Parameter determination method and device, storage medium and radiotherapy equipment

Through deep learning models, the flux map and machine parameters of the radiation therapy equipment are optimized, and the problem of inaccurate parameter determination in the prior art is solved, achieving higher individualized and accurate radiotherapy effects.

CN119034116APending Publication Date: 2024-11-29SHANGHAI UNITED IMAGING HEALTHCARE
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310619625.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-29
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

The existing radiation therapy equipment fails to accurately consider the specific dose requirements of the patient during operation, and is susceptible to the influence of individual factors of the patient, the environment and the differences in the understanding of the radiation therapy plan by the radiation therapist, resulting in inaccurate parameter determination.

Method used

The deep learning model is used to obtain the dose distribution based on the first flux map, calculate the dose distribution difference value, and determine the target parameters, including the flux map and the optimization of the machine parameters, through the pre-trained deep learning model, to improve the accuracy of the parameters.

Benefits of technology

Optimizing through deep learning methods for flux maps and machine parameters improves the individual accuracy of the radiotherapy plan, reduces data differences due to the individual patient situation and equipment parameters, and improves the accuracy of radiotherapy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119034116A_ABST
    Figure CN119034116A_ABST
Patent Text Reader

Abstract

The invention relates to a parameter determination method and device, a storage medium and radiotherapy equipment, and the method comprises the steps: obtaining the dose distribution corresponding to a first flux graph according to the first flux graph and a pre-trained first deep learning model, and enabling the first deep learning model to be used for obtaining the dose distribution in the flux graph; obtaining a dose distribution difference value by calculating a dose difference between target dose distribution and the dose distribution; and determining a target parameter according to the dose distribution difference value. According to the method, the first flux diagram is converted into the predicted dose, and the target parameter is reversely deduced and optimized according to the difference value between the predicted dose and the target dose, so that the accuracy of the target parameter is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of medical equipment, and in particular to a parameter determination method, device, storage medium and radiotherapy equipment. Background Art

[0002] During operation, existing radiotherapy equipment does not take into account the patient's specific dose requirements and is easily affected by individual patient factors, the environment, and differences in the radiotherapists' understanding of the radiotherapy plan. It also does not take into account changes in radiotherapy plan parameters such as flux maps and machine parameters, resulting in inaccurately determined parameters. Summary of the invention

[0003] In this embodiment, a parameter determination method, device, storage medium and radiotherapy equipment are provided.

[0004] In a first aspect, a parameter determination method is provided in this embodiment, and the parameter determination method includes:

[0005] According to a first flux map and a pre-trained first deep learning model, obtaining a dose distribution corresponding to the first flux map, wherein the first deep learning model is used to obtain the dose distribution in the flux map;

[0006] By calculating the dose difference between the target dose distribution and the dose distribution, a dose distribution difference value is obtained; and

[0007] The target parameter is determined according to the dose distribution difference.

[0008] In some embodiments, the target parameter is a second flux map, and determining the target parameter according to the dose distribution difference includes:

[0009] A second flux map is determined based on the dose distribution difference and a pre-trained second deep learning model, wherein the second deep learning model is used to predict the flux map corresponding to the dose distribution.

[0010] In some embodiments, determining a second flux map according to the dose distribution difference and a pre-trained second deep learning model includes:

[0011] Predicting a flux map change corresponding to the dose distribution difference according to the dose distribution difference and a pre-trained second deep learning model; and

[0012] The second flux map is determined according to the first flux map and the flux map change amount.

[0013] In some embodiments, the target parameter is a second machine parameter of a radiotherapy device, and determining the target parameter according to the dose distribution difference includes:

[0014] The second machine parameters are determined based on the dose distribution difference and a pre-trained third deep learning model, wherein the third deep learning model is used to convert the dose distribution into machine parameters of the radiotherapy equipment.

[0015] In some embodiments, the first deep learning model includes a conversion deep learning model and a prediction deep learning model, and acquiring a dose distribution corresponding to the first flux map according to the first flux map and the pre-trained first deep learning model includes:

[0016] Determining, according to the first flux map and the pre-trained deep learning model for conversion, a first machine parameter of a radiotherapy device corresponding to the first flux map, wherein the deep learning model for conversion is used to convert the flux map into the machine parameter of the radiotherapy device; and

[0017] Based on the first machine parameters and the pre-trained deep learning model for prediction, the dose distribution corresponding to the first machine parameters is predicted, wherein the deep learning model for prediction is used to predict the dose distribution corresponding to the machine parameters of the radiotherapy equipment.

