Self-adaptive aid decision-making method, system and device for patient radiotherapy and medium

Abstract

The invention discloses a self-adaptive aid decision-making method, system and device for patient radiotherapy, and a medium, and relates to the field of smart medical treatment. The method comprises the following steps: standardizing and preprocessing multi-source original data; performing automatic segmentation and change detection on tumor and normal tissues based on the processed data; delineating a tumor target region and a normal tissue in radiotherapy in a detection result; generating radiation therapy dose distribution according to the sketching result; evaluating the radiation treatment dose distribution by adopting an optimization evaluation method to generate a plurality of alternative treatment schemes; determining an optimal treatment regimen based on the plurality of alternative treatment regimens; and performing radiotherapy based on the optimal treatment scheme, monitoring anatomical changes of the patient in real time in the radiotherapy process, and dynamically adjusting the optimal treatment scheme. According to the invention, a plurality of operations of patient radiotherapy can be effectively integrated, and precision and individuation of patient radiotherapy can be realized, so that complex changes in the treatment process can be better coped with, and the overall treatment effect can be improved.

Description

Technical Field

[0001] The present application relates to the field of smart medical care, and in particular to an adaptive decision-making assistance method, system, device and medium for patient radiotherapy. Background Art

[0002] Radiation therapy is one of the main means of treating many cancers. It uses high-energy rays to destroy the DNA (Deoxyribonucleic Acid) of cancer cells, thereby inhibiting or eliminating tumors. However, the challenge of radiation therapy is how to irradiate tumor tissue to the greatest extent while minimizing damage to surrounding normal tissues. Traditional radiation therapy usually relies on static treatment plans developed before the start of treatment, which are usually based on the location and shape of the tumor in the initial imaging data. However, tumors and normal tissues may undergo significant changes during treatment, including tumor shrinkage, displacement, morphological changes, or organ movement due to changes in patient weight, breathing or other physiological activities. These changes may lead to unplanned radiation dose distribution, thereby reducing the effectiveness of treatment and increasing the risk of side effects.

[0003] Existing radiotherapy technologies include static radiotherapy, image-guided radiotherapy, adaptive radiotherapy, etc., all of which have their limitations. In traditional static radiotherapy, once the treatment plan is formulated, it usually remains unchanged throughout the treatment cycle. However, the physiological changes of patients make it difficult for these static plans to effectively cope with the dynamic changes of tumors and normal tissues during treatment, and lack the ability to adjust in real time. Image-guided radiotherapy is a widely used technology that verifies the patient's position and the shape of the tumor by using imaging techniques [such as cone beam CT (Computed Tomography), ultrasound, and MRI (Magnetic Resonance Imaging)] before or during each treatment. IGRT (Image Guide Radiation Therapy) is usually still based on a pre-established static plan, and only corrects the patient's position through simple displacement adjustments. Although this method can partially improve the accuracy of treatment, it still has the problem of being unable to track and adjust tumor changes during treatment, requiring a lot of manual intervention, and being unable to optimize dose distribution more finely. Adaptive Radiotherapy (ART) is an advanced radiotherapy technology that aims to overcome the limitations of traditional radiotherapy and image-guided radiotherapy by dynamically adjusting the treatment plan according to the patient's physiological changes during the treatment process. The basic idea of ​​ART is to evaluate the changes in tumors and normal tissues based on real-time or staged imaging data and adjust the radiotherapy plan accordingly. However, the existing ART technology also has problems in practical application, such as complex data processing, a large number of manual adjustments that are time-consuming and error-prone, and a lack of intelligent decision-making support.

[0004] Based on the above description, known technologies have significant limitations when facing the dynamic changes of the patient's anatomical structure during radiotherapy. Traditional static treatment plans and existing image-guided technologies fail to fully meet the needs of dynamic adjustment. Although adaptive radiotherapy has theoretical advantages, it is still subject to many technical challenges in practical applications. For example, it is impossible to effectively integrate operations such as data acquisition, image analysis, dose optimization, real-time adjustment, and intelligent decision support, and it is impossible to achieve precision and personalization of patient radiotherapy, and it is impossible to better cope with the complex changes in the treatment process. Summary of the invention

[0005] The purpose of this application is to provide an adaptive decision-making support method, system, device and medium for patient radiotherapy, which can effectively integrate operations such as data acquisition, image analysis, dose optimization, real-time adjustment and intelligent decision support, so as to achieve precision and personalization of patient radiotherapy, so as to better cope with complex changes in the treatment process and improve the overall effect of treatment.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] In a first aspect, the present application provides an adaptive decision-making assistance method for radiotherapy of a patient, comprising:

[0008] Acquiring raw data for radiotherapy of patients; the raw data includes multimodal imaging data, treatment history data, biomarker data and individual patient information;

[0009] Performing standardization and preprocessing on the raw data to obtain processed data;

[0010] Performing automatic segmentation and change detection of tumor and normal tissue based on the processed data to obtain detection results;

[0011] outlining the tumor target area and normal tissue in radiotherapy in the detection results to obtain an outlining result;

[0012] generating a radiation therapy dose distribution according to the delineation result;

[0013] Using an optimization evaluation method to evaluate the radiation therapy dose distribution to obtain an evaluation result, and generating a plurality of alternative treatment plans based on the evaluation result;

[0014] determining an optimal treatment regimen based on a plurality of said alternative treatment regimens;

[0015] Implement radiotherapy based on the optimal treatment plan, and monitor the patient's anatomical changes in real time during radiotherapy to obtain the change results;

[0016] The optimal treatment plan is dynamically adjusted based on the changing results.

[0017] Optionally, the raw data is subjected to standardization and preprocessing to obtain processed data, including:

[0018] Performing standardization processing on the original data to obtain standardized data;

[0019] Using an image processing algorithm to eliminate noise and artifacts in the standardized data to obtain intermediate data;

[0020] The data in the intermediate data is aligned using a registration algorithm to obtain the processed data.

[0021] Optionally, automatic segmentation and change detection of tumor and normal tissue are performed based on the processed data to obtain detection results, including:

[0022] A deep learning algorithm is used to automatically segment tumors and normal tissues and detect changes based on the processed data to obtain the detection results.

