An illumination scheme generation method, apparatus, device, and storage medium

Abstract

The application discloses a kind of illumination scheme generation method, device, equipment and storage medium.The method comprises: information extraction is carried out to treatment resource parameter and target area prescription parameter, and treatment resource information, target area geometric information and prescription requirement are obtained;The parameter space of illumination scheme is constructed;Each candidate incident direction is carried out single beam parameter optimization, and single direction optimization result set is obtained;Single direction optimization result set is carried out direction combination and dose distribution, and the best combined illumination scheme is obtained;Three-dimensional dose distribution verification is carried out to the best combined illumination scheme, and according to the verification result iterative optimization until meeting clinical acceptance standard, determine as target illumination scheme.The application is improved by the cascade architecture of parameter space construction, single direction optimization and direction combination dose distribution The degree of automation and optimization efficiency of illumination scheme generation are improved, the target area dose distribution precision is improved and normal tissue dose exposure is reduced, and it is suitable for radiotherapy plan design.

Description

Technical Field

[0001] This application relates to the field of radiotherapy technology, and in particular to a method, apparatus, device and storage medium for generating irradiation plans. Background Technology

[0002] Radiation therapy is one of the main treatments for malignant tumors, with more than half of all malignant tumor patients requiring it. With the development of modern radiation therapy technology, precision radiotherapy techniques such as intensity-modulated radiotherapy (IMRT) and volumetric rotating IMRT have become mainstream in clinical practice. These techniques, through precise control of linear accelerators, ensure that the radiation dose distribution closely matches the shape of the tumor target area while minimizing radiation to surrounding normal tissues and organs at risk. The design of the irradiation plan is the core element in achieving precision radiotherapy. Essentially, it involves determining the incident direction, energy parameters, and dose contribution of the radiation beam based on the patient's anatomical structure, ensuring sufficient prescription dose coverage for the target area while controlling the radiation dose to normal tissues within clinically acceptable limits.

[0003] Currently, the generation of irradiation protocols mainly relies on inverse optimization methods within treatment planning systems. Existing methods typically decompose the problem into several independent sub-problems and solve them sequentially. For example, the beam angle is first selected based on experience or heuristic algorithms, and then the beam intensity or weights at the fixed angle are optimized. This step-by-step, sequential approach has significant shortcomings: firstly, there is a strong coupling relationship between beam angle selection and dose allocation, making it difficult to obtain a globally optimal solution using a sequential strategy of fixing the angle first and then optimizing the dose; secondly, the angle selection process is highly dependent on the personal experience of the planner, which is not only time-consuming and labor-intensive but also results in significant differences and inconsistencies in plan quality between different planners and different medical institutions. Furthermore, existing methods often lack a systematic three-dimensional dose distribution verification and iterative optimization mechanism after generating the irradiation protocol, making it difficult to fully guarantee the accuracy of target area dose distribution and the level of protection of normal tissues during the protocol generation stage.

[0004] Therefore, how to automatically generate high-quality irradiation plans that balance target dose coverage and normal tissue protection, and improve the optimization efficiency and planning quality of irradiation plan generation, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] In view of this, the irradiation plan generation method, apparatus, device, and storage medium provided in this application can automatically generate high-quality irradiation plans that balance target area dose coverage and normal tissue protection, thereby improving the optimization efficiency and planning quality of irradiation plan generation. The irradiation plan generation method, apparatus, device, and storage medium provided in this application are implemented as follows: This application provides a method for generating an irradiation scheme, comprising: Information extraction and processing are performed on treatment resource parameters and target area prescription parameters to obtain treatment resource information, target area geometric information and prescription requirements. The treatment resource information includes the range of available electron beam energy parameters, the range of available incident direction parameters and the range of beam dose adjustment. The target area geometric information includes the target area contour and spatial location information. The prescription requirements include the target area expected dose distribution index and the dose constraint index for external tissues of the target area. The treatment resource information, the target area geometry information, and the prescription requirements are processed to construct a parameter space to obtain the parameter space of the irradiation scheme. The parameter space includes the beam pulse number parameter, the energy parameter of each pulse, the incident direction parameter of each pulse, and the dose contribution parameter of each pulse. The candidate incident directions in the parameter space are each subjected to single-beam parameter optimization processing to obtain a single-direction optimization result set covering multiple candidate incident angles. The single-beam parameter optimization is to search for the optimal single-beam parameter combination that makes the target area dose distribution meet the preset conditions in each candidate incident direction. The single-direction optimization result set is subjected to joint optimization processing of direction combination and dose contribution to obtain the optimal combined irradiation scheme. The joint optimization of direction combination and dose contribution involves selecting several incident directions from all candidate incident directions and assigning corresponding dose contributions to each direction. The optimal combination irradiation scheme was subjected to three-dimensional dose distribution verification processing to obtain the verification results; The verification results and the optimal combination irradiation scheme are judged and iterated to obtain the target irradiation scheme. If the verification results do not meet the clinical acceptance criteria, the constraint boundary of the parameter space is adjusted according to the dose deviation information in the verification results, and the single beam parameter optimization processing, the joint optimization processing of direction combination and dose contribution, and the three-dimensional dose distribution verification processing are re-executed until the clinical acceptance criteria are met.

[0006] In some embodiments, the parameter space construction process performed on the treatment resource information, the target area geometry information, and the prescription requirements to obtain the parameter space of the irradiation plan includes: The available energy parameter range, available incident direction parameter range, and beam dose adjustment range in the treatment resource information are subjected to boundary extraction processing to obtain the value boundary information of the optimization variables; Spatial mapping processing is performed on the target area contour and spatial position information in the target area geometry information to obtain the geometric constraint information of the target area in three-dimensional space; The target area desired dose distribution index and the dose constraint index of external tissues in the prescription requirements are quantified to obtain dose optimization target information; The boundary information, geometric constraint information, and dose optimization target information are subjected to variable definition and spatial combination processing to obtain an initial parameter space containing the beam pulse number variable, the pulse energy variable, the pulse incident direction variable, and the pulse dose contribution variable. The initial parameter space is subjected to physical realizability constraint verification to obtain the parameter space of the irradiation scheme.

[0007] In some embodiments, the information extraction processing of treatment resource parameters and target area prescription parameters to obtain treatment resource information, target area geometric information, and prescription requirements includes: The treatment resource parameters are analyzed using equipment capability analysis to obtain the range of available electron beam energy parameters, the range of available incident direction parameters, and the range of beam dose adjustment. The medical image data in the target area prescription parameters are subjected to target area segmentation processing to obtain the target area outline and spatial location information; The clinical prescription data in the target area prescription parameters are subjected to dose target analysis to obtain the target area expected dose distribution index and the dose constraint index on external tissues of the target area. The target region contour and spatial location information are processed for organ at risk identification to obtain the organ at risk contour and its location information, and the organ at risk contour and its location information are used as the associated geometric data of the dose constraint index of the external tissue of the target region. The available electron beam energy parameter range, the available incident direction parameter range, the beam dose adjustment range, the target area contour and spatial location information, the target area expected dose distribution index, the dose constraint index for external tissues of the target area, and the contour and location information of the organs at risk are correlated and integrated to obtain treatment resource information, target area geometric information, and prescription requirements.

[0008] In some embodiments, the step of performing single-beam parameter optimization on each candidate incident direction in the parameter space to obtain a single-direction optimization result set covering multiple candidate incident angles includes: The candidate incident directions are subjected to direction traversal processing to obtain a single incident direction to be optimized; The beam parameters under the single incident direction are subjected to initial parameter setting processing to obtain an initial beam parameter combination, wherein the beam parameters include electron beam energy parameters, beam spot size parameters and source-target distance parameters; The initial beam parameter combination is subjected to a staged search process to obtain the optimal beam parameter combination; The target area dose deposition efficiency index is calculated and processed for the optimal beam parameter combination to obtain the single-direction optimization result under the single incident direction. The optimization results of each single direction corresponding to all candidate incident directions are aggregated to obtain a set of single-direction optimization results covering multiple candidate incident angles.

