Radiation delivery plan determination method and device and storage medium

Abstract

The invention relates to a radiation delivery plan determination method and device and a storage medium, and the method comprises the steps: obtaining a to-be-evaluated radiation delivery plan and corresponding parameter information; inputting the to-be-evaluated radiation delivery plan and the parameter information into a scoring model to obtain a score of the to-be-evaluated radiation delivery plan; the scoring model is obtained by training according to a first sequence of a plurality of radiation delivery plan samples and parameter information corresponding to the plurality of radiation delivery plan samples; the first sequence is obtained by sequencing the plurality of radiation delivery plan samples by a user according to the parameter information; and determining a target radiation delivery plan according to the score of the radiation delivery plan to be evaluated. Through the method and the device, the problems that the quality evaluation of the radiation delivery plan is inaccurate and a proper radiation delivery plan cannot be selected in the prior art are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of radiation delivery, and in particular to a method, device and storage medium for determining a radiation delivery plan. BACKGROUND

[0002] In the field of radiation delivery (such as radiotherapy, radiation processing, radiation inspection, etc.), a suitable radiation delivery plan needs to be formulated or selected according to the situation to avoid insufficient or excessive irradiation, etc. Therefore, the quality of the radiation delivery plan needs to be accurately evaluated to achieve accurate and efficient radiation delivery.

[0003] As an example, in the field of radiotherapy, due to individual differences of patients receiving radiation delivery and different experiences of evaluation subjects (such as hospitals, doctors, and physicists, etc.), different evaluation subjects have certain differences in requirements for evaluation indexes of radiation delivery plans. At present, there is usually a method based on fixed dose indexes to evaluate the quality of the final radiation delivery plan. The fixed dose index scoring method does not take into account the individual differences of patients, and for cases with different degrees of difficulty of radiation delivery plans, this method cannot fully reflect the quality of the radiation delivery plan, thereby leading to inaccurate quality evaluation of the radiation delivery plan and the inability to select a suitable radiation delivery plan. SUMMARY

[0004] In this embodiment, a method, device and storage medium for determining a radiation delivery plan are provided to solve the problem of inaccurate quality evaluation of the radiation delivery plan in the prior art.

[0005] In a first aspect, a method for determining a radiation delivery plan is provided in this embodiment, comprising:

[0006] obtaining a radiation delivery plan to be evaluated and corresponding parameter information;

[0007] inputting the radiation delivery plan to be evaluated and the parameter information into a scoring model to obtain a score of the radiation delivery plan to be evaluated; the scoring model is trained according to a first ranking of a plurality of radiation delivery plan samples and parameter information corresponding to the plurality of radiation delivery plan samples; the first ranking is obtained by a user ranking the plurality of radiation delivery plan samples according to the parameter information; and

[0008] determining a target radiation delivery plan according to the score of the radiation delivery plan to be evaluated.

[0009] In some embodiments, determining the target radiation delivery plan according to the score of the radiation delivery plan to be evaluated comprises:

[0010] when the score of the radiation delivery plan to be evaluated is above a preset value, determining the radiation delivery plan to be evaluated as the target radiation delivery plan.

[0011] In some embodiments, the parameter information comprises one or more of the following: region of interest information, index information, radiation delivery plan execution scenario information, and user profile information.

[0012] In a second aspect, a radiation delivery plan determination method is provided in the embodiments, comprising:

[0013] obtaining training samples, the training samples comprising a plurality of radiation delivery plan samples and parameter information corresponding to each radiation delivery plan sample;

[0014] obtaining a first ranking of the plurality of radiation delivery plan samples, the first ranking of the plurality of radiation delivery plan samples being obtained by a user ranking the plurality of radiation delivery plan samples according to the parameter information;

[0015] constructing a scoring model, and inputting the plurality of radiation delivery plan samples and the parameter information corresponding to each radiation delivery plan sample into the scoring model respectively to obtain a plurality of scores corresponding to the plurality of radiation delivery plan samples; and

[0016] updating the scoring model according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples to obtain a trained scoring model, the trained scoring model being used to determine a target radiation delivery plan.

[0017] In some embodiments, updating the scoring model according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples to obtain a trained scoring model comprises:

[0018] ranking the plurality of radiation delivery plan samples according to the plurality of scores corresponding to the plurality of radiation delivery plan samples to obtain a second ranking of the plurality of radiation delivery plan samples, the ranking rules of the first ranking and the second ranking being the same; and

[0019] updating the scoring model when the first ranking of the plurality of radiation delivery plan samples is inconsistent with the second ranking.

[0020] In some embodiments, updating the scoring model according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples to obtain a trained scoring model comprises:

[0021] constructing a loss function according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples; and

[0022] updating the scoring model according to the loss function to obtain the trained scoring model.

[0023] In some embodiments, the target radiotherapy delivery plan is determined according to the trained scoring model, comprising:

[0024] obtaining a radiotherapy delivery plan to be evaluated and corresponding parameter information;

[0025] inputting the radiotherapy delivery plan to be evaluated and the parameter information into the trained scoring model to obtain a score of the radiotherapy delivery plan to be evaluated;

[0026] determining the target radiotherapy delivery plan according to the score of the radiotherapy delivery plan to be evaluated.

[0027] In a third aspect, a device for determining a radiotherapy delivery plan is provided in the present embodiment, the device comprising:

[0028] a data obtaining module configured to obtain a radiotherapy delivery plan to be evaluated and corresponding parameter information;

[0029] a plan scoring module configured to input the radiotherapy delivery plan to be evaluated and the parameter information into a scoring model to obtain a score of the radiotherapy delivery plan to be evaluated; the scoring model is trained according to a first ranking of a plurality of radiotherapy delivery plan samples and parameter information corresponding to the plurality of radiotherapy delivery plan samples; the first ranking is obtained by a user ranking the plurality of radiotherapy delivery plan samples according to the parameter information; and

[0030] a plan determining module configured to determine a target radiotherapy delivery plan according to the score of the radiotherapy delivery plan to be evaluated.

