Auxiliary decision-making model for improving radiotherapy effect

Abstract

The invention discloses an auxiliary decision-making model for improving a radiotherapy effect, and relates to the technical field of radiotherapy. Comprising an image data collection module, an analysis module, an initial radiotherapy dose distribution diagram generation sub-module, a dose acquisition sub-module, a dose optimization sub-module, a radiotherapy mode formulation module, a tumor position determination sub-module, a normal tissue position determination sub-module, a distance calculation sub-module, an optimal dose value calculation sub-module and a radiotherapy mode determination sub-module. On the basis that a dose distribution diagram is generated based on an existing model, a dose value is optimized to obtain an optimized dose, meanwhile, the tumor position of a patient is analyzed, the distance between the tumor and adjacent normal tissue is analyzed, and the optimal dose is obtained through calculation. The dosage optimization ensures that a tumor area is effectively irradiated, meanwhile, the damage to surrounding normal tissues is reduced to the maximum extent, a dynamic adjustment mechanism is realized, and the treatment accuracy is improved while key organs are protected.

Description

Technical Field

[0001] The present invention relates to the field of radiotherapy technology, and in particular to an auxiliary decision-making model for improving radiotherapy effects. Background Art

[0002] In modern medicine, radiotherapy is one of the important means of treating malignant tumors. However, how to accurately control the radiotherapy dose to ensure that the tumor area is effectively irradiated while minimizing the impact on surrounding healthy tissues has always been a difficult problem in clinical practice. Traditional radiotherapy planning relies on the doctor's experience and manual operation. This method is not only time-consuming and labor-intensive, but also due to the lack of individualized considerations, it is often difficult to achieve ideal treatment effects. In recent years, with the rapid development of artificial intelligence technology, especially the application of deep learning algorithms, new approaches have been provided to solve the above problems. Among the existing technologies, an artificial intelligence-assisted decision-making model that combines individual patient information with historical case data has been proposed and applied in clinical practice. The workflow of the model is as follows: 1. Data collection: First, collect basic information such as the patient's imaging data (such as CT and MRI images), pathology reports, and medical history. These data form the basis of model input.

[0003] 2. Deep Learning Analysis: The collected data is input into a pre-trained deep neural network. After learning from a large number of historical cases, the network has the ability to identify different tumor characteristics and recommend corresponding treatment plans.

[0004] 3. Personalized treatment recommendations: Deep neural networks select the most appropriate radiotherapy plan from existing treatment models based on the patient's specific situation and provide detailed dose allocation recommendations. 4. Physician Decision Support: Ultimately, doctors develop personalized radiotherapy plans based on the model's recommendations and their own expertise and experience. This process not only improves treatment accuracy but also significantly reduces planning time.

[0005] While existing technologies have improved the efficiency and effectiveness of radiotherapy to a certain extent, some shortcomings remain in practical application. Existing technologies use deep learning to predict a dose distribution map. Based on the dose distribution map's recommended values ​​and the patient's and tumor's pathological conditions, doctors adjust the dose values ​​based on their own experience to generate a radiotherapy plan. However, differences in experience and judgment among doctors can affect the consistency and accuracy of treatment plans. Furthermore, the generated dose distribution map is static and cannot be dynamically adjusted based on patient response and treatment progress. The lack of a mechanism for real-time adjustments based on actual conditions can affect treatment effectiveness and result in ineffective radiotherapy. Summary of the Invention

[0006] The purpose of the present invention is to provide an auxiliary decision-making model for improving the effect of radiotherapy, so as to solve the problem that the dose value cannot be adjusted in real time.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an auxiliary decision-making model for improving radiotherapy effects, comprising the following modules: Image data collection module: used to collect patient image data; Analysis module: connected to the image data collection module, used to analyze the image data using a deep neural network to obtain tumor areas and normal tissue areas; Radiotherapy dose distribution map generation module: connected to the analysis module, used to generate the radiotherapy dose optimization distribution map based on the segmentation results; The radiotherapy dose optimization distribution map generation module includes the following submodules: Initial radiotherapy dose distribution map generation submodule: used to generate the initial radiotherapy dose distribution map; Dose acquisition submodule: used to obtain the dose contribution in the tumor area and normal tissue area; Dose optimization submodule: used to determine the optimized dose based on dose contribution; Radiotherapy mode formulation module: connected to the radiotherapy dose distribution map generation module, used to formulate radiotherapy mode; The radiotherapy mode formulation module includes the following submodules: Tumor location determination submodule: used to determine the tumor location; Normal tissue position determination submodule: used to determine the normal tissue position; Distance calculation submodule: used to calculate the distance between the tumor location and the normal tissue location; Optimal dose value calculation submodule: used to calculate the optimal dose value based on the distance between the tumor position and the normal tissue position; Radiotherapy mode determination submodule: used to determine the radiotherapy mode according to the optimal dose value.

