A prediction model for tumor radiotherapy prognosis based on multimodal information fusion

Abstract

The present invention discloses a prediction model for tumor radiotherapy prognosis by multimodal information fusion, and relates to the technical field of radiotherapy prognosis prediction. By introducing multimodal information fusion, not only traditional imaging features are taken into account, but also the microscopic biological behavior of the tumor is deeply excavated. By accurately extracting these key parameters and calculating the corresponding scoring values, a health index that comprehensively reflects the overall health status of the tumor is finally obtained based on the scoring values. The present invention can monitor the dynamic evolution of the tumor during the treatment process in real time. Dynamic monitoring enables the model to more accurately predict the future development trend of the tumor. Secondly, the health index, as a comprehensive evaluation indicator, not only reflects the current state of the tumor, but also predicts its future evolution trend. By taking multiple key parameters into consideration, the health index can comprehensively quantify the overall health status of the tumor and provide a more comprehensive and accurate prognosis assessment.

Description

Technical Field

[0001] The present invention relates to the technical field of prognosis prediction, and in particular to a prediction model for tumor radiotherapy prognosis by multi-modal information fusion. Background Art

[0002] Tumor radiotherapy (RT) is one of the important means of cancer treatment and is widely used in the treatment of various malignant tumors. However, the effect of radiotherapy varies from patient to patient. Some patients respond well to radiotherapy, while others may develop drug resistance or adverse reactions. Therefore, accurate prediction of the prognosis of tumor radiotherapy is crucial for developing personalized treatment plans and improving patient survival and quality of life. Existing tumor prognosis assessment methods mainly rely on clinical pathological characteristics (such as tumor stage, histological type, lymph node metastasis, etc.), as well as some simple biomarkers (such as serum markers). In recent years, with the development of imaging technology and molecular biology, more and more studies have shown that multimodal information fusion can more accurately assess the prognosis of tumors and provide strong support for personalized treatment.

[0003] Existing prognostic models directly use imaging features or biomarker data as input for supervised or unsupervised learning. Tumor growth is not just a process of volume increase, but is also accompanied by changes in microscopic biological behaviors such as cell proliferation, necrosis, angiogenesis, metabolic activity, and immune response. These changes in microscopic behaviors are closely related to the prognosis of tumors, but existing models usually only focus on the macroscopic characteristics of tumors and ignore the changes in these microscopic behaviors. Therefore, the model cannot fully capture the complex biological behaviors of tumors, resulting in low prediction accuracy. Summary of the invention

[0004] The purpose of the present invention is to provide a prediction model for tumor radiotherapy prognosis by multimodal information fusion, so as to solve the problem that the complex biological behavior of tumors cannot be fully captured, resulting in low prediction accuracy.

[0005] To achieve the above object, the present invention provides the following technical solution: a prediction model for tumor radiotherapy prognosis based on multimodal information fusion, comprising the following modules:

[0006] Data acquisition module: used to obtain imaging data and biomarker data of patients before, during and after treatment;

[0007] Preprocessing module: connected to the data acquisition module, used for preprocessing the image data and biomarker data;

[0008] Fusion module: connected to the preprocessing module, used to fuse the preprocessed image data with the preprocessed biomarker data to obtain a comprehensive data set;

[0009] Key parameter extraction module: connected with the fusion module, used to extract key parameters from the image data of the comprehensive data set; the key parameters include: tumor volume, growth rate, cell proliferation index, necrotic area ratio, angiogenesis degree, metabolic activity, and immune response intensity;

[0010] Rating value calculation module: connected with the key parameter extraction module, used to calculate the rating value for each key parameter;

[0011] Health index calculation module: connected to the score value calculation module, used to calculate the health index according to each score value;

[0012] Prediction model: connected with the fusion module and the health index calculation module, used to train the prediction model with the comprehensive data set and the health index as the training set, and obtain the trained prediction model; the health index is used as the prediction label; the imaging data and biomarker data of the patient to be predicted are obtained, and the health index is output;

[0013] Prognosis evaluation module: connected to the prediction model, used to generate a prognosis evaluation plan based on the output health index.

[0014] Furthermore, the scoring value calculation module includes the following submodules:

[0015] Tumor volume score calculation submodule: used to calculate the tumor volume score;

[0016] Growth rate score calculation submodule: used to calculate the growth rate score;

[0017] Cell proliferation index score calculation submodule: used to calculate the cell proliferation index score;

[0018] Necrotic area ratio score value calculation submodule: used to calculate the necrotic area ratio index score value;

[0019] Angiogenesis degree score value calculation submodule: used to calculate the angiogenesis degree index score value;

[0020] Metabolic activity score calculation submodule: used to calculate the metabolic activity index score;

[0021] Immune response intensity score value calculation submodule: used to calculate the immune response intensity score value.