[0031] In a fourth aspect, a device for determining a radiotherapy delivery plan is provided in the present embodiment, the device comprising:

[0032] a training sample obtaining module configured to obtain training samples; the training samples comprising a plurality of radiotherapy delivery plan samples and parameter information corresponding to each radiotherapy delivery plan sample;

[0033] a ranking obtaining module configured to obtain a first ranking of the plurality of radiotherapy delivery plan samples; the first ranking of the plurality of radiotherapy delivery plan samples is obtained by a user ranking the plurality of radiotherapy delivery plan samples according to the parameter information;

[0034] a constructing module configured to construct a scoring model, and a training module configured to input the plurality of radiotherapy delivery plan samples and the parameter information corresponding to each radiotherapy delivery plan sample into the scoring model respectively to obtain a plurality of scores corresponding to the plurality of radiotherapy delivery plan samples; and

[0035] The updating module is configured to update the scoring model according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples, to obtain a trained scoring model, which is used to determine the target radiation delivery plan.

[0036] In a fifth aspect, a computer-readable storage medium is provided in the present embodiment, which stores instructions that, when executed by a processor, cause the processor to perform the radiation delivery plan determination method of the first aspect or the second aspect.

[0037] Compared with the prior art, the radiation delivery plan determination method, device and storage medium provided in the present embodiment train the scoring model according to the ranking and parameter information of the user on the radiation delivery plan, score one or more radiation delivery plans to be evaluated according to the trained scoring model, and determine the target radiation delivery plan. The trained scoring model can score the radiation delivery plan to be evaluated individually, so that the score of the radiation delivery plan to be evaluated is more accurate, and thus the determined target radiation delivery plan is more accurate. The problems in the prior art that the fixed dose index scoring method does not consider individual differences of patients, and cannot fully reflect the quality of the radiation delivery plan for different cases of the radiation delivery plan, resulting in inaccurate quality evaluation of the radiation delivery plan and the inability to select a suitable radiation delivery plan are solved.

[0038] Details of one or more embodiments of the present application are presented in the following drawings and description to make other features, objects and advantages of the present application more apparent. BRIEF DESCRIPTION OF DRAWINGS

[0039] The drawings described herein are intended to provide further understanding of the present application, and form a part of the present application. The illustrative embodiments of the present application and their description serve to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0040] Figure 1 is a hardware structure block diagram of a terminal for performing a radiation delivery plan determination method of an embodiment of the present application;

[0041] Figure 2 is a flowchart of a radiation delivery plan determination method of an embodiment of the present application;

[0042] Figure 3 is a flowchart of a radiation delivery plan determination method of an embodiment of the present application;

[0043] Figure 4 is a schematic diagram of a scoring model training process of an embodiment of the present application;

[0044] Figure 5This is a flowchart of a training process of a scoring model according to an embodiment of the present application;

[0045] Figure 6 is a schematic diagram of scoring a radiotherapy plan using a trained scoring model according to an embodiment of the present application;

[0046] Figure 7 is a structural block diagram of a device for determining a radiation delivery plan according to an embodiment of the present application;

[0047] Figure 8 This is a structural block diagram of a device for determining a radiation delivery plan according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] 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.

[0049] Unless otherwise defined, the technical terms or scientific terms involved in this application should have the general meaning understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "an", "a", "the", "these" and the like in this application do not indicate 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 direct or indirect. The "plurality" involved in this application refers to two or more. "And / or" describes the relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, and B exists alone. Generally, the character " / " indicates that the related objects are in an "or" relationship. The terms "first," "second," "third," etc. used in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0050] 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 FIG. 1 is a hardware structure diagram of a terminal for executing a method for determining a radiation delivery plan according to an embodiment of the present application. Figure 1 As shown, the terminal may include one or more ( Figure 1The terminal shown in FIG. 1 includes only one processor 102 and a memory 104 for storing data, wherein the processor 102 can include, but is not limited to, a processing device such as a microprocessor (MCU) or a programmable logic device (FPGA). The terminal can further include a transmission device 106 for communication function and an input / output device 108. Those skilled in the art can understand that Figure 1 The structure shown in FIG. 1 is only schematic and does not limit the structure of the terminal. For example, the terminal can include more or fewer components than those shown in FIG. 1, or have a different configuration from that shown in FIG. 1. Figure 1 Figure 1 The structure shown in FIG. 1 is only schematic and does not limit the structure of the terminal. For example, the terminal can include more or fewer components than those shown in FIG. 1, or have a different configuration from that shown in FIG. 1.

[0051] The memory 104 can be used to store computer programs, such as software programs of application software and modules, such as a computer program corresponding to a radiation delivery plan determination method in the present embodiment. The processor 102 can execute various functional applications and data processing by running the computer programs stored in the memory 104, i.e., implement the method described above. The memory 104 can include a high-speed random access memory and can further include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some examples, the memory 104 can further include a memory remotely arranged with respect to the processor 102, and these remote memories can be connected to the terminal through a network. Examples of the network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0052] The transmission device 106 is used to receive or send data via a network. The network includes a wireless network provided by a communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (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 (RF) module which is used to communicate with the Internet in a wireless manner.

[0053] The present embodiment provides a radiation delivery plan determination method, Figure 2 is a flowchart of a radiation delivery plan determination method according to an embodiment of the present application, as shown in FIG. 2, the flow includes the following steps: Figure 2

[0054] In step S210, a radiation delivery plan to be evaluated and corresponding parameter information are acquired.

