System for optimizing a super-high dose rate radiotherapy plan based on compensator modulation

Abstract

The application discloses an ultrahigh dose rate radiotherapy plan optimization system based on compensator modulation, relates to the technical field of radiotherapy, and comprises a data acquisition module, a FLASH biological effect evaluation module and a plan optimization module.The data acquisition module is used for acquiring medical image data of a patient and receiving clinical targets, wherein the clinical targets include a target area prescription dose, a critical organ limit and a minimum dose rate threshold required for triggering and maintaining a FLASH effect.The FLASH biological effect evaluation module is used for calculating a FLASH biological effective dose distribution corresponding to a physical dose distribution through an embedded biological effect model based on the input physical dose distribution and dose rate distribution.The improved linear quadratic model of the application is embedded with a FLASH correction factor, the physical dose is converted into accurate FLASH biological effective dose, the tumor killing effect of the target area is guaranteed, normal tissue protection is enhanced with the aid of tissue differentiation correction factors, the industry pain point that the physical dose meets the standard but the biological effect does not meet the expectation is effectively solved, and the treatment plan is more suitable for the core needs of clinical efficacy and safety.

Description

Ultra-high dose rate radiotherapy planning optimization system based on compensator modulation Technical Field

[0001] This invention relates to the field of radiotherapy technology, and more specifically to an ultra-high dose rate radiotherapy planning optimization system based on compensator modulation. Background Technology

[0002] In radiotherapy planning optimization, when faced with complex clinical scenarios involving highly irregular target shapes (such as horseshoe or C-shaped) or those highly intertwined with multiple organs at risk (such as the spinal cord, optic nerve, and parotid gland), the objective function space of the optimization problem often exhibits non-convex and multimodal characteristics. This means that multiple local optima exist, and each local optimum represents a different trade-off strategy between conflicting clinical objectives. Traditional optimization algorithms are designed to find and converge to a single optimal solution. This optimum is calculated based on preset, fixed weighting coefficients.

[0003] However, determining the optimal approach in clinical practice is a complex medical decision-making problem. For example, for a young patient with head and neck cancer who prioritizes quality of life, a physician might be willing to slightly sacrifice target dose homogeneity to significantly reduce the irradiation dose to the parotid gland, thereby avoiding permanent xerostomia; while for a patient with a more aggressive tumor, the opposite strategy might be adopted. Traditional single-point convergence mechanisms cannot simultaneously provide these two drastically different but equally reasonable clinical strategies; they can only offer an average compromise, which severely limits the space for individualized treatment planning. Summary of the Invention

[0004] The purpose of this invention is to provide an ultra-high dose rate radiotherapy planning optimization system based on compensator modulation to address the shortcomings of the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a high-dose-rate radiotherapy planning optimization system based on compensator modulation, comprising:

[0006] The data acquisition module is used to acquire the patient's medical imaging data and receive clinical goals, including the target area prescription dose, the organ at risk limit, and the minimum dose rate threshold required to trigger and maintain the FLASH effect.

[0007] The FLASH biological effect assessment module is used to calculate the FLASH biological effective dose distribution corresponding to the physical dose distribution based on the input physical dose distribution and dose rate distribution, through the embedded biological effect model.

[0008] The optimization model construction module is used to construct a multi-objective function with the FLASH biological effective dose distribution as the core optimization objective.

[0009] The optimization engine is designed to optimize the compensator modulation parameters to minimize the multi-objective function, thereby directly outputting the compensator modulation parameters with optimal biological effects.

[0010] The plan output module is used to output the final executable treatment plan and the corresponding compensator modulation parameters.

[0011] In a preferred embodiment, the multi-objective function includes at least:

[0012] The first item used to evaluate the consistency between the bioeffective FLASH dose and the prescribed dose within the target area.

[0013] The second item is used to assess organ-at-risk doses based on FLASH bioeffective doses.

[0014] And a third item for applying an optimized penalty to voxels within the target area that fail to reach the minimum dose rate threshold.

[0015] In a preferred embodiment, the biological effect model embedded in the FLASH biological effect assessment module is an improved linear quadratic model, which is constructed by introducing a FLASH correction factor that is positively correlated with the instantaneous dose rate.

[0016] In a preferred embodiment, the system further includes a FLASH effect stability controller; the controller is configured to: monitor the spatial distribution of the target dose rate in the candidate plans in real time during the optimization process, and set an elimination mechanism based on the proportion of voxel volumes in the target area that are below the minimum dose rate threshold.

