Radiotherapy intelligent multi-target automatic planning method and system

Abstract

The application discloses a kind of radiotherapy intelligent multi-target area automatic planning method and system, wherein, method includes: based on target lesion division radiotherapy target area, set the number of beams of each target area, predict the beam direction of each target area by angle prediction model;Optimize each target area according to priority order, and the dose distribution of previous target area is included in current target area optimization objective function;For the dose distribution of each target area optimization, dose superposition is carried out, and the superposition dose is evaluated;After evaluation, start radiotherapy, in the treatment gap between target area, through new CT image, evaluate dose distribution deviation, dynamically adjust subsequent target area plan.The application of a kind of intelligent multi-target area automatic planning method and system, by using a sequential optimization plan design and considering cumulative dose influence and positioning deviation influence simultaneously, timely adjust plan in the treatment process, greatly improve the treatment effect, provide a new solution for multi-target area plan design.

Description

Technical Field

[0001] This invention relates to the field of radiotherapy planning technology, specifically to an intelligent multi-target area automatic planning method and system for radiotherapy. Background Technology

[0002] In the field of radiotherapy planning, designing efficient and precise treatment plans for complex cases with multiple independent target volumes (e.g., primary lesions and metastases) has always been a challenge. Currently, the industry's standard solutions for multi-target volume planning mainly fall into two categories: single-plan coverage of all target volumes and zoned planning.

[0003] The single-plan coverage of all target areas approach refers to setting up irradiation fields that simultaneously cover all target areas within a single treatment plan, and using an optimization algorithm to perform a comprehensive optimization of the dose requirements for all target areas and organs at risk in one go. The advantage of this approach is that treatment is completed in one session, resulting in high efficiency. However, because its optimizer needs to simultaneously balance dose targets for multiple target areas and protection constraints for multiple organs at risk, the optimization space becomes complex, computationally burdensome, and it is often difficult to simultaneously meet all clinical requirements. Especially when target areas are dispersed or have complex relationships with organs at risk, it is easy to sacrifice the protection of organs at risk in order to meet the dose requirements of some target areas, or to result in insufficient doses for some target areas. In other words, it suffers from the inherent disadvantages of complex optimizer trade-offs and high pressure on organ at risk protection.

[0004] To overcome the shortcomings of single-plan treatments, partitioned treatment plans were developed. This approach breaks down multi-target volume therapy into multiple sub-plans, with each treatment targeting and optimizing only one or a portion of the target volume. This simplifies the optimization of individual plans, helping to better control the dose to each target volume and protect adjacent organs at risk. However, this approach also has significant drawbacks: because treatment needs to be performed multiple times, errors inevitably occur in patient positioning and internal anatomical structures (such as organ movement and deformation) during each treatment. These errors cause deviations in the dose distribution calculated from different sub-plans when actually superimposed, resulting in dose cold spots (insufficient dose) or hot spots (excessive dose) at target volume boundaries or overlapping areas, severely impacting the accuracy and safety of treatment.

[0005] In existing technologies, neither the overall optimized single-plan scheme nor the step-by-step regional planning scheme has been able to effectively solve the balance problem between optimization efficiency, dose accuracy and robustness to positioning errors in multi-target radiotherapy planning.

[0006] Therefore, there is an urgent need in this field for a novel intelligent multi-target area automatic planning method and system that can balance optimized efficiency, accuracy of final superimposed dose, and effective response to positioning errors during treatment.

[0007] In view of this, this invention patent is hereby proposed. Summary of the Invention

[0008] To address the aforementioned problems, this invention provides an intelligent multi-target volume automatic planning method and system for radiotherapy, specifically employing the following technical solution:

[0009] A method for intelligent multi-target automated radiotherapy planning includes:

[0010] The target area for radiotherapy is divided based on the target lesion, the number of beams in each target area is set, and the beam direction of each target area is predicted by the angle prediction model.

[0011] Each target region is optimized in order of priority, and the cumulative dose distribution of the previous target region is incorporated into the objective function of the current target region optimization.

[0012] Dose stacking is performed on the optimized dose distribution for each target region, and the stacked dose is evaluated.

[0013] After the assessment is passed, radiotherapy begins. During the treatment intervals within the target area, dose distribution deviations are assessed using new CT images, and subsequent target area plans are dynamically adjusted.

[0014] As an optional embodiment of the present invention, in a radiotherapy intelligent multi-target automatic planning method of the present invention, the input of the angle prediction model is the target area and organs at risk delineation, target area priority, vertical direction of the line connecting the centers of adjacent target areas, and the number of planned beams.

[0015] The output of the angle prediction model is the beam angle for each planned beam.

[0016] The loss function of the angle prediction model is the predicted angle θ. pred With actual angle θ true The loss includes the bias and the angle between the predicted angle direction and the perpendicular direction of the line connecting the centers of adjacent target areas. The loss function is:

[0017] L total =λ1L gt +λ2L line ;

[0018]

[0019] L line =(cos(δ) line )) α α>0;

[0020]

[0021] Where λ1 and λ2 are adjustable weights, Δ(a, b) represents the angle between angles a and b, and α is Lline The adjustment factor of the loss function; the larger α is, the larger the value of the loss function. L is the angle between the line connecting the centers of the target area and the x-axis; gt The loss function ensures that the predicted angle is as close as possible to the actual angle, L line The loss function ensures that the overlapping area between beams from different target regions is minimized.