[0055] ​​In this step, the to-be-evaluated radiotherapy plan and the parameter information corresponding to the to-be-evaluated radiotherapy plan can be obtained from a radiotherapy plan database. The radiotherapy plan database pre-stores a plurality of radiotherapy plans and parameter information corresponding to the radiotherapy plans. The to-be-evaluated radiotherapy plan and the parameter information corresponding to the to-be-evaluated radiotherapy plan can also be obtained according to the input information of the user. The to-be-evaluated radiotherapy plan obtained here can include one or more to-be-evaluated radiotherapy plans, and the corresponding parameter information obtained also includes parameter information corresponding to one or more to-be-evaluated radiotherapy plans. The to-be-evaluated radiotherapy plan and other radiotherapy plans involved herein can be or include one or more of a radiotherapy plan, a radiation processing plan, a radiation flaw detection plan, etc., which are not specifically limited here. The parameter information can include one or more of region of interest information, index information, radiotherapy plan execution scene information, user portrait information, etc., which are not specifically limited here. The user in the embodiment described herein can be a device operator, a medical staff, a physicist, etc., which are not specifically limited here.

[0056] In step S220, the to-be-evaluated radiotherapy plan and the parameter information are input into a scoring model to obtain a score of the to-be-evaluated radiotherapy plan. The scoring model is trained according to a first ranking of a plurality of radiotherapy plan samples and parameter information corresponding to the plurality of radiotherapy plan samples. The first ranking is obtained by a user ranking the plurality of radiotherapy plan samples according to the parameter information.

[0057] In this step, the to-be-evaluated radiotherapy plan and the corresponding parameter information obtained are input into the scoring model for scoring to obtain a score of each to-be-evaluated radiotherapy plan. The user ranks the plurality of radiotherapy plan samples according to the parameter information of each radiotherapy plan sample to obtain a first ranking of the plurality of radiotherapy plan samples, and then trains the scoring model according to the obtained first ranking of the plurality of radiotherapy plan samples and the parameter information corresponding to the plurality of radiotherapy plan samples to obtain the trained scoring model. The first ranking here can be obtained by the user ranking the plurality of radiotherapy plan samples according to the region of interest or the dose index of interest. The scoring model here can be various types of artificial intelligence models. Exemplary scoring models can include but are not limited to a convolutional neural network model, a generative adversarial network model, a deep reinforcement learning model, etc., or a combination thereof.

[0058] In step S230, a target radiotherapy plan is determined according to the score of the to-be-evaluated radiotherapy plan.

[0059] In this step, the to-be-evaluated radiotherapy delivery plans can be sorted according to the scores of the to-be-evaluated radiotherapy delivery plans to obtain a sorting result, and one or more to-be-evaluated radiotherapy delivery plans that meet a preset condition in the sorting are determined as target radiotherapy delivery plans. In addition, one or more to-be-evaluated radiotherapy delivery plans with a score higher than (or greater than) a preset value can also be determined as target radiotherapy delivery plans according to the scores of the to-be-evaluated radiotherapy delivery plans. For example, if the preset value is M points, one or more to-be-evaluated radiotherapy delivery plans with a score higher than M points are determined as target radiotherapy delivery plans. Here, M can be set according to an empirical value or user demand.

[0060] Through the above steps, the scoring model is trained according to the sorting and parameter information of the radiotherapy delivery plans by the user, and one or more to-be-evaluated radiotherapy delivery plans are scored according to the trained scoring model, so as to determine target radiotherapy delivery plans. The trained scoring model can score the to-be-evaluated radiotherapy delivery plans individually, so that the scores of the to-be-evaluated radiotherapy delivery plans are more accurate, and thus the determined target radiotherapy delivery plans are more accurate. The problems in the prior art that the fixed dose index scoring method does not take into account individual differences of patients, and for cases with different radiotherapy delivery plan difficulty levels, the quality of the radiotherapy delivery plans cannot be fully reflected, resulting in inaccurate quality evaluation of the radiotherapy delivery plans and the inability to select appropriate radiotherapy delivery plans are solved.

[0061] In some embodiments, the target radiotherapy delivery plan is determined according to the score of the to-be-evaluated radiotherapy delivery plan, including: when the score of the to-be-evaluated radiotherapy delivery plan is greater than or equal to a preset value, the to-be-evaluated radiotherapy delivery plan is determined as the target radiotherapy delivery plan.

[0062] The radiotherapy delivery plan to be evaluated input into the scoring model can be one or multiple. When the radiotherapy delivery plan to be evaluated input into the scoring model is one, the score of the one radiotherapy delivery plan to be evaluated is compared with a preset value, and when the score of the one radiotherapy delivery plan to be evaluated is higher than the preset value, the radiotherapy delivery plan to be evaluated is determined as the target radiotherapy delivery plan. When the radiotherapy delivery plan to be evaluated input into the scoring model is multiple, the score of each radiotherapy delivery plan to be evaluated in at least part of the radiotherapy delivery plans to be evaluated is obtained, and the score of each radiotherapy delivery plan to be evaluated in at least part of the radiotherapy delivery plans to be evaluated is compared with a preset value, and one or more radiotherapy delivery plans to be evaluated with a score higher than the preset value are determined as the target radiotherapy delivery plan. Here, the one or more radiotherapy delivery plans to be evaluated with a score higher than the preset value can be all the radiotherapy delivery plans to be evaluated with a score higher than the preset value, or the radiotherapy delivery plan to be evaluated with the highest score among the multiple radiotherapy delivery plans to be evaluated with a score higher than the preset value is determined as the target radiotherapy delivery plan. The preset value can be customized by the user or determined according to the training result of the scoring model.

[0063] In this embodiment, the target radiotherapy delivery plan is determined by comparing the score obtained according to the radiotherapy delivery plan to be evaluated through the scoring model with the preset value, so that the determined target radiotherapy delivery plan is more accurate.

[0064] The embodiment also provides a radiotherapy delivery plan determination method, Figure 3 is a flowchart of a radiotherapy delivery plan determination method according to an embodiment of the present application, as shown in the figure, the flow includes the following steps: Figure 3

[0065] In step S310, a training sample is obtained; the training sample includes multiple radiotherapy delivery plan samples and parameter information corresponding to each radiotherapy delivery plan sample.