[0023] Optionally, generating a radiation therapy dose distribution according to the delineation result includes:

[0024] The radiation therapy dose distribution is generated according to the delineation result using a dose calculation algorithm.

[0025] Optionally, the dose calculation algorithm includes at least a Monte Carlo algorithm and a pencil beam algorithm.

[0026] Optionally, the optimization evaluation method includes at least a reinforcement learning algorithm and a genetic algorithm.

[0027] In a second aspect, the present application provides an adaptive decision support system for patient radiotherapy, comprising:

[0028] A data acquisition and preprocessing module, used to obtain raw data for radiotherapy of patients, and perform standardization and preprocessing on the raw data to obtain processed data; the raw data includes multimodal imaging data, treatment history data, biomarker data and individual patient information;

[0029] An image analysis and segmentation module, connected to the data acquisition and preprocessing module, is used to automatically segment tumors and normal tissues and detect changes based on the processed data to obtain detection results;

[0030] A tumor target area delineation module, connected to the image analysis and segmentation module, is used to delineate the tumor target area and normal tissue in radiotherapy in the detection results to obtain a delineation result;

[0031] a dose optimization module, connected to the tumor target area delineation module, and used to generate a radiation therapy dose distribution according to the delineation result;

[0032] a treatment plan evaluation and optimization module, connected to the dose optimization module, for evaluating the radiation therapy dose distribution using an optimization evaluation method to obtain an evaluation result, generating a plurality of alternative treatment plans based on the evaluation result, and determining an optimal treatment plan based on the plurality of alternative treatment plans;

[0033] The real-time image guidance module is connected to the treatment plan evaluation and optimization module, and is used to implement radiotherapy based on the optimal treatment plan, and to monitor the patient's anatomical changes in real time during the radiotherapy process, obtain the change results, and dynamically adjust the optimal treatment plan based on the change results.

[0034] Optionally, the treatment plan evaluation and optimization module includes:

[0035] A graphical interface for comparing the expected therapeutic effects of multiple alternative treatment options and displaying the comparison results in a graphical interface; the expected therapeutic effects include tumor control rate, normal tissue dose, and potential side effects;

[0036] An interactive analysis tool for determining an optimal treatment regimen based on a plurality of said alternative treatment regimens.

[0037] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-described adaptive decision support methods for patient radiotherapy.

[0038] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned adaptive auxiliary decision-making methods for patient radiotherapy.

[0039] According to the specific embodiments provided in this application, this application has the following technical effects:

[0040] The present application provides an adaptive decision-making assistance method, system, device and medium for radiotherapy of patients. By dynamically adjusting the optimal treatment plan based on the change results, the patient's radiotherapy can be precise and personalized, so as to better cope with the complex changes in the treatment process and improve the overall effect of the treatment. In addition, the system provided by the present application can effectively integrate operations such as data acquisition, image analysis, dose optimization, real-time adjustment and intelligent decision support by using a modular setting. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 A flowchart of an adaptive auxiliary decision-making method for radiotherapy of a patient provided in one embodiment of the present application;

[0043] Figure 2 A schematic diagram of functional modules of an adaptive auxiliary decision system for radiotherapy of patients provided in another embodiment of the present application.

[0044] Figure 3 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0046] With the advancement of technology, especially the development of artificial intelligence technology, there is an urgent need for a more intelligent and automated adaptive radiotherapy decision-making support mechanism in clinical practice. This mechanism should be able to provide real-time analysis and optimization suggestions in all aspects of radiotherapy to adapt to the dynamic changes of patients, improve the accuracy and personalization of patient treatment, and reduce the workload of doctors to ensure the safety and effectiveness of treatment. By introducing deep learning and machine learning technologies into radiotherapy, the new mechanism can play an important role in image analysis, tumor target delineation, dose optimization, real-time image guidance and decision support, and achieve comprehensive treatment adjustments.

[0047] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0048] The adaptive auxiliary decision-making method for patient radiotherapy provided in the embodiment of the present application can be implemented using an independent server or a server cluster composed of multiple servers, or a cloud server.

[0049] In an exemplary embodiment, Figure 1 As shown, an adaptive auxiliary decision method for patient radiotherapy is provided, which is executed by a computer device, specifically, it can be executed by a computer device such as a terminal or a server alone, or it can be executed by a terminal and a server together. In the embodiment of the present application, the method is applied to a server as an example for explanation, including the following steps 100 to 108. Among them:

[0050] Step 100: Acquire raw data for radiotherapy of patients. The raw data includes multimodal imaging data, treatment history data, biomarker data and individual patient information. Among them, the multimodal imaging data of patients is collected by imaging equipment such as CT, MRI, PET, etc., to provide detailed anatomical information for subsequent treatment planning.

[0051] Step 101: Standardize and preprocess the original data to obtain processed data.

[0052] Step 102: Automatically segment tumor and normal tissue and detect changes based on the processed data to obtain detection results.

[0053] Step 103: Outline the tumor target area and normal tissue in radiotherapy in the detection results to obtain the outline results. The normal tissue includes normal lung tissue, major blood vessels (such as pulmonary artery, pulmonary vein, etc.), major bronchi and branches, pleura, heart tissue, esophagus, lymph nodes and nerves in the mediastinum, etc.

[0054] Step 104: Generate radiation therapy dose distribution according to the delineation result.

[0055] Step 105: Using an optimization evaluation method to evaluate the radiation therapy dose distribution to obtain an evaluation result, and generating multiple alternative treatment plans based on the evaluation result.

[0056] Step 106: Determine an optimal treatment plan based on multiple alternative treatment plans.

[0057] Step 107: Perform radiotherapy based on the optimal treatment plan, and monitor the patient's anatomical changes in real time during the radiotherapy to obtain change results.

[0058] Step 108: Dynamically adjust the optimal treatment plan based on the change results.

[0059] In another exemplary embodiment of the present application, in order to ensure data consistency, eliminate deviations caused by differences in different imaging devices or imaging parameters, and to improve image clarity and resolution, in this embodiment, the implementation process of the above step 101 may include:

[0060] (1) Standardize the original data to obtain standardized data to ensure data consistency and eliminate deviations caused by different imaging devices or imaging parameters.