[0008] In another embodiment, an expected dose and a peak dose are constructed for the tumor region; An optimization function is constructed based on the dose contribution within the tumor area, the dose contribution within the normal tissue area, the expected dose, and the peak dose. The dose in the initial radiotherapy dose distribution map is adjusted to the optimized dose through an iterative optimization algorithm.

[0009] In another embodiment, the optimization function is: ; Where F is the optimization function, is the dose contribution within the tumor region, Dose contribution to normal tissue regions, The expected dose, is the peak dose, 、 are weight coefficients respectively.

[0010] In another embodiment, the tumor location includes: the center coordinates and radius of the tumor region.

[0011] In another embodiment, the normal tissue position is the coordinates of the boundary pixel points of the normal tissue area.

[0012] In another embodiment, the optimal dosage value is calculated as follows: ; Where, Pixels in the tumor area The optimal dose value, Pixels in the tumor area The optimal dose value, k is the adjustment factor, is the center coordinate of the tumor area, and R is the radius of the tumor area.

[0013] In another embodiment, a safety distance is preset; Obtain the adjacent normal tissue area around the tumor area, calculate the distance between the center coordinates of the tumor area and the coordinates of the boundary pixels of the adjacent normal tissue area, and screen out the minimum distance from the distances; based on the minimum distance and the safety distance, dynamically adjust the adjustment factor: when the minimum distance is less than the safety distance, reduce the adjustment factor; when the minimum distance is greater than or equal to the safety distance, increase the adjustment factor.

[0014] In another embodiment, the minimum distance is: ; Where, is the minimum distance, min is the minimum value operation, is the coordinate of the boundary pixel point adjacent to the normal tissue area, and O is the boundary of the normal tissue area.

[0015] In another embodiment, the adjustment factor is: ; ; Where, is the adjustment factor when the minimum distance is less than the safety distance, is the adjustment factor when the minimum distance is greater than or equal to the safety distance, For a safe distance, is the baseline value.

[0016] Technical effects and advantages of the present invention: The present invention generates a dose distribution map based on an existing model and optimizes the dose value to obtain an optimized dose. At the same time, the patient's tumor location and the distance between the tumor and adjacent normal tissue are analyzed to calculate the optimal dose. In this way, the present invention can significantly improve the effect of radiotherapy, which is specifically reflected in the following aspects: First, dose optimization ensures that the tumor area is effectively irradiated while minimizing damage to surrounding normal tissues; second, the dose distribution is adjusted in real time according to the distance between the tumor and normal tissue to achieve a dynamic adjustment mechanism, ensuring that the accuracy of treatment is improved while protecting key organs; in addition, the automated process greatly reduces the need for manual intervention and improves the efficiency and consistency of treatment planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of the structure of a decision-making support model for improving radiotherapy effects provided in this embodiment; Figure 2 This is a structural block diagram of a radiotherapy dose optimization distribution map generation module of an auxiliary decision model for improving radiotherapy effects provided in this embodiment; Figure 3 This is a structural block diagram of a radiotherapy mode formulation module of an auxiliary decision-making model for improving radiotherapy effects provided in this embodiment. DETAILED DESCRIPTION

[0018] The following will be combined with the accompanying drawings to clarify the technical solutions in the embodiments of the present invention; although it is clear that the embodiments described are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] This paper proposes a decision-support model for improving radiotherapy outcomes. By integrating multiple modules, including imaging data collection, deep learning analysis, dose distribution map generation, and radiotherapy pattern development, it achieves high-precision tumor segmentation and personalized dose optimization. This model can effectively handle complex tumor structures, reduce the impact on surrounding vital organs, and significantly improve the efficiency and effectiveness of radiotherapy. By dynamically adjusting the dose values ​​in the dose distribution map, it ensures that the tumor area receives an adequate dose while protecting surrounding normal tissue, thus providing a more accurate and effective radiotherapy decision-support tool.

[0020] like Figure 1 As shown, this embodiment discloses an auxiliary decision model for improving radiotherapy effects, including the following modules: Image data collection module: used to collect patient image data.