[0066] In this step, the training sample including multiple radiotherapy delivery plan samples and parameter information corresponding to each radiotherapy delivery plan sample is obtained. The training sample can be obtained from a radiotherapy delivery plan database or constructed by the user according to experience values.

[0067] In step S320, a first ranking of multiple radiotherapy delivery plan samples is obtained; the first ranking of multiple radiotherapy delivery plan samples is obtained by ranking the multiple radiotherapy delivery plan samples according to the parameter information by the user.

[0068] ​In this step, the user sorts the plurality of radiation delivery plan samples according to the parameter information of each radiation delivery plan sample, to obtain a first sorting of the plurality of radiation delivery plan samples. The first sorting here can also be the order obtained by the user sorting the plurality of radiation delivery plan samples according to the parameter information and experience, etc.

[0069] Step S330, constructing a scoring model.

[0070] The constructed scoring model is configured to output a score corresponding to a radiotherapy delivery plan according to an input radiotherapy delivery plan and corresponding parameter information. The input of the scoring model is a radiotherapy delivery plan sample and parameter information corresponding to the radiotherapy delivery plan sample, and the output of the scoring model is a plurality of scores corresponding to a plurality of input radiotherapy delivery plan samples. The scoring model is trained to obtain a trained scoring model. For example, the scoring model includes an input layer, a hidden layer, and an output layer. The input layer is configured to receive a plurality of radiotherapy delivery plan samples and parameter information corresponding to each radiotherapy delivery plan sample. The hidden layer and the output layer are configured to train the scoring model using a deep learning algorithm. For example, the scoring model includes a backbone network and a plurality of branches. The backbone network is configured to extract a corresponding feature vector according to an input radiotherapy delivery plan and corresponding parameter information. Each branch network further processes each feature vector output by the backbone network to output a plurality of feature information corresponding to each feature vector. The backbone network is used as a feature extractor, and the output of the last fully connected layer in the backbone network is regarded as a feature vector of the radiotherapy delivery plan and corresponding parameter information. The network structure of the backbone network can be selected according to different application scenarios and requirements. For example, a deep convolutional neural network model such as Alex Net, VGG, Res Net, Google Net, or Inception Net can be selected. The network structure of the backbone network can be further selected according to different performance requirements, such as different numbers of convolutional layers, different numbers of pooling layers, different activation functions, etc. Each branch network includes at least one convolutional layer and a corresponding activation function layer. Different branch networks are configured by different network structures to obtain feature information of confidence S (indicating the accuracy and reliability of the feature point corresponding to the feature vector), position P (indicating the position of the feature point corresponding to the feature vector), and descriptor F (indicating the uniqueness and distinguishability of the feature point corresponding to the feature vector). The constructed scoring model can further extract features from the feature vector extracted by the backbone network through different branch networks, thereby obtaining more feature information of each feature vector, and making the score of the scoring model more accurate. The network structure of each branch can be selected according to different performance requirements, such as different numbers of convolutional layers and different activation functions. For example, in some embodiments, a sigmoid activation function can be selected, and in some other embodiments, a ReLU activation function or other activation functions can be selected.The scoring model takes the language model transformer as an example. Certain parameter information such as index requirements (text) can be input to help the scoring model output better results. A prompt box interface is set for the scoring model. The region of interest or dose index that needs to be focused on can be input in the prompt box of the scoring model to obtain a plan score that pays more attention to these indexes. If no input is made, the scoring model outputs a general mode score. The region of interest or dose index that can be appropriately sacrificed can be input in the dose prediction model prompt box as a negative index requirement. The negative index requirement is input to the scoring model, so that the scoring model reduces the proportion of the influence of these indexes on the final score. The language model can be replaced by a convolutional neural network or other network model. The input index (text) information needs to be processed into a specified length vector or image information for network processing.

[0071] In step S340, the plurality of radiation delivery plan samples and the parameter information corresponding to each radiation delivery plan sample are respectively input into the scoring model to obtain a plurality of scores corresponding to the plurality of radiation delivery plan samples. As an example, the scoring model can score each radiation delivery plan sample according to the input plurality of radiation delivery plan samples and their parameter information, thereby obtaining a plurality of scores.

[0072] In step S350, the scoring model is updated according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples, to obtain a trained scoring model, and the trained scoring model is used to determine the target radiation delivery plan.

[0073] In this step, the scoring model is updated based on the first ranking of the plurality of radiation delivery plan samples and the multiple scores corresponding to the plurality of radiation delivery plan samples. The trained scoring model can be obtained by ranking the plurality of radiation delivery plan samples based on the multiple scores corresponding to the plurality of radiation delivery plan samples to obtain a second ranking of the plurality of radiation delivery plan samples. The second ranking can be a ranking obtained by sorting the plurality of radiation delivery plan samples in descending order based on the scores, or a ranking obtained by sorting the plurality of radiation delivery plan samples in ascending order based on the scores, such as placing radiation delivery plan samples with higher scores at the top or lower scores at the top. The second ranking can have the same sorting logic as the first ranking, such as placing radiation delivery plan samples that are more suitable for the patient or better meet the user's needs at the top, or placing radiation delivery plan samples that are more suitable for the patient or better meet the user's needs at the bottom. If the first ranking of the plurality of radiation delivery plan samples is inconsistent with the second ranking, the scoring model is updated. As an example, the parameters, structure, etc. of the scoring model can be updated. The sorting rules of the first and second rankings are the same. The scoring model can be updated based on the first ranking of the plurality of radiation delivery plan samples and the multiple scores corresponding to the plurality of radiation delivery plan samples to obtain a trained scoring model. Alternatively, the scoring model can be updated by ranking the plurality of radiation delivery plan samples based on the multiple scores corresponding to the plurality of radiation delivery plan samples to obtain a second ranking of the plurality of radiation delivery plan samples, comparing the first ranking and the second ranking to obtain a difference between the first ranking and the second ranking, and updating the scoring model when the difference is greater than a first preset value. The first preset value can be set empirically. Optionally, the difference can be considered in updating the scoring model. For example, the difference can be used as an indicator in an optimization function for updating and optimizing the scoring model, thereby ensuring that the plan ranking obtained based on the scoring model is as close as possible to the plan ranking determined by the user. Alternatively, the scoring model can be updated based on the first ranking of the plurality of radiation delivery plan samples and the multiple scores corresponding to the plurality of radiation delivery plan samples to obtain a trained scoring model by obtaining a number of scoring model update iterations. When the number of iterations is greater than a second preset value and the difference between the first ranking and the second ranking is less than the first preset value, the updating is stopped to obtain a trained scoring model. The second preset value can be set empirically, such as 30,000, 40,000, or 50,000 iterations. A radiation delivery plan to be evaluated and its corresponding parameter information are obtained; the radiation delivery plan to be evaluated and its parameter information are input into a trained scoring model, which then outputs a score for the radiation delivery plan to be evaluated. A target radiation delivery plan can then be determined based on the score of the radiation delivery plan to be evaluated, either automatically by a computing device or manually by a user.