[0022] As an optional embodiment of the present invention, in a radiotherapy intelligent multi-target volume automatic planning method of the present invention, the optimization of each target volume according to priority includes:

[0023] Prioritize each target region based on the total prescription dose for each target region.

[0024] The lower the total prescription dose in the target area, the higher the priority of target area optimization; the higher the total prescription dose in the target area, the lower the priority of target area optimization.

[0025] As an optional embodiment of the present invention, in a radiotherapy intelligent multi-target volume automatic planning method of the present invention, the step of incorporating the cumulative dose distribution of the previous target volume into the current target volume optimization objective function includes:

[0026] After the current primary target area plan is optimized, the accumulated dose distribution is included as a fixed dose in the objective function of the current primary target area plan. The objective function includes:

[0027]

[0028] F k (ω)=min ω max f f k (ω j );

[0029] Among them, f k (ω j ) represents the objective function of the biased or unbiased scenario in the k-th target area plan, ω represents the intensity map, j represents the j-th scenario, and F k (ω) represents the objective function for robust optimization of the k-th target area plan, D i D represents organ point dose. s The summation represents the cumulative dose at organ points corresponding to the planned dose distribution of the previous target area, D. p The H function represents the prescribed dose or limit for the target area and organs at risk.

[0030] As an optional embodiment of the present invention, in a radiotherapy intelligent multi-target volume automatic planning method of the present invention, the constraint function H satisfies: if the constraint requires <D p ,but When the constraint function H is applied, the value of the constraint function H is 1; otherwise, it is 0.

[0031] As an optional embodiment of the present invention, in a radiotherapy intelligent multi-target volume automatic planning method of the present invention, the optimization of each target volume according to priority includes:

[0032] After the current level target area plan optimization is completed, the system automatically saves the plan information and switches the prescription, beam and constraint information. At the same time, the accumulated dose distribution is put into the next level target area plan optimization as a fixed dose to continue optimization until all target area plans are optimized.

[0033] As an optional embodiment of the present invention, in a smart multi-target automatic radiotherapy planning method of the present invention, the dose superposition of the dose distribution optimized for each target region and the evaluation of the superposition dose include:

[0034] Assess whether each target area meets the prescription coverage requirements, assess whether the cumulative dose to each organ at risk exceeds the clinical limit, and identify any dose hotspots or colds.

[0035] If the assessment requirements are met, the treatment plan will be used for subsequent validation; otherwise, the constraints of each target area will be readjusted.

[0036] As an optional embodiment of the present invention, in a smart multi-target volume automatic planning method for radiotherapy, the step of assessing dose distribution deviation through new CT images and dynamically adjusting subsequent target volume planning during the treatment interval between target volumes includes:

[0037] Radiotherapy is administered to each target area in order of priority.

[0038] Before the irradiation plan for the current target area begins, CT images are reacquired each time the irradiation plan for the previous target area ends. The original plan for the current target area is then transferred to the newly acquired CT images for dose calculation, and the dose distribution deviation is evaluated to determine whether it exceeds the preset threshold.

[0039] If the preset threshold is not exceeded, the existing original irradiation plan will continue to be used for the current target area; otherwise, the current and subsequent target area plans will be re-optimized based on the newly acquired CT images.

[0040] As an optional embodiment of the present invention, in a radiotherapy intelligent multi-target automatic planning method of the present invention, the step of migrating the original plan of the current target area to the currently reacquired CT image for dose calculation includes: registering the original CT image and the currently reacquired CT image, copying the current target area plan data, including plan information and beam information, to the currently reacquired CT image, performing dose calculation, and obtaining the dose distribution of the current target area on the currently reacquired CT image;

[0041] The assessment of whether the dose distribution deviation exceeds the preset threshold includes: using a dose difference tool to calculate the difference between the dose distribution on the original CT image and the currently reacquired CT image, and checking whether the global maximum deviation dose is less than the preset threshold. If it is less than the preset threshold, the current plan does not need to be re-optimized; otherwise, the current plan and subsequent plans will be re-optimized and adjusted on the currently reacquired CT image.

[0042] The process of re-optimizing the current and subsequent target area plans based on the newly acquired CT images includes: using the original plan optimization results as the initial values ​​for re-optimization, and the accumulated irradiation dose as the fixed dose, fine-tuning the optimization results, and quickly optimizing a new plan.

[0043] This invention also provides an intelligent multi-target volume automatic planning system for radiotherapy, comprising:

[0044] The intelligent field setting module divides the target area based on the target lesion, sets the number of beams for each target area, and predicts the beam direction of each target area through the angle prediction model.

[0045] The iterative optimization module optimizes each target region in order of priority and incorporates the cumulative dose distribution of the previous target region into the current target region's optimization objective function.

