal direction, the gray-level value of a pixel is 100 and the gray-level value of the neighboring pixel is 120, then update G The corresponding values ​​in the LCM are recorded for each frequency component. These include matrix symmetry, energy, and entropy, to facilitate subsequent texture feature extraction. The energy and entropy values ​​of the GLCM are calculated. If the entropy value of the high-frequency component is 1.2 and that of the low-frequency component is 0.5, this indicates that the high-frequency component has more detailed information. An appropriate texture feature extraction method is selected. The statistical characteristics of the gray-level co-occurrence matrix (such as contrast, correlation, energy, and entropy) are typically used to characterize the texture characteristics of different frequency components. A formula for calculating texture features is determined to facilitate the extraction of effective information from the GLCM. Statistical analysis is performed on the GLCM of each frequency component to extract the texture features at each frequency. The contrast and entropy values ​​of the GLCM are calculated to describe the texture complexity of the image. If the contrast of the high-frequency image is 0.8 and that of the low-frequency image is 0.3, this indicates that the texture of the high-frequency image is more complex. Texture feature data for each frequency is recorded to provide a foundation for subsequent deep convolutional learning. The statistical features of each frequency are summarized to form a feature table for subsequent analysis and modeling.Select an appropriate convolutional neural network (CNN) model, typically a pre-trained model such as VGG16 or ResNet, for transfer learning to extract deep texture features of metastases. Determine the shape of the input layer and parameters of the convolutional layers to accommodate features of different frequency components. Train the deep convolutional network with GLCM and texture features of each frequency component as input to extract deep-level texture features. Input the feature maps of high and low frequency components through multiple layers of convolution and pooling to obtain deep features for each frequency. Record the extracted deep texture features and evaluate the effectiveness of convolution learning, such as through classification accuracy or feature reconstruction error. Select an appropriate difference identification method, typically statistical tests such as t-test or ANOVA, or machine learning classifiers such as support vector machines (SVM), to evaluate the texture feature differences between metastases and surrounding normal tissue. Determine the key features identified, such as the deep texture features of metastases and the feature values of surrounding tissue. Compare the deep texture features of metastases and surrounding normal tissue to identify significantly different features. Through statistical analysis, determine which features are significantly higher or lower in tumor tissue than in normal tissue. If the texture feature entropy of metastases is 1.5 and that of normal tissue is 0.8, it indicates that there is a significant difference in texture complexity between the two. Record the identification results of texture difference features, including significance level and difference degree, for subsequent analysis. If the p-value is less than 0.05, it indicates that there is a significant difference between metastases and normal tissue, which needs further analysis of its clinical significance. Select an appropriate microstructure identification method, typically image segmentation techniques such as threshold segmentation, region growing or machine learning segmentation algorithms, to identify the microstructure of the surrounding tissue of metastases. Determine the parameters and thresholds of the segmentation algorithm to ensure accurate identification of microstructures. Based on the identified texture difference features, extract the microstructure of the surrounding tissue of metastases to identify potential tumor metastatic cells. If cell clusters smaller than 5μm are found at the boundary of metastases, they can be determined as potential metastatic cells. Record the results of microstructure identification, including the number, location and distribution characteristics of potential tumor metastatic cells, to support subsequent clinical decision-making. If 10 potential metastatic cells are identified at the boundary of metastases, further pathological analysis and evaluation are needed.

[0088] In this embodiment, the metastasis boundary box is subjected to multi-scale decomposition to obtain metastasis sub-images of different frequency components;

[0089] The pixel gray level of the metastasis sub-image of different frequency components is traversed to extract the gray level co-occurrence matrix of each frequency;

[0090] According to the metastasis sub-image of different frequency components, texture analysis is performed for each frequency to generate metastasis texture features of each frequency;

[0091] Deep convolution learning is performed on the gray level co-occurrence matrix of each frequency and the transfer tumor texture feature of each frequency, so as to extract deep transfer tumor texture features;

[0092] According to the deep transfer tumor texture feature, a difference in the texture feature of the peripheral tissue of the transfer tumor boundary box is recognized, and a texture difference feature of the transfer tumor and the peripheral normal tissue is extracted;

[0093] Based on the texture difference feature, a peripheral component micro detail is recognized, so as to extract a potential tumor transfer cell.