[0009] In some embodiments, the joint optimization processing of the unidirectional optimization result set with directional combination and dose contribution to obtain the optimal combined irradiation scheme includes: The single-direction optimization results of each candidate incident direction in the single-direction optimization result set are processed to construct a direction combination space, thereby obtaining a direction combination search space; The dose allocation variable space is obtained by defining dose allocation variables for each direction combination in the direction combination search space. The optimal dose allocation scheme and the corresponding combined dose distribution index under each directional combination are obtained by jointly optimizing and solving the directional combination search space and the dose contribution allocation variable space. The optimal dose allocation scheme and the combined dose distribution index are comprehensively evaluated and sorted to obtain the best combined irradiation scheme.

[0010] In some embodiments, the three-dimensional dose distribution verification process for the optimal combined irradiation scheme to obtain the verification results includes: The optimal combination irradiation scheme is subjected to three-dimensional dose calculation processing to obtain the three-dimensional dose distribution data corresponding to the optimal combination irradiation scheme. The three-dimensional dose distribution data is compared with the target area expected dose distribution index in the prescription requirements on a voxel-by-voxel basis to obtain the target area dose distribution deviation data. The three-dimensional dose distribution data is compared with the dose constraint index for external tissues in the prescription requirements on a voxel-by-voxel basis to obtain dose distribution deviation data for external tissues. The dose distribution deviation data of the target area and the dose distribution deviation data of the tissues outside the target area are statistically processed to obtain the verification results.

[0011] An irradiation scheme generation apparatus provided in this application includes: The extraction module is used to extract and process information from treatment resource parameters and target area prescription parameters to obtain treatment resource information, target area geometric information and prescription requirements. The treatment resource information includes the range of available electron beam energy parameters, the range of available incident direction parameters and the range of beam dose adjustment. The target area geometric information includes the target area contour and spatial location information. The prescription requirements include the target area expected dose distribution index and the dose constraint index for external tissues of the target area. The processing module is used to perform parameter space construction processing on the treatment resource information, the target area geometry information and the prescription requirements to obtain the parameter space of the irradiation plan. The parameter space includes the beam pulse number parameter, the energy parameter of each pulse, the incident direction parameter of each pulse and the dose contribution parameter of each pulse. The processing module is also used to perform single-beam parameter optimization processing on the candidate incident directions in the parameter space to obtain a single-direction optimization result set covering multiple candidate incident angles. The single-beam parameter optimization is to search for the optimal single-beam parameter combination that makes the target area dose distribution meet the preset conditions in each candidate incident direction. The processing module is also used to perform joint optimization processing of direction combination and dose contribution on the single-direction optimization result set to obtain the best combined irradiation scheme. The joint optimization of direction combination and dose contribution is to select several incident directions from all candidate incident directions and assign corresponding dose contributions to each direction. The processing module is also used to perform three-dimensional dose distribution verification processing on the optimal combination irradiation scheme to obtain verification results; The determination module is used to judge and iteratively process the verification results and the optimal combination irradiation scheme to obtain the target irradiation scheme. In the case that the verification results do not meet the clinical acceptance criteria, the constraint boundary of the parameter space is adjusted according to the dose deviation information in the verification results, and the single beam parameter optimization processing, the joint optimization processing of direction combination and dose contribution, and the three-dimensional dose distribution verification processing are re-executed until the clinical acceptance criteria are met.

[0012] The computer device provided in this application includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the method described in this application.

[0013] The computer-readable storage medium provided in this application embodiment stores a computer program thereon, which, when executed by a processor, implements the method provided in this application embodiment.

[0014] This application provides an irradiation plan generation method, apparatus, device, and storage medium. By extracting information from treatment resource parameters and target area prescription parameters, treatment resource information, target area geometric information, and prescription requirements are obtained. A parameter space for the irradiation plan is constructed. Single-beam parameters are optimized for each candidate incident direction to obtain a single-direction optimization result set. The single-direction optimization result set is then combined with directions and dose allocation to obtain the optimal combined irradiation plan. The optimal combined irradiation plan is verified for three-dimensional dose distribution, and iterative optimization is performed based on the verification results until clinical acceptance criteria are met, at which point it is determined as the target irradiation plan. This allows for the automated generation of high-quality irradiation plans that balance target area dose coverage and normal tissue protection, improving the optimization efficiency and planning quality of irradiation plan generation, and solving the technical problems mentioned in the background art. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A schematic diagram illustrating the implementation flow of an irradiation scheme generation method provided in an embodiment of this application; Figure 2 This is a schematic diagram of an irradiation scheme generation device provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0018] The following description of some technologies involved in the embodiments of this application is provided to aid understanding and should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and conciseness, some descriptions of well-known functions and structures are omitted in the following description.

[0019] Figure 1 This is a schematic flowchart illustrating the implementation of an irradiation scheme generation method provided in an embodiment of this application, including steps 101 to 106. Wherein, Figure 1This is merely one execution order shown in the embodiments of this application and does not represent the only execution order for an irradiation scheme generation method. Where the final result can be achieved, Figure 1 The steps shown can be performed in parallel or in reverse order.

[0020] Step 101: Extract and process information from the treatment resource parameters and target area prescription parameters to obtain treatment resource information, target area geometric information, and prescription requirements.

[0021] In this embodiment, treatment resource parameters refer to capability description data of radiotherapy equipment (such as linear accelerators or laser tail field accelerators), which are derived from equipment parameter configuration files or equipment parameters input by the user through a graphical interface. Target prescription parameters refer to medical imaging data and dose prescription data derived from clinical prescription systems, typically stored in DICOM (Digital Imaging and Communications in Medicine) format, containing images of the patient's anatomical structures and treatment goals set by the physician.

[0022] In this step, information extraction and processing of treatment resource parameters are performed. The purpose is to transform the raw equipment parameters and prescription data into structured, standardized information that can be directly used by subsequent optimization algorithms, providing a data foundation for the construction of the parameter space. Since different accelerator models have different energy output capabilities, gantry rotation ranges, and dose adjustment accuracies, and prescription formats vary among different medical institutions, this step requires unified parsing and formatting to eliminate data heterogeneity.

[0023] Specifically, the treatment resource parameters are analyzed using equipment capabilities to obtain the range of available electron beam energy parameters, the range of available incident direction parameters, and the range of beam dose adjustment; additionally, the laser tail field accelerator's specific hardware constraints are analyzed: beam repetition frequency range, minimum pulse time interval, and pulse timing programmable control permissions, which serve as the underlying boundaries for timing parameter optimization.

[0024] Equipment capability analysis refers to retrieving the minimum and maximum output energy values ​​(or a list of available energy channels) of the accelerator from the equipment parameter configuration file to obtain the range of available electron beam energy parameters. For example, for a linear accelerator equipped with six energy levels (6 MeV, 9 MeV, 12 MeV, 16 MeV, 20 MeV, and 25 MeV), the analyzed range of available energy parameters is (6, 9, 12, 16, 20, 25) MeV; for a laser tail field accelerator, the continuously adjustable energy range of 15–50 MeV can be obtained. The range of allowable gantry rotation angles (typically 0°–360°) and treatment bed rotation angles (typically -90°–90°), or a list of incident angles achievable by the deflecting magnets, is then retrieved to obtain the range of available incident direction parameters. The beam intensity adjustment level or continuous adjustment range (e.g., single-pulse dose 0.01–0.5 Gy) of the device is read to obtain the beam dose adjustment range; simultaneously, hardware timing constraints are extracted: minimum pulse interval 0.1 ms, repetition frequency 1–10 kHz, and maximum continuous pulse duration per treatment, as the basis for subsequent FLASH timing constraints. Optionally, for laser tail field accelerators, the beam repetition frequency range (e.g., 1–10 kHz) and minimum pulse time interval limit (e.g., 0.1 ms) can also be read as supplementary constraints for subsequent parameter space.