[0046] The dose assessment module performs dose superposition for the optimized dose distribution of each target area and evaluates the superposition dose.

[0047] The dynamic adjustment module initiates radiotherapy after successful evaluation. During treatment intervals within the target area, dose distribution deviations are assessed using new CT images, and subsequent target planning is dynamically adjusted. This invention provides an intelligent multi-target automatic planning method for radiotherapy, which, compared to existing technologies, achieves the following significant technical advantages:

[0048] The overall quality and efficiency of multi-target zone planning have been improved: By combining intelligent field setting (angle prediction model) with sequential optimization that considers cumulative dose, automated and intelligent design of multi-target zone planning has been achieved. This method not only avoids the complex trade-offs encountered in global optimization of a single plan, but also fundamentally avoids the dose hotspot and colds problems that may occur in zone planning due to inaccurate dose superposition through precise sequential dose control. Thus, while ensuring plan quality, the efficiency of plan design has been significantly improved.

[0049] Global optimization of the radiation field direction was achieved: the angle prediction model not only aimed at prediction accuracy, but also innovatively introduced a loss function designed to reduce the overlap of radiation fields in different target areas. This makes the radiation field direction output by the model not only the optimal solution for a single target area, but also consider the interaction between multiple target areas from a global perspective, thereby generating a better and more reasonable radiation field layout for the overall irradiation scheme, laying a solid foundation for obtaining high-quality dose distribution.

[0050] This ensures the accuracy and reliability of the final superimposed dose: the sequential optimization process incorporates the cumulative dose generated in previous target regions as a fixed quantity into the optimization objective function of the current target region, ensuring that the optimization of each sub-plan is performed under a known and fixed background dose. This "incremental" optimization method can more realistically simulate the final actual dose distribution, keeping the optimization objective consistent with the final clinical evaluation objective (i.e., the total dose after superimposing doses from all target regions), thereby greatly improving the accuracy and clinical feasibility of the final dose distribution.

[0051] This method enhances the robustness and adaptability of the treatment process: by introducing a dynamic adjustment mechanism for treatment intervals based on new CT images, it effectively addresses dose deviations caused by patient positioning errors or changes in anatomical structures. By assessing dose distribution deviations and determining whether re-optimization is necessary, online validation and adaptive adjustment of the treatment plan are achieved. This effectively overcomes the inherent limitations of traditional zonal planning, which relies on a "one-and-done" approach, significantly improving treatment accuracy and safety, and ensuring dosimetric precision throughout the entire process from planning to execution.

[0052] The efficiency of re-optimization has been optimized: when re-optimization is required in the dynamic adjustment stage, the original planned result is used as the initial value and fine-tuned. This "hot start" optimization strategy can quickly converge to the new optimal solution, minimizing the additional time consumption caused by plan adjustments and ensuring the smoothness of clinical workflow.

[0053] In summary, the intelligent multi-target area automatic planning method and system of the present invention greatly improves the treatment effect by adopting a sequential optimization planning design that simultaneously considers the effects of cumulative dose and positioning deviation, and adjusts the plan in a timely manner during the treatment process, thus providing a new solution for multi-target area planning design. Attached image description:

[0054] Figure 1 A flowchart of an intelligent multi-target automatic planning method for radiotherapy according to an embodiment of the present invention;

[0055] Figure 2 A block diagram of a smart multi-target automatic planning system for radiotherapy according to an embodiment of the present invention. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0057] Therefore, the following detailed description of embodiments of the present invention is not intended to limit the scope of the claimed invention, but merely illustrates some embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0058] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the present invention can be combined with each other.

[0059] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0060] In the description of this invention, it should be noted that the terms "upper," "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. These terms are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0061] See Figure 1 As shown in this embodiment, a method for intelligent multi-target volume automatic planning in radiotherapy includes:

[0062] The target area for radiotherapy is divided based on the target lesion, the number of beams in each target area is set, and the beam direction of each target area is predicted by the angle prediction model.

[0063] Each target region is optimized in order of priority, and the cumulative dose distribution of the previous target region is incorporated into the objective function of the current target region optimization.

[0064] Dose stacking is performed on the optimized dose distribution for each target region, and the stacked dose is evaluated.

[0065] After the assessment is passed, radiotherapy begins. During the treatment intervals within the target area, dose distribution deviations are assessed using new CT images, and subsequent target area plans are dynamically adjusted.

[0066] The intelligent multi-target volume automatic planning method for radiotherapy provided in this invention has the following significant technical effects compared to the prior art:

[0067] The overall quality and efficiency of multi-target zone planning have been improved: By combining intelligent field setting (angle prediction model) with sequential optimization that considers cumulative dose, automated and intelligent design of multi-target zone planning has been achieved. This method not only avoids the complex trade-offs encountered in global optimization of a single plan, but also fundamentally avoids the dose hotspot and colds problems that may occur in zone planning due to inaccurate dose superposition through precise sequential dose control. Thus, while ensuring plan quality, the efficiency of plan design has been significantly improved.