[0018] In some embodiments, determining the second machine parameter according to the dose distribution difference and a pre-trained third deep learning model includes:

[0019] Determining, based on the dose distribution difference and a pre-trained third deep learning model, a change in a machine parameter of the radiotherapy device corresponding to the dose distribution difference; and

[0020] The second machine parameter is determined according to the first machine parameter and the machine parameter change.

[0021] In some of the embodiments, the second flux map is the sum of the first flux map and the flux map variation.

[0022] In some of the embodiments, after determining the second flux map according to the dose distribution difference and the pre-trained second deep learning model, the method further comprises:

[0023] Determining a first machine parameter of a radiotherapy device corresponding to the second flux map according to the second flux map and the pre-trained deep learning model for conversion, wherein the deep learning model for conversion is used to convert the flux map into the machine parameter of the radiotherapy device; and

[0024] Predicting a dose distribution corresponding to the first machine parameter according to the first machine parameter and the pre-trained deep learning model for prediction, wherein the deep learning model for prediction is used to predict a dose distribution corresponding to the machine parameter of the radiotherapy device;

[0025] obtaining a dose distribution difference by calculating a dose difference between a target dose distribution and a dose distribution corresponding to the first machine parameter; and

[0026] The second machine parameter is determined according to the dose distribution difference and a pre-trained third deep learning model.

[0027] In a second aspect, a parameter determination device is provided in this embodiment, and the parameter determination device includes:

[0028] A first determination module, configured to obtain a dose distribution corresponding to a first flux map according to a first flux map and a pre-trained first deep learning model, wherein the first deep learning model is used to obtain the dose distribution in the flux map;

[0029] The second determination module is used to obtain a dose distribution difference by calculating the dose difference between the target dose distribution and the dose distribution; and determine the target parameter according to the dose distribution difference.

[0030] According to a third aspect, a radiotherapy device is provided in this embodiment, wherein machine parameters of the radiotherapy device are determined according to the parameter determination method described in the first aspect.

[0031] In a fourth aspect, a computer-readable storage medium is provided in this embodiment, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the parameter determination method described in the first aspect are implemented.

[0032] Compared with the prior art, a parameter determination method, device, storage medium and radiotherapy equipment provided in this embodiment convert the first flux map into a predicted dose, and reversely optimize the target parameters according to the difference between the predicted dose and the target dose, thereby improving the accuracy of the target parameters.

[0033] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0035] Figure 1is a hardware structure block diagram of a terminal that executes a parameter determination method according to an embodiment of the present application;

[0036] Figure 2 is a flow chart of a parameter determination method according to an embodiment of the present application;

[0037] Figure 3 is a flow chart of a parameter determination method according to a specific embodiment of the present application;

[0038] Figure 4 is a flow chart of another parameter determination method according to a specific embodiment of the present application;

[0039] Figure 5 is a flow chart of another parameter determination method according to a specific embodiment of the present application;

[0040] Figure 6 is a flow chart of another parameter determination method according to a specific embodiment of the present application;

[0041] Figure 7 is a flow chart of another parameter determination method according to a specific embodiment of the present application;

[0042] Figure 8 It is a structural block diagram of a parameter determination device according to an embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0044] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the", "these" and the like in this application do not represent quantitative restrictions, and they can be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether directly or indirectly. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. Usually, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0045] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. For example, running on a terminal, Figure 1 is a hardware structure block diagram of a terminal that executes a parameter determination method according to an embodiment of the present application. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown in the figure) processor 102 and memory 104 for storing data, wherein processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. The above terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0046] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to a parameter determination method in the present embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0047] The transmission device 106 is used to receive or send data via a network. The above network includes a wireless network provided by the communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0048] In this embodiment, a parameter determination method is provided. Figure 2 FIG. 1 is a flow chart of a parameter determination method according to an exemplary embodiment of the present application. Figure 2 As shown, the process includes the following steps:

[0049] Step S210, according to the first flux map and the pre-trained first deep learning model, obtain the dose distribution corresponding to the first flux map, wherein the first deep learning model is used to obtain the dose distribution in the flux map. The first flux map can be an initial flux map, but can also be an optimized or processed flux map.

[0050] Specifically, a first flux map of the current case is obtained, and the first flux map is input into a pre-trained first deep learning model to obtain a dose distribution corresponding to the first flux map. The first deep learning model here is used to obtain the dose distribution in the flux map.

[0051] Step S220, obtaining a dose distribution difference by calculating a dose difference between the target dose distribution and the dose distribution corresponding to the first flux map; and determining a target parameter according to the dose distribution difference.