[0025] The medical image data in the target area prescription parameters are segmented to obtain the target area contour and spatial location information.

[0026] Target region segmentation refers to separating voxels belonging to the target region from the background in medical images and extracting the precise geometric boundary of the target region. This processing can be carried out by any one or more combinations of the following methods: (1) a segmentation method based on grayscale threshold, which extracts the target region by setting the upper and lower limits of CT values; (2) a segmentation method based on region growing, which expands from the seed point to adjacent voxels and includes voxels with similar grayscale features into the target region; (3) a segmentation method based on active contour model, which converges to the target region boundary through deformable curves driven by the image energy function; (4) a segmentation network based on deep learning, which inputs CT (Computed Tomography) or MR (Magnetic Resonance Imaging) images into a pre-trained U-Net or nnUNet segmentation model, outputs a probability map of each voxel belonging to the target region, and then obtains the precise contour of the target region through threshold processing (e.g., probability ≥ 0.5) and connected component analysis.

[0027] Taking a deep learning-based segmentation method as an example, the processing flow is as follows: First, the CT image is resampled to an isotropic resolution (e.g., 1mm×1mm×1mm) and grayscale normalization is performed (the CT value is truncated to the range of [-200, 200] HU and normalized to the interval [0, 1]); then, the normalized image is input into the trained nnUNet segmentation network, which adopts an encoder-decoder architecture. The encoder extracts multi-scale features through convolution and pooling operations, and the decoder restores the spatial resolution and outputs pixel-level classification results through upsampling and skip connections; finally, the connected component analysis is performed on the probability map output by the network, the largest connected component is retained as the target area, isolated false positive areas are removed, and the accurate contour of the target area is obtained.

[0028] After segmentation, the target region contour is mapped from the pixel coordinate system of the medical image to a three-dimensional spatial coordinate system (such as a patient coordinate system with the isocenter as the origin), obtaining the spatial location information of the target region, including the geometric center coordinates of the target region (e.g., x=0.5mm, y=12.3mm, z=-3.8mm), volume (e.g., 35.6cm³), and three-dimensional spatial extent (i.e., the minimum and maximum values ​​of the target region on each coordinate axis). The mapping process is completed using parameters such as pixel spacing and image position patient stored in the DICOM header file of the medical image.

[0029] The clinical prescription data in the target area prescription parameters are analyzed for dose targets to obtain the expected dose distribution index of the target area and the dose constraint index of the external tissues of the target area. For the FLASH radiotherapy scenario, additional hard timing constraints within the prescription are analyzed: the lower limit of instantaneous dose rate (≥40Gy / s), the upper limit of total irradiation duration, and the pulse timing arrangement requirements, which serve as the basis for timing optimization and dose rate verification.

[0030] Dosage target parsing refers to extracting numerical dosage targets and constraints from clinical prescription data. Specifically, clinical prescription data typically exists in the form of structured prescription forms or unstructured text. For structured forms, the values ​​of each field are directly read; for unstructured text, the dosage values ​​and constraints are parsed using natural language processing.

[0031] The target volume expected dose distribution indicators obtained after analysis include: minimum prescription dose of the target volume (e.g., D95 ≥ 95% of the prescription dose, meaning at least 95% of the target volume receives irradiation of no less than 95% of the prescription dose), maximum permissible dose of the target volume (e.g., Dmax ≤ 115% of the prescription dose), target volume dose uniformity requirements (e.g., uniformity index ≤ 0.1), and target volume dose conformity requirements (e.g., conformity index ≥ 0.7), etc.

[0032] The dose constraint indicators for external tissues obtained after analysis include: maximum dose limits for each organ at risk (e.g., spinal cord Dmax≤5Gy, brainstem Dmax≤6Gy), average dose limits (e.g., average dose of bilateral parotid glands ≤3Gy, average dose of whole lung ≤2Gy), and dose-volume constraints (e.g., V20≤30%, meaning no more than 20% of the organ volume receives a dose ≥30Gy).

[0033] FLASH-specific timing constraint analysis content: instantaneous dose rate threshold, total irradiation duration limit, and pulse minimum interval limit are simultaneously incorporated into the prescription constraint system.

[0034] The target area contour and spatial location information are processed to identify organs at risk, and the contours and location information of organs at risk are obtained. The contours and location information of organs at risk are then used as the associated geometric data of dose constraint indicators for tissues outside the target area.

[0035] The identification and processing of organs at risk can employ automatic or semi-automatic segmentation methods. Taking automatic segmentation as an example, the CT image is input into a pre-trained multi-organ segmentation model (such as nnUNet or TotalSegmentator). This model outputs segmentation results for multiple organs, from which organs at risk (such as the spinal cord, brainstem, optic nerve, parotid gland, lung, heart, etc.) that are spatially adjacent to the target area are selected, obtaining the contour and spatial location information of each organ at risk. After segmentation, the contours of each organ at risk are associated and bound with the corresponding dose constraint indices obtained in step 101. For example, the spinal cord contour is associated with the constraint "spinal cord Dmax≤5Gy" to form a complete dose constraint-geometric data pair.

[0036] Finally, the available electron beam energy parameter range, available incident direction parameter range, beam dose adjustment range, timing hardware constraints, target area contour and spatial location information, target area expected dose distribution index, dose constraint index for external tissues of the target area, FLASH timing dose rate constraints, and the contour and location information of organs at risk are correlated and integrated. The data is then stored in a structured format according to a unified data format to obtain treatment resource information, target area geometric information, and prescription requirements.

[0037] Data association and integration processing includes: establishing spatial location relationships between the target area and organs at risk (such as calculating the shortest distance and overlap volume between the target area and each organ at risk); binding the dose constraint indicators of each organ at risk to its corresponding geometric contour data; and binding FLASH timing and instantaneous dose rate constraints to the laser accelerator timing hardware boundaries to form a complete prescription constraint data structure. The structured data format can be JSON, XML, or a dedicated binary format, and must include at least resource information fields (equipment type, energy range, orientation range, dose adjustment range, pulse timing hardware boundaries), geometric information fields (contour point sets or mask data of the target area and organs at risk), and prescription information fields (dose target, spatial dose constraints, and a list of FLASH timing dose rate constraints). This standardized output provides directly readable input for the subsequent construction of the parameter space.

[0038] Step 102: Perform parameter space construction processing on the treatment resource information, target area geometry information, and prescription requirements to obtain the parameter space of the irradiation plan.

[0039] In this embodiment, the parameter space refers to the domain of all optimization variables in the subsequent optimization solution. It is a multi-dimensional search space that includes parameters such as the number of beam pulses, the energy parameters of each pulse, the incident direction parameters of each pulse, the dose contribution parameters of each pulse, and the timing arrangement parameters of each pulse. The core innovation of constructing the parameter space is to simultaneously incorporate pulse timing variables to meet the rigid timing control requirements of FLASH radiotherapy and fully cover the pulse timing parameter optimization chain. The purpose of constructing the parameter space is to transform clinical problems and equipment capabilities into a structured mathematical optimization problem, clarifying the value range and physical constraint boundaries of all adjustable variables. For laser tail field accelerator scenarios, since each pulse can be independently programmed, the parameter space also needs to include the timing arrangement parameters of each pulse to meet the instantaneous dose rate requirements of FLASH radiotherapy.