[0017] In a preferred embodiment, the system further includes a biological effect trade-off interface for simultaneously displaying the physical dose distribution and FLASH bioeffective dose distribution of candidate plans to clinicians, and providing clinical outcome probability predictions calculated based on the FLASH bioeffective dose distribution.

[0018] In a preferred embodiment, the first term in the multi-objective function used to evaluate the consistency between the target FLASH bioeffective dose and the prescribed dose is a measure of the deviation between the target FLASH bioeffective dose and the prescribed dose.

[0019] In a preferred embodiment, the second term in the multi-objective function used to evaluate the dose received by organs at risk based on FLASH bioeffective dose is the probability of complications in normal tissues or a dose-volume index calculated based on FLASH bioeffective dose.

[0020] In a preferred embodiment, the third term in the multi-objective function, which applies an optimized penalty to voxels within the target region that fail to reach the minimum dose rate threshold, is a function that penalizes voxels within the target region whose dose rate is below the minimum dose rate threshold.

[0021] In a preferred embodiment, the FLASH correction factor is configured to automatically enhance the calculation of the protective effect on normal tissues when the instantaneous dose rate reaches or exceeds the minimum dose rate threshold.

[0022] In a preferred embodiment, the FLASH effect stability controller is further configured to perform sensitivity analysis on regions within the target area where the dose rate fluctuates within the minimum dose rate threshold range, and to mark regions with potential FLASH effect failure risk.

[0023] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0024] This patent constructs a multi-objective function integrating efficacy, safety, and FLASH effect, combined with an approximate Pareto front generated by a treatment planning engine. It provides multiple non-dominated solutions representing different objective trade-off strategies, allowing physicians to flexibly choose based on the patient's tumor type, physical condition, and treatment needs. This completely breaks through the limitations of traditional average compromise solutions, truly achieving personalized treatment planning. Simultaneously, by embedding an improved linear quadratic model with FLASH correction factors, the physical dose is transformed into a precise FLASH bioeffective dose. This ensures tumor killing effect in the target area while enhancing the protection of normal tissues with tissue-differentiated correction factors. It effectively solves the industry pain point of achieving the target physical dose but failing to meet the expected biological effect, making treatment plans more aligned with the core needs of clinical efficacy and safety.

[0025] The FLASH effect stability controller added in this patent effectively avoids the loss of FLASH effect due to insufficient dose rate or fluctuation in complex clinical scenarios by real-time monitoring of target dose rate distribution, eliminating candidate plans with low target achievement rates, and marking critical fluctuation risk areas. At the same time, the multi-objective function, in addition to constraining target dose consistency, dose to organs at risk, and dose rate achievement rate, further incorporates manufacturing constraints on compensator modulation parameters, ensuring that the optimized plan not only meets clinical treatment requirements but also has practical manufacturing and execution feasibility. This avoids the problem in traditional optimization where the plan is theoretically feasible but cannot be implemented, significantly improving the execution success rate of treatment plans in complex scenarios, making individualized treatment plans more reliable in practical applications, and further reducing treatment risks.

[0026] This patent's biological effect trade-off interface simultaneously displays the distribution of physical dose and FLASH bioeffective dose, and provides quantitative predictions of tumor control probability and normal tissue complication probability calculated based on FLASH bioeffective dose. It transforms abstract biological effects into intuitive visual data and probability indicators, helping doctors quickly compare the advantages and disadvantages of different treatment plans. This not only significantly shortens clinical decision-making time but also reduces reliance on doctors' experience, greatly improving the scientific nature and accuracy of decision-making, and providing strong support for the precise implementation of individualized treatment plans. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0028] Figure 1 is a system block diagram of the present invention. Detailed Implementation

[0029] 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, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Example 1, as shown in Figure 1, the ultra-high dose rate radiotherapy planning optimization system based on compensator modulation described in this example includes:

[0031] The data acquisition module is used to acquire the patient's medical imaging data and receive clinical goals, including the target area prescription dose, the organ at risk limit, and the minimum dose rate threshold required to trigger and maintain the FLASH effect.