[0022] Furthermore, the tumor volume score value calculation submodule includes the following units:

[0023] Actual tumor volume acquisition unit: used to acquire the actual tumor volume of the current patient;

[0024] Tumor volume maximum value acquisition unit: used to obtain the maximum value of the tumor volume;

[0025] A tumor volume acquisition unit at a current time point: used to acquire the tumor volume at a current time point;

[0026] Initial time point tumor volume acquisition unit: used to acquire the tumor volume at the initial time point;

[0027] Tumor volume average value acquisition unit: used to obtain the tumor volume average value:

[0028] Tumor volume score value calculation unit: used to calculate the tumor volume score value according to the actual tumor volume of the current patient, the maximum tumor volume, the tumor volume at the current time point, the tumor volume at the initial time point, and the average tumor volume:

[0029] ;

[0030] In the formula, represents the tumor volume score, represents the actual tumor volume of the current patient, represents the maximum tumor volume, represents pi, , represents the tumor volume at the current time point t, represents the tumor volume at the initial time point, represents the mean tumor volume. , Represent the weight coefficients respectively.

[0031] Furthermore, the growth rate score value calculation submodule includes the following units:

[0032] Tumor growth rate acquisition unit: used to obtain the tumor growth rate of the current patient;

[0033] Tumor growth rate maximum value acquisition unit: used to obtain the maximum value of tumor growth rate;

[0034] Tumor growth rate average value acquisition unit: used to obtain the tumor growth rate average value;

[0035] Tumor growth rate standard deviation acquisition unit: used to acquire the tumor growth rate standard deviation;

[0036] Growth rate score value calculation unit: used to calculate the growth rate score value according to the tumor growth rate, the maximum tumor growth rate, the average tumor growth rate, and the standard deviation of the tumor growth rate:

[0037] ;

[0038] In the formula, Represents the growth rate score, represents the current patient's tumor growth rate, represents the maximum tumor growth rate, represents the average tumor growth rate, represents the standard deviation of tumor growth rate, Represents the adjustment factor.

[0039] Furthermore, the cell proliferation index score value calculation submodule includes the following units:

[0040] Cell proliferation index acquisition unit: used to obtain cell proliferation index;

[0041] Cell proliferation index maximum value acquisition unit: used to obtain the maximum value of the cell proliferation index;

[0042] Cell proliferation index score value calculation unit: used to calculate the cell proliferation index score value according to the cell proliferation index and the maximum value of the cell proliferation index:

[0043] ;

[0044] In the formula, Represents the cell proliferation index score, represents the cell proliferation index, represents the maximum value of cell proliferation index, represents pi, , Represents the adjustment coefficient.

[0045] Furthermore, the necrosis area ratio score value calculation submodule includes the following units:

[0046] Necrotic area ratio acquisition unit: used to obtain the necrotic area ratio;

[0047] Necrotic area ratio score value calculation unit; used to calculate the necrotic area ratio score value according to the necrotic area ratio, the actual tumor volume of the current patient, and the cell proliferation index:

[0048] ;

[0049] In the formula, represents the necrotic area ratio score, N represents the necrotic area ratio, represents the adjustment slope, represents the adjustment range, and a and b represent the adjustment factors respectively.

[0050] Furthermore, the metabolic activity score calculation submodule includes the following units:

[0051] Metabolic activity acquisition unit: used to acquire metabolic activity;

[0052] Metabolic activity maximum value acquisition unit: used to obtain the maximum value of metabolic activity;

[0053] Metabolic activity average value acquisition unit: used to obtain the metabolic activity average value;

[0054] Metabolic activity standard deviation acquisition unit: used to acquire the metabolic activity standard deviation;

[0055] The metabolic activity score value calculation unit is used to calculate the metabolic activity score value according to the maximum metabolic activity value, the average metabolic activity value and the standard deviation of the metabolic activity.

[0056] Furthermore, the immune response intensity score value calculation submodule includes the following units:

[0057] Immune response intensity acquisition unit: used to acquire immune response intensity;

[0058] Immune response intensity maximum value acquisition unit: used to acquire the maximum value of immune response intensity;

[0059] Immune response intensity score value calculation unit: used to calculate the immune response intensity score value according to the immune response intensity, the maximum immune response intensity, and the metabolic activity:

[0060] ;

[0061] In the formula, Represents the immune response intensity score, Represents the strength of the immune response. represents the maximum intensity of immune response, M represents metabolic activity, and e represents the control factor of metabolic activity.