[0040] Specifically, boundary extraction processing is performed on the range of available energy parameters, the range of available incident direction parameters, the range of beam dose adjustment, and the hardware boundary of pulse timing in the treatment resource information to obtain the value boundary information of the optimization variables; the timing constraint boundary of FLASH prescription is extracted simultaneously as a time-series variable-specific constraint.

[0041] Boundary extraction processing refers to directly reading or calculating the value range of each optimization variable from the structured data output in step 101. Specifically, this includes: extracting the minimum and maximum values ​​of energy (e.g., 15 MeV and 50 MeV) or a list of available energy channels (e.g., 6, 9, 12, 16, 20, 25) MeV from the range of available energy parameters, as the value boundaries of the energy optimization variables; and extracting the minimum and maximum values ​​of the gantry angle (e.g., 0° and 360°) and the treatment bed angle (e.g., -90° and 90°) from the range of available incident direction parameters, or a list of available incident directions, as the value boundaries of the incident direction optimization variables. For the case using a discrete direction list, the discrete variables numbered 1 to N for each direction can have value boundaries of {1, 2, ..., N}. Extract the minimum and maximum values ​​of dose adjustment (e.g., 0.01 Gy and 0.5 Gy) from the beam dose adjustment range as the value boundaries of the dose contribution optimization variables; extract the pulse timing variable boundaries from the timing hardware constraints: minimum pulse interval 0.1 ms, single pulse timestamp value range, and upper limit of total pulse duration; and the timing constraint boundary of synchronously loading FLASH instantaneous dose rate ≥ 40 Gy / s.

[0042] Spatial mapping processing is performed on the target area contour and spatial location information in the target area geometry information to obtain the geometric constraint information of the target area in three-dimensional space.

[0043] Spatial mapping processing includes: transforming the target area contour from the image coordinate system to the planning coordinate system (such as a coordinate system with the isocenter as the origin), calculating the spatial range of the target area (minimum / maximum coordinate values, such as x∈[-15mm, 25mm], y∈[-20mm, 30mm], z∈[-10mm, 20mm]), volume (such as 35.6cm³), and spatial positional relationship between the target area and each organ at risk (such as the shortest distance between the target area and the spinal cord is 5mm, and the overlap volume with the left parotid gland is 0cm³), and generating the geometric constraint information required for subsequent dose calculation and optimization.

[0044] The target area desired dose distribution index, the dose constraint index of external tissues in the prescription requirements, and the FLASH time-series dose rate constraint are quantified to obtain dose optimization target information.

[0045] The quantification process refers to transforming the dose targets and constraints described in the clinical prescription text into a numerical objective function that can be calculated by the optimization algorithm. Specifically, for the target area expected dose distribution index, it is transformed into a target area dose penalty term in the optimization objective function. The basic rule is: when the calculated dose of the target area voxel is lower than the prescription requirement dose, a positive penalty is applied, and the larger the deviation, the heavier the penalty; when the target area voxel dose reaches or exceeds the prescription requirement, no penalty is applied. For FLASH timing constraints, a timing penalty term is constructed: timing penalties are applied when the instantaneous dose rate is lower than 40 Gy / s, the pulse interval is less than the hardware minimum threshold, and the total irradiation duration exceeds the prescription upper limit, achieving coordinated optimization of spatial dose and timing dose rate. For the dose constraint index of external tissues in the target area, it is transformed into a normal tissue dose penalty term in the optimization objective function. The basic rule is: when the calculated dose of the organ at risk voxel exceeds its corresponding dose constraint limit, a positive penalty is applied, and the larger the exceedance, the heavier the penalty; when the voxel dose is lower than the constraint limit, no penalty is applied. Through the above transformation method, the qualitative dosage description in clinical prescription requirements is transformed into a numerically computable optimization objective function, so that subsequent optimization algorithms can be numerically solved based on this objective function.

[0046] The boundary information, geometric constraint information, dose optimization target information, and time sequence optimization target information are defined and spatially combined to obtain an initial parameter space containing variables such as the number of beam pulses, the energy of each pulse, the incident direction of each pulse, the dose contribution of each pulse, and the time sequence arrangement of each pulse.

[0047] The variable definition and spatial combination processing includes: defining the beam pulse quantity variable N as an integer variable, with its value range set according to equipment capabilities and clinical needs (e.g., 30≤N≤300). Defining each pulse energy variable E_j (j=1, 2, ..., N) as either a discrete variable (selected from the list of available energy channels) or a continuous variable (taking values ​​within the available energy range). Defining each pulse incident direction variable D_j as a discrete variable (selected from a set of candidate directions, such as from 36 discrete directions). Defining each pulse dose contribution variable W_j as a continuous variable, with its value boundary being the beam dose adjustment range (e.g., 0.01≤W_j≤0.5Gy). Adding a new definition for each pulse timing arrangement variable (representing the pulse emission timestamp), whose value is limited by the beam repetition frequency (e.g., 1~10kHz corresponding to a time interval of 0.1~1ms) and the FLASH instantaneous dose rate threshold constraint. Combining all variables according to the data structure of "pulse number-energy-direction-dose" to form an N×5 dimensional initial parameter space.

[0048] Optionally, for the laser tail field accelerator scenario, a timing arrangement variable T_j for each pulse (representing the emission timestamp of the j-th pulse) is also defined, the value of which is limited by the beam repetition frequency (e.g., 1 to 10 kHz corresponds to a time interval of 0.1 to 1 ms).

[0049] The initial parameter space is subjected to physical realizability constraint verification to obtain the parameter space of the irradiation scheme.

[0050] The physical feasibility constraint verification process refers to checking whether each combination of variables in the initial parameter space meets the physical limitations of the equipment and the hard constraints of FLASH timing. The verification content includes, but is not limited to: (1) whether the energy parameters of each pulse are within the energy range that the equipment can provide; (2) whether the incident direction parameters of each pulse are within the angle range allowed by the equipment; (3) whether the dose contribution parameters of each pulse are within the dose adjustment range of the equipment; (4) whether the number of beam pulses does not exceed the maximum number of pulses in a single irradiation of the equipment; (5) timing-specific verification: whether the time interval between each pulse is not less than the minimum pulse interval of the equipment (e.g., 0.1ms), whether the total irradiation time does not exceed the maximum continuous operating time allowed by the equipment, and whether the instantaneous dose rate at a single moment meets the requirement of FLASH≥40Gy / s. After eliminating the combination of variables that does not meet any of the above physical / timing constraints, the irradiation scheme parameter space with complete coverage of spatial dose and FLASH timing is finally obtained.

[0051] Step 103: Perform single-beam parameter optimization on the candidate incident directions in the parameter space to obtain a single-direction optimization result set.

[0052] In this embodiment, single-beam parameter optimization refers to the process of searching for the optimal combination of single-beam parameters that satisfies preset conditions for the dose distribution in the target area, while fixing a single candidate incident direction. The candidate incident direction is generated from the incident direction parameters in the parameter space according to a preset discretization step size. The preset conditions include maximizing the target area dose deposition efficiency or minimizing the dose outside the target area.

[0053] The purpose of this step is to pre-calculate the "optimal single-direction performance index of each direction," the fundamental unit for subsequent cross-directional combination optimization, thereby reducing the dimensionality of the direction combination problem to a discrete selection problem. Its necessity lies in the fact that without pre-calculating the single-direction performance of each direction, the beam parameters in each direction need to be re-optimized for each direction combination scheme evaluated in subsequent direction combination optimization, leading to a surge in computational complexity (number of combinations × number of optimization times for each direction parameter). By pre-optimizing the single-direction parameters in this step, subsequent direction combination optimization only needs to read the pre-set optimal results for each direction and assign dose weights, significantly reducing computational complexity.