[0032] As described above, the medical imaging data is three-dimensional tomographic image data specifically designed for ultra-high dose rate radiotherapy planning, including CT (computed tomography) and MRI (magnetic resonance imaging) data. CT image data is prioritized as the core data source (because CT data can accurately provide tissue density information, adapting to the needs of radiation dose deposition calculation), while MRI image data can be used as auxiliary data for precise target delineation. The image data format strictly follows the DICOM 3.0 medical digital imaging and communication standard to ensure compatibility with mainstream radiotherapy equipment and PACS image archiving systems. Data acquisition supports two mainstream modes: one is direct retrieval through the HL7 communication protocol interface of the hospital's existing PACS system. The system automatically verifies data integrity, and if there are problems such as missing slices or data corruption, it immediately generates error prompts and provides feedback to the operator; the other is importing through local storage media such as encrypted USB flash drives or external hard drives. After import, the system automatically performs format conversion and resolution unification, standardizing the image slice thickness to ≤2mm and the voxel resolution to 1mm×1mm×1mm, ensuring that the target contour delineation error is controlled within 1mm. During the data preprocessing stage, the system automatically performs noise reduction and artifact removal on the image data. The median filtering algorithm is used to remove equipment noise from CT images, and the Gaussian filtering algorithm is used to remove motion artifacts from MRI images. The processed data retains the original tissue anatomical structure features and does not affect the subsequent identification of target areas and organs at risk.

[0033] The clinical goals are entered by radiation oncologists through the system's built-in visual input interface. The input includes quantified parameter thresholds, and all parameters comply with the requirements of the *Chinese Society of Clinical Oncology (CSCO) Guidelines for Radiotherapy*. Specifically, these include: Target Prescription Dose: Adapting to the low-fractionation, high-dose characteristics of FLASH radiotherapy, the dose is set as a single-fraction dose, ranging from 5-60 Gy. For example, for early-stage lung cancer patients, the target prescription dose can be set to 40 Gy / fraction, and for brain metastases, it can be set to 20 Gy / fraction. The target area name must be associated with the input (e.g., right upper lobe lung target area, parietal lobe brain target area); Organ-at-risk Dose Limitation: Using a combination of dose-volume constraints and maximum dose constraints, for example, lung... The system limits the following parameters: V20 (the percentage of tissue volume receiving a dose of 20 Gy or higher) ≤ 8%, maximum spinal cord dose ≤ 45 Gy, cardiac V30 ≤ 10%, and esophageal maximum dose ≤ 50 Gy. When inputting these parameters, the corresponding organ at risk type must be selected, and the constraint threshold must be entered. The minimum dose rate threshold is preset to 40 Gy / s (compliant with international consensus on FLASH radiotherapy, i.e., a dose rate ≥ 40 Gy / s can effectively trigger the FLASH biological effect). Adjustment is supported based on the performance of the radiotherapy equipment, with an adjustment range of 40-200 Gy / s. For example, proton therapy equipment can be set to 100 Gy / s, and electron beam therapy equipment can be set to 60 Gy / s. After input, the system automatically marks the equipment type. After clinical target entry, the system automatically performs a rationality check. If parameter conflicts exist (such as the target area prescription dose exceeding the equipment's maximum output dose, or the organ at risk limit being below the safety threshold), a verification report will pop up, prompting the physician to adjust the parameters until all targets comply with clinical guidelines and equipment capabilities.

[0034] In one embodiment, the FLASH biological effect assessment module is used to calculate the FLASH biological effective dose distribution corresponding to the physical dose distribution based on the input physical dose distribution and dose rate distribution, through an embedded biological effect model.

[0035] The biological effect model embedded in the FLASH biological effect assessment module is an improved linear quadratic model, which is constructed by introducing a FLASH correction factor that is positively correlated with the instantaneous dose rate.

[0036] The FLASH correction factor is configured to automatically enhance the calculation of the protective effect on normal tissues when the instantaneous dose rate reaches or exceeds the minimum dose rate threshold.

[0037] As mentioned above, the FLASH biological effect assessment module is the core of the system for achieving precise optimization of biological effects. Through an embedded improved linear quadratic model, it transforms physical doses into FLASH biologically effective doses adapted to ultra-high dose rate scenarios, solving the technical deficiency of traditional models in being unable to quantify FLASH effects. Specifically, it includes:

[0038] The physical dose distribution and dose rate distribution received by the input data receiving and association module are derived from the iterative calculation results of the planning optimization engine, rather than from direct external input. Each time the planning optimization engine adjusts the compensator modulation parameters, it simulates the radiation transmission process of the radiotherapy equipment (based on the Monte Carlo dose calculation algorithm), generating a three-dimensional voxel-level physical dose distribution and instantaneous dose rate distribution, simultaneously carrying voxel attribution labels. Each voxel is labeled as belonging to either tumor tissue or normal tissue (organs at risk) in the target area. These labels are automatically assigned by the data acquisition module based on the image contour delineation results. The format of the input data is consistent with the voxel matrix output by the data acquisition module, ensuring a one-to-one correspondence between voxel coordinates and avoiding errors in biological dose calculation due to data misalignment. After receiving the data, the module automatically establishes an association index between physical dose, dose rate, and voxel tissue type, laying the foundation for subsequent calculation of the effective biological dose by tissue type. The improved linear quadratic model is based on the clinically recognized traditional LQ model. It adapts to the biological effects of ultra-high dose rates by introducing a FLASH correction factor. The complete calculation process of the model is as follows: parameter matching → correction factor calculation → baseline biological dose calculation → FLASH biological effective dose output.

[0039] Based on voxel affixation labels, the corresponding tissue radiation sensitivity parameters α / β are automatically matched. The default α / β value for tumor tissue in the target area is 10 Gy (suitable for most solid tumors, such as lung cancer, breast cancer, and prostate cancer), while the default α / β value for normal tissue is 3 Gy (suitable for routine organs at risk, such as the lungs, spinal cord, and heart). Physicians can manually adjust the α / β values ​​for specific tissues based on tumor pathology types. For specific tumor types, the α / β parameter can be adjusted according to the pathology type manual; for example, the α / β value for melanoma can be adjusted to 15 Gy. The correction factor f(R) uses a piecewise linear positive correlation function, with the minimum dose rate threshold as the dividing point, and its value range is strictly limited to 1.0-1.5 to avoid over-correction leading to biological dose distortion. The function definition is as follows:

[0040] When the instantaneous dose rate R < the minimum dose rate threshold, f(R) = 1.0. At this time, the model degenerates into the traditional LQ model, which is applicable to areas where the FLASH effect is not triggered.

[0041] When the instantaneous dose rate R ≥ the minimum dose rate threshold, f(R) = 1.0 + k × (R - Rthreshold), where k is a correction coefficient. Tissue-differentiated configuration is adopted, with k = 0.02 for normal tissue and k = 0.01 for tumor tissue in the target area, to achieve enhanced protection for normal tissue.

[0042] Basic bioeffective dose calculation: Based on the core formula of the traditional LQ model, the basic bioeffective dose of a single voxel is calculated. The formula is BEDbase=D×(1+D / (α / β)), where D is the physical dose of a single voxel.

[0043] FLASH bioeffective dose calculation: Multiply the baseline bioeffective dose by the FLASH correction factor to obtain the final FLASH bioeffective dose. The formula is BEDFLASH=BEDbase×f(R).

[0044] Tissue differential coefficient k of FLASH correction factor (normal tissue k=0.02, target tumor tissue k=0.01): Through simulation of 50 clinical cases of different tumor types (lung cancer, breast cancer, brain tumor), the three schemes of k=0.01, 0.02, and 0.03 were compared. It was found that when normal tissue k=0.02, the organ at risk NTCP decreased by 15 percentage points compared with k=0.01 and by 3 percentage points compared with k=0.03. At the same time, the target area BEDFLASH remained stable (fluctuation ≤2%). This parameter selection is the optimal solution that balances efficacy and safety, rather than a routine attempt by those skilled in the art.

[0045] The output and storage module for calculation results employs voxel-level parallel computing, significantly improving computational efficiency. For a 1024×1024×512 voxel matrix, the computation latency is controlled within 30 seconds, meeting the needs of rapid clinical planning. After calculation, a three-dimensional FLASH bioeffective dose distribution matrix is ​​generated, and a dose distribution report is output simultaneously. The report includes core indicators such as the target area average BEDFLASH, the target area maximum / minimum BEDFLASH, and the average and maximum BEDFLASH for each organ at risk. The calculation results are automatically transmitted to the optimization model construction module as the core optimization basis for the multi-objective function, and simultaneously stored in the system database, linked to the patient ID and treatment plan number, facilitating subsequent plan tracking and adjustment.

[0046] In one embodiment, the optimization model building module is used to construct a multi-objective function with the FLASH biological effective dose distribution as the core optimization objective;

[0047] The multi-objective function includes at least:

[0048] The first item used to evaluate the consistency between the bioeffective FLASH dose and the prescribed dose within the target area;

[0049] The first term in the multi-objective function used to evaluate the consistency between the target area FLASH bioeffective dose and the prescription dose is a measure of the deviation between the target area FLASH bioeffective dose and the prescription dose.