[0068] Global optimization of the radiation field direction was achieved: the angle prediction model not only aimed at prediction accuracy, but also innovatively introduced a loss function designed to reduce the overlap of radiation fields in different target areas. This makes the radiation field direction output by the model not only the optimal solution for a single target area, but also consider the interaction between multiple target areas from a global perspective, thereby generating a better and more reasonable radiation field layout for the overall irradiation scheme, laying a solid foundation for obtaining high-quality dose distribution.

[0069] This ensures the accuracy and reliability of the final superimposed dose: the sequential optimization process incorporates the cumulative dose generated in previous target regions as a fixed quantity into the optimization objective function of the current target region, ensuring that the optimization of each sub-plan is performed under a known and fixed background dose. This "incremental" optimization method can more realistically simulate the final actual dose distribution, keeping the optimization objective consistent with the final clinical evaluation objective (i.e., the total dose after superimposing doses from all target regions), thereby greatly improving the accuracy and clinical feasibility of the final dose distribution.

[0070] This method enhances the robustness and adaptability of the treatment process: by introducing a dynamic adjustment mechanism for treatment intervals based on new CT images, it effectively addresses dose deviations caused by patient positioning errors or changes in anatomical structures. By assessing dose distribution deviations and determining whether re-optimization is necessary, online validation and adaptive adjustment of the treatment plan are achieved. This effectively overcomes the inherent limitations of traditional zonal planning, which relies on a "one-and-done" approach, significantly improving treatment accuracy and safety, and ensuring dosimetric precision throughout the entire process from planning to execution.

[0071] The efficiency of re-optimization has been optimized: when re-optimization is required in the dynamic adjustment stage, the original planned result is used as the initial value and fine-tuned. This "hot start" optimization strategy can quickly converge to the new optimal solution, minimizing the additional time consumption caused by plan adjustments and ensuring the smoothness of clinical workflow.

[0072] In summary, the intelligent multi-target area automatic planning method and system of this invention, by adopting a sequential optimization planning design that simultaneously considers the effects of cumulative dose and positioning deviation, and by adjusting the plan in a timely manner during treatment, greatly improves the treatment effect and provides a new solution for multi-target area planning design.

[0073] As an optional implementation of this embodiment, in the intelligent multi-target automatic planning method for radiotherapy described in this embodiment, the input of the angle prediction model is the delineation of the target area and organs at risk, the target area priority (the lower the prescribed radiotherapy dose, the higher the priority), the vertical direction of the line connecting the centers of adjacent target areas, and the number of planned beams.

[0074] The output of the angle prediction model is the beam angle for each planned beam.

[0075] The loss function of the angle prediction model is the predicted angle θ. pred With actual angle θ true The deviation and the angle loss between the predicted angle direction and the perpendicular direction of the line connecting the centers of adjacent target areas; the larger the angle, the greater the loss. The greater the loss, the better the loss function becomes:

[0076] L total =λ1L gt +λ2L line ;

[0077]

[0078] L line =(cos(δ) line )) α , α>0;

[0079]

[0080] Where λ1 and λ2 are adjustable weights, Δ(a, b) represents the angle between angles a and b, and α is L line The adjustment factor of the loss function; the larger α is, the larger the value of the loss function. L is the angle between the line connecting the centers of the target area and the x-axis; gt The loss function ensures that the predicted angle is as close as possible to the actual angle, L line The loss function ensures that the overlapping area between beams from different target regions is minimized.

[0081] As an optional implementation of this embodiment, in the intelligent multi-target volume automatic planning method for radiotherapy described in this embodiment, the optimization of each target volume according to priority includes:

[0082] Prioritize each target region based on the total prescription dose for each target region.

[0083] The lower the total prescription dose in the target area, the higher the priority of target area optimization; the higher the total prescription dose in the target area, the lower the priority of target area optimization.

[0084] The reason for optimizing target plans with low total prescription doses first and then optimizing those with high total prescription doses is that if target plans with low total prescription doses are optimized later, the already accumulated fixed doses will inevitably increase the optimization difficulty and worsen the optimization effect.

[0085] Furthermore, in the intelligent multi-target volume automatic planning method for radiotherapy of this embodiment, incorporating the cumulative dose distribution of previous target volumes into the current target volume optimization objective function includes:

[0086] After the current primary target area plan is optimized, the accumulated dose distribution is included as a fixed dose in the objective function of the current primary target area plan. The objective function includes:

[0087]

[0088] F k (ω)=min ω max f f k (ω j );

[0089] Among them, f k (ω j ) represents the objective function of the biased or unbiased scenario in the k-th target area plan, ω represents the intensity map, j represents the j-th scenario, and F k (ω) represents the objective function for robust optimization of the k-th target area plan, D i D represents organ point dose. s The summation represents the cumulative dose at organ points corresponding to the planned dose distribution of the previous target area, D. p The H function represents the prescribed dose or limit for the target area and organs at risk.

[0090] Specifically, the constraint function H satisfies: if the constraint requires <D p ,but When the constraint function H is applied, the value of the constraint function H is 1; otherwise, it is 0.

