Radiotherapy plan generation system, method and equipment based on intelligent agent and storage medium

Abstract

The invention discloses an agent-based radiotherapy plan generation system, method and device and a storage medium, and the system comprises a dosimeter agent, a TPS agent, a physicist agent and an optimization parameter consultant tool, the dosimeter agent is used for generating initial optimization parameters of a radiotherapy plan according to patient information and a treatment target, the optimization parameters are iteratively adjusted according to the feedback optimization parameter adjustment strategy; the TPS agent is used for interacting with the TPS system, performing plan optimization on the optimization parameters generated by the dosimeter agent in the TPS system, and outputting a dose result; the physicist agent is used for evaluating a dose result output by the TPS agent, identifying an aspect needing to be improved and generating feedback data; and the optimization parameter consultant tool is used for generating an optimization parameter adjustment strategy according to the feedback data of the physicist agent. The automation level, the planning quality and the efficiency of the radiotherapy planning can be effectively improved, and the flexibility and the intelligence of the planning are enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of radiotherapy, and particularly relates to an intelligent agent-based radiotherapy plan generation system, method, device, and storage medium. Background Art

[0002] Radiation treatment planning is a crucial part of cancer treatment, and its goal is to effectively control tumors while maximizing the protection of surrounding normal tissues (organs at risk, OAR). Traditional radiation treatment planning mainly relies on manual operation, and dosimetrists perform iterative adjustments in the treatment planning system (TPS), which is a resource-intensive and highly experience-dependent process. The quality and efficiency of manual planning are affected by the personal experience and workflow of dosimetrists, and it is difficult to ensure the standardization and consistency of the plan.

[0003] To improve the efficiency and quality of radiation treatment planning, automated planning methods have emerged. Early automated planning methods mainly relied on protocol-based automatic iterative optimization (PB-AIO), which performed iterative optimization through predefined clinical target templates, but their flexibility and adaptability to different patient anatomies were limited. Subsequently, the knowledge-based planning (KBP) method used historical patient data to guide the planning process, but it relied on a pre-constructed knowledge base, and feature engineering required professional knowledge. In recent years, the emergence of deep learning (DL) models has brought revolutionary progress to automated planning. Deep supervised learning (DSL) models can directly predict dose distributions from patient anatomies, but their performance is limited by the quality and scope of the training data. Deep reinforcement learning (DRL) methods learn and optimize through the interaction between the intelligent agent and the TPS environment, but their training process requires a large number of TPS interactions and lacks the domain knowledge and reasoning ability of human experts.

[0004] Although existing automated planning methods have made certain progress, there are still limitations in terms of planning quality, efficiency, and flexibility. Especially in complex cases and scenarios that require fine optimization, the performance of automated planning still needs to be improved. Therefore, there is an urgent need for a more intelligent, efficient, and flexible automated radiation treatment planning method. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes an intelligent agent-based radiotherapy plan generation system, method, device, and storage medium.

[0006] To achieve the above object, the technical solution of the present invention is as follows:

[0007] In the first aspect, the present invention discloses an intelligent agent-based radiotherapy plan generation system, including:

[0008] A dosimetrist agent, which is used to generate initial optimization parameters for a radiotherapy plan according to patient information and treatment goals, and adjust strategies based on the feedback optimization parameters to iteratively adjust the optimization parameters;

[0009] A TPS agent, which is used to interact with the TPS system, execute plan optimization with the optimization parameters generated by the dosimetrist agent in the TPS system, and output dose results;

[0010] A physicist agent, which is used to evaluate the dose results output by the TPS agent, identify aspects that need improvement, and generate feedback data;

[0011] An optimization parameter advisor tool, which is used to generate an optimization parameter adjustment strategy according to the feedback data of the physicist agent.

[0012] Based on the above technical solutions, the following improvements can also be made:

[0013] As a preferred solution, the radiotherapy plan generation system further includes:

[0014] A retrieval tool, which is used to retrieve similar historical plans from the historical plan library according to patient information;

[0015] And the dosimetrist agent can generate initial optimization parameters for the radiotherapy plan according to patient information, treatment goals, and the retrieved similar historical plans.