[0061] (2) Use image processing algorithms (such as convolution filtering or adaptive filtering) to eliminate noise and artifacts in the standardized data and obtain intermediate data to improve image clarity and resolution.

[0062] (3) Use the registration algorithm to align the data in the intermediate data to obtain processed data to ensure the spatial consistency of the images at each time point, providing an accurate basis for subsequent analysis.

[0063] In adaptive radiotherapy, the registration algorithm can help identify changes in tumors and surrounding normal tissues by spatially aligning images at different time points or in different modalities, providing an important basis for adjusting the treatment plan. The following is a detailed implementation of a common image registration algorithm based on affine transformation and mutual information, which is suitable for the registration of multi-modal images such as CT and MRI, where:

[0064] Step 1: Data preparation. First, two sets of medical image data for registration are required: reference images and images to be registered. The reference image is the benchmark image for registration, usually the patient's initial image data (i.e. treatment history data, such as CT or MRI). The image to be registered is a real-time or staged image of the current treatment, usually obtained before or during each treatment, and may not be completely consistent with the reference image due to patient position, tumor growth, etc.

[0065] Step 2: Preprocessing. Before registration, some preprocessing operations are required on the image to improve the accuracy of registration:

[0066] 1) De-noising: Remove noise from the image through filters (such as Gaussian filtering or median filtering).

[0067] 2) Image normalization: Normalize the grayscale value of the image to make the intensity range of images of different modalities consistent and reduce the grayscale deviation caused by imaging differences.

[0068] 3) Image cropping: only retain the key areas that need to be registered to reduce the amount of calculation.

[0069] Step 3: Registration model selection. During actual patient radiotherapy, different registration models can be used to describe the geometric transformation between the reference image and the image to be registered. Commonly used registration models include rigid registration (RigidRegistration) and affine registration (Affine Registration). In rigid registration (Rigid Registration), it is assumed that the image to be registered has only been translated and rotated relative to the reference image, and scale changes or nonlinear deformations are not considered. Affine registration (Affine Registration) allows the image to be registered to be translated, rotated, scaled, and sheared relative to the reference image, and is suitable for situations where there are certain shape changes but no complex deformations are involved. In order to handle more complex anatomical changes, the affine registration model is used in this embodiment.

[0070] Step 4: Similarity metric selection. The key to registration is to define a similarity metric to measure the degree of alignment of two images. For multimodal images (such as CT and MRI), the commonly used similarity metric is mutual information (MI). Mutual information is a metric based on information entropy that can measure the statistical dependence of two images in grayscale values. The better the two images are aligned, the greater their mutual information.

[0071] Step 5: Optimization algorithm selection. In order to find the optimal affine transformation parameters that maximize the mutual information, it is necessary to select an optimization algorithm. Commonly used optimization algorithms include gradient descent and random optimization algorithms (such as Powell's method). Among them, the gradient descent method calculates the gradient of the similarity measure and gradually adjusts the parameters in the affine transformation matrix to find the optimal solution. The random optimization algorithm (such as the Powell method) searches for different transformation parameters step by step, does not rely on gradient information, and is suitable for processing complex multimodal image registration problems. This embodiment adopts the Powell optimization method to effectively avoid local optimal problems and is suitable for multimodal image registration.

[0072] Step 6: Implement registration based on the selected optimization algorithm:

[0073] Step 1: Initialize the affine transformation matrix.

[0074] The initial affine transformation matrix is ​​set to the unit matrix, indicating that the reference image and the image to be registered completely coincide with each other in the initial state.

[0075] Step 2: Calculate the initial mutual information.

[0076] Use the current affine transformation matrix to project the image to be registered into the reference image space and calculate the initial mutual information value.

[0077] Step 3: Iterative optimization.

[0078] The parameters in the affine transformation matrix are gradually adjusted through the Powell optimization algorithm. After each adjustment, the mutual information value between the image to be registered and the reference image is recalculated, and the changes in mutual information before and after the adjustment are compared. If the new mutual information value is greater than the mutual information value of the previous round, the new affine transformation matrix is ​​accepted. Otherwise, the parameters are adjusted until the mutual information value converges or the maximum number of iterations is reached.

[0079] Step 4: When the change in mutual information is lower than the set threshold (for example, less than 10 -4 ), or when the optimization process reaches the maximum number of iterations, the optimization is stopped and the final affine transformation matrix is ​​output.

[0080] Step 5: Image resampling.

[0081] According to the final affine transformation matrix, the image to be registered is resampled so that it is spatially aligned with the reference image.

[0082] Step 7: Post-processing and verification.

[0083] Step 1: Image overlay and verification.

[0084] The registered image to be registered is superimposed with the reference image and the registration effect is verified by visualization. If the two images are well aligned on key structures (such as tumor margins, organ contours, etc.), the registration is successful.

[0085] Step 2: Evaluation of registration accuracy.

[0086] Quantitative indicators such as target coincidence (Dice coefficient) and mean surface distance (MSD) are used to evaluate the registration accuracy. The Dice coefficient is used to measure the similarity between two images. The closer the value is to 1, the higher the coincidence. MSD calculates the average distance between the boundaries of the key structures of the two images after registration. The smaller the distance, the better the registration effect.

[0087] Step 8: Feedback and use of results.

[0088] After registration is completed, the verified registered images will be transmitted to downstream modules for subsequent tumor target delineation, dose planning, and real-time treatment adjustment.

[0089] In summary, the registration algorithm can accurately align images of different modalities or at different time points, providing high-precision anatomical references for subsequent radiotherapy. It is particularly suitable for the registration of multimodal images (such as CT and MRI), which helps to track the dynamic changes of tumors and normal tissues in real time, thereby optimizing treatment plans and reducing the dose to normal tissues during treatment.

[0090] In another exemplary embodiment of the present application, in order to automatically identify tumor areas, key normal organs and other important anatomical structures, in this embodiment, deep learning algorithms (such as convolutional neural networks CNN, U-Net) can be used to automatically segment and detect changes in tumors and normal tissues based on processed data to obtain detection results.

[0091] In this embodiment, by integrating multimodal imaging information such as CT, MRI, and PET, a deep learning model can be used to achieve comprehensive analysis of information under different modalities and enhance the recognition accuracy of tumors and normal tissues.