[0052] Specifically, the target dose distribution of the current case is obtained, the dose difference between the target dose distribution and the dose distribution generated in step S210 is calculated, the dose distribution difference between the two is obtained, and the target parameters corresponding to the case are determined according to the dose distribution difference. The target parameters here can be parameters of the radiotherapy plan such as flux map and machine parameters. That is, the flux map or machine parameters corresponding to the current case are optimized through the dose distribution difference.

[0053] In this embodiment, the first flux map is converted into a predicted dose, and the target parameter is reversely inferred and optimized according to the difference between the predicted dose and the target dose, thereby improving the accuracy of the target parameter.

[0054] In some of the embodiments, the target parameter in step S220 is a second flux map. Step S220 determines the target parameter based on the dose distribution difference, including: determining the second flux map based on the dose distribution difference and a pre-trained second deep learning model, wherein the second deep learning model is used to predict the flux map corresponding to the dose distribution.

[0055] Specifically, the dose distribution difference between the target dose distribution calculated in step S220 and the dose distribution corresponding to the first flux map is input into the second deep learning model, and the second flux map corresponding to the current case is determined according to the output result of the second deep learning model. The second flux map here is the optimized new flux map.

[0056] In this embodiment, the possible change value of the flux map is predicted by the difference between the dose distribution corresponding to the current flux map and the target dose distribution, which plays a role in making the predicted flux map correspond to the individual-specific optimization target. In the process of radiotherapy optimization, this application predicts the intermediate results of optimization through a deep learning method (the intermediate results of optimization here include flux map results, machine parameters, machine jump counts, etc., which can also be understood as the output of each deep learning model in the embodiment of this application), rather than directly predicting the final result or the final result of a certain optimization module (such as flux map optimization results, direct machine parameter optimization results). The purpose is to eliminate the problem of poor prediction and optimization effect caused by differences in radiotherapy data caused by factors such as individual patient conditions, physicists, and equipment parameters. Compared with existing radiotherapy optimization methods based on artificial intelligence, it can better achieve individualized and precise radiotherapy goals.

[0057] In some of the embodiments, a second flux map is determined based on the dose distribution difference and a pre-trained second deep learning model, including: predicting a flux map change corresponding to the dose distribution difference based on the dose distribution difference and the pre-trained second deep learning model; and determining the second flux map based on the first flux map and the flux map change.

[0058] Specifically, the dose distribution difference between the target dose distribution calculated in step S220 and the dose distribution corresponding to the first flux map is input into the second deep learning model to obtain the flux map change corresponding to the dose distribution difference, and the second flux map is determined according to the first flux map and the flux map change. The flux map change is the flux map change predicted according to the dose distribution difference, and the first flux map and the flux map change are summed to obtain the second flux map, that is, the first flux map and the flux map change predicted according to the dose distribution difference are summed to obtain the optimized new flux map.

[0059] In some of the embodiments, the target parameter is a second machine parameter of the radiotherapy equipment, and the target parameter is determined according to the dose distribution difference, including: determining the second machine parameter according to the dose distribution difference and a pre-trained third deep learning model, wherein the third deep learning model is used to convert the dose distribution into the machine parameter of the radiotherapy equipment.

[0060] Specifically, the dose distribution difference between the target dose distribution calculated in step S220 and the dose distribution corresponding to the first flux map is input into the pre-trained third deep learning model, and the second machine parameter is determined according to the output result of the third deep learning model. The second machine parameter here is the optimized new machine parameter.

[0061] In some embodiments, the first deep learning model includes a deep learning model for conversion and a deep learning model for prediction, and according to the first flux map and the pre-trained first deep learning model, obtaining the dose distribution corresponding to the first flux map includes: determining the first machine parameter of the radiotherapy device corresponding to the first flux map according to the first flux map and the pre-trained deep learning model for conversion, wherein the deep learning model for conversion is used to convert the flux map into the machine parameter of the radiotherapy device; and predicting the dose distribution corresponding to the first machine parameter according to the first machine parameter and the pre-trained deep learning model for prediction, wherein the deep learning model for prediction is used to predict the dose distribution corresponding to the machine parameter of the radiotherapy device. The first machine parameter may be an initial machine parameter, but may also be an optimized or processed machine parameter.