[0091] As an optional implementation of this embodiment, in a radiotherapy intelligent multi-target volume automatic planning method of this embodiment, the optimization of each target volume according to priority order includes:

[0092] After the current level target area plan optimization is completed, the system automatically saves the plan information and switches the prescription, beam and constraint information. At the same time, the accumulated dose distribution is put into the next level target area plan optimization as a fixed dose to continue optimization until all target area plans are optimized.

[0093] This embodiment of the intelligent multi-target volume automatic planning method for radiotherapy, which "automatically performs sequential optimization according to priority order", brings the following significant technical effects:

[0094] This approach achieves full-process automation, improving the efficiency and consistency of treatment planning. By automatically saving plans, switching treatment parameters (prescription, beam, constraints), and importing cumulative doses into the next stage of optimization, it constructs a seamless automated workflow. This completely avoids operational errors and inconsistencies caused by human intervention that may occur during traditional manual switching and setting processes, greatly reducing manual operation time and tediousness, and ensuring the high efficiency and standardization of multi-target area treatment planning.

[0095] This ensures precise delivery of cumulative dose and improves the reliability of the final dose distribution. By incorporating the cumulative dose from the previous target region optimization as a fixed background dose into the optimization function of subsequent target regions, the optimization of each sub-plan is based on "known and immutable" previous doses. This design means that the final summed dose is no longer a simple post-hoc addition, but is precisely considered and actively controlled during the optimization process. This fundamentally guarantees the accuracy and clinical predictability of the final total dose distribution in multi-target region planning, effectively avoiding dose hotspots and colds.

[0096] The solution process has been optimized, accelerating the overall convergence speed. This sequential optimization method decomposes the complex multi-target global optimization problem into a series of controlled single-target optimization subproblems. Each subproblem has a clear optimization objective, resulting in a relatively simplified search space. Furthermore, subsequent optimizations can be "warm-started" based on the results of previous optimizations (such as initial intensity map values), which helps the optimization algorithm converge to a high-quality solution more quickly, thereby shortening the overall computation time while ensuring plan quality.

[0097] As an optional implementation of this embodiment, in a smart multi-target automatic radiotherapy planning method of this embodiment, the dose superposition of the optimized dose distribution for each target region and the evaluation of the superposition dose include:

[0098] Assess whether each target area meets the prescription coverage requirements, assess whether the cumulative dose to each organ at risk exceeds the clinical limit, and identify any dose hotspots or colds.

[0099] If the assessment requirements are met, the treatment plan will be used for subsequent validation; otherwise, the constraints of each target area will be readjusted.

[0100] This embodiment defines in detail the specific evaluation criteria and feedback mechanism. This evaluation is not a simple dose summation display, but a systematic and quantitative quality control and decision-making process. Its core lies in the rigorous clinical compliance verification of the final comprehensive dose distribution of the multi-target plan.

[0101] Specifically, the evaluation process includes checks on the following three levels:

[0102] Target coverage assessment: This examines whether each target volume meets clinical prescribing requirements under its own dose distribution. For example, does it meet key indicators such as "95% of the target volume receives 100% of the prescribed dose"? This assessment ensures that the treatment intensity for each lesion is sufficient.

[0103] Organ-at-risk protection assessment: This is the most crucial part of the assessment. The system calculates the cumulative dose to the organ at risk from all doses generated by the optimized target area plan and determines whether the total dose exceeds the clinically set safety limits (e.g., maximum spinal cord dose <45 Gy, average intestinal dose <30 Gy, etc.). This directly relates to treatment safety and avoids the risk of neglecting the total OAR dose due to targeted therapy.

[0104] Dose distribution uniformity assessment: This examines the overall dose distribution after stacking for dose hotspots and cold spots. Cold spots (typically referring to doses significantly lower than the prescription within or around the target area) may lead to tumor control failure; hot spots (typically referring to abnormally high doses in organs at risk or normal tissues) may increase the risk of complications. This assessment ensures the conformity and uniformity of the overall dose distribution.

[0105] This embodiment performs dose superposition for the optimized dose distribution of each target region, evaluates the superposition dose, and the resulting technical effects are mainly reflected in:

[0106] This approach ensures the overall quality and clinical safety of the final treatment plan. By implementing comprehensive, cumulative dose-based assessments, the focus shifts from the quality of isolated individual plans to the global, final overall dose distribution quality. It fundamentally eliminates the serious risk of seemingly perfect sub-plans resulting in OAR overdose or dose abnormalities when combined, providing crucial assurance for the safety and effectiveness of multi-target therapy.

[0107] An automated "evaluation-feedback" optimization loop was constructed, enhancing the intelligence level of plan generation. By directly linking evaluation results with constraint adjustments, the entire system possesses self-correcting capabilities. When a plan fails to meet requirements, the system provides clear directions for improvement (constraint adjustment) and supports automated re-optimization. This significantly reduces the time spent on manual trial and error and reliance on experience, guiding the workflow to quickly converge to a clinically usable optimal plan.