[0062] Furthermore, the angiogenesis degree score value calculation submodule includes the following units:

[0063] Angiogenesis degree acquisition unit: used for acquiring the degree of angiogenesis;

[0064] Angiogenesis maximum value acquisition unit: used for acquiring the maximum value of angiogenesis;

[0065] Angiogenesis degree score calculation unit: used to calculate the angiogenesis degree score according to the angiogenesis degree, the maximum angiogenesis degree, the metabolic activity, and the immune response intensity:

[0066] ;

[0067] In the formula, Represents the angiogenesis score. Represents the degree of angiogenesis, Represents the maximum degree of angiogenesis, , Represent the adjustment weights respectively.

[0068] Furthermore, the health index is:

[0069] ;

[0070] Where H represents the health index.

[0071] Technical effects and advantages of the present invention:

[0072] The present invention comprehensively sublimates the macroscopic characteristics of the tumor by introducing multimodal information fusion, not only considering the traditional imaging features, but also deeply exploring the microscopic biological behavior of the tumor (such as cell proliferation index, necrotic area ratio, angiogenesis degree, metabolic activity and immune response intensity). By accurately extracting these key parameters and calculating the corresponding score values, a health index that comprehensively reflects the overall health status of the tumor is finally obtained based on the score value. The present invention can monitor the dynamic evolution of the tumor during treatment in real time, including volume changes, cell proliferation rate, development of necrotic areas, degree of angiogenesis, and changes in metabolic activity. This dynamic monitoring enables the model to more accurately predict the future development trend of the tumor, avoiding the prediction bias caused by relying solely on static data in the prior art. Secondly, as a comprehensive evaluation indicator, the health index not only reflects the current state of the tumor, but also predicts its future evolution trend. By taking multiple key parameters into consideration, the health index can comprehensively quantify the overall health status of the tumor and provide a more comprehensive and accurate prognostic evaluation. Compared with the practice of focusing only on a specific feature in the prior art, the health index of the present invention can better capture the complex biological behavior of the tumor and improve the accuracy of prognosis prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 A schematic diagram of the structure of a prediction model for tumor radiotherapy prognosis using multimodal information fusion provided in this embodiment. DETAILED DESCRIPTION

[0074] The following will be combined with the drawings in the embodiments of the present invention to clarify the technical solutions in the embodiments of the present invention; it is clearly described in detail. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0075] like Figure 1 As shown, this embodiment discloses a prediction model for tumor radiotherapy prognosis based on multimodal information fusion, including the following modules:

[0076] Data acquisition module: used to obtain imaging data and biomarker data of patients before, during and after treatment.

[0077] Obtain CT, MRI, and PET-CT images of patients before, during, and after treatment, then segment the tumor area of ​​CT, MRI, and PET-CT, and obtain image data from the segmented images. Biomarker data is obtained through blood tests, tissue biopsies, circulating tumor cells (CTCs), and circulating tumor DNA (ctDNA) tests.

[0078] Preprocessing module: connected to the data acquisition module, used for preprocessing the image data and biomarker data.

[0079] Fusion module: connected to the preprocessing module, used to fuse the preprocessed image data with the preprocessed biomarker data to obtain a comprehensive data set.

[0080] The comprehensive dataset includes imaging and biomarker data for each patient before, during, and after treatment.

[0081] Key parameter extraction module: connected to the fusion module, used to extract key parameters from the image data of the comprehensive data set; the key parameters include: tumor volume, growth rate, cell proliferation index, necrotic area ratio, angiogenesis degree, metabolic activity, and immune response intensity.

[0082] Rating value calculation module: connected to the key parameter extraction module, used to calculate the rating value for each key parameter.

[0083] The rating value calculation module includes the following submodules:

[0084] Tumor volume score calculation submodule: used to calculate the tumor volume score.

[0085] The tumor volume score calculation submodule includes the following units:

[0086] Actual tumor volume acquisition unit: used to acquire the actual tumor volume of the current patient.

[0087] The latest imaging data of the patient is retrieved from the hospital's imaging database to ensure that the timestamp of the data is consistent with the current time point. Advanced image segmentation algorithms (such as deep learning models or traditional threshold segmentation methods) are used to automatically segment the tumor area in the image and extract the three-dimensional contour of the tumor. Based on the segmentation results, the volume of the tumor is calculated based on the tumor cross-sectional area and layer thickness of the current layer image. The actual tumor volume is calculated based on the latest imaging data and reflects the size of the tumor at the current time point.