[0054] Another key aspect of this step is the adoption of a fast-slow graded dose calculation architecture. The coarse-grained search stage calls the fast analytical pen beam model, while the high-precision Monte Carlo engine is only invoked in the final fine-tuning stage. This hierarchical matching of computational accuracy requirements further reduces computational consumption. Since the number of candidate incident directions is typically large (36–72 in a typical scenario of this application), directly performing high-precision Monte Carlo simulation optimization for each direction would result in unacceptable computational costs. Therefore, this application employs a staged search strategy: a coarse search quickly locates the optimal parameter range, followed by a fine search for refined optimization, thereby significantly reducing the number of dose calculation engine calls while maintaining optimization accuracy.

[0055] Specifically, the candidate incident directions are traversed to obtain a single incident direction to be optimized.

[0056] The direction traversal process includes: sequentially selecting one direction from the candidate incident direction set as the single incident direction to be optimized in a preset order (such as from smallest to largest rack angle, or a random order), until all candidate directions have been processed. During the traversal, the optimization result of each direction is recorded independently and does not affect each other, so this traversal process can be executed in parallel to further improve efficiency.

[0057] The initial parameters of the beam under a single incident direction are set to obtain the initial beam parameter combination, which includes electron beam energy parameters, beam spot size parameters, and source-target distance parameters.

[0058] The initial parameter setting process can be carried out in any of the following ways: (1) Set the energy parameter to the middle value of the available energy range (e.g., 32.5 MeV when the available range is 15-50 MeV), set the beam spot size parameter to the default value of the device (e.g., 3 mm), and set the source-target distance parameter to the standard treatment distance of the device (e.g., 100 cm); (2) Estimate based on the geometric characteristics of the target area, estimate the initial energy value matching it based on the center depth of the target area, and estimate the initial beam spot size based on the lateral size of the target area; (3) Set the initial value based on historical experience data or prior knowledge. In this embodiment, the first method is preferred because it is simple, universal and does not require additional calculation. The initial beam parameter combination is searched in stages to obtain the optimal beam parameter combination.

[0059] The phased search process leverages the physical characteristic that the target area dose deposition efficiency varies with beam parameters (such as energy and beam spot size) in a single-peak or quasi-single-peak pattern. By sequentially executing multiple search phases, it reduces the number of calls to the dose calculation engine.

[0060] Specifically, the phased search process includes the following three progressive steps: The first stage is the coarse search stage. It traverses the parameter space with a large parameter step size to quickly locate the parameter range that maximizes the dose deposition efficiency in the target area. For example, it traverses the available energy range of 15–50 MeV with a step size of 5 MeV (i.e., calculating the dose deposition efficiency at 8 energy points: 15, 20, 25, 30, 35, 40, 45, and 50 MeV sequentially), and traverses the beam size range of 2–5 mm with a step size of 1 mm (i.e., 2, 3, 4, and 5 mm, a total of 4 beam size points), totaling approximately 8 × 4 = 32 dose simulations. This quickly determines the energy range (e.g., 30–40 MeV) and beam size range (e.g., 2.5–3.5 mm) with the highest efficiency. This stage does not require high precision; it is only used to narrow down the subsequent search range.

[0061] The second stage is the fine-search stage. Within the optimal parameter range determined by the coarse search, a fine-search is performed with smaller parameter step sizes. For example, the energy is finely searched in the range of 30–40 MeV with a step size of 1 MeV (a total of 11 energy points), and the beam size is finely searched in the range of 2.5–3.5 mm with a step size of 0.3 mm (a total of 4 beam size points). A total of approximately 11 × 4 = 44 dose simulations are performed to simultaneously optimize the energy and beam size parameters, gradually approaching the optimal parameter combination.

[0062] The third stage is the source-target distance fine-tuning stage. Based on the optimal energy and beam size determined in the first two stages, the source-target distance parameters are fine-tuned with a small step size (e.g., 0.5 cm), with a search range of 95–105 cm (a total of 21 step size points), requiring approximately 21 dose simulations to further optimize the dose deposition efficiency.

[0063] The total number of dose simulations across the three phases is approximately 32 + 44 + 21 = 97. Compared to performing a dense mesh search across the entire parameter space (e.g., 15–50 MeV with a 1 MeV step, 2–5 mm with a 0.3 mm step, and 95–105 cm with a 0.5 cm step, totaling 36 × 11 × 21 = 8316 simulations), the phased search reduces the number of simulations by approximately 98.8%. It should be noted that the above figures are for illustrative purposes only; the actual number of simulations depends on the size of the parameter space, but the reduction achieved by the phased search is significant.

[0064] The dose calculation engine can employ Monte Carlo simulation programs (such as Geant4, EGSnrc, etc.), analytical dose calculation algorithms (such as the pen beam model), or fast dose prediction models based on machine learning. This embodiment preferably uses Monte Carlo simulation programs because they offer the highest accuracy and are considered the "gold standard" for radiotherapy dose calculation.

[0065] The target area dose deposition efficiency index is calculated and processed for the optimal beam parameter combination to obtain the single-direction optimization result under a single incident direction.

[0066] The target area dose deposition efficiency index is used to measure the deposition efficiency of beam energy in the target area under the incident direction. It can be any one or more of the following indices: (1) the ratio of the deposited dose in the target area to the total incident dose. The higher the value, the higher the energy utilization efficiency; (2) the ratio of the average dose in the target area to the maximum dose. The closer the value is to 1, the more uniform the dose distribution in the target area; (3) the ratio of the dose in the target area to the dose outside the target area. The higher the value, the better the target selectivity. In this embodiment, index (1) is preferred as the main evaluation index. The single-direction optimization results include at least: the optimal beam parameter combination (optimal energy value E_opt, optimal beam spot size σ_opt, optimal source-target distance SAD_opt) in the incident direction, the corresponding target area dose deposition efficiency index η, the corresponding dose distribution overview outside the target area, and the reference pulse timing interval of the FLASH timing adapted to the direction.

[0067] The optimization results for each of the candidate incident directions are aggregated to obtain a set of single-direction optimization results.

[0068] The aggregation process includes summarizing the optimization results of all candidate incident directions according to a unified data format of "direction number-direction angle-optimal parameter-efficiency index" to form a structured dataset for subsequent cross-direction combined optimization. For example, for 36 candidate directions, the single-direction optimization result set contains 36 records, each containing fields such as direction index, gantry angle, treatment bed angle, optimal energy, optimal beam size, optimal source-target distance, dose deposition efficiency, and out-of-target dose profile.

[0069] Step 104: Perform directional combination and dose allocation processing on the single-direction optimization result set to obtain the optimal combined irradiation scheme.

[0070] In this embodiment, direction combination and dose allocation refer to selecting several incident directions from all candidate incident directions and assigning corresponding dose contributions to each direction to form an optimal combination irradiation scheme that includes multiple directions and their corresponding dose contributions. In this embodiment, direction selection and dose allocation are solved collaboratively in the same optimization process, not sequentially. This step needs to solve two coupled sub-problems: first, "which directions to select," and second, "how much dose each direction contributes." Since there is a strong coupling relationship between direction selection and dose allocation (the optimal dose allocation differs under different direction combinations), this application treats them as a joint optimization problem for unified solution.

[0071] Specifically, the single-direction optimization results of each candidate incident direction in the single-direction optimization result set are processed to construct a direction combination space, thus obtaining the direction combination search space.