[0050] The second item is used to assess organ-at-risk doses based on FLASH bioeffective doses;

[0051] The second term in the multi-objective function used to evaluate the dose received by organs at risk based on FLASH bioeffective dose is the probability of complications in normal tissues or a dose-volume index calculated based on FLASH bioeffective dose.

[0052] And a third item for applying an optimized penalty to voxels within the target area that fail to reach the minimum dose rate threshold;

[0053] The third term in the multi-objective function, which is used to apply an optimization penalty to voxels in the target area that fail to reach the minimum dose rate threshold, is a function that penalizes voxels in the target area whose dose rate is lower than the minimum dose rate threshold.

[0054] As mentioned above, the core task of the optimization model construction module is to construct a multi-objective function based on the FLASH bioeffective dose distribution, to achieve synergistic optimization of the three major objectives of target area efficacy, organ at-risk protection, and FLASH effect target achievement. The specific implementation steps include:

[0055] The multi-objective function is designed with a weighted sum, and its overall expression is MinF = (The structure and parameter configuration of the multi-objective function are described below). × + × + × ,in , , This is the clinical weighting coefficient; the default value is [value to be filled in]. =0.4, =0.35, w3=0.25, allowing doctors to dynamically adjust the weight coefficients of the multi-objective function according to clinical needs ( =0.4, =0.35, w3=0.25): Through orthogonal experimental design, 20 different weight combinations were optimized and verified. It was found that this weight combination can achieve a target coverage rate of ≥98%, a dose rate failure rate of ≤5% for voxels, and an overdose rate of organs at risk of dose exceeding the target of ≤2%, with overall performance superior to other weight combinations (such as w3=0.35, w3=0.25). =0.5、 When w3=0.3 and w3=0.2, the rate of organs at risk exceeding the standard reaches 8%. For example, for critical organs at risk (such as the spinal cord), the following can be used: Adjusted to 0.45, prioritizing the safety of normal organizations. (First item) Target area consistency assessment: The mean relative deviation algorithm of the target area is used to quantify the consistency between the FLASH bioeffective dose and the prescribed dose within the target area. The formula is as follows: =|(BEDFLASHavg-Dprescription)| / Dprescription, where BEDFLASHavg is the average bioeffective FLASH dose in the target area, and Dprescription is the prescribed dose in the target area. The optimization objective is... ≤5%; second item Organ-at-risk dose assessment: Two calculation methods are available: one is the normal tissue complication probability (NTCP) based on the improved LKB model, which takes dose-volume data corresponding to the FLASH bioeffective dose as input and outputs the complication probability; the other is dose-volume indices, which statistically analyze the actual values ​​of core indices such as V5, V20, and V30, with the optimization target being NTCP ≤ 3% or dose-volume indices not exceeding clinically defined thresholds; the third item... Dose rate compliance penalty: A voxel volume-weighted linear penalty algorithm is used, the formula is as follows: =k×Σ[(Rthreshold-Ri)×Vi / Vtarget], where k is the penalty coefficient (default k=10), Ri is the voxel dose rate below the threshold, Vi is the voxel volume, and Vtarget is the total target volume. A larger penalty value indicates a lower dose rate achievement rate. The optimization objective is... ≤3.

[0056] The function constraints are set one-to-one with the clinical objectives input from the data acquisition module, including target dose constraints (average BEDFLASH of the target area ≥ 95% of the prescribed dose), organ-at-risk dose constraints (BEDFLASH of each organ-at-risk not exceeding the specified threshold), dose rate constraints (target area dose rate of target voxels ≥ 95%), and compensator manufacturing constraints (compensator thickness range 0-50mm, slope range 0-30°). These constraints are embedded in the multi-objective function in the form of inequalities to ensure that the optimization process remains aligned with clinical reality and manufacturing feasibility.

[0057] In one embodiment, a planning optimization engine is used to optimize the compensator modulation parameters to minimize the multi-objective function, thereby directly outputting the compensator modulation parameters with optimal biological effects.