[0062] Specifically, the first flux map obtained is input into a pre-trained deep learning model for conversion to obtain a first machine parameter of the radiotherapy device corresponding to the first flux map, where the first machine parameter is the first machine parameter predicted according to the first flux map, and the deep learning model for conversion is used to convert the flux map into the machine parameter of the radiotherapy device. After obtaining the first machine parameter, the first machine parameter is input into a pre-trained deep learning model for prediction to obtain a dose distribution corresponding to the first machine parameter, where the dose distribution is also the dose distribution corresponding to the first flux map, and the deep learning model for prediction is used to predict the dose distribution corresponding to the machine parameter of the radiotherapy device.

[0063] In some of the embodiments, determining the second machine parameter based on the dose distribution difference and a pre-trained third deep learning model includes: determining the machine parameter change of the radiotherapy equipment corresponding to the dose distribution difference based on the dose distribution difference and a pre-trained third deep learning model; and determining the second machine parameter based on the first machine parameter and the machine parameter change.

[0064] Specifically, when the target parameter is the second machine parameter of the radiotherapy equipment, the dose distribution difference between the target dose distribution and the dose distribution corresponding to the first machine parameter obtained by a pre-trained prediction deep learning model is calculated and input into a pre-trained third deep learning model to obtain the machine parameter change of the radiotherapy equipment corresponding to the dose distribution difference. The machine parameter change here is the machine parameter change predicted according to the dose distribution difference. The first machine parameter and the machine parameter change are summed to obtain the second machine parameter, that is, the first machine parameter and the machine parameter change predicted according to the dose distribution difference are summed to obtain the optimized new machine parameter.

[0065] In this embodiment, the dose distribution corresponding to the current machine parameters and the difference between the target dose distribution are used to predict the change in the optimized machine parameters, so as to make the predicted optimized machine parameters correspond to the individual-specific optimization targets. In the process of radiotherapy optimization, this application predicts the intermediate results of optimization through a deep learning method (the intermediate results of optimization here include flux map results, machine parameters, machine jump counts, etc., which can also be understood as the output of each deep learning model in the embodiment of this application), rather than directly predicting the final result or the final result of a certain optimization module (such as flux map optimization results, direct machine parameter optimization results). The purpose is to eliminate the problem of poor prediction and optimization effect caused by differences in radiotherapy data caused by factors such as individual patient conditions, physicists, and equipment parameters. Compared with existing radiotherapy optimization methods based on artificial intelligence, it can better achieve individualized and precise radiotherapy goals.

[0066] In some embodiments, the first flux map can be calculated by the first machine parameter and the optimization target. The first flux map is calculated by the relationship between the machine parameter and the flux map and the preset optimization target. The flux map is optimized by the first machine parameter and the optimization target using a preset algorithm to obtain the first flux map. The preset algorithm here can be a gradient descent method, a coordinate descent method, a Newton iteration method, a least squares method, etc., which are not specifically limited here.

[0067] In some of the embodiments, the first flux map, the second flux map, and the flux map variation are two-dimensional data, and the dose distribution corresponding to the flux map and the target dose distribution are three-dimensional data, and the two-dimensional flux map and the three-dimensional dose distribution are converted through the trained deep learning model. The two-dimensional flux map is optimized to facilitate rapid parameter optimization.

[0068] It should be noted that the deep learning models involved in the embodiments of the present application, such as the first deep learning model, the second deep learning model, the third deep learning model, the deep learning model for conversion, and the deep learning model for prediction, etc., can adopt UNet (unity net) model, ResNet (Residual Network) model, DenseNet (densely connected network) model or GAN (Generative Adversarial Networks) model and other deep learning models, which are not specifically limited here; the optimization target involved in the embodiments of the present application can be the target value involved in the target table or optimization target table input by the user, or it can be the target dose distribution generated according to the target table or optimization target table input by the user, which is not specifically limited here.

[0069] In this embodiment, a radiotherapy device is provided, and the machine parameters of the radiotherapy device are determined according to the parameter determination method described in the above embodiments.

[0070] The embodiments of the present application are described and illustrated below by means of non-limiting exemplary specific embodiments.

[0071] Figure 3 is a flow chart of a parameter determination method according to a specific embodiment of the present application. Figure 3 As shown, the process includes the following steps:

[0072] Step S310, obtaining a dose distribution corresponding to the first flux map according to the first flux map and a pre-trained first deep learning model, wherein the first deep learning model is used to obtain the dose distribution in the flux map.

[0073] Specifically, the target dose distribution and the first flux map of the current case are obtained, and the first flux map is converted into its corresponding dose distribution through the first deep learning model.