[0094] In this embodiment, the gray level co-occurrence matrix (GLCM) is a tool used to describe the spatial relationship between the gray levels of pixels in an image. By calculating the gray value pairs of adjacent pixels, the texture information of the image can be obtained. The features of GLCM are used to quantify the texture characteristics of the image, such as contrast, correlation, energy, and entropy. Select the sub-image of the metastatic tumor with different frequency components, and by traversing each pixel, record its gray value and the gray value of the adjacent pixel, and construct the corresponding GLCM. First, determine the direction (such as horizontal, vertical, and diagonal) and distance (usually set to 1 pixel) of the GLCM calculation. For a certain high-frequency component sub-image, traverse each pixel and check the gray value of its adjacent pixels. If the current pixel gray is 100 and the adjacent pixel gray is 120, increase the count at the (100, 120) position of the GLCM. In this way, the GLCM of each frequency component is generated. Record the GLCM of each frequency component and calculate its statistical features, such as symmetry, energy, and entropy, etc. These data will provide the necessary information for subsequent texture feature analysis. If the GLCM energy value of the high-frequency component is 0.75 and the low-frequency component is 0.30, it indicates that the texture features of the high-frequency component are more abundant, suggesting that there may be more complex organizational structures. Use the statistical features of GLCM (such as contrast, correlation, energy, entropy, etc.) to represent the texture features of different frequency components. These features effectively describe the texture complexity and consistency of the image. Calculate the GLCM of each frequency component and extract its texture features. The calculation formula usually includes: contrast: measures the sum of squares of gray differences. Correlation: describes the linear relationship between gray values. Energy: the sum of squares of all elements in the GLCM, reflecting the uniformity of the image. Entropy: reflects the complexity of image information. For the GLCM of the high-frequency component, the contrast is 0.85 and the correlation is 0.6, which indicates that the texture complexity of this component is higher. Select a suitable convolutional neural network (CNN) model, usually use a pre-trained model (such as VGG16, ResNet) for transfer learning, to extract deep texture features from different frequency components. Input the GLCM and texture features of each frequency component into the convolutional neural network for training, and extract deep texture features. Through multiple layers of convolution and pooling operations, higher-level texture feature representations are gradually abstracted. Input the feature maps of high-frequency and low-frequency components into the CNN, and finally obtain the deep features of each frequency through the connection between layers. Use statistical tests (such as t-test, ANOVA) or machine learning classifiers (such as support vector machine SVM) to evaluate the differences in texture features between metastatic tumors and surrounding normal tissues. Compare the deep texture features of metastatic tumors and surrounding normal tissues to identify significant differences. Use statistical analysis to determine which features are significantly higher or lower in tumor tissue than in normal tissue. If the deep texture feature entropy value of the metastatic tumor is 1.8 and that of the normal tissue is 0.7, it indicates that there is a significant difference in texture complexity between the two.The results of identifying the texture difference features, including the significance level and the difference degree, are recorded for subsequent analysis. If the p-value is less than 0.05, it indicates that there is a significant difference between the metastatic tumor and the normal tissue, and further analysis of its clinical significance is needed. Image segmentation techniques, such as threshold segmentation, region growing or machine learning segmentation algorithms, are used to identify the microstructure in the tissue surrounding the metastatic tumor. More accurate segmentation can be combined with texture difference features. For the identified texture difference features, the selected segmentation algorithm is applied to extract the microstructure of the tissue surrounding the metastatic tumor and identify potential tumor metastatic cells. Using the region growing algorithm, the texture difference features are segmented to identify cell clusters smaller than 5 μm, which are judged as potential metastatic cells. The results of microstructure identification, including the number, location and distribution characteristics of potential tumor metastatic cells, are recorded to support subsequent clinical decision-making. If 10 potential metastatic cells are identified at the boundary of the metastatic tumor, further pathological analysis and evaluation are needed.