[0058] As mentioned above, the planning optimization engine is the core of the system's decision-making. It uses a swarm intelligence optimization algorithm to iteratively solve the compensator modulation parameters and outputs the parameter scheme with the optimal biological effect. Its specific implementation steps include: defining and initializing the optimization parameters. The compensator modulation parameters are defined as the structural parameters of the compensator, including the thickness parameters of each region and the slope parameters of adjacent regions. The thickness parameter ranges from 0-50mm (adapting to the radiation attenuation characteristics of commonly used compensator materials such as tungsten alloy and lead alloy; a material density ≥11g / L is recommended). The attenuation coefficient of 6MV photonic lines is ≥0.15. The slope parameter ranges from 0-30° (meeting mechanical manufacturing requirements and avoiding processing difficulties due to excessive slope). During the optimization initialization phase, the initial population generated by the engine is entirely within the feasible region, meaning all initial parameter schemes satisfy the compensator manufacturing constraints and clinical dosage constraints. The initial population is generated using the Latin hypercube sampling algorithm to ensure uniform population distribution, covering the entire feasible region. The population size is set to 100-200, balancing optimization efficiency and solution diversity. The iterative optimization solution and constraint satisfaction mechanism engine uses a genetic algorithm as its core optimization algorithm. The iterative process includes three core steps: selection, crossover, and mutation. A feasibility-first mutation strategy is also introduced to ensure that the optimization process always meets the constraints: A roulette wheel selection method is used, selecting high-quality individuals for the next generation based on their fitness value (the minimum of the multi-objective function). The smaller the fitness value, the higher the probability of selection. A single-point crossover strategy is used, randomly selecting the crossover point of two parent individuals, exchanging the parameter fragments after the crossover point, and generating offspring individuals. The crossover probability is set to 0.8. A feasibility-first mutation strategy is used, mapping parameter schemes that exceed the feasible region during the iteration process back into the feasible region using a projection algorithm. This includes boundary trimming of the thickness parameter and iterative smoothing of the slope parameter, ensuring that the mutated individuals still meet the manufacturing constraints. The iteration termination condition sets a maximum number of iterations (default 200) and a convergence threshold (when the fitness value change of the population over 20 consecutive generations is ≤0.01). The iteration terminates when either condition is met, and the current optimal parameter scheme is output. After the iteration of selecting and determining the optimal parameter scheme terminates, the engine selects non-dominated solutions from the final population to form an approximate Pareto front for doctors to refer to and select. If multiple non-dominated solutions exist, the engine calculates the comprehensive evaluation index of each solution (based on a weighted summation of weight coefficients), outputs the scheme with the smallest comprehensive evaluation index as the default optimal solution, and retains all non-dominated solutions, allowing doctors to manually adjust them according to clinical preferences.

[0059] In one embodiment, the plan output module is used to output the final executable treatment plan and the corresponding compensator modulation parameters.

[0060] As described above, the planning output module is responsible for integrating and optimizing the results, outputting treatment plans and compensator manufacturing documents that can be directly used for clinical execution. Specifically, these include:

[0061] The output of the integration and standardization module includes two parts: an executable treatment plan and compensator modulation parameters, as detailed below:

[0062] Executable treatment plans include: irradiation field parameters (irradiation angle, radiation energy such as 6MV / 10MV photon beams), physical dose distribution reports of the target area and organs at risk, FLASH bioeffective dose distribution reports, and dose validation indicators (gamma pass rate ≥95%, meeting the clinical treatment plan validation standards). The report format supports PDF export, which is convenient for archiving and clinical review.

[0063] Compensator modulation parameters: Output in a standardized manufacturing file format (STL format), containing the three-dimensional structural data of the compensator, including thickness distribution maps of each region, slope annotations and material recommendations (such as tungsten alloy, lead alloy). The file can be directly imported into 3D printers, CNC machine tools and other processing equipment without secondary format conversion.

[0064] The output verification and feedback module performs final verification of the output content. Verification includes parameter completeness (ensuring no missing parameters), dosage rationality (ensuring the target area and organs at risk receive the required doses), and manufacturing feasibility (ensuring the compensator parameters meet processing requirements). Upon successful verification, the output is displayed to the physician through the system interface and can also be transmitted to the radiotherapy equipment control console and compensator manufacturing workshop via network interface. If verification fails, the module sends error information to the planning optimization engine, prompting re-optimization until the output fully meets clinical and manufacturing requirements.

[0065] In one embodiment, the system further includes a FLASH effect stability controller; the controller is configured to: monitor the spatial distribution of the target dose rate in the candidate plan in real time during the optimization process, and set an elimination mechanism based on the proportion of voxel volumes in the target area that are below the minimum dose rate threshold;

[0066] The FLASH effect stability controller is also configured to perform sensitivity analysis on regions within the target area where the dose rate fluctuates within the minimum dose rate threshold range, and to mark regions with potential FLASH effect failure risk.