[0108] This achieves quantitative and objective quality control, reducing subjective differences. The evaluation standard is based on clear quantitative indicators (coverage, dose limits), replacing the traditional method that relies on the personal experience of physicists. This makes the assessment of plan quality more objective, consistent, and repeatable, which is conducive to the standardization of treatment standards.

[0109] In summary, this refined evaluation process is the core element of this invention in ensuring the high clinical usability of its automatically generated plans. It closely integrates intelligent optimization with strict clinical standards and is key to achieving the goal of "fast and good" radiotherapy planning.

[0110] As an optional implementation of this embodiment, in a smart multi-target volume automatic planning method for radiotherapy in this embodiment, the step of assessing dose distribution deviation through new CT images and dynamically adjusting subsequent target volume planning during the treatment interval between target volumes includes:

[0111] Radiotherapy is administered to each target area in order of priority.

[0112] Before the irradiation plan for the current target area begins, CT images are reacquired each time the irradiation plan for the previous target area ends. The original plan for the current target area is then transferred to the newly acquired CT images for dose calculation, and the dose distribution deviation is evaluated to determine whether it exceeds the preset threshold.

[0113] If the preset threshold is not exceeded, the existing original irradiation plan will continue to be used for the current target area; otherwise, the current and subsequent target area plans will be re-optimized based on the newly acquired CT images.

[0114] The dynamic adjustment mechanism described in this embodiment brings crucial real-time adaptability and robustness to the intelligent multi-target area automatic planning method of the present invention. Its technical effects are mainly reflected in the following aspects:

[0115] This method effectively compensates for positioning errors and anatomical changes, ensuring precise dose delivery. Between treatment fractions, the patient's positioning and internal anatomical structures (such as organ movement, deformation, or tumor regression) may change. By re-acquiring CT images and comparing dose calculations before irradiating the current target area, this method actively detects the actual impact of these changes on dose distribution (i.e., dose deviation), thus basing the treatment plan on the latest and most accurate anatomical foundation. This significantly improves the geometric accuracy of dose delivery and avoids the pitfalls of haphazard dose delivery.

[0116] This represents an upgrade from "static planning" to "dynamic adaptive treatment," significantly improving treatment safety. This mechanism breaks away from the traditional radiotherapy plan, which is statically executed throughout the entire treatment process once it is formulated. By automatically judging preset dose deviation thresholds, the system can intelligently decide whether to continue with the original plan or initiate a re-optimization. This is equivalent to installing a "safety sentinel" in the treatment process; it only continues execution when the dose deviation is within an acceptable range, otherwise it corrects it promptly, effectively preventing the risk of over-radiation or under-radiation to patients due to accumulated errors, thus ensuring the bottom line of treatment safety.

[0117] The clinical workflow has been optimized, balancing the needs for accuracy and efficiency. This dynamic adjustment strategy is both efficient and targeted. Instead of blindly rescanning and optimizing every target area, it uses threshold-based screening. In most cases, treatment can proceed quickly if there is no significant deviation, ensuring efficiency; re-optimization of the current and subsequent target areas is triggered only when necessary (deviation exceeding the threshold). Furthermore, re-optimization can be fine-tuned based on the original plan's results, avoiding a complete overhaul and minimizing the time required for adjustments, making this high-precision adaptive treatment clinically feasible.

[0118] This enhances the robustness of treatment for complex cases. For complex cases requiring multi-target, multi-fractionation treatment, this dynamic adjustment mechanism significantly improves the overall treatment regimen's tolerance to various uncertainties (i.e., robustness). It ensures that the final dose distribution can be closely adjusted to track the patient's actual condition, providing strong technical support for achieving the expected therapeutic effect.

[0119] In summary, this dynamic adjustment step is the core technological guarantee that ensures the high-quality plans of this invention can be executed accurately, safely, and efficiently in complex real-world treatment environments, and is a key breakthrough in achieving personalized, adaptive radiotherapy.

[0120] Specifically, in this embodiment of a smart multi-target volume automatic planning method for radiotherapy:

[0121] The step of migrating the original target area plan to the newly acquired CT image for dose calculation includes: registering the original CT image and the newly acquired CT image, copying the current target area plan data, including plan information and beam information, to the newly acquired CT image, performing dose calculation, and obtaining the dose distribution of the current target area on the newly acquired CT image.

[0122] The assessment of whether the dose distribution deviation exceeds the preset threshold includes: using a dose difference tool to calculate the difference between the dose distribution on the original CT image and the currently reacquired CT image, and checking whether the global maximum deviation dose is less than the preset threshold. If it is less than the preset threshold, the current plan does not need to be re-optimized; otherwise, the current plan and subsequent plans will be re-optimized and adjusted on the currently reacquired CT image.

[0123] The process of re-optimizing the current and subsequent target area plans based on the newly acquired CT images includes: using the original plan optimization results as the initial values ​​for re-optimization, and the accumulated irradiation dose as the fixed dose, fine-tuning the optimization results, and quickly optimizing a new plan.