[0088] Tumor volume maximum value acquisition unit: used to acquire the maximum value of the tumor volume.

[0089] The image data at each time point were segmented and the volume was calculated to obtain the tumor volume at each time point. The tumor volumes at all time points were compared and the maximum value was selected as the maximum tumor volume. The maximum tumor volume was extracted from the data at multiple time points throughout the treatment process and reflected the maximum size of the tumor during treatment.

[0090] Current time point tumor volume acquisition unit: used to acquire the tumor volume at the current time point.

[0091] Initial time point tumor volume acquisition unit: used to acquire the tumor volume at the initial time point.

[0092] The current time point refers to the specific time point when the model is predicted, usually the time point of the most recent imaging examination. The volume of the tumor at the current time point is calculated based on the tumor cross-sectional area and layer thickness at the current time point.

[0093] The initial time point usually refers to the time point before the patient starts receiving radiotherapy or other treatments, reflecting the baseline status of the tumor before treatment. The volume of the tumor at the initial time point is calculated based on the tumor cross-sectional area and layer thickness at the initial time point.

[0094] Tumor volume average value acquisition unit: used to acquire the tumor volume average value.

[0095] The mean values ​​were calculated based on the tumor volume at each time point.

[0096] Tumor volume score value calculation unit: used to calculate the tumor volume score value according to the actual tumor volume of the current patient, the maximum tumor volume, the tumor volume at the current time point, the tumor volume at the initial time point, and the average tumor volume:

[0097] ;

[0098] In the formula, represents the tumor volume score, represents the actual tumor volume of the current patient, represents the maximum tumor volume, represents pi, , represents the tumor volume at the current time point t, represents the tumor volume at the initial time point, represents the mean tumor volume. , Represent the weight coefficients respectively.

[0099] Further, It reflects the ratio of the current tumor volume to the maximum tumor volume. When the ratio is close to 1, it means that the tumor volume is large, and when it is much smaller than When the ratio is close to 0, it means that the tumor volume is small. This ratio can intuitively reflect the relative size of the tumor and thus assess the severity of the tumor.

[0100] It is a periodic function whose value varies between (-1, 1). This part introduces periodic fluctuations and uses a sinusoidal function to simulate the periodic characteristics of tumor volume changes, which can reflect the fluctuation of tumor volume during the treatment process.

[0101] , the difference between the current tumor volume and the initial tumor volume is measured by an exponential function and normalized to the mean tumor volume. and When the difference is not large, the value of the exponential function is close to 1, indicating that the tumor volume does not change much. and When the difference is large, the value of the exponential function decreases significantly, indicating a large change in tumor volume. This part can capture the degree of tumor volume change, especially the change relative to the initial volume.

[0102] Growth rate score calculation submodule: used to calculate the growth rate score.

[0103] The growth rate score calculation submodule includes the following units:

[0104] Tumor growth rate acquisition unit: used to obtain the tumor growth rate of the current patient.

[0105] Tumor growth rate, also based on the segmentation results to obtain the tumor growth rate. Specifically, determine the time interval between two adjacent time points The tumor at two time points was calculated by the following formula , The growth rate between :

[0106] .

[0107] Tumor growth rate maximum value acquisition unit: used to obtain the maximum value of tumor growth rate.

[0108] According to the tumor growth rates at multiple time points, the maximum tumor growth rate is obtained.

[0109] Tumor growth rate average value acquisition unit: used to obtain the tumor growth rate average value.

[0110] Tumor growth rate standard deviation acquisition unit: used to obtain the tumor growth rate standard deviation.

[0111] Growth rate score value calculation unit: used to calculate the growth rate score value according to the tumor growth rate, the maximum tumor growth rate, the average tumor growth rate, and the standard deviation of the tumor growth rate:

[0112] ;

[0113] In the formula, Represents the growth rate score, Represents the current patient's tumor growth rate, represents the maximum tumor growth rate, represents the average tumor growth rate, represents the standard deviation of tumor growth rate, Represents the adjustment factor.

[0114] , which reflects the ratio of the current tumor growth rate to the maximum tumor growth rate. , the ratio is close to 1, indicating a high tumor growth rate. , the ratio is close to 0, indicating a low tumor growth rate.