[0095] In this embodiment, the specific steps of step S6 are:

[0096] According to the multi-cycle immune response intensity map and the potential tumor metastatic cells, the metastasis hotspot regions are distinguished, and a plurality of metastasis hotspot regions are marked;

[0097] The metastasis tumor situation is rolling predicted for the plurality of metastasis hotspot regions and the time sequence metastasis trajectory map, and a plurality of time window metastasis tumor prediction situations are obtained;

[0098] According to the metastasis tumor prediction situations of the plurality of time windows, the metastasis time points and the diffusion trend are analyzed, and a metastasis tumor prediction time axis is constructed.

[0099] In this embodiment, the metastasis hot spot region refers to the area with a higher risk of tumor metastasis, which is usually closely related to the intensity of immune response in the body and the distribution of metastatic cells. At this stage, the goal is to identify and label these areas for subsequent analysis and prediction. Collect multi-cycle immune response intensity maps and potential tumor metastatic cell distribution data. The immune response intensity map is usually obtained through PET scanning or immunohistochemical methods, while the distribution of metastatic cells is extracted through previous image analysis techniques. Using image processing techniques such as threshold segmentation and region growing algorithms, analyze the immune response intensity map to identify areas with intensity higher than a certain threshold (e.g., 0.6). Combine the location of potential tumor metastatic cells and perform overlap analysis to screen out metastasis hot spot regions. Set the intensity threshold to 0.6 and identify 3 hot spot regions, A, B, and C. For each region, record its coordinates, area, and immune response intensity. Record the labeled metastasis hot spot regions and their characteristic data in the database to provide a basis for subsequent metastasis trend prediction. If the immune response intensity of region A is 0.75 and the area is 20mm², this information will be used for subsequent analysis to help judge its metastasis risk. Collect metastasis tumor image data at multiple time points to construct a metastasis trajectory graph. This graph shows the changes of the tumor on the time axis, which can clearly reflect the growth and spread trend of the tumor. Use image registration techniques to align images at different time points to ensure comparison in the same coordinate system. Choose appropriate time series prediction models, such as ARIMA (Autoregressive Integrated Moving Average Model) or LSTM (Long Short-Term Memory Network), to perform rolling prediction on the changes of metastasis hot spot regions. These models need to be trained with historical data. Determine the parameters of the model, such as the size of the time window (e.g., 3 months) and the prediction step (e.g., 1 month), to facilitate effective prediction. Use the trained model to perform rolling prediction on the identified metastasis hot spot regions and record the metastasis tumor trend at each time window. If the prediction result of a hot spot region shows that the volume changes in the next 3 months are 10%, 20%, and 25%, respectively, record these changes. The predicted volume changes of region A are 10cm³, 12cm³, and 15cm³, showing the growth trend of the tumor. Record the metastasis tumor prediction trend at multiple time windows, including volume, spread range, and potential metastasis path. These information will provide the basis for subsequent metastasis time point and spread trend analysis. If the metastasis tumor volume increases from 10cm³ to 12cm³ in the first time window, it indicates that the tumor is growing in that time period. Based on the prediction results of multiple time windows, identify possible metastasis time points. These time points are usually nodes where the tumor volume or spread rate changes significantly. Set a volume change threshold (e.g., 20%) to identify key metastasis time points. If the tumor volume increases from 10cm³ to 15cm³ (50% increase) in a certain time window, mark that time point as an important metastasis node.Analyze the identified metastasis time points to assess the tumor's spread trend. Use trend analysis methods such as linear regression or polynomial fitting to analyze the rate and direction of tumor volume change. Through trend analysis, if it is found that the tumor spreads significantly faster in a certain direction than in other directions, record that direction as the main diffusion path. According to the metastasis time points and the spread trend, construct a metastasis tumor prediction timeline, marking key time nodes and diffusion paths. If the predicted metastasis time points are T1, T2, and T3, and the diffusion paths are A, B, then clearly mark this information on the timeline. Record the constructed metastasis tumor prediction timeline, including time points, volume changes, diffusion paths, and their clinical significance. Ensure the integrity and traceability of the data.