[0124] The following describes a specific example of the present invention:

[0125] A method for intelligent multi-target volume automatic planning in radiotherapy, the specific implementation steps of which are as follows:

[0126] 1) A type of head tumor is divided into two target regions (target region 1 and target region 2), and the planned number of beams and prescriptions for each target region are set (the prescription for target region 1 is less than the prescription for target region 2);

[0127] 2) Using the independently developed MB-UNet model that has completed angle prediction task training, the planned beam directions for the two target areas are predicted;

[0128] 3) Optimize the target area plans for target area 1 and target area 2 sequentially. When optimizing the target area plan for target area 2, use the dose distribution optimized for target area 1 as the fixed dose in the objective function of target area 2 for optimization.

[0129] 4) Evaluate the cumulative dose distribution of the planned dose distribution in target areas 1 and 2, and check whether the target coverage meets the requirements (95% of the target volume covers 100% of the target dose), whether the organs at risk meet the clinical limit requirements, and whether there are dose hotspots or colds in the cumulative dose. If the evaluation requirements are met, the plan can be used for subsequent validation treatment; otherwise, it should be re-optimized in sequence.

[0130] 5) Set the maximum dose deviation threshold. Before executing the target area 2 target area plan, compare the maximum dose deviation value between the dose calculation result of the plan on the new CT and the original plan result. If the maximum dose deviation value exceeds the set threshold, execute the original plan. Otherwise, re-optimize the target area 2 target area plan based on the new CT. That is, use the original plan optimization result as the initial value for re-optimization, fine-tune the optimization result, and quickly optimize the new plan.

[0131] See Figure 2 As shown, this embodiment also provides a smart multi-target volume automatic planning system for radiotherapy, including:

[0132] The intelligent field setting module divides the target area based on the target lesion, sets the number of beams for each target area, and predicts the beam direction of each target area through the angle prediction model.

[0133] The iterative optimization module optimizes each target region in order of priority and incorporates the cumulative dose distribution of the previous target region into the current target region's optimization objective function.

[0134] The dose assessment module performs dose superposition for the optimized dose distribution of each target area and evaluates the superposition dose.

[0135] The dynamic adjustment module initiates radiotherapy after successful evaluation. During treatment intervals within the target area, dose distribution deviations are assessed using new CT images, and subsequent target area planning is dynamically adjusted. This embodiment also provides a computer-readable storage medium storing a computer-executable program. When executed, the computer-executable program implements the intelligent multi-target automatic planning method for radiotherapy as described above.

[0136] The computer-readable storage medium described in this embodiment may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable storage medium may also be any readable medium other than a readable storage medium, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0137] This embodiment also provides an electronic device, including a processor and a memory, wherein the memory is used to store a computer-executable program, and when the computer program is executed by the processor, the processor executes the aforementioned intelligent multi-target automatic planning method for radiotherapy.

[0138] The electronic device is manifested in the form of a general-purpose computing device. It may contain one or more processors that work collaboratively. This invention also does not preclude distributed processing, meaning that processors may be distributed across different physical devices. The electronic device of this invention is not limited to a single entity, but may also be the sum of multiple physical devices.

[0139] The memory stores a computer-executable program, typically machine-readable code. The computer-readable program can be executed by the processor to enable the electronic device to perform the method of the present invention, or at least some steps of the method.

[0140] The memory includes volatile memory, such as random access memory (RAM) and / or cache memory, and may also be non-volatile memory, such as read-only memory (ROM).

[0141] It should be understood that the electronic device of the present invention may also include elements or components not shown in the examples above. For example, some electronic devices also include display units such as a display screen, and some electronic devices also include human-computer interaction elements such as buttons and keyboards. Any electronic device capable of executing a computer-readable program in its memory to implement the method of the present invention or at least some steps of the method can be considered as an electronic device covered by the present invention.

[0142] From the above description of the embodiments, those skilled in the art will readily understand that the present invention can be implemented by hardware capable of executing specific computer programs, such as the system of the present invention, and the electronic processing unit, server, client, mobile phone, control unit, processor, etc. included in the system. The present invention can also be implemented by computer software that executes the methods of the present invention, for example, by control software executed by a microprocessor, electronic control unit, client, server, etc. However, it should be noted that the computer software executing the methods of the present invention is not limited to execution in one or a specific set of hardware entities; it can also be implemented in a distributed manner by unspecified hardware. For computer software, the software product can be stored in a computer-readable storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or distributed across a network, as long as it enables electronic devices to execute the methods according to the present invention.

[0143] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described herein. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the specific embodiments described above. Therefore, any modifications or equivalent substitutions to the present invention, as well as all technical solutions and improvements that do not depart from the spirit and scope of the invention, are covered within the scope of the claims of the present invention.

Claims

1. A smart multi-target volume automatic planning system for radiotherapy, characterized in that, include: The intelligent field setting module divides the target area based on the target lesion, sets the number of beams for each target area, and predicts the beam direction of each target area through the angle prediction model. The iterative optimization module optimizes each target region in order of priority and incorporates the cumulative dose distribution of the previous target region into the current target region's optimization objective function. The dose assessment module performs dose superposition for the optimized dose distribution of each target area and evaluates the superposition dose. The dynamic adjustment module initiates radiotherapy after successful evaluation. During the treatment intervals within the target area, dose distribution deviations are assessed using new CT images, and subsequent target area plans are dynamically adjusted.