[0115] This part uses an exponential function to measure the difference between the current tumor growth rate and the average tumor growth rate and normalizes it to the standard deviation; it can capture the changing trend of tumor growth rate, especially the change relative to the average value. When G is close to , the value of the exponential function is close to 1, indicating that the tumor growth rate is close to the average value. If the difference is large, the exponential function value will decrease significantly, indicating that the tumor growth rate deviates greatly from the average value.

[0116] Cell proliferation index score calculation submodule: used to calculate the cell proliferation index score.

[0117] The cell proliferation index score calculation submodule includes the following units:

[0118] Cell proliferation index acquisition unit: used to obtain cell proliferation index.

[0119] The cell proliferation index is obtained through biomarker data. Specifically, tumor tissue samples are obtained from patients during surgery or puncture. Ensure that the quality and quantity of the samples are sufficient for subsequent testing. Anti-Ki-67 antibodies are used to stain tumor tissue sections. Ki-67 is a nuclear protein that is only expressed in proliferating cells, so the cell proliferation activity can be evaluated by staining intensity and the proportion of positive cells. After the tumor tissue sections are fixed, dewaxed, and antigen repaired, anti-Ki-67 antibodies are added for incubation. Subsequently, secondary antibodies and color developers are used for color development, and finally the staining results are observed under a microscope. The proportion of Ki-67-positive cells is automatically identified and calculated using image analysis software (such as Aperio ImageScope, QuPath, etc.) to obtain the cell proliferation index:

[0120] .

[0121] Cell proliferation index maximum value acquisition unit: used to obtain the maximum value of the cell proliferation index.

[0122] The maximum cell proliferation index is the maximum value of the cell proliferation index during the entire treatment process. Cell proliferation index data at multiple time points are obtained from the patient's medical records, including PI values ​​before, during, and after treatment. The cell proliferation indexes at all time points are compared, and the maximum value is selected as the maximum cell proliferation index.

[0123] Cell proliferation index score value calculation unit: used to calculate the cell proliferation index score value according to the cell proliferation index and the maximum value of the cell proliferation index:

[0124] ;

[0125] In the formula, Represents the cell proliferation index score, represents the cell proliferation index, represents the maximum value of cell proliferation index, represents pi, , Represents the adjustment coefficient.

[0126] This part reflects the ratio of the current cell proliferation index to the maximum cell proliferation index. , the ratio is close to 1, indicating a high cell proliferation index. When the PI is small, the ratio is close to 0, indicating a low cell proliferation index. This ratio can be used to intuitively reflect the relative level of the current cell proliferation index.

[0127] This part introduces periodic fluctuations through a sine function to simulate the changing trend of the cell proliferation index. , the value of the sine function fluctuates between (-1, 1). The value of the sine function will significantly affect the overall score value, causing the score value to fluctuate within a certain range. This part can capture the changing trend of the cell proliferation index, especially the change relative to the maximum value.

[0128] Necrotic area ratio score value calculation submodule: used to calculate the necrotic area ratio index score value.

[0129] The necrosis area ratio score value calculation submodule includes the following units:

[0130] Necrotic area ratio acquisition unit: used to obtain the necrotic area ratio.

[0131] The proportion of necrotic area is also processed based on the segmentation results. Specifically, the radiologist or pathologist manually or semi-automatically outlines the tumor boundary and necrotic area based on professional knowledge. The overall volume of the tumor and the volume of the necrotic area are accurately measured through three-dimensional reconstruction technology. Based on the overall volume of the tumor and the volume of the necrotic area, the proportion of the necrotic area is obtained. By comparing the image data at different time points, the changes in the proportion of the necrotic area are verified to ensure the reliability of the measurement results.

[0132] Necrotic area ratio score value calculation unit; used to calculate the necrotic area ratio score value according to the necrotic area ratio, the actual tumor volume of the current patient, and the cell proliferation index:

[0133] ;

[0134] In the formula, represents the necrotic area ratio score, N represents the necrotic area ratio, represents the adjustment slope, represents the adjustment range, and a and b represent the adjustment factors respectively.

[0135] , which reflects the influence of the proportion of necrotic area on the score value. When N is large, 1-N is close to 0, resulting in a small overall score value. When N is small, 1-N is close to 1, resulting in a large overall score value.

[0136] This part reflects the influence of tumor volume and cell proliferation index on the score value. When larger, When the cell proliferation index PI is large, the whole score is small, thus reducing the score value. is smaller, resulting in a larger overall score, thus increasing the rating value.

[0137] Angiogenesis degree score calculation submodule: used to calculate the angiogenesis degree index score.

[0138] Metabolic activity score calculation submodule: used to calculate the metabolic activity index score.