[0074] Step S320, obtaining a first dose distribution difference by calculating a dose difference between the target dose distribution and the dose distribution corresponding to the first flux map.

[0075] Specifically, the first flux map is converted into its corresponding dose distribution through the first deep learning model, and the difference is calculated with the target dose distribution to obtain a dose distribution difference, i.e., a first dose distribution difference. The first dose distribution difference here is the dose distribution difference calculated according to the target dose distribution and the dose distribution corresponding to the first flux map as described in the aforementioned embodiment.

[0076] Step S330: determining a second flux map according to the first dose distribution difference and a pre-trained second deep learning model.

[0077] Specifically, the first dose distribution difference is input into the pre-trained second deep learning model, and the change value of the flux map corresponding to the first dose distribution difference is predicted. The change value of the flux map is summed with the first flux map to obtain a second flux map, and the second flux map is a new flux map as an output result. Among them, the second deep learning model is used to predict the flux map corresponding to the dose distribution. In practical applications, machine parameters can be set according to the second flux map. Through the second deep learning model, the change value of the flux map corresponding to the first dose distribution difference is predicted, and the deep learning model is used to save the rounds of flux map optimization.

[0078] In this embodiment, the possible change value of the flux map is predicted by the difference between the dose distribution corresponding to the current flux map and the target dose distribution, which plays a role in making the predicted flux map correspond to the individual-specific optimization target. In the process of radiotherapy optimization, this application predicts the intermediate results of optimization through a deep learning method (the intermediate results of optimization here include flux map results, machine parameters, machine jump counts, etc., which can also be understood as the output of each deep learning model in the embodiment of this application), rather than directly predicting the final result or the final result of a certain optimization module (such as flux map optimization results, direct machine parameter optimization results). The purpose is to eliminate the problem of poor prediction and optimization effect caused by differences in radiotherapy data caused by factors such as individual patient conditions, physicists, and equipment parameters. Compared with existing radiotherapy optimization methods based on artificial intelligence, it can better achieve individualized and precise radiotherapy goals.

[0079] Figure 4 is a flow chart of another parameter determination method according to a specific embodiment of the present application. Figure 4 As shown, the process includes the following steps:

[0080] Step S410: Determine a first machine parameter of a radiotherapy device corresponding to the first flux map according to the first flux map and a pre-trained deep learning model for conversion.

[0081] Specifically, a first flux map and a target dose distribution of the current case are obtained, and the first flux map is input into a deep learning model for conversion (also referred to as a fourth deep learning model) to obtain a first machine parameter of the radiotherapy device corresponding to the first flux map. The deep learning model for conversion is used to convert the flux map into a machine parameter of the radiotherapy device.

[0082] Step S420, predicting the dose distribution corresponding to the first machine parameter based on the first machine parameter and a pre-trained prediction deep learning model (also referred to as a fifth deep learning model).

[0083] Specifically, the first machine parameter obtained in step S410 is input into a pre-trained prediction deep learning model to obtain a predicted dose distribution corresponding to the first machine parameter. The prediction deep learning model is used to predict the dose distribution corresponding to the machine parameter of the radiotherapy device.

[0084] Step S430, obtaining a second dose distribution difference by calculating the dose difference between the dose distribution corresponding to the first machine parameter and the target dose distribution.

[0085] Specifically, the predicted dose distribution corresponding to the first machine parameter is subtracted from the target dose distribution to obtain a dose difference, i.e., a second dose distribution difference. The second dose distribution difference here is the dose distribution difference calculated according to the target dose distribution and the dose distribution corresponding to the first machine parameter as described in the above embodiment.

[0086] Step S440, determining a second machine parameter according to the second dose distribution difference and a pre-trained third deep learning model.

[0087] Specifically, the second dose distribution difference is input into the pre-trained third deep learning model, and the change amount of the machine parameter corresponding to the second dose distribution difference is predicted. The change amount of the machine parameter and the first machine parameter are summed to obtain the second machine parameter, which is the new machine parameter as an output result, or used for further machine learning for optimization. The change amount of the machine parameter corresponding to the second dose distribution difference is predicted by the third deep learning model, and the deep learning model is used to save the rounds of machine parameter optimization.