[0092] In addition, based on the patient's continuous imaging data, it can automatically detect changes in the shape, volume, and position of the tumor and surrounding normal tissue, generate a change report, and provide a basis for real-time adjustment.

[0093] In another exemplary embodiment of the present application, in the above step 103, the image analysis and automatic segmentation results can be used to automatically outline the tumor target area (GTV, CTV, PTV) in radiotherapy and mark important normal tissues.

[0094] In another exemplary embodiment of the present application, clinicians are allowed to automatically modify the delineated target area according to the specific conditions of the patient to ensure the clinical rationality of the delineation results. Based on this, in the above step 104, the dose distribution of radiotherapy can be calculated based on the delineation results of the tumor target area and normal tissue using advanced dose calculation algorithms (including but not limited to Monte Carlo algorithms and pencil beam algorithms) to ensure that the tumor receives sufficient treatment dose while minimizing the dose to normal tissue.

[0095] Taking the Monte Carlo algorithm as an example, the steps are as follows:

[0096] Step 1: Model initialization: When applying the Monte Carlo algorithm for dose calculation, it is first necessary to construct a three-dimensional model of the patient's anatomical structure and a physical model of the radiation.

[0097] 1. Patient anatomical model establishment. Use medical imaging data such as CT, MRI or PET to reconstruct the patient's anatomical structure in three dimensions. After registration and segmentation, the data is divided into different types of tissue regions (such as tumors, bones, soft tissues, air, etc.). These tissue types have different physical properties, such as density and composition materials, which will affect the propagation and absorption of radiation. The patient's anatomical structure is divided into voxels. The size of each voxel is determined by the image resolution, and the common size is between 1 and 3 mm.

[0098] 2. Physical modeling of the radiation source. Construct a physical model of the radiation source in the radiotherapy equipment, such as the photon or electron beam model of the linear accelerator (LINAC). The energy distribution, beam shape, intensity, and initial direction of the beam need to be modeled in detail. Initialize the beam emission parameters, including the energy, angle, beam shape, and beam intensity distribution of the beam.

[0099] Step 2: Physical interaction modeling. The Monte Carlo algorithm simulates the propagation of radiation in tissues particle by particle and accurately calculates the interaction between radiation and tissues. Therefore, it is necessary to model the main physical interaction processes between radiation and matter, including the interaction between photons and matter and the interaction between electrons and matter.

[0100] Step 3: Particle trajectory simulation. The core of the Monte Carlo algorithm is to simulate the propagation of rays (photons or electrons) in human tissues through random sampling and particle tracking. This process can be divided into the following sub-steps:

[0101] 1. Particle emission: A particle is randomly emitted from a ray source. The initial position, energy, direction and type (photon or electron) of the particle are determined according to the physical model of the ray source.

[0102] 2. Free path calculation. Use random numbers to generate the free path of particles in the tissue, that is, the distance that particles can travel before physical interaction occurs. The free path l is usually calculated by the formula Where R is a random number uniformly distributed between 0 and 1, and μ is the linear attenuation coefficient of the tissue.

[0103] 3. Interaction type selection: When the particle reaches the end of the free path, a random number is used to determine the type of interaction between the ray and the substance. The probability of different interaction types varies according to the energy and tissue material.

[0104] 4. Energy deposition and particle path update. Calculate the energy lost by the particle in the interaction and deposit it to the corresponding voxel. For incomplete energy transfer processes such as Compton scattering, the particle will continue to propagate with reduced energy and change direction. If the particle energy drops to a certain threshold, it is considered that the particle stops propagating. Otherwise, continue to repeat the free path calculation and interaction process until the particle leaves the tissue or the energy is exhausted.

[0105] Step 4: Dose calculation: By simulating the trajectories of a large number of particles, the energy deposition in each voxel can be obtained, and then the dose distribution in radiotherapy can be calculated.

[0106] 1. Energy deposition statistics.

[0107] During the simulation, each time a particle interacts with tissue, energy is deposited into a specific voxel, and the system records the sum of the energy deposition of all particles in each voxel.

[0108] 2. Dose calculation formula.

[0109] Dose D is the energy absorbed by unit mass of tissue, measured in Gray (Gy), 1Gy = 1J / kg, and the dose calculation formula is: Among them, E total is the total energy deposited by all particles in the voxel, and m is the mass of the voxel.

[0110] 3. Dose distribution map generation.

[0111] The dose values ​​in all voxels are combined into a 3D dose distribution map, which clearly shows the dose distribution of radiation in tumors and normal tissues.

[0112] Step 5: Precision control and convergence detection. The accuracy of the Monte Carlo algorithm depends on the number of simulated particles. In order to ensure the accuracy of dose calculation, it is usually necessary to simulate enough particles until the results converge.

[0113] Step 6: Post-processing and dose optimization. The generated three-dimensional dose distribution map usually needs further processing to remove abnormal data points, smooth transition areas, etc. This ensures the availability and interpretability of the dose distribution. In adaptive radiotherapy, the angle, intensity, and shape of the radiation beam can also be optimized based on the difference between the dose distribution calculated by the Monte Carlo algorithm and the planned dose to ensure that the tumor receives sufficient dose and normal tissue is protected.

[0114] Step 7: Output and verification of results. To verify the dose distribution, the calculated dose distribution is compared with the actual treatment plan to ensure that the simulation results are consistent with the clinical goals. If there is a significant deviation, the beam parameters will be adjusted and the dose calculation will be re-performed until the expected goals are met. The final dose distribution is presented in a visual form for the visualization of the results. The doctor can intuitively view the dose situation in the tumor area and normal tissue to ensure the safety and effectiveness of the treatment.

[0115] In another exemplary embodiment of the present application, the optimization evaluation method includes at least a reinforcement learning algorithm and a genetic algorithm.

[0116] In another exemplary embodiment of the present application, in step 105, multiple alternative treatment plans (i.e., radiotherapy dose planning plans) can be generated by using reinforcement learning, genetic algorithms, or other optimization algorithms. By simulating and analyzing the effects of each alternative treatment plan, the optimal treatment plan is selected to ensure the best treatment effect and the least side effects. For example:

[0117] (1) During each radiotherapy session, the patient's current anatomical status is monitored through real-time imaging, which is accurately aligned with the images in the original treatment plan to detect the displacement and changes of tumors and normal tissues.