[0139] The metabolic activity score calculation submodule includes the following units:

[0140] Metabolic activity acquisition unit: used to obtain metabolic activity .

[0141] PET-CT can visualize the metabolic activity of a tumor by injecting a radiolabeled glucose analog (such as FDG). Areas of high uptake indicate metabolically active cells.

[0142] Metabolic activity maximum value acquisition unit: used to obtain the maximum value of metabolic activity .

[0143] Metabolic activity average value acquisition unit: used to obtain the average value of metabolic activity .

[0144] Metabolic activity standard deviation acquisition unit: used to obtain metabolic activity standard deviation .

[0145] The metabolic activity score calculation unit is used to calculate the metabolic activity score based on the maximum metabolic activity, the average metabolic activity, and the standard deviation of the metabolic activity. :

[0146] . is the influence coefficient.

[0147] Immune response intensity score value calculation submodule: used to calculate the immune response intensity score value.

[0148] The immune response intensity score calculation submodule includes the following units:

[0149] Immune response intensity acquisition unit: used to acquire immune response intensity.

[0150] Electron emission tomography (PET-CT) can show the infiltration of immune cells and evaluate the strength of immune response by injecting radiolabeled immune cells or cytokine analogs. Use professional image analysis software (such as MIM, Fiji, etc.) to quantitatively analyze the immune cell uptake in PET-CT images to obtain the strength of immune response.

[0151] Immune response intensity maximum value acquisition unit: used to acquire the maximum value of immune response intensity.

[0152] Immune response intensity score value calculation unit: used to calculate the immune response intensity score value according to the immune response intensity, the maximum immune response intensity, and the metabolic activity:

[0153] ;

[0154] In the formula, Represents the immune response intensity score, Represents the strength of the immune response. represents the maximum intensity of immune response, M represents metabolic activity, and e represents the control factor of metabolic activity.

[0155] The angiogenesis degree score calculation submodule includes the following units:

[0156] Angiogenesis degree acquisition unit: used for acquiring the angiogenesis degree.

[0157] Select a tumor, normal tissue or other tissue sample of interest, and fix the tissue with an appropriate fixative (such as 4% paraformaldehyde) to maintain its structural integrity. Then dehydrate, clear and other treatments are performed for subsequent imaging. Use specific antibodies to label vascular endothelial cells. Commonly used markers include CD31, CD34, VEGF receptors, etc. These markers can help clearly identify vascular structures in images. Confocal microscopy can provide high-resolution optical sectioning images, which are particularly suitable for thinner tissue sections (usually no more than 1 mm). By scanning layer by layer, a series of two-dimensional images can be obtained. For larger tissue samples or entire organs, micro CT scanning is a non-invasive imaging method. It can provide high-resolution three-dimensional images, which are particularly suitable for hard tissues (such as bones) or calcified tissues. Then perform denoising, registration, and segmentation. Based on the segmented image, a surface reconstruction algorithm (such as Marching Cubes) is used to generate a three-dimensional surface model of the blood vessel. This method can intuitively display the morphology and distribution of blood vessels. Use volume rendering techniques (such as maximum intensity projection MIP, surface occlusion rendering SOR, etc.) to generate a three-dimensional volume image of the blood vessel. This method can more comprehensively display the internal structure and branch complexity of blood vessels. The vascular density and branch complexity are quantitatively analyzed based on three-dimensional reconstruction, and finally the degree of angiogenesis is obtained by summing the vascular density and branch complexity.

[0158] Angiogenesis maximum value acquisition unit: used to acquire the maximum value of angiogenesis.

[0159] Angiogenesis degree score calculation unit: used to calculate the angiogenesis degree score according to the angiogenesis degree, the maximum angiogenesis degree, the metabolic activity, and the immune response intensity:

[0160] ;

[0161] In the formula, represents the angiogenesis score, X represents the angiogenesis degree, Represents the maximum degree of angiogenesis, , Represent the adjustment weights respectively.

[0162] Health index calculation module: connected to the score value calculation module, used to calculate the health index according to each score value;

[0163] ;

[0164] In the formula, H represents the health index, and the value range of H is [0, 1].

[0165] Prediction model: connected with the fusion module and the health index calculation module, used to use the comprehensive data set and the health index as training sets, train the prediction model through the training sets, and obtain a trained prediction model; the health index is used as the prediction label; the imaging data and biomarker data of the patient to be predicted are obtained, and the health index is output.

[0166] Prognosis evaluation module: connected to the prediction model, used to generate a prognosis evaluation plan based on the output health index.