[0088] In this embodiment, the dose distribution corresponding to the current machine parameters and the difference between the target dose distribution are used to predict the change in the optimized machine parameters, so as to make the predicted optimized machine parameters correspond to the individual-specific optimization targets. In the process of radiotherapy optimization, this application predicts the intermediate results of optimization through a deep learning method (the intermediate results of optimization here include flux map results, machine parameters, machine jump counts, etc., which can also be understood as the output of each deep learning model in the embodiment of this application), rather than directly predicting the final result or the final result of a certain optimization module (such as flux map optimization results, direct machine parameter optimization results). The purpose is to eliminate the problem of poor prediction and optimization effect caused by differences in radiotherapy data caused by factors such as individual patient conditions, physicists, and equipment parameters. Compared with existing radiotherapy optimization methods based on artificial intelligence, it can better achieve individualized and precise radiotherapy goals.

[0089] Figure 5 is a flow chart of another parameter determination method according to a specific embodiment of the present application. Figure 5 As shown, the process includes the following steps:

[0090] Step S510, obtaining a dose distribution corresponding to the first flux map according to the first flux map and a pre-trained first deep learning model, wherein the first deep learning model is used to obtain the dose distribution in the flux map.

[0091] Specifically, the target dose distribution and the first flux map of the current case are obtained, and the first flux map is converted into its corresponding dose distribution through the first deep learning model.

[0092] Step S520, obtaining a first dose distribution difference by calculating a dose difference between the target dose distribution and the dose distribution corresponding to the first flux map.

[0093] Specifically, the first flux map is converted into its corresponding dose distribution through the first deep learning model, and the difference is calculated with the target dose distribution to obtain a dose distribution difference, i.e., a first dose distribution difference. The first dose distribution difference here is the dose distribution difference calculated according to the target dose distribution and the dose distribution corresponding to the first flux map as described in the aforementioned embodiment.

[0094] Step S530: determining a second flux map according to the first dose distribution difference and a pre-trained second deep learning model.

[0095] Specifically, the first dose distribution difference is input into the pre-trained second deep learning model, and the change value of the flux map corresponding to the first dose distribution difference is predicted. The change value of the flux map is summed with the first flux map to obtain a second flux map, and the second flux map is a new flux map as an output result. Among them, the second deep learning model is used to predict the flux map corresponding to the dose distribution. In practical applications, machine parameters can be set according to the second flux map.

[0096] Step S540: Determine a first machine parameter of the radiotherapy device corresponding to the second flux map according to the second flux map and a pre-trained deep learning model for conversion.

[0097] Specifically, the second flux map is input into the deep learning model for conversion to obtain the first machine parameter of the radiotherapy device corresponding to the second flux map, wherein the deep learning model for conversion is used to convert the flux map into the machine parameter of the radiotherapy device.

[0098] Step S550, predicting the dose distribution corresponding to the first machine parameter according to the first machine parameter and a pre-trained prediction deep learning model.

[0099] Specifically, the first machine parameter obtained in step S540 is input into a pre-trained prediction deep learning model to obtain a predicted dose distribution corresponding to the first machine parameter. The prediction deep learning model is used to predict the dose distribution corresponding to the machine parameter of the radiotherapy device.

[0100] Step S560, obtaining a second dose distribution difference by calculating the dose difference between the dose distribution corresponding to the first machine parameter and the target dose distribution.

[0101] Specifically, the predicted dose distribution corresponding to the first machine parameter is subtracted from the target dose distribution to obtain a dose difference, i.e., a second dose distribution difference. The second dose distribution difference here is the dose distribution difference calculated according to the target dose distribution and the dose distribution corresponding to the first machine parameter as described in the above embodiment.

[0102] Step S570, determining a second machine parameter according to the second dose distribution difference and a pre-trained third deep learning model.

[0103] Specifically, the second dose distribution difference is input into a pre-trained third deep learning model to predict the change in the machine parameter, and the change in the machine parameter is summed with the first machine parameter to obtain the second machine parameter, which is the new machine parameter as the output result.

[0104] In this embodiment, the possible change value of the flux map is predicted by the difference between the dose distribution corresponding to the current flux map and the target dose distribution, so as to make the predicted flux map correspond to the individual-specific optimization target; the change amount of the optimized machine parameters is predicted by the difference between the dose distribution corresponding to the current machine parameters and the target dose distribution, so as to make the predicted optimized machine parameters correspond to the individual-specific optimization target. In the process of radiotherapy optimization, the present application predicts the intermediate results of optimization through a deep learning method (the intermediate results of optimization here include flux map results, machine parameters, machine jump count, etc., which can also be understood as the output of each deep learning model in the embodiment of the present application), rather than directly predicting the final result or the final result of a certain optimization module (such as flux map optimization results, direct machine parameter optimization results), in order to eliminate the problem of poor prediction and optimization effect caused by differences in radiotherapy data caused by factors such as individual patient conditions, physicists, and equipment parameters, and can better achieve individualized and precise radiotherapy goals compared to existing radiotherapy optimization methods based on artificial intelligence.