[0074] The score model is trained according to the ranking and parameter information of the user on the radiation delivery plan, the one or more radiation delivery plans to be evaluated are scored according to the trained score model, and the target radiation delivery plan is determined. The trained score model can score the radiation delivery plan to be evaluated individually, so that the score of the radiation delivery plan to be evaluated is more accurate, so that the determined target radiation delivery plan is more accurate, and the problems in the prior art that the fixed dose index scoring method does not consider individual differences of patients, cannot fully reflect the quality of the radiation delivery plan for different cases of the radiation delivery plan, and thus leads to inaccurate quality evaluation of the radiation delivery plan and cannot select a suitable radiation delivery plan are solved.

[0075] In some embodiments, the score model is updated according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples to obtain the trained score model, including: constructing a loss function according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples; and updating the score model according to the loss function to obtain the trained score model.

[0076] In this step, the first score of the first radiation delivery plan and the second score of the second radiation delivery plan are obtained; the position of the first radiation delivery plan in the first ranking is in front of the position of the second radiation delivery plan in the first ranking; and a loss function is constructed according to the difference between the first score and the second score. Alternatively, the plurality of radiation delivery plan samples are ranked according to the plurality of scores corresponding to the plurality of radiation delivery plan samples to obtain a third ranking, and a loss function is constructed according to the difference between the first ranking and the third ranking.

[0077] In this embodiment, the loss function is constructed according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples, and the score model is updated according to the loss function to obtain the trained score model, thereby improving the training speed of the score model.

[0078] In some other embodiments, the score model is updated according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples to obtain the trained score model, including: calculating the semantic similarity between two adjacent radiation delivery plan samples in the first ranking, calculating the difference between the scores of the two adjacent radiation delivery plan samples, constructing an objective function according to the semantic similarity and the difference, and updating the score model according to the objective function to obtain the trained score model.

[0079] In the embodiment, a target function is constructed according to semantic similarity between two adjacent radiotherapy plan samples in the first ranking and a difference value between scores of the two adjacent radiotherapy plan samples, the scoring model is updated according to the target function, and a trained scoring model is obtained, thereby improving the training speed of the scoring model.

[0080] The radiotherapy plan is exemplarily described below as a radiotherapy plan.

[0081] Training data preparation: For a case, three-dimensional dose field data of multiple radiotherapy plans are collected or obtained on the same set of image data (for example, CT images, MR images, etc.) and the region of interest delineation (the multiple plans can be multiple plans formulated or determined by a physicist under different optimization objectives and / or weights; or the radiotherapy plans can be automatically generated by using existing technologies, such as image generation technologies and deep learning technologies, etc.). The multiple radiotherapy plans are ranked by a senior physicist or doctor to obtain a first ranking, but specific scores do not need to be given. The standard for giving the ranking or ranking can be arbitrary, and only as an example, the standard can be or include a dose distribution preferred situation for a certain type of patient or a specific patient considered by a user based on historical experience, academic research conclusions, etc. (for example, for some patients, the dose of some sensitive organs at risk is lower the better, and the dose of some other insensitive organs at risk can be slightly higher), whether the dose of the irradiated target region is close to an optimal value (that is, neither too low nor too high), the workload brought by the radiotherapy plan to the radiotherapy equipment, and / or the treatment efficiency of the radiotherapy plan, etc. The optimal value can be obtained by experiment or set according to the user's experience. After the ranking is given, the region of interest or dose index that the doctor or physicist focuses on when ranking the multiple plans of the current case can be labeled, and alternatively or additionally, the region of interest or dose index that is sacrificed to a certain extent can also be labeled, and then these can be used as index requirements, that is, parameter information.

[0082] Scoring model training process: When the scoring model is trained, each plan of the same case is input into the scoring model in order according to the ranking, and if there are index requirements, the index requirements are input from the channels corresponding to the index requirements. The scoring model will output a score for each input plan in order, and if the ranking of the output score does not match the ranking of the plan, the loss function will increase the penalty value. The purpose of the loss function is to maximize the difference between the scores of the plans ranked in the front and the plans ranked in the back, thereby prompting the model to learn the corresponding ranking mode. The training process is repeated until the model converges. The scoring model training process is as follows: Figure 4As shown, after a plurality of radiotherapy plans and index requirements are input into the scoring model, a plurality of scores corresponding to the plurality of radiotherapy plans are output, for example, N scores corresponding to N radiotherapy plans are output after the scoring model inputs N radiotherapy plans and index requirements, and further, for example, a first score corresponding to a first radiotherapy plan, a second score corresponding to a second radiotherapy plan, and an Nth score corresponding to an Nth radiotherapy plan are output after the scoring model inputs a first radiotherapy plan, a second radiotherapy plan, …, and an Nth radiotherapy plan and index requirements. A ranking loss function is constructed according to the obtained scores, and the scoring model is updated according to the constructed ranking loss function.