[0021] Imaging data includes CT and MRI images. Patient CT and MRI images are extracted from the hospital information system, ensuring they are in a uniform format (e.g., DICOM). Denoising is performed on the images to remove noise interference and normalize them to ensure pixel values ​​are within a uniform range. Image registration is performed to ensure spatial consistency between images of different modalities. Preprocessed images are stored in a database for easy access by subsequent modules.

[0022] Analysis module: connected to the image data collection module, used to analyze the image data using a deep neural network to obtain the tumor area and normal tissue area.

[0023] Load preprocessed CT and MRI images from the database. Initialize a deep learning-based convolutional neural network (CNN), such as the U-Net architecture. Load pretrained weights to accelerate convergence and improve segmentation accuracy. Use convolutional and pooling layers to extract low-level to high-level features of the image step by step. Apply an attention mechanism at each level to enhance the feature representation of important regions. Use a multi-scale feature fusion module to integrate feature information from different levels, merge high-level and low-level features, and improve segmentation detail and accuracy. Input the image into the deep neural network to generate a preliminary segmentation result. The output result is a binary mask that identifies the tumor area. Morphological operations (such as dilation and erosion) are applied to optimize the segmentation result. Boundary smoothing is performed to reduce jagged effects. The final segmentation result is displayed for use by the subsequent dose distribution map generation module.

[0024] like Figure 2 As shown, the radiotherapy dose distribution map generation module is connected to the analysis module and is used to generate a radiotherapy dose optimization distribution map based on the segmentation results.

[0025] The system consists of three submodules: initial radiation dose distribution map generation submodule, dose acquisition submodule, and dose optimization submodule. These submodules generate highly accurate radiation dose distribution maps, ensuring that the tumor area is effectively irradiated while minimizing damage to surrounding normal tissues.

[0026] The radiotherapy dose optimization distribution map generation module includes the following submodules: Initial radiotherapy dose distribution map generation submodule: used to generate the initial radiotherapy dose distribution map.

[0027] Load high-precision tumor segmentation results from the analysis module and set the target dose and maximum allowable dose. These parameters can be set according to clinical standards. Generate an initial dose distribution map using a pretrained deep learning model (such as U-Net): The input includes the tumor segmentation results and the normal tissue segmentation results, and the output is an initial dose distribution map. The initial dose distribution map shows the dose value for each pixel.

[0028] Dose acquisition submodule: used to obtain the dose contribution in the tumor area and normal tissue area.

[0029] ; ; Where, is the dose contribution within the tumor region, Dose contribution to normal tissue regions, is the pixel point in the tumor area, is the pixel point in the normal tissue area, T is the tumor area, and N is the normal tissue area.

[0030] Dose optimization submodule: used to determine the optimized dose based on dose contribution.

[0031] The expected dose and peak dose are constructed for the tumor region. Based on the dose contribution within the tumor region, the dose contribution within the normal tissue region, the expected dose, and the peak dose, an optimization function is constructed to balance the dose to the tumor region and the dose to normal tissue. The dose in the initial radiotherapy dose distribution map is adjusted to the optimized dose through an iterative optimization algorithm.

[0032] Furthermore, the dose values ​​in the dose distribution map are adjusted using a gradient descent method or other optimization algorithm. The goal is to find a set of dose values ​​in the dose distribution map that minimizes the entire optimization function and obtains the optimized dose. The dose distribution map is updated based on the optimized dose values ​​to obtain the optimized dose distribution map.

[0033] ; Where F is the optimization function, is the dose contribution within the tumor region, Dose contribution to normal tissue regions, is the target dose, is the peak dose, 、 are weight coefficients respectively.

[0034] Ensure that the dose contribution of the tumor area is close to the target dose. When the dose contribution is close to the target dose, this item is close to 0, indicating that the tumor area has received an appropriate dose.

[0035] Ensure that the dose contribution of surrounding normal tissue does not exceed the maximum permissible dose. , which is close to 0, indicating that the surrounding normal tissues are not exposed to excessive radiation.

[0036] Radiotherapy mode formulation module: connected to the radiotherapy dose distribution map generation module, used to formulate radiotherapy mode; like Figure 3 As shown, the radiotherapy mode formulation module includes the following submodules: Tumor location determination submodule; used to determine the tumor location.

[0037] The tumor location includes: the center coordinates of the tumor area and radius R.

[0038] , ; .

[0039] i is the x-coordinate of the pixel point in the tumor area, j is the y-coordinate of the pixel point in the tumor area, is the number of pixels in the tumor area.

[0040] Normal tissue position determination submodule: used to determine the normal tissue position.

[0041] The normal tissue position is the coordinates of the boundary pixels of the normal tissue area, which is obtained through edge detection algorithm.