[0072] The direction combination space consists of all possible combinations formed by selecting a number of directions from all candidate incident directions. The number of directions K in the combination can be preset according to clinical needs (e.g., selecting 3 to 8 directions), or it can be automatically determined as an optimization variable during the solution process. When the number of candidate directions is M and the number of selected directions is K, the total number of direction combinations is C(M, K). When M=72 and K=5, C(72, 5)≈13 million combinations, which is a huge number and requires the use of intelligent optimization algorithms for efficient search.

[0073] The dose allocation variable space is obtained by defining dose allocation variables for each directional combination in the directional combination search space.

[0074] For each directional combination, a dose contribution variable w_k (k=1, 2, ..., K) is defined for each incident direction in the combination. This variable represents the dose weight or dose value contributed by that direction to the total irradiation dose. Simultaneously, pulse timing allocation variables for each direction are defined, matching the timing reference interval of the single-direction optimization output in step 103. The value space of all dose contribution variables constitutes the dose contribution allocation variable space corresponding to that directional combination. The value range of each variable is the beam dose adjustment range (e.g., 0.01≤w_k≤0.5Gy), and the sum of the dose contributions from all directions equals the total prescribed dose.

[0075] The optimal dose allocation scheme and corresponding combined dose distribution index under each directional combination are obtained by jointly optimizing the directional combination search space and the dose contribution allocation variable space.

[0076] The joint optimization solution process employs intelligent optimization algorithms, including but not limited to simulated annealing, genetic algorithms, particle swarm optimization, or Bayesian optimization.

[0077] Taking the simulated annealing algorithm as an example, its execution process is as follows: (1) Initialization: Set the initial temperature, termination temperature, temperature decay coefficient, and number of iterations at each temperature. Randomly generate an initial direction combination (randomly select a preset number of directions from all candidate directions) and randomly assign an initial dose contribution that meets the total dose requirement to each direction as the current solution.

[0078] (2) Perturbation to generate new solutions: At the current temperature, the current solution is randomly perturbed to generate new solutions. The perturbation methods include three equally probable operations: ① Replace direction: Randomly replace one direction in the combination with the unselected direction; ② Add or remove direction: Randomly add or remove one direction (within the preset minimum and maximum number of directions); ③ Adjust dose: Randomly select one direction to adjust its dose contribution, and adjust other directions accordingly to ensure the total dose is conserved; ④ Adjust timing: Fine-tune the pulse timestamp within the optimal timing interval of a single direction to verify the instantaneous dose rate.

[0079] (3) Calculate the overall score: Call the dose calculation engine (which can use a pen beam model or fast Monte Carlo) to calculate the dose distribution of the new solution, and calculate the overall score according to the preset evaluation criteria. The evaluation criteria include at least: spatial distribution characteristics between directions (such as minimum angle between directions, uniformity of three-dimensional spatial coverage), single-beam dose deposition efficiency in each direction (taken from the single-direction optimization results in step 103), and normal tissue burden on the beam path. The overall score is obtained by weighting and summing the scores of each sub-item according to their corresponding weight ratios.

[0080] (4) Acceptance decision: Compare the comprehensive scores of the new solution and the current solution. If the score of the new solution is better than that of the current solution, the new solution is accepted; if the score of the new solution is worse than that of the current solution, the worse solution is accepted with a certain probability. This probability decreases as the temperature decreases, thus allowing more worse solutions to escape local optima in the early stage of the search, and gradually tending to local optimization in the later stage of the search.

[0081] (5) Cooling: After completing all iterations at the current temperature, the temperature is reduced according to the preset attenuation coefficient.

[0082] (6) Termination judgment: Triggering dual termination conditions for iteration (core computing power optimization logic): ① Reaching the preset maximum number of iterations; ② The fluctuation of the comprehensive score in N consecutive iterations is less than the convergence tolerance (e.g., the change in comprehensive score is <0.5%). If either condition is met, the search will be terminated and the current optimal solution will be output.

[0083] By jointly optimizing the solution, the optimal dose distribution scheme under each directional combination and the corresponding combined dose distribution index (such as target area D95, V95, homogeneity index, Dmax of each organ at risk, etc.) are obtained.

[0084] The weight allocation is based on the normalized index of the optimization results in each direction, and the time allocation is based on the dynamic adjustment of the pulse interval according to the FLASH instantaneous dose rate constraint, so that the superimposed dose distribution of the combined scheme is as close as possible to the prescription requirements.

[0085] The optimal dose allocation scheme and combined dose distribution index are comprehensively evaluated and ranked to obtain the best combined irradiation scheme.

[0086] Specifically, each directional combination is comprehensively scored based on the combined dose distribution index, the degree of FLASH timing compliance, and the degree of matching with the prescription requirements. The scoring function comprehensively considers multiple dimensions such as target coverage, target homogeneity, and protection of organs at risk. The directional combination with the highest comprehensive score and its corresponding dose allocation scheme are selected as the optimal combined irradiation scheme, ranked from highest to lowest. The optimal combined irradiation scheme includes at least the energy parameters, incident direction parameters, dose contribution parameters, and pulse timing arrangement parameters for each pulse.

[0087] Step 105: Perform three-dimensional dose distribution verification processing on the optimal combination irradiation scheme to obtain the verification results.

[0088] In this embodiment of the application, three-dimensional dose distribution verification refers to performing independent dosimetric and FLASH timing dual verification on the optimal combination irradiation scheme generated in step 104, simultaneously completing three-dimensional dose distribution verification and instantaneous dose rate synchronous verification, forming a complete FLASH timing parameter optimization and dose rate verification closed-loop capability.

[0089] The necessity of this step lies in the fact that the dose calculation engine called during the joint optimization process in step 104 may employ approximate algorithms (such as the pencil beam model or fast Monte Carlo) to ensure optimization efficiency. This step, however, uses a high-precision dose calculation method for final verification, ensuring the consistency of the verification results with actual clinical practice. Through this hierarchical strategy of "low-precision rapid optimization + high-precision accurate verification," both optimization efficiency and verification accuracy are guaranteed.

[0090] Specifically, the optimal combination irradiation scheme is processed by three-dimensional dose calculation to obtain the three-dimensional dose distribution data and pulse-by-pulse instantaneous dose rate sequence corresponding to the optimal combination irradiation scheme.

[0091] Three-dimensional dose calculation is performed using Monte Carlo simulation methods (such as Geant4) in three-dimensional voxel space. The voxel grid size is set according to clinical accuracy requirements, typically 2mm×2mm×2mm or 1.5mm×1.5mm×1.5mm. The statistical error of the Monte Carlo simulation is controlled within 1% to ensure the clinical reliability of the calculation results. The calculation input is the optimal combination irradiation scheme output from step 104 (including the energy, incident direction, dose contribution parameters, and timing parameters of each pulse), and the output is uniform three-dimensional dose grid data covering the entire target area and organs at risk, with each grid point storing the absolute dose value (in Gy) at that location.

[0092] For laser tail field accelerator scenarios, three-dimensional dose calculation also includes instantaneous dose rate calculation, that is, based on the timing arrangement parameters and dose contribution parameters of each pulse, the instantaneous dose rate value (unit Gy / s) at each moment is calculated to determine whether it meets the FLASH dose rate requirement (≥40Gy / s), and the instantaneous dose rate synchronous verification is completed, thus opening up the FLASH timing optimization-verification closed loop.

[0093] Voxel-by-voxel comparison was performed between the three-dimensional dose distribution data and the target area expected dose distribution index in the prescription requirements to obtain target area dose distribution deviation data.