[0067] As mentioned above, the system also includes a FLASH effect stability controller to further enhance the safety and clinical suitability of the plan, as detailed below:

[0068] The implementation of the FLASH effect stability controller involves the controller running in real time during the planning optimization process, synchronously monitoring the spatial distribution of the target dose rate in candidate plans, including:

[0069] The elimination mechanism performs real-time statistical analysis of the proportion of voxel volumes (Rlow) below the minimum dose rate threshold within the target area. The preset elimination threshold is 5%. If Rlow ≥ 5%, the candidate plan is directly eliminated, and the reason for elimination is recorded. Sensitivity analysis and risk labeling define the minimum dose rate threshold ± 10% as the critical fluctuation range. The continuous voxel clusters within the range are identified by spatial clustering algorithm, and the coefficient of variation (CV) of the dose rate within the cluster is calculated. If CV ≥ 8%, it is marked as a potential FLASH effect failure risk area, marked with a red dashed box on the image, and fed back to the plan optimization engine for targeted parameter adjustment. The elimination threshold (5%) and sensitivity analysis coefficient of variation threshold (8%) of the FLASH effect stability controller are based on the measured data of 30 different types of radiotherapy equipment (photon / proton). When the elimination threshold is set to 5%, 92% of the FLASH effect failure plans can be filtered out, while 95% of the effective candidate plans are retained. If the threshold is adjusted to 3% or 8%, the optimization efficiency will decrease by 40% or the failure risk will increase by 30%, respectively.

[0070] In one embodiment, the system further includes a biological effect trade-off interface for simultaneously displaying the physical dose distribution and FLASH bioeffective dose distribution of candidate plans to clinicians, and providing clinical outcome probability predictions calculated based on the FLASH bioeffective dose distribution.

[0071] As described above, the implementation of the biological effect trade-off interface adopts a design of dual-distribution overlay visualization + quantified probability panel, including:

[0072] The visualization displays isodose lines of both physical dose distribution and FLASH bioeffective dose distribution simultaneously in the left window. Different dose levels are marked with different colors, and the target area and organs at risk are distinguished by outlines. Mouse hover is supported to view detailed dose data for any voxel. The right panel outputs the probability of tumor control (TCP) and the probability of normal tissue complications (NTCP) for each organ at risk, calculated based on FLASH bioeffective dose. TCP is calculated using the Poisson model, and NTCP uses a dedicated adaptation model for different organs. The probability values ​​are accurate to one decimal place, making it easy for doctors to make intuitive comparisons.

[0073] The biological effect trade-off interface is not a conventional medical data visualization function. The physical dose distribution and the FLASH biological effective dose distribution displayed on the interface are linked in real time. The latter is directly taken from the calculation results of the FLASH biological effect assessment module, rather than an independent static data display. The tumor control probability (TCP) and normal tissue complication probability (NTCP) provided by the interface are calculated based on the BEDFLASH distribution using dedicated models (Poisson model, improved LKB model), rather than general probability prediction tools. The calculation results directly serve the doctor's selection of non-dominated solutions for multi-objective functions, which is a key link in the implementation of individualized treatment plans. The interface supports mouse hover to view BEDFLASH, physical dose, and dose rate data for any voxel. This interactive function is linked with the plan optimization engine. Doctors can mark the areas that need adjustment through the interface, and the system automatically feeds back to the optimization model building module to dynamically adjust the weights of multi-objective functions, realizing a closed loop of visual decision-making → model optimization → plan update, rather than an independent display tool.

[0074] Within the technical framework of the collaborative operation of the aforementioned modules, this system precisely addresses the core pain points of traditional optimization algorithms in the background technology, such as convergence to a single optimal solution, inability to adapt to individualized clinical needs, disconnect between physical dosage and biological effects, and insufficient feasibility of plans in complex scenarios. Through multi-dimensional technological innovation, a closed-loop solution is formed: the data acquisition module provides a precise input foundation for individualized optimization; the FLASH biological effect assessment module solves the industry problem of physical dosage's inability to quantify FLASH biological damage; the multi-objective function and plan optimization engine overcome the limitations of average compromise solutions; and the FLASH effect stability controller and biological effect trade-off interface enhance the practicality of the solution from the perspectives of execution safety and decision-making efficiency, respectively. The modules do not work in isolation but are deeply interconnected through data flow. This not only solves the optimization challenges in complex clinical scenarios such as irregular target areas and multiple organs at risk, but also achieves a comprehensive improvement in the entire process from individualized customization to precise biological optimization, safe and reliable execution, and scientific and efficient decision-making. The resulting technical effects and clinical value represent a targeted breakthrough and upgrade to the shortcomings of existing solutions in the background technology.