[0083] Further, referring to Figure 5 , the training process of the scoring model includes the following steps:

[0084] In step S510, the radiotherapy plans and index requirements are sequentially input into the scoring model according to a given order based on a plurality of radiotherapy plans.

[0085] In this step, based on a plurality of radiotherapy plans, the scoring model is sequentially input for training after experienced doctors or physicists give rankings and evaluation indexes to these radiotherapy plans. For example, the user can input the regions of interest and / or dose indexes that need to be focused on and the regions of interest and / or dose indexes that can be sacrificed in the index requirement box provided by the computer system for training the scoring model. The ranking can be given by multiple doctors or physicists to reduce the influence of individual evaluation. The radiotherapy plan can include one or more of machine parameters, weights and jumps of sub-fields, field angles, field regions, dose-volume histogram (DVH), and dose distribution during radiotherapy. The index requirement can include one or more of site, physicist, and specific machine parameter requirements. The index parameter is input as a string, and the radiotherapy plan is input in the form of a three-dimensional matrix (radiotherapy dose map). The plurality of radiotherapy plans can come from: a) multiple radiotherapy plans tried by doctors; b) radiotherapy plans made by multiple doctors / physicists; c) multiple radiotherapy plans output by adjusting parameters of an existing dose prediction model; d) multiple plans output by multi-criteria optimization (MCO); and the like. The input index requirement can include hospital / doctor information.

[0086] In step S520, the scoring model gives scores to the radiotherapy plans one by one based on the radiotherapy plans and evaluation indexes.

[0087] In step S530, the radiotherapy scores are ranked, and a loss function is constructed.

[0088] In this step, the radiotherapy scores are ranked, and the ranking loss function is obtained based on the ranking loss function by comparing the ranking given by the experienced doctors / physicists, where the ranking loss function can be in the following form:

[0089]

[0090] wherein r θ is the score output by the scoring model, x is the index requirement, y better is a radiotherapy plan ranked more forwardly, y worse is a radiotherapy plan ranked more backwardly, r θ (x, y better ) is the score output by the scoring model for the radiotherapy plan ranked more forwardly, E denotes expectation, and D denotes function distribution. In formula (1), the D function distribution is [-log(σ(r θ (x, y better )-r θ (x, y worse )))]. is the expectation of the difference between the scores of the radiotherapy plan y better ranked more forwardly and the radiotherapy plan y worse ranked more backwardly under the x index requirement under the D distribution, and σ(·) is the objective function. Exemplarily, the objective function σ(·) can be used to limit the function value to the range of 0 to 1, that is, 0<σ(·)<1, 0<σ(r θ (x, y better )-r θ (x, y worse ))<1. Each loss value calculated through formula (1) is greater than 0. For all radiotherapy plans, any two are taken out, and the loss function is calculated according to formula (1), and finally all the calculated loss functions are summed as the current total loss function. The loss function used for ranking can be replaced by any function that plays a similar role.

[0091] Step S540, the loss value is reversely derived based on the loss value, and the scoring model parameter is updated.

[0092] Step S550, steps S510-S540 are repeatedly executed until the scoring model converges.

[0093] With reference to Figure 6 , the trained scoring model is used to score the radiotherapy plan, the radiotherapy plan to be evaluated and the index requirement corresponding to the radiotherapy plan are input into the scoring model, and the scoring model outputs an evaluation score of the effect of the current radiotherapy plan. Among them, the input index requirement can sacrifice the index requirement with low importance, that is, the index requirement with low importance of the radiotherapy plan is not input into the scoring model.

[0094] In some embodiments, a plurality of radiotherapy plan samples and corresponding index requirements of a specific patient are obtained, a scoring model is trained according to the plurality of radiotherapy plan samples and corresponding index requirements, the trained scoring model is used to score a radiotherapy plan to be evaluated of the patient and corresponding index requirements, so as to evaluate the radiotherapy plan for the patient. It should be noted that the method in the embodiments can also be applied to a plurality of patients, for example, a plurality of different patients with the same or similar signs or characteristics. Alternatively or additionally, a general scoring model can also be trained using a large amount of patient data, so that accurate scores can be given to the radiotherapy plan of any type of patient input into the scoring model, that is, the scoring model can be applied to various types of patients.

[0095] The scoring model in the embodiments does not depend on specific formulas or biological factors, and can more effectively reflect whether the radiotherapy plan of an individual case is close to the optimal plan that can be achieved under the current case. When training the scoring model, it is not necessary to collect radiotherapy plans with a certain score evaluation as training data, but only experienced physicists or doctors need to rank a plurality of plans corresponding to a case and provide the basis for the ranking. Instead of using a fixed evaluation function, the individual specificity can be considered, so that the final output score better reflects the effect of the radiotherapy plan. Under the premise of sufficient training data and appropriate index requirements, by using the combination of index requirements, the scoring model does not need to be fine-tuned, and can meet the plan scoring under new requirements by converting the requirements of a new site / doctor into a new index requirement combination, without the need to fine-tune the model parameters according to the data of the new site / doctor.

[0096] It should be noted that the steps shown in the above flow or the flowchart of 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 described herein.

[0097] In the embodiments, a radiotherapy delivery plan determination apparatus is also provided, which is used to implement the above embodiments and preferred embodiments, and will not be described again. The terms "module", "unit", "sub-unit" and the like used below can be a combination of software and / or hardware that implements a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware or a combination of software and hardware is also possible and contemplated.