[0105] Figure 6 is a flow chart of another parameter determination method according to a specific embodiment of the present application. Figure 6 As shown, the process includes:

[0106] Obtain the target dose distribution and the first flux map of the current case; the first flux map can be an initial flux map, but can also be an optimized or processed flux map. Input the first flux map into the first deep learning model, and convert the first flux map into its corresponding dose distribution, i.e., the predicted dose. Subtract the target dose distribution from the predicted dose to obtain the dose difference. Input the dose difference into the second deep learning model, and convert the dose difference into the predicted flux map change. Sum the predicted flux map change with the first flux map to obtain a new flux map, and end the process.

[0107] Figure 7 is a flow chart of another parameter determination method according to a specific embodiment of the present application. Figure 7 As shown, the process includes:

[0108] Obtain the target dose distribution and the first flux map of the current case; the first flux map can be an initial flux map, but can also be an optimized or processed flux map. Input the first flux map into the deep learning model for conversion, and convert the first flux map into its corresponding machine parameters, i.e., the predicted initial machine parameters. Input the predicted initial machine parameters into the deep learning model for prediction, and convert the predicted initial machine parameters into predicted doses. Subtract the target dose distribution from the predicted dose to obtain a dose difference. Input the dose difference into the third deep learning model, and convert the dose difference into a machine parameter change. Sum the predicted initial machine parameters and the machine parameter change to obtain new machine parameters, and end the process. The training process of the deep learning model in the embodiment of the present application is described below.

[0109] As an example, the training data includes but is not limited to: outlining the region of interest, obtaining the projection shape of the region of interest at each beam angle; the first flux map, the intermediate optimization result (in the form of a flux map) of the traditional flux map optimization algorithm (FMO, fluence map optimization), the final optimization result (in the form of a flux map), and the three-dimensional dose distribution corresponding to all the above flux maps (obtained by the dose calculation algorithm); the first machine parameter (including the MLC (multi-leaf collimator) blade shape and number of hops), the intermediate result (blade shape and number of hops) of the traditional machine parameter optimization algorithm, the final optimization result (blade shape and number of hops), and the three-dimensional dose distribution corresponding to all the above results (obtained by the dose calculation algorithm).

[0110] As a non-limiting example, the training process is:

[0111] The first deep learning model input is a flux map and outputs a three-dimensional dose map.

[0112] Set the initial input flux map as A0, and take n intermediate result flux maps from the traditional optimization algorithm, set as: A1, A2, ..., A n ; Set the three-dimensional dose distribution corresponding to the flux map to D0, D1,…, D n , with the corresponding A n and D n As a set of inputs and gold standards, the network is input for training to obtain the first deep learning model.

[0113] The input of the second deep learning model is the three-dimensional dose difference, and the output is the flux map change.

[0114] Set the initial input as A0, and take n intermediate result flux graphs from the traditional optimization algorithm, set as: A1, A2, ..., A n ; Set the three-dimensional dose distribution corresponding to the flux map to D0, D1,…, D n ; Take any two sets of corresponding data A k , D k and A m , D m Do poorly and get an A m-k ,D m-k As a set of inputs and a gold standard, the network is input for training to obtain a second deep learning model.

[0115] The deep learning model used for the conversion takes as input a flux map and outputs machine parameters.

[0116] The input is a flux map, and the corresponding machine parameters are obtained by converting the flux map into machine parameters using a traditional algorithm. A set of input networks are trained to obtain a deep learning model for conversion.

[0117] The prediction deep learning model takes machine parameters as input and outputs a three-dimensional dose map.

[0118] Set the initial input machine parameter to C0, and take n intermediate results from the traditional optimization algorithm, set as: C1, C2, ..., C n ; Set the three-dimensional dose distribution corresponding to the above machine parameters to D0, D1,…, D n ; With the corresponding C n and D n The network is trained for a set of inputs and a gold standard input to obtain a deep learning model for prediction.

[0119] The input of the third deep learning model is the three-dimensional dose difference, and the output is the change in machine parameters.

[0120] Set the initial input machine parameter to C0, and take n intermediate results from the traditional optimization algorithm, set as: C1, C2, ..., C n ; Set the three-dimensional dose distribution corresponding to the flux map to D0, D1,…, D n ; Take any two sets of corresponding data C k , D k and C m , D m Do worse and get C m-k ,D m-k As a set of inputs and a gold standard, the network is input for training to obtain a third deep learning model.