[0016] As a preferred solution, the radiotherapy plan generation system further includes:

[0017] An optimization trajectory analysis tool, which is used to analyze the optimization trajectory data generated during the iterative optimization process of the plan, identify effective optimization strategies and patterns, and generate an optimization trajectory adjustment strategy;

[0018] And the dosimetrist agent can iteratively adjust the optimization parameters according to the feedback optimization trajectory adjustment strategy.

[0019] As a preferred solution, the radiotherapy plan generation system further includes:

[0020] A human-computer interaction agent, which is used to implement human-computer interaction, transmit the optimization guidance strategy provided by human experts to the dosimetrist agent, and / or transmit the optimization trajectory adjustment strategy generated by the optimization trajectory analysis tool to human experts for selection.

[0021] As a preferred solution, the dosimetrist agent, the physicist agent, the optimization parameter advisor tool, and the optimization trajectory analysis tool all adopt large language models, and the dosimetrist agent and the physicist agent both have a self-reflection mechanism and can detect and correct the errors in their outputs.

[0022] In a second aspect, the present invention discloses an agent-based radiotherapy plan generation method, including:

[0023] Step S1: Obtain patient information and treatment goals;

[0024] Step S3: The dosimetrist agent generates initial optimization parameters for the radiotherapy plan based on the patient information and treatment goals;

[0025] Step S4: The TPS agent receives the optimization parameters generated by the dosimetrist agent, performs plan optimization in the TPS system with the optimization parameters, and outputs a dose result;

[0026] Step S5: The physicist agent analyzes the dose result output by the TPS agent, evaluates it according to clinical dosimetry standards, identifies aspects that need improvement, and generates feedback data;

[0027] Step S6: The optimization parameter advisor tool generates an optimization parameter adjustment strategy based on the feedback data from the physicist agent;

[0028] Step S7: The dosimetrist agent adjusts the optimization parameters according to the optimization parameter adjustment strategy;

[0029] Step S8: Repeat Step S4 - Step S7 until the physicist agent determines that the dose result output by the TPS agent meets the requirements or reaches the preset maximum number of iterations;

[0030] Step S9: Generate a radiotherapy plan according to the final optimization parameters.

[0031] As a preferred solution, the radiotherapy plan generation method further includes:

[0032] Step S2: The retrieval tool retrieves similar historical plans from the historical plan library according to the patient information;

[0033] And Step S3 is further that the dosimetrist agent generates initial optimization parameters for the radiotherapy plan based on the patient information, treatment goals, and the retrieved similar historical plans.

[0034] As a preferred solution, during the iterative optimization process of the plan, an optimization trajectory analysis tool is used to analyze the generated optimization trajectory data during the process, identify effective optimization strategies and patterns, and generate an optimization trajectory adjustment strategy;

[0035] And the dosimetrist agent can iteratively adjust the optimization parameters according to the feedback optimization trajectory adjustment strategy.

[0036] In a third aspect, the present invention discloses a computing device, including:

[0037] One or more processors;

[0038] A memory;

[0039] and one or more programs, where the one or more programs are stored in a memory and configured to be executed by one or more processors, and the one or more programs include instructions for any of the above agent-based radiotherapy plan generation methods.

[0040] In a fourth aspect, the present invention discloses a storage medium storing one or more computer-readable programs, where the one or more programs include instructions adapted to be loaded and executed by a memory for any of the above agent-based radiotherapy plan generation methods.

[0041] The present invention overcomes the problems in the prior art such as low automation degree of radiotherapy plans, limited plan quality and efficiency, etc., and discloses an agent-based radiotherapy plan generation system, method, device and storage medium, which can effectively improve the automation level, plan quality and efficiency of radiotherapy plans, and enhance the flexibility and intelligence of the plans.

[0042] Specifically, the present invention has the following beneficial effects:

[0043] First, the present invention can effectively improve the plan quality. The high-quality radiotherapy plans generated by the present invention are superior to the existing automatic plan methods in terms of target coverage and dose uniformity, and are equivalent to or even better than manual plans in terms of protecting critical organs.