[0100] In this embodiment, an image recognition-based adrenal metastasis tumor prediction analysis system is provided for performing the image recognition-based adrenal metastasis tumor prediction analysis method as described above, comprising:

[0101] An image processing module for acquiring patient adrenal PET-CT monitoring images at multiple time points and performing metastasis tumor visual detection at each time point, extracting metastasis tumor bounding boxes at multiple time points;

[0102] An immune response evaluation module for analyzing the dynamic fluctuations of the surrounding tissue microenvironment and evaluating the degree of immune influence on the boundary based on the metastasis tumor bounding boxes, and constructing a multi-cycle immune response intensity map;

[0103] An image change module for metastasis tumor change trend evolution of the metastasis tumor bounding box, obtaining metastasis tumor image change trend graphs at different time points;

[0104] A difference identification module for identifying the texture feature differences of the surrounding tissue based on the metastasis tumor bounding box, thereby extracting potential tumor metastatic cells;

[0105] A metastasis trajectory tracking module for dynamic metastasis trajectory tracking based on the adrenal PET-CT monitoring images and the metastasis tumor image change trend graphs, and performing full-process metastasis trajectory fitting to construct a time-series metastasis trajectory graph;

[0106] A situation prediction module for metastasis tumor situation rolling prediction of the time-series metastasis trajectory graph based on the multi-cycle immune response intensity map and the potential tumor metastatic cells, and constructing a metastasis tumor prediction timeline.

[0107] The present application provides comprehensive dynamic data for subsequent analysis by collecting adrenal gland PET-CT images at multiple time points, which can effectively track the changes of the tumor. The metastatic tumor and its boundary are accurately identified through image visual detection, avoiding subjective differences in manual judgment and improving detection efficiency and accuracy. The extracted metastatic tumor boundary box at multiple time points provides core data support for subsequent analysis and is the basis for analysis of other modules. By analyzing the immune response around the tumor, the trend of the tumor microenvironment can be evaluated, and the sensitivity of the tumor to the immune response can be inferred. Through the image change trend chart, the changes of the metastatic tumor at different time points can be tracked in real time, revealing the trend of tumor growth, shrinkage or stability. The change trend chart can help identify the development trend of the metastatic tumor, provide early warning and take preventive measures in advance. Texture feature difference recognition can reveal the subtle differences between tumor metastatic cells and surrounding normal tissues, improving the detection accuracy of metastatic cells. Through texture difference analysis, potential diffusion areas of metastatic tumors can be identified early, enhancing the early diagnosis ability of metastatic tumors. The path and speed of tumor metastasis can be accurately tracked, providing a clearer tumor development pattern to help evaluate the risk and severity of metastasis. Through the fitting of the metastatic trajectory chart, from the initial metastasis to the possible diffusion area. Through the analysis of immune response and tumor metastatic cells, combined with the metastatic trajectory chart, the future trend of tumor metastasis can be predicted.

[0108] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, the scope of the present application being defined by the appended claims and not by the description above, therefore all variations falling within the meaning and scope of the equivalent requirements of the application file are intended to be included in the present application.