[0121] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0122] In this embodiment, a parameter determination device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. The terms "module", "unit", "subunit" and the like used below can implement a combination of software and / or hardware of predetermined functions. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware is also possible and conceivable.

[0123] Figure 8 is a structural block diagram of a parameter determination device according to an embodiment of the present application, such as Figure 8 As shown, the device comprises:

[0124] A first determination module 610 is used to obtain a dose distribution corresponding to the first flux map according to the first flux map and a pre-trained first deep learning model, wherein the first deep learning model is used to obtain the dose distribution in the flux map;

[0125] The second determination module 620 is used to obtain a dose distribution difference by calculating the dose difference between the target dose distribution and the dose distribution; and determine the target parameter according to the dose distribution difference.

[0126] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0127] In this embodiment, an electronic device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0128] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0129] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0130] S1. Obtain a dose distribution corresponding to the first flux map according to a first flux map and a pre-trained first deep learning model, wherein the first deep learning model is used to obtain the dose distribution in the flux map.

[0131] S2, obtaining a dose distribution difference by calculating a dose difference between the target dose distribution and the dose distribution; and determining a target parameter according to the dose distribution difference.

[0132] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0133] In addition, in combination with a parameter determination method provided in the above embodiment, a storage medium may be provided in this embodiment to implement the method. The storage medium stores a computer program; when the computer program is executed by a processor, the steps of any parameter determination method in the above embodiment are implemented.

[0134] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0135] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

[0136] The term "embodiment" in this application refers to a specific feature, structure or characteristic described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is clearly or implicitly understood by those of ordinary skill in the art that the embodiments described in this application can be combined with other embodiments without conflict.

[0137] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A parameter determination method, characterized in that: The parameter determination method comprises: According to the first flux map and a pre-trained first deep learning model, obtaining a dose distribution corresponding to the first flux map; By calculating the dose difference between the target dose distribution and the dose distribution, a dose distribution difference value is obtained; and The target parameter is determined according to the dose distribution difference.

2. The parameter determination method according to claim 1, characterized in that: The target parameter is a second flux map, and determining the target parameter according to the dose distribution difference includes: A second flux map is determined according to the dose distribution difference and a pre-trained second deep learning model.

3. The parameter determination method according to claim 2, characterized in that: Determining a second flux map according to the dose distribution difference and a pre-trained second deep learning model includes: Predicting a flux map change corresponding to the dose distribution difference according to the dose distribution difference and a pre-trained second deep learning model; and The second flux map is determined according to the first flux map and the flux map change amount.

4. The parameter determination method according to claim 1, characterized in that: The target parameter is a second machine parameter of the radiotherapy device, and determining the target parameter according to the dose distribution difference includes: The second machine parameter is determined according to the dose distribution difference and a pre-trained third deep learning model.

5. The parameter determination method according to claim 4, characterized in that: The first deep learning model includes a conversion deep learning model and a prediction deep learning model, and acquiring a dose distribution corresponding to the first flux map according to the first flux map and the pre-trained first deep learning model includes: Determining a first machine parameter of a radiotherapy device corresponding to the first flux map according to the first flux map and the pre-trained deep learning model for conversion, wherein the deep learning model for conversion is used to convert the flux map into the machine parameter of the radiotherapy device; and According to the first machine parameters and the pre-trained prediction deep learning model, a dose distribution corresponding to the first machine parameters is predicted.

6. The parameter determination method according to claim 5, characterized in that: The determining of a second machine parameter according to the dose distribution difference and a pre-trained third deep learning model comprises: Determining, based on the dose distribution difference and a pre-trained third deep learning model, a change in a machine parameter of the radiotherapy device corresponding to the dose distribution difference; and The second machine parameter is determined according to the first machine parameter and the machine parameter change.

7. The parameter determination method according to claim 3, characterized in that: The second flux map is the sum of the first flux map and the flux map variation.

8. A parameter determination device, characterized in that: The parameter determination device comprises: A first determination module, configured to obtain a dose distribution corresponding to a first flux map according to a first flux map and a pre-trained first deep learning model; The second determination module is used to obtain a dose distribution difference by calculating the dose difference between the target dose distribution and the dose distribution; and determine the target parameter according to the dose distribution difference.

9. A radiotherapy device, characterized in that: The machine parameters of the radiotherapy apparatus are determined according to the parameter determination method according to any one of claims 1-7.

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 steps of the parameter determination method according to any one of claims 1 to 7 are implemented.