[0044] Second, the present invention can effectively improve the plan efficiency. The present invention can significantly reduce the number of plan iterations, shorten the plan time, improve the plan efficiency, and is equivalent to senior physicists in terms of plan efficiency.

[0045] Third, the present invention can enhance the plan flexibility. Based on a large language model, the present invention has a powerful knowledge base and reasoning ability, can adapt to different clinical scenarios and patient individual differences, and has good flexibility.

[0046] Fourth, the present invention can achieve intelligent automation. The present invention simulates the collaborative workflow of human dosimetrists and physicists, uses large language model agents for intelligent decision-making and iterative optimization, and realizes the intelligent automation of radiotherapy plans.

[0047] Fifth, the present invention can reduce the dependence on manual work. The present invention can significantly reduce the dependence on manual operations, reduce human errors, and improve the standardization and consistency of plans. Description of the Drawings

[0048] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the embodiments. It should be understood that the following drawings only show certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other relevant drawings can also be obtained based on these drawings.

[0049] Figure 1 It is a block diagram of a radiotherapy plan generation system provided by an embodiment of the present invention.

[0050] Figure 2 It is a schematic diagram of the prompting strategy of a dosimetrist agent provided by an embodiment of the present invention.

[0051] Figure 3 It is a comparison result diagram of the present method (GPT-Plan) and an automatic planning method (ECHO) in lung patients provided by an embodiment of the present invention.

[0052] Figure 4 It is a comparison diagram of the DVHs (dose volume histograms) generated by the present method (GPT-Plan) and an automatic planning method (ECHO) provided by an embodiment of the present invention.

[0053] Figure 5 It is a comparison diagram of the DVHs (dose volume histograms) generated by the present method (GPT-Plan) and a human expert provided by an embodiment of the present invention.

[0054] Wherein: 201 - retrieval tool, 202 - dosimetrist agent, 203 - TPS agent, 204 - physicist agent, 205 - optimization parameter advisor tool, 206 - optimization trajectory analysis tool, 207 - human-computer interaction agent. Detailed implementation manners

[0055] The following will detail the preferred implementation manners of the present invention with reference to the accompanying drawings.

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope protected by the present invention.

[0057] The expression of "including" an element is an "open" expression, which only means that there are corresponding components or steps and should not be construed as excluding additional components or steps.

[0058] To achieve the object of the present invention, in some embodiments of an agent-based radiotherapy plan generation system, such as Figure 1 shown, the radiotherapy plan generation system includes:

[0059] A retrieval tool 201, configured to retrieve similar historical plans from a historical plan library according to patient information (such as patient anatomical structure information, etc.);

[0060] A dosimetrist agent 202, configured to generate initial optimization parameters of a radiotherapy plan according to patient information, treatment objectives, and the retrieved similar historical plans, and iteratively adjust the optimization parameters according to the feedback optimization parameter adjustment strategy;

[0061] A TPS agent 203, configured to interact with a TPS system, execute plan optimization in the TPS system with the optimization parameters generated by the dosimetrist agent, and output a dose result;

[0062] A physicist agent 204, configured to evaluate the dose result output by the TPS agent, identify aspects that need improvement, and generate feedback data;

[0063] An optimization parameter advisor tool 205, configured to generate an optimization parameter adjustment strategy according to the feedback data of the physicist agent.

[0064] It should be noted that the above-mentioned dosimetrist agent 202, physicist agent 204, and optimization parameter advisor tool 205 all adopt a large language model (LLM).

[0065] Furthermore, both the dosimetrist agent 202 and the physicist agent 204 have a self-reflection mechanism, capable of detecting and correcting errors in their outputs.

[0066] To further optimize the implementation effect of the present invention, in some other embodiments of the radiotherapy plan generation system, the remaining characteristic technologies are the same, except that the radiotherapy plan generation system further includes:

[0067] An optimization trajectory analysis tool 206, configured to analyze the optimization trajectory data generated during the iterative optimization process of the plan, identify effective optimization strategies and patterns, and generate an optimization trajectory adjustment strategy;

[0068] And the dosimetrist agent 202 can iteratively adjust the optimization parameters according to the feedback optimization trajectory adjustment strategy.