[0098] Figure 7 is a structural block diagram of a radiotherapy delivery plan determination apparatus according to an embodiment of the present application, as shown in Figure 7 The apparatus comprises:

[0099] A data acquisition module 710 is used to acquire the radiation delivery plan to be evaluated and corresponding parameter information;

[0100] The plan scoring module 720 is configured to input the radiation delivery plan to be evaluated and the parameter information into a scoring model to obtain a score for the radiation delivery plan to be evaluated; the scoring model is trained based on a first ranking of a plurality of radiation delivery plan samples and the parameter information corresponding to the plurality of radiation delivery plan samples; the first ranking is obtained by a user ranking the plurality of radiation delivery plan samples based on the parameter information;

[0101] The plan determination module 730 is configured to determine a target radiation delivery plan according to the scores of the radiation delivery plans to be evaluated.

[0102] In some embodiments, the plan determination module 730 is further configured to determine the radiation delivery plan to be evaluated as the target radiation delivery plan when the score of the radiation delivery plan to be evaluated is above a preset value.

[0103] Figure 8 This is a structural block diagram of a device for determining a radiation delivery plan according to an embodiment of the present application. Figure 8 As shown, the device includes:

[0104] The training sample acquisition module 810 is used to acquire training samples; the training samples include multiple radiation delivery plan samples and parameter information corresponding to each radiation delivery plan sample;

[0105] A ranking acquisition module 820 is configured to acquire a first ranking of the plurality of radiation delivery plan samples; the first ranking of the plurality of radiation delivery plan samples is obtained by a user by ranking the plurality of radiation delivery plan samples according to parameter information;

[0106] Construction module 830 is used to construct a scoring model.

[0107] a training module 840 for inputting a plurality of radiation delivery plan samples and parameter information corresponding to each radiation delivery plan sample into a scoring model to obtain a plurality of scores corresponding to the plurality of radiation delivery plan samples;

[0108] The updating module 850 is used to update the scoring model according to the first ranking of the multiple radiation delivery plan samples and the multiple scores corresponding to the multiple radiation delivery plan samples to obtain a trained scoring model, and the trained scoring model is used to determine the target radiation delivery plan.

[0109] In some embodiments, the update module 850 is further configured to:

[0110] rank the plurality of radiation delivery plan samples according to the plurality of scores corresponding to the plurality of radiation delivery plan samples, to obtain a second ranking of the plurality of radiation delivery plan samples; the ranking rules of the first ranking and the second ranking are the same; and

[0111] update the scoring model when the first ranking of the plurality of radiation delivery plan samples is inconsistent with the second ranking.

[0112] In some embodiments, the updating module 850 is further configured to:

[0113] construct a loss function according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples; and

[0114] update the scoring model according to the loss function, to obtain a trained scoring model.

[0115] In some embodiments, the second determining module 860 is further configured to:

[0116] obtain a radiation delivery plan to be evaluated and corresponding parameter information;

[0117] input the radiation delivery plan to be evaluated and the parameter information into the trained scoring model, to obtain a score of the radiation delivery plan to be evaluated;

[0118] determine a target radiation delivery plan according to the score of the radiation delivery plan to be evaluated.

[0119] It should be noted that each of the above modules can be a functional module or a program module, and can be implemented by software or hardware. For the modules implemented by hardware, each of the above modules can be located in the same processor; or each of the above modules can also be located in different processors in any combination.

[0120] In this embodiment, an electronic device is also provided, which includes a memory and a processor, the memory stores a computer program, and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0121] Optionally, the electronic device can further include a transmission device and an input / output device, wherein the transmission device is connected with the processor, and the input / output device is connected with the processor.

[0122] Optionally, in this embodiment, the processor can be configured to perform the following steps by the computer program:

[0123] S1, obtaining a radiation delivery plan to be evaluated and corresponding parameter information;

[0124] S2, input the radiation delivery plan to be evaluated and the parameter information into the scoring model to obtain a score of the radiation delivery plan to be evaluated; the scoring model is trained according to the first ranking of the plurality of radiation delivery plan samples and the parameter information corresponding to the plurality of radiation delivery plan samples; the first ranking is obtained by a user ranking the plurality of radiation delivery plan samples according to the parameter information;

[0125] S3, determine the target radiation delivery plan according to the score of the radiation delivery plan to be evaluated.

[0126] Optionally, in the embodiment, the processor can be configured to execute the following steps by means of a computer program:

[0127] Q1, obtain training samples; the training samples include a plurality of radiation delivery plan samples and parameter information corresponding to each radiation delivery plan sample;

[0128] Q2, obtain a first ranking of the plurality of radiation delivery plan samples; the first ranking of the plurality of radiation delivery plan samples is obtained by a user ranking the plurality of radiation delivery plan samples according to the parameter information;

[0129] Q3, construct a scoring model;

[0130] Q4, input the plurality of radiation delivery plan samples and the parameter information corresponding to each radiation delivery plan sample into the scoring model respectively to obtain a plurality of scores corresponding to the plurality of radiation delivery plan samples;

[0131] Q5, update the scoring model according to the first ranking of the plurality of radiation delivery plan samples and the plurality of scores corresponding to the plurality of radiation delivery plan samples to obtain a trained scoring model; and

[0132] Q6, determine the target radiation delivery plan according to the trained scoring model.

[0133] It should be noted that the specific examples in the embodiment can refer to the examples described in the above embodiments and optional implementation manners, which will not be described herein again.

[0134] In addition, in combination with the radiation delivery plan determination method provided in the above embodiments, a storage medium can also be provided to implement the method in the embodiment. The storage medium has a computer program stored thereon; the computer program is executed by a processor to implement any of the radiation delivery plan determination methods in the above embodiments.

[0135] It should be understood that the specific embodiments described herein are only used to explain this application, but not to limit it. According to the embodiments provided in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application.

[0136] It is apparent that the drawings depicted are only a few examples of embodiments of this application and that a person of ordinary skill in the art would be able to adapt this application to other similar situations without paying creative thought. In addition, it is understood that although the work done in the development of this application can be complex and long, certain modifications, such as design, manufacture or production, made by a person of ordinary skill in the art based on the technical content disclosed in this application should not be considered as a lack of disclosure.