[0167] Patients are divided into different prognostic risk levels according to the numerical range of the health index H. Common risk stratification methods include:

[0168] Low risk: H ≥ 0.8:

[0169] The patient's overall health was good, with an active immune system, stable metabolism, normal tumor marker levels, and no significant abnormalities in imaging.

[0170] Prognosis assessment: The prognosis is good, with a longer expected survival period and a lower risk of recurrence.

[0171] Treatment recommendations: Conservative treatment or observation and follow-up may be considered, with regular monitoring of health status.

[0172] Medium risk: 0.6≤H<0.8:

[0173] The patient's health condition was relatively stable, but there were certain metabolic or immune dysfunctions, slightly elevated tumor markers, and slight abnormalities in imaging.

[0174] Prognosis assessment: The prognosis is general, the expected survival period is moderate, and the risk of recurrence is moderate.

[0175] Treatment recommendations: Standard treatments such as surgery, chemotherapy, radiotherapy, etc. are recommended, combined with immunotherapy or targeted therapy, and regular follow-up.

[0176] High risk: 0.4≤H<0.6:

[0177] The patients are in poor health, with significant abnormalities in metabolism or immune function, significantly elevated tumor markers, and imaging showing that the tumor is large or has metastasized.

[0178] Prognostic assessment: The prognosis is poor, the expected survival period is short, and the risk of recurrence is high.

[0179] Treatment recommendations: It is recommended to adopt an active comprehensive treatment plan, such as multidisciplinary treatment (MDT), and consider participating in clinical trials to explore new treatments.

[0180] Very high risk: H<0.4:

[0181] The patients were in extremely poor health, with severely impaired metabolic and immune functions, extremely high tumor markers, and imaging showing extensive metastasis or multi-organ involvement.

[0182] Prognosis assessment: The prognosis is extremely poor, the expected survival period is very short, and the risk of recurrence is extremely high.

[0183] Treatment recommendations: Palliative care is recommended to relieve symptoms, improve quality of life, and prolong survival. Hospice care can be considered if necessary.

Claims

1. A prediction model for tumor radiotherapy prognosis based on multimodal information fusion, characterized in that: Includes the following modules: Data acquisition module: used to obtain imaging data and biomarker data of patients before, during and after treatment; Preprocessing module: connected to the data acquisition module, used for preprocessing the image data and biomarker data; Fusion module: connected to the preprocessing module, used to fuse the preprocessed image data with the preprocessed biomarker data to obtain a comprehensive data set; Key parameter extraction module: connected with the fusion module, used to extract key parameters from the image data of the comprehensive data set; the key parameters include: tumor volume, growth rate, cell proliferation index, necrotic area ratio, angiogenesis degree, metabolic activity, and immune response intensity; Rating value calculation module: connected with the key parameter extraction module, used to calculate the rating value for each key parameter; The rating value calculation module includes the following submodules: Tumor volume score calculation submodule: used to calculate the tumor volume score; The tumor volume score calculation submodule includes the following units: Actual tumor volume acquisition unit: used to acquire the actual tumor volume of the current patient; Tumor volume maximum value acquisition unit: used to obtain the maximum value of the tumor volume; A tumor volume acquisition unit at a current time point: used to acquire the tumor volume at a current time point; Initial time point tumor volume acquisition unit: used to acquire the tumor volume at the initial time point; Tumor volume average value acquisition unit: used to obtain the tumor volume average value: Tumor volume score value calculation unit: used to calculate the tumor volume score value according to the actual tumor volume of the current patient, the maximum tumor volume, the tumor volume at the current time point, the tumor volume at the initial time point, and the average tumor volume: ； In the formula, represents the tumor volume score, represents the actual tumor volume of the current patient, represents the maximum tumor volume, represents pi, , represents the tumor volume at the current time point t, represents the tumor volume at the initial time point, represents the mean tumor volume. , Represent weight coefficients respectively; Growth rate score calculation submodule: used to calculate the growth rate score; Cell proliferation index score calculation submodule: used to calculate the cell proliferation index score; Necrotic area ratio score value calculation submodule: used to calculate the necrotic area ratio index score value; Angiogenesis degree score value calculation submodule: used to calculate the angiogenesis degree index score value; Metabolic activity score calculation submodule: used to calculate the metabolic activity index score; Immune response intensity score value calculation submodule: used to calculate the immune response intensity score value; Health index calculation module: connected to the score value calculation module, used to calculate the health index according to each score value; Prediction model: connected with the fusion module and the health index calculation module, used to train the prediction model with the comprehensive data set and the health index as the training set, and obtain the trained prediction model; the health index is used as the prediction label; the imaging data and biomarker data of the patient to be predicted are obtained, and the health index is output; Prognosis evaluation module: connected to the prediction model, used to generate a prognosis evaluation plan based on the output health index.