[0042] Distance calculation submodule: used to calculate the distance between the tumor location and the normal tissue location.

[0043] The distance between the center coordinates of the tumor region and the coordinates of the boundary pixels of the adjacent normal tissue region is calculated, and the minimum distance is selected from these distances. The purpose is to screen out the normal tissue closest to the tumor, ensuring that the radiotherapy process does not inadvertently damage critical organs. The normal tissue selected by the minimum distance is preferably critical organs such as the heart and lungs.

[0044] ; Where, is the minimum distance, min is the minimum value operation, is the coordinate of the boundary pixel point adjacent to the normal tissue area, and O is the boundary of the normal tissue area.

[0045] Optimal dose value calculation submodule: used to calculate the optimal dose value based on the distance between the tumor position and the normal tissue position; The formula for calculating the optimal dose value is: ; Where, Pixels in the tumor area The optimal dose value, Pixels in the tumor area The optimal dose value, k is the adjustment factor, is the center coordinate of the tumor area, and R is the radius of the tumor area.

[0046] Furthermore, this embodiment preferably selects the heart and lungs as key organs that need to be protected in this embodiment to avoid overdose of key organs during radiotherapy. Therefore, a safety distance is preset; referring to medical literature and clinical standards, the safety distance of each important organ is determined, the safety distance for the heart is 30 mm, and the safety distance for the lungs is 20 mm.

[0047] The adjustment factor is dynamically adjusted based on the minimum distance and safety distance: when the minimum distance is less than the safety distance, the adjustment factor is reduced to reduce the dose to critical organs; when the minimum distance is greater than or equal to the safety distance, the adjustment factor is increased to increase the dose to the tumor.

[0048] ; ; Where, is the adjustment factor when the minimum distance is less than the safety distance, is the adjustment factor when the minimum distance is greater than or equal to the safety distance, For a safe distance, is the baseline value.

[0049] Using the exponential function The adjustment factor can be reduced rapidly as the distance decreases, thereby effectively protecting key organs. The adjustment factor can be gradually increased as the distance increases, thereby improving the radiotherapy effect.

[0050] Baseline value It is the basic value of the adjustment factor, which is usually set to a reasonable initial value, referring to medical literature and clinical standards. .

[0051] This embodiment calculates the optimal dose value corresponding to the vital organs based on the vital organs, and generates the optimal dose distribution map corresponding to the vital organs based on the optimal dose value.

[0052] Radiotherapy mode determination submodule: used to determine the radiotherapy mode according to the optimal dose value.

[0053] Determine the beam angle, dose rate, number of irradiations, and other parameters based on the optimal dose distribution map. Use inverse planning technology to generate a specific radiotherapy plan based on the optimal dose distribution map. Input the optimal dose distribution map and use an optimization algorithm (such as gradient descent or other optimization algorithms) to determine the beam angle, dose rate, number of irradiations, and other parameters. For example: Multi-angle irradiation is achieved using a multileaf collimator (MLC) and a rotating gantry. Multiple beam angles (e.g., six or more) are typically selected to cover the entire tumor area. The dose rate is set based on the capabilities of the treatment equipment and the patient's specific circumstances. For example, a dose rate of 1 Gy / min is used. The number of fractions is determined based on the total treatment dose and the dose per fraction. For example, if the total treatment dose is 60 Gy and the dose per fraction is 2 Gy, the number of fractions is 30.

[0054] For example, the tumor in this example occurs in the chest, and the heart and lungs need to be protected. The specific radiotherapy plan is as follows: Beam angles: Six beam angles are used: 0°, 60°, 120°, 180°, 240°, and 300°; each angle covers the entire tumor area while avoiding direct irradiation of the heart and lungs.

[0055] Dose rate: 1 Gy / min.

[0056] Number of irradiation doses: For a total treatment dose of 60 Gy and a single 2 Gy dose, the number of irradiation doses is 30. Treatment is completed over 6 weeks, with 2 Gy doses delivered 5 days per week. A 3D dose distribution map is generated, showing high-dose areas in the tumor and low-dose areas in normal tissue.

[0057] This embodiment also generates a detailed report to help doctors understand the following: Distance between tumor and vital organs: Clarify the actual distance between the tumor and key organs.

[0058] Adjustment Factor: Displays the dynamically adjusted adjustment factor to help doctors understand the reasons and extent of dose adjustment.

[0059] Optimal dose distribution map: Visually display the optimal dose distribution and intuitively present the distribution of dose in different areas.