[0094] Voxel-by-voxel comparison processing includes: comparing the dose values ​​of voxels located within the target area in the three-dimensional dose distribution data with the expected dose values ​​required by the prescription for the target area point by point, calculating the dose deviation for each voxel, which is the difference between the actual calculated dose of the voxel and the prescription-required dose. When the voxel dose is lower than the prescription requirement, a negative deviation is generated, indicating that the voxel is under-dosed; when the voxel dose is higher than the prescription requirement, a positive deviation is generated, indicating that the voxel is over-dosed. Based on the dose deviation values ​​of each voxel, the following target area dose evaluation indicators were further statistically analyzed and calculated: (1) Target area coverage indicators, such as V95 and D95. V95 represents the proportion of the target area volume receiving more than 95% of the prescribed dose, and D95 ​​represents the minimum dose value received by at least 95% of the target area volume. The above indicators are used to evaluate whether the target area is adequately covered; (2) Target area maximum dose Dmax and its ratio to the prescribed dose are used to evaluate whether there are dose hotspots in the target area; (3) Target area uniformity index is used to evaluate the uniformity of dose distribution in the target area by comparing the dose difference between high-dose areas and low-dose areas in the target area. The closer the index is to zero, the more uniform the dose distribution in the target area; (4) Target area conformity index is used to evaluate the degree of fit between the dose distribution and the geometric shape of the target area by calculating the ratio of the volume surrounded by the prescribed dose line to the actual volume of the target area. The closer the index is to 1, the better the conformity. The above indicators together constitute the target area dose distribution deviation data.

[0095] Voxel-by-voxel comparison was performed on the three-dimensional dose distribution data and the dose constraint index for external tissues in the prescription requirements to obtain dose distribution deviation data for external tissues.

[0096] Voxel-by-voxel alignment processing includes: comparing the dose values ​​of voxels located within each organ at risk in the three-dimensional dose distribution data with the dose constraints for each organ at risk required by the prescription, and calculating the dose deviation for each voxel. Based on the deviation values ​​of each voxel, the maximum dose Dmax, average dose Dmean, and dose-volume constraint compliance status (e.g., whether V20 and V30 meet the ≤ limit values) for each organ at risk are further calculated. These indicators together constitute the dose distribution deviation data for tissues outside the target area.

[0097] The instantaneous dose rate sequence is compared with the FLASH prescription timing constraints on a time-stamp basis to obtain the timing dose rate deviation data: the number of pulses with instantaneous dose rates below 40 Gy / s, whether the total irradiation duration exceeds the limit, and whether the pulse interval violates the hardware threshold are statistically analyzed to form the FLASH timing deviation index.

[0098] Statistical index calculations were performed on the target area dose distribution deviation data, the dose distribution deviation data of tissues outside the target area, and the FLASH time-series dose rate deviation data to obtain the verification results.

[0099] The statistical indicator calculation and processing includes: calculating the statistical characteristic values ​​of dose deviation for all voxels in the target area and organs at risk, specifically including the average deviation, which reflects the overall trend of dose shift; the maximum and minimum deviation, which reflects the extreme range of dose deviation; and the standard deviation, which evaluates the dispersion and distribution consistency of dose deviation in the target area or organs at risk. The FLASH time-series comprehensive compliance rate is calculated simultaneously. Furthermore, based on the degree of deviation of each dose indicator from the prescription requirements, a comprehensive dose compliance score is calculated. This score is calculated as follows: for each dose indicator, a corresponding sub-score is assigned according to its compliance status. If an indicator is fully compliant, the sub-score is full; if it is not compliant, a value within the corresponding range is taken according to the degree of deviation. Then, the sub-scores are weighted and summed according to their corresponding weight ratios to obtain a comprehensive dose compliance score reflecting the degree of compliance of the overall dose distribution with the prescription requirements. The above statistical indicators and scores are summarized to form the verification results. For laser tailfield accelerator scenarios, the verification results also include the determination of whether the instantaneous dose rate meets the FLASH dose rate requirements (e.g., not less than 40 Gy / s).

[0100] Step 106: Determine and iterate the verification results and the optimal combination irradiation scheme to obtain the target irradiation scheme.

[0101] In this embodiment, the judgment and iterative processing refers to evaluating the verification results and continuously improving the irradiation protocol through iterative optimization when requirements are not met, until a clinically acceptable standard is reached. This step forms a complete closed loop of "optimization-simulation-verification-iteration" by back-transmitting the dose deviation information from the verification stage to the optimization stage, thereby continuously improving the quality of the protocol.

[0102] Specifically, the verification results are compared with the preset clinical acceptance criteria item by item to obtain the compliance status data of each acceptance item.

[0103] Clinical acceptance criteria include at least the following: target dose coverage indicators (e.g., V95 ≥ 95%, D95 ≥ 95% of the prescribed dose), target dose homogeneity indicators (e.g., Dmax ≤ 115% of the prescribed dose, homogeneity index HI ≤ 0.15), and external tissue dose constraint indicators (e.g., Dmax and Dmean of each organ at risk do not exceed preset limits, V20 / V30 does not exceed preset thresholds). For FLASH radiotherapy scenarios, acceptance criteria also include time-dimensional constraints such as instantaneous dose rate ≥ 40 Gy / s and total irradiation time ≤ 200 ms. Each dose indicator in the validation results is compared item by item with the corresponding acceptance criteria to obtain the compliance status data (compliant / non-compliant) for each acceptance item.

[0104] The compliance status data of each acceptance item are comprehensively judged and processed. When all acceptance items meet the standards, the optimal combination irradiation scheme is determined and verified.

[0105] When the optimal combination irradiation scheme fails to pass validation, the dose deviation data in the validation results are processed to locate the source of the deviation, and the optimization variable information to be adjusted is obtained.

[0106] Deviation source localization and processing includes analyzing the main reasons for non-compliance of each acceptance item. For example, if target area coverage is insufficient (V95 not meeting standards), the cause is determined based on the spatial location of the under-dose area in the target area, indicating a lack of effective irradiation directions in the directional combination or insufficient dose allocation. If a dose to a critical organ exceeds the standard, the cause is determined based on the spatial location of the organ and the distribution of voxels exceeding the standard, indicating that the beam path of some incident directions passes through the critical organ area. If the target area dose uniformity is poor, the cause is determined to be the need for rebalancing of dose allocation in each direction or inappropriate energy selection. Based on the deviation source localization results, the optimization variables to be adjusted are determined, including incident directions that need to be added or replaced, dose contribution parameters that need to be adjusted, and energy parameters that need to be adjusted.

[0107] Constraint update processing is performed on the optimization variable information and parameter space to be adjusted, resulting in the updated parameter space.

[0108] The constraint update process includes adjusting the value boundaries or search strategies of the corresponding optimization variables in the parameter space based on the deviation source localization results. For example, if the target area coverage is insufficient, the priority weight of candidate directions effective for under-dose regions of the target area is increased (given higher weight in the direction selection in step 104), or the upper limit of the dose contribution search for related directions is expanded (from 0.5 Gy to 0.6 Gy); if the dose to a certain organ at risk exceeds the limit, the upper limit of the dose contribution of candidate directions passing through the region of that organ at risk is reduced (tightened from 0.5 Gy to 0.3 Gy), or the direction is temporarily excluded from the candidate set; if the target area homogeneity is poor, the energy optimization range is adjusted to allow more energy combinations. After the constraint update, the updated parameter space is obtained.

[0109] Based on the updated parameter space, single-beam parameter optimization, direction combination and dose allocation, and three-dimensional dose distribution verification are performed sequentially until the verification results meet the preset clinical acceptance criteria.

[0110] That is, return to step 103 and perform optimization and verification again under the new parameter space constraints. Optionally, a maximum number of iterations (e.g., 10 times) or a convergence tolerance can be set as a termination condition. When the maximum number of iterations is reached or the difference in the comprehensive score of the consecutive iterations is less than the tolerance (e.g., the change in comprehensive score is <0.5%), the iteration is terminated and the current optimal solution is output.