[0075] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A system for optimizing ultra-high dose rate radiotherapy planning based on compensator modulation, characterized in that, include: The data acquisition module acquires the patient's medical imaging data and receives clinical objectives, including the target area prescription dose, organ-at-risk dose limit, and the minimum dose rate threshold required to trigger and maintain the FLASH effect. The FLASH bioeffect assessment module calculates the FLASH bioeffective dose distribution corresponding to the physical dose distribution based on the input physical dose distribution, dose rate distribution, and voxel tissue labels, using an embedded bioeffect model. The optimization model construction module constructs a multi-objective function with the FLASH bioeffective dose distribution as the core optimization objective. The multi-objective function includes at least: a first term to evaluate the consistency between the FLASH bioeffective dose and the prescription dose within the target area; a second term to evaluate the organ-at-risk dose based on the FLASH bioeffective dose; and a third term to impose an optimization penalty on voxels within the target area that fail to reach the minimum dose rate threshold. The planning optimization engine optimizes the compensator modulation parameters under compensator manufacturing constraints to minimize the multi-objective function, thereby directly outputting the compensator modulation parameters that satisfy manufacturing feasibility and have the optimal bioeffect. The compensator modulation parameters are the structural parameters of the compensator, including the thickness parameters of each region and the slope parameters of adjacent regions. The system includes a plan output module for outputting the final executable treatment plan and corresponding compensator modulation parameters. It also includes a FLASH effect stability controller, configured to: monitor the spatial distribution of the target dose rate in real time during optimization, and set an elimination mechanism based on the proportion of voxels below the minimum dose rate threshold within the target area. The FLASH effect stability controller is further configured to: perform sensitivity analysis on regions within the target area where the dose rate fluctuates within the minimum dose rate threshold range, mark regions with potential FLASH effect failure risk, and feed the marking results back to the plan optimization engine for targeted adjustment of the compensator modulation parameters.

2. The ultra-high dose rate radiotherapy planning optimization system based on compensator modulation according to claim 1, characterized in that: The biological effect model embedded in the FLASH biological effect assessment module is an improved linear quadratic model. The improved linear quadratic model distinguishes between tumor tissue and normal tissue in the target area based on voxel tissue labels and matches the corresponding α / β parameters according to tissue type. The improved linear quadratic model is constructed by introducing a FLASH correction factor with the minimum dose rate threshold as the boundary.

3. The ultra-high dose rate radiotherapy planning optimization system based on compensator modulation according to claim 1, characterized in that: The system also includes a biological effect trade-off interface, which simultaneously displays the physical dose distribution and FLASH bioeffective dose distribution of candidate plans to clinicians, and provides the tumor control probability and / or normal tissue complication probability calculated based on the FLASH bioeffective dose distribution.

4. The ultra-high dose rate radiotherapy planning optimization system based on compensator modulation according to claim 1, characterized in that: The first term in the multi-objective function used to evaluate the consistency between the target FLASH bioeffective dose and the prescribed dose is a measure of the deviation between the target FLASH bioeffective dose and the prescribed dose.

5. The ultra-high dose rate radiotherapy planning optimization system based on compensator modulation according to claim 1, characterized in that: The second term in the multi-objective function, used to evaluate the dose received by organs at risk based on FLASH bioeffective dose, is the probability of complications in normal tissues or a dose-volume index calculated based on FLASH bioeffective dose.

6. The ultra-high dose rate radiotherapy planning optimization system based on compensator modulation according to claim 1, characterized in that: The third term in the multi-objective function, which is used to apply an optimization penalty to voxels in the target area that fail to reach the minimum dose rate threshold, is a function that applies a weighted penalty to voxels in the target area whose dose rate is lower than the minimum dose rate threshold based on their voxel volume percentage.

7. The ultra-high dose rate radiotherapy planning optimization system based on compensator modulation according to claim 2, characterized in that: The FLASH correction factor is configured to apply a correction calculation to enhance the protective effect on normal tissues when the instantaneous dose rate reaches or exceeds the minimum dose rate threshold.

Images