[0069] Furthermore, the optimization trajectory analysis tool 206 also adopts a large language model (LLM).

[0070] To further optimize the implementation effect of the present invention, in some other embodiments of the radiotherapy plan generation system, the remaining characteristic technologies are the same, except that the radiotherapy plan generation system further includes:

[0071] The human-computer interaction agent 207 is used to implement human-computer interaction, transmit the optimization guidance strategy provided by human experts to the dosimetrist agent, and / or transmit the optimization trajectory adjustment strategy generated by the optimization trajectory analysis tool 206 to human experts for selection.

[0072] In this embodiment, a human-computer interaction agent 207 is added to allow human experts to intervene in the planning process and provide guidance or select the final plan.

[0073] Each of the above agents will be elaborated in detail below.

[0074] For the retrieval tool 201. The present invention uses the retrieval tool 201 to retrieve historical plans similar to the current patient using the similarity of patient information, providing a reference for the dosimetrist agent to accelerate the plan optimization process.

[0075] Furthermore, the retrieval tool 201 may not be limited to using principal component analysis to quantify the similarity of patient information (such as anatomical structures).

[0076] For the dosimetrist agent 202. In the radiotherapy plan generation system of the present invention, the dosimetrist agent 202 is the core component of the present invention, responsible for generating and optimizing the optimization parameters (OPs) of the treatment plan. The dosimetrist agent 202 makes decisions based on patient information, treatment goals, retrieved similar historical plans, feedback provided by the physicist agent 204, and guidance provided by the human-computer interaction agent 207.

[0077] The patient information is the patient's anatomical structure information, including but not limited to the patient's CT images, structure segmentation and other information. The treatment goals include clinical goals such as prescription dose, dose limits for organs at risk.

[0078] The dosimetrist agent 202 adopts a structured prompting strategy, such as Figure 2 shown, including System Prompt and User Prompt.

[0079] Among them, the System Prompt defines the role, task and output format of the dosimetrist agent; the User Prompt contains patient-specific information and iterative information.

[0080] The dosimetrist agent 202 uses a large language model (LLM) to generate optimization parameters and verifies and corrects them through a self-reflection mechanism.

[0081] For the TPS agent 203. The TPS agent 203 is the interface between the system and the TPS system (i.e., the treatment planning system), responsible for receiving the optimization parameters provided by the dosimetrist agent 202, performing plan optimization calculations in the TPS system, and extracting dose results. The TPS agent 203 controls and interacts with data of the TPS system through the application programming interface of the TPS system.

[0082] For the physicist agent 204. The physicist agent 204 can simulate the role of a senior physicist, responsible for evaluating the dose results generated by the TPS agent and providing feedback to the dosimetrist agent. The physicist agent evaluates the plan quality according to clinical dosimetry criteria (such as: dose volume histogram (DVH) metrics, target coverage, organ-at-risk dose, etc.), and identifies aspects that need improvement. The evaluation results of the physicist agent 204 will affect the subsequent optimization parameter adjustment direction of the dosimetrist agent 202.

[0083] For the optimization parameter advisor tool 205. The optimization parameter advisor tool 205 recommends feasible optimization parameter adjustment strategies based on the feedback provided by the physicist agent 204 and combines the reasoning ability of the large language model. The optimization parameter advisor tool 205 can provide multiple adjustment strategies to balance target coverage and organ-at-risk protection, such as adjusting the objective function weight, dose constraints, etc.

[0084] For the optimization trajectory analysis tool 206. The optimization trajectory analysis tool 206 analyzes the optimization trajectory data generated during the system's iterative optimization process, including optimization parameters, dose results, evaluation feedback, etc. The optimization trajectory analysis tool 206 utilizes the analysis ability of the large language model to identify effective optimization strategies and patterns, and provides optimization suggestions for the dosimetrist agent 202 to guide the subsequent optimization direction.