2. The prediction model for tumor radiotherapy prognosis based on multimodal information fusion according to claim 1, characterized in that: The growth rate score calculation submodule includes the following units: Tumor growth rate acquisition unit: used to obtain the tumor growth rate of the current patient; Tumor growth rate maximum value acquisition unit: used to obtain the maximum value of tumor growth rate; Tumor growth rate average value acquisition unit: used to obtain the tumor growth rate average value; Tumor growth rate standard deviation acquisition unit: used to acquire the tumor growth rate standard deviation; Growth rate score value calculation unit: used to calculate the growth rate score value according to the tumor growth rate, the maximum tumor growth rate, the average tumor growth rate, and the standard deviation of the tumor growth rate: ； In the formula, Represents the growth rate score, Represents the current patient's tumor growth rate, represents the maximum tumor growth rate, represents the average tumor growth rate, represents the standard deviation of tumor growth rate, Represents the adjustment factor.

3. The prediction model for tumor radiotherapy prognosis based on multimodal information fusion according to claim 2, characterized in that: The cell proliferation index score calculation submodule includes the following units: Cell proliferation index acquisition unit: used to obtain cell proliferation index; Cell proliferation index maximum value acquisition unit: used to obtain the maximum value of the cell proliferation index; Cell proliferation index score value calculation unit: used to calculate the cell proliferation index score value according to the cell proliferation index and the maximum value of the cell proliferation index: ； In the formula, Represents the cell proliferation index score, represents the cell proliferation index, represents the maximum value of cell proliferation index, represents pi, , Represents the adjustment coefficient.

4. The prediction model for tumor radiotherapy prognosis based on multimodal information fusion according to claim 3, characterized in that: The necrosis area ratio score value calculation submodule includes the following units: Necrotic area ratio acquisition unit: used to obtain the necrotic area ratio; Necrotic area ratio score value calculation unit; used to calculate the necrotic area ratio score value according to the necrotic area ratio, the actual tumor volume of the current patient, and the cell proliferation index: ； In the formula, represents the necrotic area ratio score, N represents the necrotic area ratio, represents the adjustment slope, represents the adjustment range, and a and b represent the adjustment factors respectively.

5. The prediction model for tumor radiotherapy prognosis based on multimodal information fusion according to claim 4, characterized in that: The metabolic activity score calculation submodule includes the following units: Metabolic activity acquisition unit: used to acquire metabolic activity; Metabolic activity maximum value acquisition unit: used to obtain the maximum value of metabolic activity; Metabolic activity average value acquisition unit: used to obtain the metabolic activity average value; Metabolic activity standard deviation acquisition unit: used to acquire the metabolic activity standard deviation; The metabolic activity score value calculation unit is used to calculate the metabolic activity score value according to the maximum metabolic activity value, the average metabolic activity value and the standard deviation of the metabolic activity.

6. The prediction model for tumor radiotherapy prognosis based on multimodal information fusion according to claim 5, characterized in that: The immune response intensity score calculation submodule includes the following units: Immune response intensity acquisition unit: used to acquire immune response intensity; Immune response intensity maximum value acquisition unit: used to acquire the maximum value of immune response intensity; Immune response intensity score value calculation unit: used to calculate the immune response intensity score value according to the immune response intensity, the maximum immune response intensity, and the metabolic activity: ； In the formula, Represents the immune response intensity score, Represents the strength of the immune response. represents the maximum intensity of immune response, M represents metabolic activity, and e represents the control factor of metabolic activity.

7. The prediction model for tumor radiotherapy prognosis based on multimodal information fusion according to claim 6, characterized in that: The angiogenesis degree score calculation submodule includes the following units: Angiogenesis degree acquisition unit: used for acquiring the degree of angiogenesis; Angiogenesis maximum value acquisition unit: used for acquiring the maximum value of angiogenesis; Angiogenesis degree score calculation unit: used to calculate the angiogenesis degree score according to the angiogenesis degree, the maximum angiogenesis degree, the metabolic activity, and the immune response intensity: ； In the formula, Represents the angiogenesis score. Represents the degree of angiogenesis, Represents the maximum degree of angiogenesis, , Represent the adjustment weights respectively.

Images