[0118] (2) Based on the results of real-time image analysis, the direction, shape, and dose distribution of the radiation beam can be dynamically adjusted to adapt to the immediate changes in the patient's anatomical structure.

[0119] (3) Calculate and evaluate the dose distribution in the current treatment plan in real time, pay special attention to the dose received by the tumor area and normal tissue, identify possible areas of excessive or insufficient dose, and obtain dose assessment results.

[0120] Among them, the computer can identify possible areas of over- or under-dosing, specifically:

[0121] 1. Use image-guided technology (such as CT, MRI, PET, etc.) to obtain the patient's anatomical structure and tumor location in real time to ensure the timeliness and accuracy of the data.

[0122] 2. By applying accurate dose calculation algorithms (such as Monte Carlo algorithm, radiation transfer model, etc.), the actual dose distribution of the tumor area and normal tissue is calculated based on the beam parameters (such as angle, intensity, etc.) in the treatment plan.

[0123] 3. Determine the maximum permissible dose and minimum effective dose by setting dose thresholds for tumors and normal tissues. For example, setting the tumor target dose lower than 95% of the prescribed dose indicates insufficient dose. Setting the tumor target dose higher than 110% of the prescribed dose indicates excessive dose. Such thresholds can be set based on clinical standards and individual patient conditions.

[0124] 4. By comparing the real-time dose distribution with the preset dose threshold, the system automatically identifies areas with excessive or insufficient doses. The system then provides feedback to the doctor, who then analyzes and recommends possible treatment adjustments to ensure that the tumor target area receives sufficient dose while avoiding excessive radiation to normal tissues.

[0125] (4) Generate multiple alternative treatment plans based on real-time dose assessment results. Through optimization models such as reinforcement learning or machine learning, the potential effects of different alternative treatment plans can be predicted to help select the best treatment path.

[0126] By analyzing and referencing historical treatment data and outcomes of similar cases, data-driven recommendations are provided for current patient treatment plans, further enhancing the precision of personalized treatment.

[0127] Comparing the expected treatment effects of multiple alternative treatment options, including tumor control rate, normal tissue dose, and potential side effects, allows clinicians to simulate the potential treatment outcomes of different decision-making pathways to explore the possible impact of different treatment decisions and thus select the optimal treatment option.

[0128] Furthermore, during the treatment implementation, the treatment process is monitored through real-time images and dose data to ensure that the treatment execution is consistent with the plan. If any deviation is found, the system will immediately remind and provide adjustment suggestions. The deviation between the actual treatment and the plan is automatically detected. When the deviation exceeds the predetermined range, an alarm is generated and the doctor is advised to adjust the treatment plan. Based on the data accumulated during the treatment process, the model and algorithm are continuously optimized to improve the accuracy and effect of future treatments. At the same time, these feedback data can also be used to generate personalized treatment plans for subsequent patients.

[0129] Furthermore, taking the reinforcement learning algorithm as an example, the specific process of generating multiple dose plans through reinforcement learning is as follows:

[0130] Step 1: Model initialization.

[0131] 1. Environmental modeling: Use the patient's CT, MRI or PET imaging data to build a three-dimensional anatomical model, including detailed information about the tumor and surrounding normal tissue. Pre-calculate the dose distribution under different radiation beam parameters through methods such as the Monte Carlo algorithm, which is used as environmental feedback for the reinforcement learning model.

[0132] 2. Agent initialization: Initialize a model based on deep reinforcement learning (DRL). Typical model structures are deep Q-network (DQN) or deep policy gradient method (Policy Gradient), which can handle continuous state and action space.

[0133] 3. Action space definition: The possible treatment parameters are used as part of the action space, for example: Beam angle range: 0 to 360 degrees. Beam energy range: 5 to 15 MV. Beam shape: The beam shape is controlled by a dynamic multi-leaf grating (MLC).

[0134] Step 2: Reinforcement learning training process.

[0135] 1. Initial state acquisition: Starting from the initial state of the patient (i.e., image data and anatomical model), the 3D position, size, density and other information of the tumor have been determined by the image analysis module, and the initial dose distribution map can be used as a starting point.

[0136] 2. Action selection: At each step, the agent selects an action based on the current state (e.g., changing the beam angle or adjusting the dose intensity). In the early stages, the agent can balance exploration (i.e., trying different actions) and exploitation (i.e., choosing the best action currently known) through an ε-greedy policy. For example: select a random action with probability ∈. Select the best action in the current state with probability 1-∈.

[0137] 3. State transfer: After the agent performs an action, the environment changes. For example, after the beam angle changes, the dose distribution of the tumor and normal tissue will be adjusted. At this time, the environment will update the state (i.e., the new image data and dose distribution map).

[0138] 4. Reward calculation: Calculate the reward for the current action. The reward function can be set according to the following formula: R = w1 × tumor coverage score - w2 × normal tissue damage score. Among them, w1 and w2 are weight coefficients used to balance the treatment effect and side effects.

[0139] 5. Q-value or policy update: Based on the current reward, the reinforcement learning algorithm updates the Q-value (or policy function) so that the agent tends to choose actions that bring higher cumulative rewards in future decisions.

[0140] Use the Bellman equation to update the Q value. The Bellman equation is expressed as:

[0141]

[0142] Among them, Q'(S t ,A t ) is the updated Q value, α is the learning rate, γ is the discount factor, R t is the current instant reward, is the optimal Q value in the next state, Q(S t ,A t ) is the Q value before the update, A' is the action in the next state, A t It is the action at the present moment.

[0143] 6. Repeated training: Repeat the above process, and the agent will continuously adjust its strategy during the training process. Through multiple interactions with the environment, the agent can gradually learn to choose the best action in different states to maximize the cumulative reward.

[0144] Step 3: Generate multiple radiation therapy dose plans. After the reinforcement learning model is trained to a certain level, the agent can generate different alternative treatment plans. Each alternative treatment plan represents a different combination of treatment parameters, specifically different beam angles, energies, and dose distributions.