[0085] For the human-computer interaction agent 207. The human-computer interaction agent 207 is an optional component of this system, allowing human experts to intervene in the automated planning process. During the plan optimization process, human experts can provide guidance to the dosimetrist agent 202 through the human-computer interaction agent 207, such as specifying the optimization direction, adjusting the optimization strategy, etc. After the plan optimization is completed, human experts can also review the candidate plans in the optimization trajectory adjustment strategy identified by the optimization trajectory analysis tool 206 through the human-computer interaction agent 207 and finally select the final plan.

[0086] The present invention adopts a self-reflection mechanism to detect and correct possible errors or inconsistent outputs of the large language model agent. The self-reflection mechanism includes two stages:

[0087] 1) Rule verification: Based on predefined expert rules, verify whether the optimization parameters generated by the dosimetrist agent 202 comply with specifications and logic, such as parameter format, value range, structure name, etc.

[0088] 2) Mirror LLM Review: Use an independent mirror large language model agent to conduct a strategic review of the optimization strategy of the dosimetrist agent 202, and evaluate the overall rationality, goal consistency, priority order, etc. of the plan.

[0089] In some other embodiments, the present invention discloses an agent-based radiotherapy plan generation method, including:

[0090] Step S101: Obtain patient information (such as patient anatomical structure information, etc.) and treatment goals;

[0091] Step S102: The retrieval tool 201 retrieves similar historical plans from the historical plan library according to the patient information.

[0092] Step S103: The dosimetrist agent 202 generates initial optimization parameters for the radiotherapy plan according to the patient information, treatment goals, and the retrieved similar historical plans.

[0093] Step S104: The TPS agent 203 receives the optimization parameters generated by the dosimetrist agent 202, and executes plan optimization in the TPS system with the optimization parameters, and outputs a dose result.

[0094] Step S105: The physicist agent 204 analyzes the dose result output by the TPS agent 203, and evaluates according to the clinical dosimetry criteria, identifies aspects that need improvement, and generates feedback data.

[0095] Step S106: The optimization parameter advisor tool 205 generates an optimization parameter adjustment strategy according to the feedback data of the physicist agent 204.

[0096] Step S107: The dosimetrist agent 202 adjusts the optimization parameters according to the optimization parameter adjustment strategy.

[0097] Step S108: Repeat Step S104 - Step S107 until the physicist agent 204 determines that the dose result output by the TPS agent 203 meets the requirements, or reaches a preset maximum number of iterations.

[0098] Step S109: Generate a radiotherapy plan according to the final optimization parameters.

[0099] In order to further optimize the implementation effect of the present invention, in some other implementation manners of the radiotherapy plan generation method, the remaining characteristic technologies are the same, and the difference is that during the plan iterative optimization process, an optimization trajectory analysis tool 206 is used to analyze the generated optimization trajectory data during the process, identify effective optimization strategies and patterns, and generate an optimization trajectory adjustment strategy.

[0100] Moreover, the dosimetrist agent 202 can adjust the strategy according to the feedback optimization trajectory and iteratively adjust the optimization parameters.

[0101] To further optimize the implementation effect of the present invention, in some other embodiments of the radiotherapy plan generation method, the remaining characteristic technologies are the same, except that in the process of plan iterative optimization, a human-computer interaction agent 207 is used to implement human-computer interaction, and the optimization guidance strategy provided by a human expert is transmitted to the dosimetrist agent, and / or the optimization trajectory adjustment strategy generated by the optimization trajectory analysis tool 206 is transmitted to the human expert for selection.

[0102] To verify the effectiveness of the present invention, experimental verification was carried out on lung cancer and cervical cancer cases, and the GPT-4o model was selected as the basic LLM.

[0103] The present invention has been experimentally verified on lung cancer and cervical cancer cases, and the experimental results show that the system can generate high-quality radiotherapy plans and has high planning efficiency and flexibility.

[0104] Specifically, the comparison results of the method of the present invention (i.e., Figure 3 GPT-Plan in Figure 3 and the automatic planning method (ECHO) in lung patients are as

[0105] shown, and the numbers in the figure represent the average values of the dosimetric indices. Figure 3 As

[0106] shown, both the method of the present invention and the automatic planning method have generated clinically acceptable plans. However, the present invention demonstrates superior target coverage, achieving significantly higher D95 (an increase of 4.75%, p < 0.05), while maintaining comparable or better OAR protection. The GPT plan also achieved significantly better dose uniformity within the target, as evidenced by a lower HI (a reduction of 49.52%, p < 0.05). Notably, GPT-Plan significantly reduced the average dose (a reduction of 14.31%, p < 0.05) and the impact on healthy lung tissue of V5.0 (a reduction of 17.60%, p < 0.05), indicating the potential to reduce lung toxicity.