[0109] The above description is merely that of specific embodiments of the present application, enabling a person skilled in the art to understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An adrenal metastasis prediction analysis method based on image recognition, characterized by, The method comprises the following steps: Step S1: Collecting adrenal gland PET-CT monitoring images of a patient at multiple time points, and performing metastasis tumor visual detection at each time point to extract metastasis tumor bounding boxes at multiple time points; Step S2: According to the metastasis tumor bounding boxes, performing surrounding tissue microenvironment dynamic fluctuation analysis and boundary immune influence degree evaluation, and constructing a multi-cycle immune response intensity diagram; Step S3: Performing metastasis tumor change trend evolution on the metastasis tumor bounding boxes to obtain metastasis tumor image change trend diagrams at different time points; Step S4: According to the metastasis tumor bounding boxes, performing surrounding tissue texture feature difference recognition to extract potential tumor metastasis cells; Step S5: According to the adrenal gland PET-CT monitoring images and the metastasis tumor image change trend diagrams, performing dynamic metastasis trajectory tracking, and performing whole-process metastasis trajectory fitting to construct a time-series metastasis trajectory diagram; Step S6: According to the multi-cycle immune response intensity diagram and the potential tumor metastasis cells, performing metastasis tumor situation rolling prediction on the time-series metastasis trajectory diagram to construct a metastasis tumor prediction time axis.

2. The image recognition-based adrenal metastasis prediction analysis method of claim 1, wherein, The specific steps of step S1 are: Collecting adrenal gland PET-CT monitoring images of a patient at multiple time points; Performing sharpening enhancement and histogram equalization on the PET-CT monitoring images to construct a globally optimized monitoring image; Performing adaptive filtering and noise reduction on the globally optimized monitoring image to obtain a filtering and noise reduction optimized image; Performing metastasis tumor visual detection at each time point on the filtering and noise reduction optimized image to mark the positions of metastasis tumors in the image; According to the positions of metastasis tumors in the image, performing bounding box segmentation to extract metastasis tumor bounding boxes at multiple time points. 3.The image recognition-based adrenal metastasis prediction analysis method of claim 2, wherein The specific steps of performing sharpening enhancement and histogram equalization on the PET-CT monitoring images to construct a globally optimized monitoring image are: Performing Gaussian blur filtering on the PET-CT monitoring images to obtain a blurred filtered image; Calculating the pixel difference between the blurred filtered image and the PET-CT monitoring image to obtain a sharpening mask; According to the sharpening mask, performing sharpening enhancement processing on the PET-CT monitoring image to construct a sharpening enhanced monitoring image; Calculating the gray level histogram of the sharpening enhanced monitoring image; Counting the occurrence frequency of each gray level of the gray level histogram; According to the occurrence frequency of each gray level, calculating the pixel cumulative frequency to obtain a cumulative distribution function; Performing normalization processing on the cumulative distribution function to obtain a gray level mapping range parameter; Performing global pixel equalization conversion on the sharpening enhanced monitoring image through the gray level mapping range parameter to construct a globally optimized monitoring image. 4.The image recognition-based adrenal metastasis prediction analysis method of claim 1, wherein, The specific steps of step S2 are: According to the metastasis tumor bounding boxes, performing quantitative analysis on the change of blood vessel density of the surrounding tissue to extract blood vessel density change characteristics; Identifying the infiltration situation at the boundary of the metastasis tumor bounding box; Performing boundary tissue metabolism level change calculation on the metastasis tumor bounding box to generate metabolism level change data; According to the blood vessel density change characteristics, the boundary infiltration situation, and the metabolism level change data, performing surrounding tissue microenvironment dynamic fluctuation analysis to generate surrounding tissue microenvironment change rules; According to the change rule of the microenvironment of the surrounding tissue, the influence degree of the boundary immunity is evaluated to generate an influence degree evaluation value of the boundary immunity; According to the influence degree evaluation value of the boundary immunity, the time sequence immune interaction intensity evolution of the metastatic tumor boundary box is performed to construct a multi-cycle immune reaction intensity graph. 5.The image recognition-based adrenal metastasis prediction analysis method of claim 1, wherein The specific steps of step S3 are as follows: The three-dimensional morphological analysis of the metastatic tumor boundary box is performed to generate metastatic tumor three-dimensional morphological features; The morphological change features of the metastatic tumor three-dimensional morphological features at multiple time points are identified; The metastatic tumor density at multiple time points is calculated according to the metastatic tumor boundary box to generate metastatic tumor densities at multiple time points; The spatial density distribution of the metastatic tumor densities at multiple time points is mined to generate metastatic tumor spatial density distribution features; Based on the morphological change features and the metastatic tumor spatial density distribution features, the metastatic tumor change trend evolution is performed to obtain metastatic tumor image change trend graphs at different time points. 6.The image recognition-based adrenal metastasis prediction analysis method of claim 1, wherein The specific steps of step S4 are as follows: The metastatic tumor boundary box is subjected to multi-scale decomposition processing to obtain metastatic tumor sub-images of different frequency components; The pixel gray level of the metastatic tumor sub-images of different frequency components is traversed to extract a gray level co-occurrence matrix of each frequency; The texture analysis of the metastatic tumor sub-images of different frequency components is performed to generate metastatic tumor texture features of each frequency; Deep convolution learning is performed on the gray level co-occurrence matrix of each frequency and the metastatic tumor texture features of each frequency to extract deep metastatic tumor texture features; The surrounding tissue texture feature difference of the metastatic tumor boundary box is identified according to the deep metastatic tumor texture features to extract texture difference features of the metastatic tumor and the surrounding normal tissue; Based on the texture difference features, the surrounding component micro details are identified to extract potential tumor metastatic cells. 7.The image recognition-based adrenal metastasis prediction analysis method of claim 1, wherein The specific steps of step S5 are as follows: The metastatic tumor position points at different time points are marked according to the adrenal gland PET-CT monitoring images; The dynamic metastasis trajectory is extracted by performing dynamic metastasis trajectory tracking based on the metastatic tumor image change trend graph and the metastatic tumor position points at different time points; The metastasis speed and the metastasis direction of the dynamic metastasis trajectory are calculated; The whole-process metastasis trajectory fitting is performed according to the metastasis speed and the metastasis direction to construct a time sequence metastasis trajectory graph. 8.The image recognition-based adrenal metastasis prediction analysis method of claim 1, wherein, The specific steps of step S6 are as follows: The metastasis hot spot areas are distinguished according to the multi-cycle immune reaction intensity graph and the potential tumor metastatic cells to mark multiple metastasis hot spot areas; The metastatic tumor situation rolling prediction is performed on the multiple metastasis hot spot areas and the time sequence metastasis trajectory graph to obtain metastatic tumor prediction situations at multiple time windows; The metastasis time points and the diffusion trend are analyzed according to the metastatic tumor prediction situations at multiple time windows to construct a metastatic tumor prediction time axis.