[0060] Recommended radiotherapy mode: The optimal beam angle, dose rate, number of irradiations and other parameters are recommended for the doctor's reference.

[0061] Risk assessment: Analyze potential risks and side effects and provide preventive measures and monitoring plans.

[0062] The radiotherapy mode determination submodule loads the segmentation results and optimal dose values, calculates the minimum distance, and dynamically adjusts the adjustment factor to ultimately determine the optimal dose value. Based on the optimal dose value, the radiotherapy mode is determined, including parameters such as beam angle, dose rate, and number of exposures. A detailed radiotherapy plan is generated using inverse planning techniques, including multiple beam angles, appropriate dose rates, and number of exposures. These measures assist physicians in making radiotherapy decisions, enhance radiotherapy response, improve outcomes, and minimize harm to patients.

Claims

1. A decision-making support model for improving radiotherapy effects, characterized in that: Includes the following modules: Image data collection module: used to collect patient image data; Analysis module: connected to the image data collection module, used to analyze the image data using a deep neural network to obtain the tumor area and normal tissue area; Radiotherapy dose distribution map generation module: connected with the analysis module, used to generate a radiotherapy dose optimization distribution map according to the segmentation results; The radiotherapy dose optimization distribution map generation module includes the following submodules: Initial radiotherapy dose distribution map generation submodule: used to generate initial radiotherapy dose distribution map; Dose acquisition submodule: used to obtain dose contributions in tumor areas and normal tissue areas; Dose optimization submodule: used to determine the optimized dose based on dose contribution; Radiotherapy mode formulation module: connected to the radiotherapy dose distribution map generation module, used to formulate radiotherapy mode; The radiotherapy mode formulation module includes the following sub-modules: Tumor location determination submodule: used to determine the tumor location; Normal tissue position determination submodule: used to determine the normal tissue position; Distance calculation submodule: used to calculate the distance between the tumor location and the normal tissue location; Optimal dose value calculation submodule: used to calculate the optimal dose value according to the distance between the tumor position and the normal tissue position; Radiotherapy mode determination submodule: used to determine the radiotherapy mode according to the optimal dose value.

2. The auxiliary decision model for improving radiotherapy effect according to claim 1, characterized in that: Construct expected dose and peak dose for the tumor region; An optimization function is constructed based on the dose contribution in the tumor area, the dose contribution in the normal tissue area, the expected dose, and the peak dose. The dose in the initial radiotherapy dose distribution map is adjusted to the optimized dose through an iterative optimization algorithm.

3. The auxiliary decision model for improving radiotherapy effect according to claim 2, characterized in that: The optimization function is: ； Where F is the optimization function, is the dose contribution within the tumor region, is the dose contribution to the normal tissue region, is the expected dose, is the peak dose, , are weight coefficients respectively.

4. The auxiliary decision model for improving radiotherapy effect according to claim 1, characterized in that: The tumor position includes: the center coordinates and radius of the tumor area.

5. The auxiliary decision model for improving radiotherapy effect according to claim 4, characterized in that: The normal tissue position is the coordinates of the boundary pixel points of the normal tissue area.

6. The auxiliary decision model for improving radiotherapy effect according to claim 1, characterized in that: The optimal dose value is calculated as follows: ； In the formula, Pixels in the tumor area The optimal dose value, Pixels in the tumor area The optimal dose value, k is the adjustment factor, is the center coordinate of the tumor area, and R is the radius of the tumor area.

7. The auxiliary decision model for improving radiotherapy effect according to claim 6, characterized in that: Preset safety distance; The adjacent normal tissue area around the tumor area is obtained, the distance between the center coordinates of the tumor area and the coordinates of the boundary pixel points of the adjacent normal tissue area is calculated, and the minimum distance is screened out from the distances; based on the minimum distance and the safety distance, the adjustment factor is dynamically adjusted: when the minimum distance is less than the safety distance, the adjustment factor is reduced; when the minimum distance is greater than or equal to the safety distance, the adjustment factor is increased.

8. The auxiliary decision model for improving radiotherapy effect according to claim 7, characterized in that: The minimum distance is: ； In the formula, is the minimum distance, min is the minimum value operation, are the coordinates of the boundary pixels adjacent to the normal tissue area, and O is the boundary of the normal tissue area.

9. The auxiliary decision model for improving radiotherapy effect according to claim 8, characterized in that: The adjustment factor is: ； ； In the formula, is the adjustment factor when the minimum distance is less than the safety distance, is the adjustment factor when the minimum distance is greater than or equal to the safety distance, For a safe distance, is the benchmark value.

Images