[0106] Furthermore, for a representative lung cancer case, the comparison of the DVHs (dose-volume histograms) generated by GPT-Plan and ECHO is as Figure 4 shown. GPT-Plan achieved superior target coverage compared to ECHO and comparable or better results in protecting the OAR (organ at risk region).

[0107] Furthermore, for a representative cervical cancer case, the comparison of the DVHs generated by GPT-Plan and those generated by human experts is as Figure 5As shown, GPT-Plan demonstrates a plan quality comparable to that of senior physicists and better protects the OAR than junior physicists.

[0108] In addition, in some other embodiments, the present invention also discloses a computing device, including:

[0109] One or more processors;

[0110] A memory;

[0111] And one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for the method for determining the dominant vertices of the alternating group network disclosed in the above embodiments.

[0112] The processor may include one or more processing cores, such as: a 4-core processor, an 8-core processor, etc. The processor may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor may also include a main processor and a coprocessor. The main processor is a processor for processing data in the wake state, also known as the CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor may be integrated with a GPU (Graphics Processing Unit), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor may also include an AI (Artificial Intelligence) processor, and the AI processor is used to process computational operations related to machine learning.

[0113] The memory may include one or more computer-readable storage media, and the computer-readable storage media may be non-transitory. The memory may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices and flash storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory is used to store at least one instruction, and the at least one instruction is used to be executed by the processor to implement the method for determining the dominant vertices of the alternating group network provided in the method embodiments of the present invention.

[0114] In addition, the computing device may optionally further include: a peripheral device interface and at least one peripheral device. The processor, the memory, and the peripheral device interface may be connected through a bus or signal lines. Each peripheral device may be connected to the peripheral device interface through a bus, signal lines, or a circuit board. Schematically, the peripheral devices include, but are not limited to: a radio frequency circuit, a touch display screen, an audio circuit, and a power supply, etc.

[0115] The present invention overcomes the problems in the prior art such as low automation level of radiotherapy planning, limited plan quality and efficiency, etc., and discloses a radiotherapy plan generation system, method, device, and storage medium based on an agent, which can effectively improve the automation level, plan quality, and efficiency of radiotherapy planning, and enhance the flexibility and intelligence of the plan.

[0116] Specifically, the present invention has the following beneficial effects:

[0117] First, the present invention can effectively improve the plan quality. The high-quality radiotherapy plan generated by the present invention is superior to the existing automatic plan methods in terms of target coverage rate and dose uniformity, and is equivalent to or even better than the manual plan in terms of protecting critical organs.

[0118] Second, the present invention can effectively improve the plan efficiency. The present invention can significantly reduce the number of plan iterations, shorten the plan time, improve the plan efficiency, and is equivalent to that of a senior physicist in terms of plan efficiency.

[0119] Third, the present invention can enhance the plan flexibility. Based on a large language model, the present invention has a powerful knowledge base and reasoning ability, can adapt to different clinical scenarios and patient individual differences, and has good flexibility.

[0120] Fourth, the present invention can achieve intelligent automation. The present invention simulates the collaborative work process of human dosimetrists and physicists, uses a large language model agent for intelligent decision-making and iterative optimization, and realizes the intelligent automation of radiotherapy planning.

[0121] Fifth, the present invention can reduce the dependence on manual work. The present invention can significantly reduce the dependence on manual operations, reduce human errors, and improve the standardization and consistency of the plan.

[0122] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of the present invention claimed is defined by the appended claims and their equivalents.