[0145] 1. Standard plan: The agent generates a standard plan based on the optimized strategy. This plan may have good tumor coverage and low normal tissue damage, and is the most recommended plan.

[0146] 2. Tumor enhancement plan: Generate a plan to enhance the dose in the tumor area, increase the dose in the tumor area, but may slightly increase the dose to normal tissue. This type of plan is suitable for situations where a stronger tumor killing effect is required.

[0147] 3. Normal tissue protection plan: The reinforcement learning model generates a plan that focuses more on protecting normal tissue. This plan minimizes the radiation dose to critical organs, but the dose to the tumor area is slightly reduced, which is suitable for patients who are particularly sensitive to side effects.

[0148] 4. Time optimization plan: Under the premise of ensuring the treatment effect, the intelligent agent can generate a plan with the best treatment time, reduce the complexity and duration of treatment, and is suitable for scenarios of efficient and fast treatment.

[0149] Step 4: Compare and select multiple treatment options. Present the generated multiple alternative treatment options to the doctor and provide a detailed comparative analysis, including:

[0150] 1. Dose distribution diagram: shows the dose distribution of tumor and normal tissue in each scheme.

[0151] 2. Key indicators: such as tumor coverage rate, normal tissue protection rate, treatment duration, etc.

[0152] 3. Risk assessment: potential side effects or risks of each option.

[0153] Through the above process, the doctor can finally choose the best plan to implement based on the patient's individual needs and treatment goals.

[0154] In one embodiment provided by the present application, the process of generating multiple alternative treatment plans based on the real-time dose assessment results can be completed by the radiotherapy plan library Plan Library. The concept of the plan library is: a database containing "historical, current patient" radiotherapy plans based on different tumor types, locations and patient characteristics, which records information such as treatment plans, dose distributions, effects and side effects of various types of patients. The patient treatment plans are standardized and classified, including commonly used dose distributions, beam parameters, and treatment techniques (such as IMRT, VMAT, etc.) for quick retrieval. Based on this, the process of generating alternative treatment plans using the plan library is roughly as follows:

[0155] 1. Establish a corresponding plan library for the current patient in advance based on tumor type, movement status, and patient characteristics.

[0156] 2. By analyzing the dose distribution acquired in real time, focus on the dose received by the tumor area and normal tissue, and identify possible areas with excessive or insufficient doses.

[0157] 3. Based on the real-time dose assessment results, combined with the similarity between the current dose distribution and the historical plan, patient characteristics (such as tumor location, type, volume, motion status, etc.), and risk assessment of side effects, select alternative treatment plans suitable for the current patient condition from the plan library. Based on the screening process and results, multiple alternative treatment plans can be generated, each of which is optimized for the current dose assessment. Among them, alternative treatment plans include different beam angles, dose intensities, treatment techniques, etc.

[0158] 4. Perform dose evaluation on the generated alternative treatment options to ensure that each option meets the standards for tumor control and normal tissue protection, and use a real-time feedback mechanism to further optimize the alternative treatment options to ensure that the optimal treatment option can quickly respond to changes in the patient during treatment.

[0159] 5. Promptly feedback the changes in step 4 to the doctor and provide the doctor with a detailed analysis of alternative options, including the advantages and disadvantages, potential side effects and therapeutic effects of each treatment option, to assist the doctor in making further treatment decisions.

[0160] Based on the same inventive concept, the embodiment of the present application also provides an adaptive auxiliary decision system for patient radiotherapy (referred to as the system) for implementing the adaptive auxiliary decision method for patient radiotherapy involved in the above-mentioned. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more embodiments of the adaptive auxiliary decision system for patient radiotherapy provided below can refer to the limitations of the adaptive auxiliary decision method for patient radiotherapy above, and will not be repeated here.

[0161] In an exemplary embodiment, Figure 2 As shown, an adaptive decision support system for radiotherapy of a patient is provided, comprising:

[0162] The data acquisition and preprocessing module is used to obtain the original data used for radiotherapy of patients, and perform standardization and preprocessing on the original data to obtain processed data. The original data includes multimodal image data, treatment history data, biomarker data and individual patient information. Among them, the data acquisition and preprocessing module can include image denoising, standardization, biomarker data processing and multimodal image registration and other functions.

[0163] The image analysis and segmentation module is connected to the data acquisition and preprocessing module to automatically segment and detect changes in tumors and normal tissues based on the processed data to obtain detection results. The image analysis and segmentation module includes multimodal fusion processing, automatic segmentation of tumors and normal tissues using machine learning models or other segmentation models, and other functions.

[0164] The tumor target delineation module is connected to the image analysis and segmentation module to delineate the tumor target and normal tissue in radiotherapy in the detection results to obtain the delineation results. Among them, the tumor target delineation module continuously optimizes the automatic delineation model by learning a large amount of labeled data, thereby improving the accuracy and adaptability of target area recognition.

[0165] The dose optimization module is connected with the tumor target area delineation module and is used to generate the radiation therapy dose distribution according to the delineation results.

[0166] The treatment plan evaluation and optimization module is connected to the dose optimization module to evaluate the radiation therapy dose distribution using an optimization evaluation method to obtain evaluation results, generate multiple alternative treatment plans based on the evaluation results, and determine the optimal treatment plan based on multiple alternative treatment plans. The treatment plan evaluation and optimization module can assist in generating optimization plans and provide data-driven personalized treatment recommendations by utilizing past treatment data and results of similar patients.

[0167] The real-time image guidance module is connected to the treatment plan evaluation and optimization module to implement radiotherapy based on the optimal treatment plan, and to monitor the patient's anatomical changes in real time during radiotherapy, obtain the change results, and dynamically adjust the optimal treatment plan based on the change results.

[0168] As an optional implementation, the treatment plan evaluation and optimization module includes:

[0169] The graphical interface is used to compare the expected treatment effects of multiple alternative treatment options and display the comparison results in a graphical interface. The expected treatment effects include tumor control rate, normal tissue dose, and potential side effects.

[0170] Interactive analysis tool for determining the optimal treatment plan based on multiple treatment alternatives.

[0171] Among them, by implementing this embodiment, the system allows clinicians to simulate the potential treatment outcomes of different decision paths. Through interactive analysis tools, clinicians can explore the possible impacts of different alternative treatment options and thus select the optimal treatment option.