[0137] The word "embodiment" in this application refers to the specific features, structures or characteristics described in conjunction with the embodiments can be included in at least one embodiment of this application. The phrase appears in various places in the specification does not necessarily mean the same embodiment, nor does it mean independence or alternative to other embodiments. It is clear 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.

[0138] The above-described embodiments only express several implementation manners of this application, which are described in detail and specifically, but should not be understood as a limitation on the scope of patent protection. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, a number of modifications and improvements can be made, which are within the scope of protection of this application. Therefore, the scope of protection of this application should be subject to the appended claims.

Claims

1. A radiation delivery plan determination method, characterized by, The method comprises: obtaining a radiotherapy plan to be evaluated and corresponding parameter information; inputting the radiotherapy plan to be evaluated and the parameter information into a scoring model to obtain a score of the radiotherapy plan to be evaluated; the scoring model is trained according to a first ranking of a plurality of radiotherapy plan samples and parameter information corresponding to the plurality of radiotherapy plan samples; the first ranking is obtained by a user ranking the plurality of radiotherapy plan samples according to the parameter information; and determining a target radiotherapy plan according to the score of the radiotherapy plan to be evaluated. The method of determining a target radiotherapy plan according to the score of the radiotherapy plan to be evaluated comprises:

2. The radiation delivery plan determination method of claim 1, wherein, when the score of the radiotherapy plan to be evaluated is above a preset value, determining the radiotherapy plan to be evaluated as the target radiotherapy plan. The parameter information comprises one or more of the following: region of interest information, index information, radiotherapy plan execution scene information and user portrait information.

3. The radiation delivery plan determination method of claim 1, wherein, The method comprises:

4. A method for determining a radiation delivery plan, characterized in that: obtaining training samples; the training samples comprise a plurality of radiotherapy plan samples and parameter information corresponding to each radiotherapy plan sample; obtaining a first ranking of the plurality of radiotherapy plan samples; the first ranking of the plurality of radiotherapy plan samples is obtained by a user ranking the plurality of radiotherapy plan samples according to the parameter information; constructing a scoring model; inputting the plurality of radiotherapy plan samples and the parameter information corresponding to each radiotherapy plan sample into the scoring model respectively to obtain a plurality of scores corresponding to the plurality of radiotherapy plan samples; and updating the scoring model according to the first ranking of the plurality of radiotherapy plan samples and the plurality of scores corresponding to the plurality of radiotherapy plan samples to obtain a trained scoring model, which is used to determine a target radiotherapy plan. The method of updating the scoring model according to the first ranking of the plurality of radiotherapy plan samples and the plurality of scores corresponding to the plurality of radiotherapy plan samples to obtain a trained scoring model comprises:

5. The radiation delivery plan determination method of claim 4, wherein, ranking the plurality of radiotherapy plan samples according to the plurality of scores corresponding to the plurality of radiotherapy plan samples to obtain a second ranking of the plurality of radiotherapy plan samples; the ranking rules of the first ranking and the second ranking are the same; and when the first ranking of the plurality of radiotherapy plan samples is inconsistent with the second ranking, updating the scoring model. The method of updating the scoring model according to the first ranking of the plurality of radiotherapy plan samples and the plurality of scores corresponding to the plurality of radiotherapy plan samples to obtain a trained scoring model comprises:

6. The radiation delivery plan determination method of claim 4, wherein, constructing a loss function according to the first ranking of the plurality of radiotherapy plan samples and the plurality of scores corresponding to the plurality of radiotherapy plan samples; and updating the scoring model according to the loss function to obtain the trained scoring model. The method of determining a target radiotherapy plan according to the trained scoring model comprises:

7. The radiation delivery plan determination method of claim 4, wherein, ​ obtain a radiotherapy delivery plan to be evaluated and corresponding parameter information; input the radiotherapy delivery plan to be evaluated and the parameter information into a trained scoring model to obtain a score of the radiotherapy delivery plan to be evaluated; determine a target radiotherapy delivery plan according to the score of the radiotherapy delivery plan to be evaluated.

8. A device for determining a radiation delivery plan, characterized in that: The apparatus comprises: a data obtaining module configured to obtain a radiotherapy delivery plan to be evaluated and corresponding parameter information; a plan scoring module configured to input the radiotherapy delivery plan to be evaluated and the parameter information into a scoring model to obtain a score of the radiotherapy delivery plan to be evaluated; the scoring model is trained according to a first ranking of a plurality of radiotherapy delivery plan samples and parameter information corresponding to the plurality of radiotherapy delivery plan samples; the first ranking is obtained by a user ranking the plurality of radiotherapy delivery plan samples according to the parameter information; and a plan determining module configured to determine a target radiotherapy delivery plan according to the score of the radiotherapy delivery plan to be evaluated.

9. A device for determining a radiation delivery plan, characterized in that: The apparatus comprises: a training sample obtaining module configured to obtain training samples; the training samples comprise a plurality of radiotherapy delivery plan samples and parameter information corresponding to each radiotherapy delivery plan sample; a ranking obtaining module configured to obtain a first ranking of the plurality of radiotherapy delivery plan samples; the first ranking of the plurality of radiotherapy delivery plan samples is obtained by a user ranking the plurality of radiotherapy delivery plan samples according to the parameter information; a constructing module configured to construct a scoring model; a training module configured to input the plurality of radiotherapy delivery plan samples and the parameter information corresponding to each radiotherapy delivery plan sample into the scoring model respectively to obtain a plurality of scores corresponding to the plurality of radiotherapy delivery plan samples; and an updating module configured to update the scoring model according to the first ranking of the plurality of radiotherapy delivery plan samples and the plurality of scores corresponding to the plurality of radiotherapy delivery plan samples to obtain a trained scoring model; the trained scoring model is used to determine a target radiotherapy delivery plan.

10. A computer-readable storage medium, characterized in that instructions stored in the memory, when executed by the processor, cause the processor to perform the method for determining a radiotherapy delivery plan according to any one of claims 1 to 7.

Images