9. An adrenal metastasis prediction analysis system based on image recognition, characterized by, The image recognition-based adrenal gland metastatic tumor prediction analysis method according to claim 1 comprises: An image processing module is configured to collect adrenal gland PET-CT monitoring images of a patient at multiple time points, and perform metastatic tumor visual detection at each time point to extract metastatic tumor boundary boxes at multiple time points; An immune response evaluation module is configured to perform surrounding tissue microenvironment dynamic fluctuation analysis and boundary immune influence degree evaluation according to the metastatic tumor boundary box, and construct a multi-cycle immune response intensity diagram; An image change module is configured to perform metastatic tumor change trend evolution on the metastatic tumor boundary box, and obtain a metastatic tumor image change trend diagram at different time points; A difference identification module is configured to perform surrounding tissue texture feature difference identification according to the metastatic tumor boundary box, and extract potential tumor metastatic cells; A metastasis trajectory tracking module is configured to perform dynamic metastasis trajectory tracking according to the adrenal gland PET-CT monitoring image and the metastatic tumor image change trend diagram, perform full-process metastasis trajectory fitting, and construct a time-series metastasis trajectory diagram; A situation prediction module is configured to perform metastatic tumor situation rolling prediction on the time-series metastasis trajectory diagram according to the multi-cycle immune response intensity diagram and the potential tumor metastatic cells, and construct a metastatic tumor prediction time axis.

Images