2. The intelligent multi-target automatic planning system for radiotherapy according to claim 1, characterized in that, The inputs to the angle prediction model are target area and organ at risk delineation, target area priority, vertical direction of the line connecting the centers of adjacent target areas, and the number of planned beams. The output of the angle prediction model is the beam angle for each planned beam. The loss function of the angle prediction model is the predicted angle. From a practical perspective The loss includes the bias and the angle between the predicted angle direction and the perpendicular direction of the line connecting the centers of adjacent target areas. The loss function is: ； ； ； ； ； in, , For adjustable weights, This represents the angle between two angles, a and b. for Adjustment factor of the loss function The larger the value, the larger the loss function value. The angle between the line connecting the centers of the target area and the x-axis; The loss function ensures that the predicted angle is as close as possible to the actual angle. The loss function ensures that the overlapping area between beams from different target regions is minimized.

3. The intelligent multi-target automatic planning system for radiotherapy according to claim 1, characterized in that, The optimization of each target region according to priority includes: Prioritize each target region based on the total prescription dose for each target region. The lower the total prescription dose in the target area, the higher the priority of target area optimization; the higher the total prescription dose in the target area, the lower the priority of target area optimization.

4. The intelligent multi-target automatic planning system for radiotherapy according to claim 3, characterized in that, The step of incorporating the cumulative dose distribution of the previous target region into the current target region optimization objective function includes: After the current primary target area plan is optimized, the accumulated dose distribution is included as a fixed dose in the objective function of the current primary target area plan. The objective function includes: ； ； in, Let represent the objective function for either the biased or unbiased scenario of the k-th level target area plan. This represents the intensity map, where j represents the j-th scene. Let the objective function be the robustness optimization function for the k-th target area. Indicates organ point dose, The summation represents the cumulative dose at organ points corresponding to the planned dose distribution of the previous target region. The H function represents the prescribed dose or limit for the target area and organs at risk.

5. The intelligent multi-target automatic planning system for radiotherapy according to claim 4, characterized in that, The constraint function H satisfies: if the constraint requires ,but When the constraint function H is applied, the value of the constraint function H is 1; otherwise, it is 0.

6. The intelligent multi-target automatic planning system for radiotherapy according to claim 4, characterized in that, The optimization of each target region according to priority includes: After the current level target area plan optimization is completed, the system automatically saves the plan information and switches the prescription, beam and constraint information. At the same time, the accumulated dose distribution is put into the next level target area plan optimization as a fixed dose to continue optimization until all target area plans are optimized.

7. The intelligent multi-target automatic planning system for radiotherapy according to claim 1, characterized in that, The dose distribution optimized for each target region is superimposed, and the superimposed dose is evaluated, including: Assess whether each target area meets the prescription coverage requirements, assess whether the cumulative dose to each organ at risk exceeds the clinical limit, and identify any dose hotspots or colds. If the assessment requirements are met, the treatment plan will be used for subsequent validation; otherwise, the constraints of each target area will be readjusted.

8. The intelligent multi-target automatic planning system for radiotherapy according to claim 1, characterized in that, The method of assessing dose distribution deviation through new CT images and dynamically adjusting subsequent target area plans during the treatment interval within the target area includes: Radiotherapy is administered to each target area in order of priority. Before the irradiation plan for the current target area begins, CT images are reacquired each time the irradiation plan for the previous target area ends. The original plan for the current target area is then transferred to the newly acquired CT images for dose calculation, and the dose distribution deviation is evaluated to determine whether it exceeds the preset threshold. If the preset threshold is not exceeded, the existing original irradiation plan will continue to be used for the current target area; otherwise, the current and subsequent target area plans will be re-optimized based on the newly acquired CT images.

9. The intelligent multi-target automatic planning system for radiotherapy according to claim 8, characterized in that, The step of migrating the original target area plan to the newly acquired CT image for dose calculation includes: registering the original CT image and the newly acquired CT image, copying the current target area plan data, including plan information and beam information, to the newly acquired CT image, performing dose calculation, and obtaining the dose distribution of the current target area on the newly acquired CT image. The assessment of whether the dose distribution deviation exceeds the preset threshold includes: using a dose difference tool to calculate the difference between the dose distribution on the original CT image and the currently reacquired CT image, and checking whether the global maximum deviation dose is less than the preset threshold. If it is less than the preset threshold, the current plan does not need to be re-optimized; otherwise, the current plan and subsequent plans will be re-optimized and adjusted on the currently reacquired CT image. The process of re-optimizing the current and subsequent target area plans based on the newly acquired CT images includes: using the original plan optimization results as the initial values ​​for re-optimization, and the accumulated irradiation dose as the fixed dose, fine-tuning the optimization results, and quickly optimizing a new plan.

Images