Claims

1. Agent-based radiotherapy plan generation system, comprising: Dosimetrist agent, configured to generate initial optimization parameters of a radiotherapy plan according to patient information and treatment objectives, and adjust the optimization parameters iteratively according to the feedback optimization parameter adjustment strategy; TPS agent, configured to interact with the TPS system, execute plan optimization with the optimization parameters generated by the dosimetrist agent in the TPS system, and output dose results; Physicist agent, configured to evaluate the dose results output by the TPS agent, identify aspects that need improvement, and generate feedback data; Optimization parameter advisor tool, configured to generate an optimization parameter adjustment strategy according to the feedback data of the physicist agent.

2. The radiotherapy plan generation method according to claim 1, wherein The radiotherapy plan generation system further comprises: Retrieval tool, configured to retrieve similar historical plans from the historical plan library according to patient information; And the dosimetrist agent is capable of generating initial optimization parameters of a radiotherapy plan according to patient information, treatment objectives, and the retrieved similar historical plans.

3. The radiotherapy plan generation method according to claim 1, wherein The radiotherapy plan generation system further comprises: Optimization trajectory analysis tool, configured to analyze the optimization trajectory data generated during the iterative optimization of the plan, identify effective optimization strategies and patterns, and generate an optimization trajectory adjustment strategy; And the dosimetrist agent is capable of iteratively adjusting the optimization parameters according to the feedback optimization trajectory adjustment strategy.

4. The radiotherapy plan generation system according to claim 3, wherein The radiotherapy plan generation system further comprises: Human-computer interaction agent, configured to implement human-computer interaction, transmit the optimization guidance strategy provided by a human expert to the dosimetrist agent, and / or transmit the optimization trajectory adjustment strategy generated by the optimization trajectory analysis tool to the human expert for selection.

5. The radiotherapy plan generation system according to claim 4, characterized in that, The dosimetrist agent, physicist agent, optimization parameter advisor tool, and optimization trajectory analysis tool all adopt large language models, and the dosimetrist agent and physicist agent both have a self-reflection mechanism and are capable of detecting and correcting errors in their outputs.

6. Agent-based radiotherapy plan generation method, comprising: Step S1: Obtain patient information and treatment objectives; Step S3: The dosimetrist agent generates initial optimization parameters of a radiotherapy plan according to patient information and treatment objectives; Step S4: The TPS agent receives the optimization parameters generated by the dosimetrist agent, executes plan optimization with the optimization parameters in the TPS system, and outputs dose results; Step S5: The physicist agent analyzes the dose results output by the TPS agent, evaluates according to clinical dosimetry standards, identifies aspects that need improvement, and generates feedback data; Step S6: The optimization parameter advisor tool generates an optimization parameter adjustment strategy according to the feedback data of the physicist agent; Step S7: The dosimetrist agent adjusts the optimization parameters according to the optimization parameter adjustment strategy; Step S8: Repeat Step S4 - Step S7 until the physicist agent determines that the dose results output by the TPS agent meet the requirements or reach a preset maximum number of iterations; Step S9: Generate a radiotherapy plan according to the final optimization parameters.

7. The radiotherapy plan generation method according to claim 6, wherein The radiotherapy plan generation method further comprises: Step S2: The retrieval tool retrieves similar historical plans from the historical plan library according to patient information; And the step S3 further includes that the dosimetrist agent generates initial optimization parameters of the radiotherapy plan according to the patient information, treatment objectives, and retrieved similar historical plans.

8. The radiotherapy plan generation method according to claim 6 or 7, characterized in that, During the iterative optimization process of the plan, an optimization trajectory analysis tool is used to analyze the optimization trajectory data generated during the process, identify effective optimization strategies and patterns, and generate an optimization trajectory adjustment strategy. And the dosimetrist agent can iteratively adjust the optimization parameters according to the feedback optimization trajectory adjustment strategy.

9. A computing device, characterized in that, Including: One or more processors; A memory; And one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, and the one or more programs include instructions of the agent-based radiotherapy plan generation method according to any one of claims 6-8 above.

10. A storage medium, characterized in that, The storage medium stores one or more computer-readable programs, and the one or more programs include instructions, and the instructions are adapted to be loaded and executed by the memory to perform the agent-based radiotherapy plan generation method according to any one of claims 6-8 above.

Images