[0172] As an optional implementation, the system provided by the present application further includes:

[0173] The decision support module is used to provide detailed analysis and suggestions for optimization plans. The decision support module combines the optimization plan with the current treatment progress to provide targeted clinical pathway knowledge to help doctors make the best decision in complex situations.

[0174] The result verification and feedback module is used to monitor the treatment process in real time and make subsequent adjustments based on the feedback data. The result verification and feedback module verifies the execution of the treatment plan through real-time data collection and analysis, and issues an alarm or provides adjustment suggestions when the deviation exceeds the predetermined range.

[0175] Based on the above description, the obtained system can be connected to the hospital information system and radiotherapy equipment through the network to achieve real-time data interaction and integration of automated workflow.

[0176] Furthermore, during the treatment process, the treatment process is monitored through real-time images and dose data to ensure that the treatment is carried out in accordance with the plan. If any deviation is found, the system will immediately remind and provide adjustment suggestions.

[0177] The system automatically detects the deviation between the actual treatment and the plan. When the deviation exceeds the predetermined range, an alarm is generated and the doctor is advised to adjust the treatment plan.

[0178] The system continuously optimizes its own models and algorithms based on the data accumulated during the treatment process to improve the accuracy and effectiveness of future treatments. At the same time, the system uses this feedback data to generate personalized treatment plans for subsequent patients.

[0179] In an exemplary embodiment, the process of implementing radiotherapy for a patient is described by taking the above-mentioned adaptive decision support system for radiotherapy for a patient as an example. Specifically:

[0180] Before the initial treatment, patients undergo imaging examinations such as CT, MRI, and PET, and relevant equipment is used to collect multimodal imaging data of patients to obtain detailed anatomical structures and tumor locations. These imaging data are transmitted to the imaging data acquisition unit of ART-ADMS.

[0181] The system standardizes the received CT, MRI, PET and other data, including standardization of resolution and grayscale values, to ensure data consistency across different devices and imaging modes.

[0182] The system denoises the image by applying a convolution filter to eliminate noise in the image, improve image clarity, and provide higher quality data for subsequent automatic segmentation and analysis.

[0183] The system uses a multimodal registration algorithm to accurately align CT and MRI images in space. The registered image data is integrated into a unified image data set ready for subsequent analysis.

[0184] After completing the preprocessing, the system starts the image analysis and segmentation module. Based on the trained deep learning model (including but not limited to U-Net), the system automatically identifies and segments the patient's lung tumor area (GTV) and key normal tissue.

[0185] After segmentation is completed, the system automatically generates the tumor target volume (GTV), clinical target volume (CTV) and planning target volume (PTV). Clinicians can review these automatically generated target volumes through the interactive interface and make manual adjustments when necessary to ensure the accuracy and clinical rationality of the target volumes.

[0186] The system uses a dose-related algorithm to calculate the dose. Based on the results of the target area delineation, the system calculates the optimal radiation dose distribution to ensure that the tumor area (PTV) receives sufficient treatment dose while keeping the dose to normal tissue within a safe range.

[0187] The system generates multiple alternative treatment plans through reinforcement learning algorithms or other optimization algorithms. These alternative treatment plans further optimize the dose plan by changing the angle, energy and dose distribution of the radiation beam to improve the treatment effect and reduce the risk of damage to normal tissue. The system simulates the effects of different alternative treatment plans on tumor control rate and normal tissue damage, and generates a comparative report for clinicians to choose.

[0188] The system will present the comparison results of multiple alternative treatment plans in a graphical interface, including the dose distribution diagram of each plan, tumor control rate prediction, and normal tissue dose analysis, etc. Clinicians can intuitively compare the advantages and disadvantages of different alternative treatment plans.

[0189] The system assesses the potential risks of each alternative treatment option, including the risk of side effects from excessive doses, particularly radiation exposure to the heart and healthy lung tissue, and recommends strategies to reduce the risks.

[0190] Based on the plan comparison and risk assessment report provided by the system and the patient's specific situation, the clinician ultimately determines the best treatment plan (and optimal treatment plan). The system locks this plan into a treatment plan and generates relevant treatment execution instructions.

[0191] In another exemplary embodiment, the process of implementing the patient's radiotherapy is described by taking the above-mentioned adaptive decision support system for patient radiotherapy as an example. Specifically:

[0192] During a course of radiation therapy, the patient received real-time image-guided therapy (IGRT). Using cone beam CT, the system acquired the current anatomical image data of the patient on the treatment bed.

[0193] The system accurately registers the real-time images with the reference images in the initial treatment plan. The purpose of registration is to detect whether the current tumor position and shape are consistent with the initial plan, especially for the displacement of tumors and normal tissues caused by breathing, body position changes or other physiological factors.

[0194] The registration results show that the tumor position has shifted slightly and the shape of normal tissue has also changed slightly. By analyzing these changes, the system automatically adjusts the direction, shape and dose distribution of the radiation beam to adapt to the current state of the tumor and tissue. The adjusted treatment plan is implemented immediately to ensure that the tumor still receives sufficient dose while protecting normal tissue.

[0195] The purpose of analyzing these changes is to ensure that radiotherapy can adapt to the current state of tumors and normal tissues. The specific process can be roughly divided into the following steps:

[0196] 1. Analysis of changes in tumor position. This includes calculating the displacement (translation, rotation, etc.) of the tumor area between the real-time image and the reference image through the registration results. Shape change assessment uses shape descriptors (such as contours, volumes, etc.) to compare tumor shape changes and identify tumor expansion and contraction.

[0197] 2. Analysis of normal tissue changes. This includes boundary comparison, comparing the normal tissue boundaries in real-time images with those in reference images, and identifying changes in their shape and position. Structural change assessment, identifying changes in the morphology of key normal tissues (such as lungs, blood vessels, etc.), and ensuring that these changes do not cause excessive radiation damage.

[0198] 3. Quantitative analysis. Quantify the detected displacement and shape changes, calculate the specific values ​​of the changes, such as the movement distance of the tumor center, the percentage of volume change, etc., and evaluate the potential impact on radiotherapy based on the degree of change, including tumor coverage and the degree of protection of normal tissues.

[0199] 4. Adjustment of radiotherapy plan. 1) Adjustment in direction: According to the new position of the tumor, adjust the direction of the radiation beam to ensure that the tumor is within the irradiation range. 2) Adjustment in shape: Use the new shape data of the tumor to adjust the shape of the radiation beam to better fit the contour of the tumor. 3) Recalculation of dose distribution: Use relevant dose calculation algorithms (such as Monte Carlo algorithm) to recalculate the new dose distribution to ensure that the tumor receives sufficient treatment dose while reducing radiation to normal tissues.

[0200] 5. Real-time feedback detection of analysis. During the patient's subsequent treatment, the changes in tumors and normal tissues are continuously monitored so that further adjustments can be made when necessary. Based on the above analysis results, the system makes real-time decisions and automatically adjusts the treatment plan to optimize the treatment effect and reduce the risk of side effects.

[0201] During the treatment, the system continuously monitors the deviation between the imaging data and the planned data. When it detects that the tumor position deviation exceeds the predetermined threshold, the system issues an alarm and automatically pauses the treatment, waiting for further instructions from the clinician.

[0202] Clinicians view the cause of the deviation through the system interface and make temporary adjustments to the treatment plan based on real-time imaging data. The system updates the treatment plan based on the doctor's adjustments and continues to implement it. The treatment data after each adjustment will be recorded and fed back to the system for further optimization of the model and improvement of the accuracy of future treatments.

[0203] All images, dose distribution and adjustment data during the treatment process are recorded by the system. The system uses this data to continuously update and train its built-in optimization learning model to improve future treatment recommendation and decision support capabilities. The system also predicts the patient's long-term efficacy based on this data and recommends adjustments to subsequent treatment strategies when necessary.

[0204] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store video tag processing data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an adaptive auxiliary decision method for patient radiotherapy is implemented.

[0205] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0206] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0207] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0208] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0209] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0210] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0211] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0212] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. An adaptive decision-making aid method for radiotherapy of a patient, characterized in that: The adaptive auxiliary decision-making method for patient radiotherapy comprises: Acquiring raw data for radiotherapy of patients; the raw data includes multimodal imaging data, treatment history data, biomarker data and individual patient information; Performing standardization and preprocessing on the raw data to obtain processed data; Performing automatic segmentation and change detection of tumor and normal tissue based on the processed data to obtain detection results; outlining the tumor target area and normal tissue in radiotherapy in the detection results to obtain an outlining result; generating a radiation therapy dose distribution according to the delineation result; Using an optimization evaluation method to evaluate the radiation therapy dose distribution to obtain an evaluation result, and generating a plurality of alternative treatment plans based on the evaluation result; determining an optimal treatment regimen based on a plurality of said alternative treatment regimens; Implement radiotherapy based on the optimal treatment plan, and monitor the patient's anatomical changes in real time during radiotherapy to obtain the change results; The optimal treatment plan is dynamically adjusted based on the changing results.

2. The adaptive decision-making aid method for patient radiotherapy according to claim 1, characterized in that: The raw data is subjected to standardization and preprocessing to obtain processed data, including: Performing standardization processing on the original data to obtain standardized data; Using an image processing algorithm to eliminate noise and artifacts in the standardized data to obtain intermediate data; The data in the intermediate data is aligned using a registration algorithm to obtain the processed data.

3. The adaptive auxiliary decision method for patient radiotherapy according to claim 1, characterized in that: Automatic segmentation and change detection of tumor and normal tissue are performed based on the processed data to obtain detection results, including: A deep learning algorithm is used to automatically segment tumors and normal tissues and detect changes based on the processed data to obtain the detection results.

4. The adaptive decision-making aid method for radiotherapy of a patient according to claim 1, characterized in that: Generating a radiation therapy dose distribution according to the delineation result includes: The radiation therapy dose distribution is generated according to the delineation result using a dose calculation algorithm.

5. The adaptive auxiliary decision method for patient radiotherapy according to claim 4, characterized in that: The dose calculation algorithm includes at least a Monte Carlo algorithm and a pencil beam algorithm.

6. The adaptive auxiliary decision method for patient radiotherapy according to claim 1, characterized in that: The optimization evaluation method at least includes a reinforcement learning algorithm and a genetic algorithm.

7. An adaptive decision support system for patient radiotherapy, characterized in that: The adaptive auxiliary decision system for patient radiotherapy comprises: A data acquisition and preprocessing module, used to obtain raw data for radiotherapy of patients, and perform standardization and preprocessing on the raw data to obtain processed data; the raw data includes multimodal imaging data, treatment history data, biomarker data and individual patient information; An image analysis and segmentation module, connected to the data acquisition and preprocessing module, is used to automatically segment tumors and normal tissues and detect changes based on the processed data to obtain detection results; A tumor target area delineation module, connected to the image analysis and segmentation module, is used to delineate the tumor target area and normal tissue in radiotherapy in the detection results to obtain a delineation result; a dose optimization module, connected to the tumor target area delineation module, and used to generate a radiation therapy dose distribution according to the delineation result; a treatment plan evaluation and optimization module, connected to the dose optimization module, for evaluating the radiation therapy dose distribution using an optimization evaluation method to obtain an evaluation result, generating a plurality of alternative treatment plans based on the evaluation result, and determining an optimal treatment plan based on the plurality of alternative treatment plans; The real-time image guidance module is connected to the treatment plan evaluation and optimization module, and is used to implement radiotherapy based on the optimal treatment plan, and to monitor the patient's anatomical changes in real time during the radiotherapy process, obtain the change results, and dynamically adjust the optimal treatment plan based on the change results.

8. The adaptive decision support system for patient radiotherapy according to claim 7, characterized in that: The treatment plan evaluation and optimization module includes: A graphical interface for comparing the expected therapeutic effects of multiple alternative treatment options and displaying the comparison results in a graphical interface; the expected therapeutic effects include tumor control rate, normal tissue dose, and potential side effects; An interactive analysis tool for determining an optimal treatment regimen based on a plurality of said alternative treatment regimens.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the adaptive auxiliary decision method for radiotherapy of a patient as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the adaptive auxiliary decision-making method for radiotherapy of a patient described in any one of claims 1 to 6 is implemented.

Images