[0111] When the verification results meet the preset clinical acceptance criteria, the optimal combination irradiation scheme obtained in the current iteration is determined as the target irradiation scheme.

[0112] The target illumination scheme includes at least the energy parameters, incident direction parameters, and dose contribution parameters for each pulse. For laser backfield accelerator scenarios, the target illumination scheme also includes the timing arrangement parameters of each pulse, which can be directly output as an accelerator control script (such as CSV format or dedicated binary format) to drive the laser backfield accelerator to perform illumination without the need for manual secondary parameter conversion.

[0113] The embodiments of this application, through the technical solutions described in the above specific embodiments, have achieved the following beneficial effects: (i) Significantly reduce computational complexity and hardware computing power requirements. This application adopts a divide-and-conquer strategy of "firstly, independently optimizing in each direction, and then solving in combination across directions" to decompose the ultra-high-dimensional joint optimization problem into low-dimensional subproblems for sequential solution, thus avoiding an exponential explosion of computing power. At the same time, it adopts a phased search strategy based on physical priors to significantly reduce the number of calls to high-precision Monte Carlo simulation, enabling the method to run efficiently on industrial-grade computing equipment.

[0114] (ii) Expanding the optimization dimensions to adapt to emerging clinical needs. This application incorporates pulse timing parameters into a unified parameter space to achieve synergistic optimization of spatial dose distribution and temporal dose rate distribution. This enables the solution to directly adapt to emerging clinical needs such as FLASH ultra-high dose rate radiotherapy, breaking through the dimensional limitations of traditional methods that only optimize spatial distribution.

[0115] (III) Constructing a closed-loop iterative mechanism to ensure the quality and robustness of the protocol. This application forms a complete closed loop of "optimization-simulation-verification-iteration" by back-transmitting the dose deviation information from the validation stage to the optimization stage. This can adaptively handle cases of varying complexity and ensure that the final protocol stably meets the clinical acceptance criteria.

[0116] (iv) Achieving a fully automated process from prescription to execution. This application achieves a fully automated process from clinical prescription input to executable irradiation plan output, from information extraction, parameter space construction, unidirectional optimization, directional combination and dose allocation to verification iteration. No manual intervention is required, avoiding the problem of varying plan quality from person to person, and opening up the engineering link from "physical adjustability" to "clinical executability".

[0117] While this application provides the method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-inventive labor. The order of steps listed in this embodiment is merely one possible execution order among many and does not represent the only execution order. In actual device or client product execution, the methods shown in this embodiment or the accompanying drawings can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0118] like Figure 2 As shown in the illustration, this application also provides an irradiation scheme generation apparatus 200. The apparatus includes: The extraction module 201 is used to extract and process information from treatment resource parameters and target area prescription parameters to obtain treatment resource information, target area geometric information and prescription requirements. The treatment resource information includes the range of available electron beam energy parameters, the range of available incident direction parameters and the range of beam dose adjustment. The target area geometric information includes the target area contour and spatial location information. The prescription requirements include the target area expected dose distribution index and the dose constraint index for external tissues of the target area.

[0119] The processing module 202 is used to construct the parameter space of the treatment resource information, target area geometry information and prescription requirements to obtain the parameter space of the irradiation plan. The parameter space includes the beam pulse number parameter, the energy parameter of each pulse, the incident direction parameter of each pulse and the dose contribution parameter of each pulse.

[0120] The processing module 202 is also used to perform single-beam parameter optimization processing on the candidate incident directions in the parameter space to obtain a single-direction optimization result set. The single-beam parameter optimization is to search for the optimal single-beam parameter combination that makes the target area dose distribution meet the preset conditions in each candidate incident direction.

[0121] The processing module 202 is also used to perform direction combination and dose allocation processing on the single-direction optimization result set to obtain the optimal combination irradiation scheme. The direction combination and dose allocation is to select several incident directions from all candidate incident directions and allocate corresponding dose contributions to each direction.

[0122] The processing module 202 is also used to perform three-dimensional dose distribution verification processing on the optimal combination irradiation scheme to obtain the verification results.

[0123] The determination module 203 is used to determine and iteratively process the verification results and the optimal combination irradiation scheme to obtain the target irradiation scheme.

[0124] Some modules in the apparatus described in this application can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0125] The apparatus or module described in the above embodiments can be implemented by a computer chip or physical entity, or by a product with a certain function. For ease of description, the above apparatus is described by dividing it into various modules according to their functions. When implementing the embodiments of this application, the functions of each module can be implemented in one or more software and / or hardware. Of course, a module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0126] The methods, apparatus, or modules described in this application can be implemented in a computer-readable program code manner. The controller can be implemented in any suitable manner, such as a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of a memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code manner, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included within it for implementing various functions can also be considered as structures within the hardware component. Alternatively, the device used to implement various functions can be viewed as either a software module that implements the method or a structure within a hardware component.

[0127] This application also provides an apparatus, the apparatus comprising: a processor; a memory for storing processor-executable instructions; wherein, when the processor executes the executable instructions, it implements the method described in this application.

[0128] This application also provides a non-volatile computer-readable storage medium storing a computer program or instructions thereon, which, when executed, enables the method described in this application embodiment to be implemented.

[0129] Furthermore, in the various embodiments of the present invention, each functional module can be integrated into a processing module, or each module can exist independently, or two or more modules can be integrated into a single module.

[0130] The aforementioned storage media include, but are not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Cache, Hard Disk Drive (HDD), or Memory Card. The memory can be used to store computer program instructions.

[0131] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary hardware. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, or it can be embodied in the process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0132] The various embodiments described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. All or part of this application can be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, mobile communication terminals, multiprocessor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0133] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of this application.

Claims

1. A method for generating an irradiation scheme, characterized in that, include: Information extraction and processing are performed on treatment resource parameters and target area prescription parameters to obtain treatment resource information, target area geometric information and prescription requirements. The treatment resource information includes the range of available electron beam energy parameters, the range of available incident direction parameters and the range of beam dose adjustment. The target area geometric information includes the target area contour and spatial location information. The prescription requirements include the target area expected dose distribution index and the dose constraint index for external tissues of the target area. The treatment resource information, the target area geometry information, and the prescription requirements are processed to construct a parameter space to obtain the parameter space of the irradiation scheme. The parameter space includes the beam pulse number parameter, the energy parameter of each pulse, the incident direction parameter of each pulse, and the dose contribution parameter of each pulse. The candidate incident directions in the parameter space are each subjected to single-beam parameter optimization processing to obtain a single-direction optimization result set covering multiple candidate incident angles. The single-beam parameter optimization is to search for the optimal single-beam parameter combination that makes the target area dose distribution meet the preset conditions in each candidate incident direction. The single-direction optimization result set is subjected to joint optimization processing of direction combination and dose contribution to obtain the optimal combined irradiation scheme. The joint optimization of direction combination and dose contribution involves selecting several incident directions from all candidate incident directions and assigning corresponding dose contributions to each direction. The optimal combination irradiation scheme was subjected to three-dimensional dose distribution verification processing to obtain the verification results; The verification results and the optimal combination irradiation scheme are judged and iterated to obtain the target irradiation scheme. If the verification results do not meet the clinical acceptance criteria, the constraint boundary of the parameter space is adjusted according to the dose deviation information in the verification results, and the single beam parameter optimization processing, the joint optimization processing of direction combination and dose contribution, and the three-dimensional dose distribution verification processing are re-executed until the clinical acceptance criteria are met.

2. The method according to claim 1, characterized in that, The parameter space construction process, which involves processing the treatment resource information, the target area geometry information, and the prescription requirements to obtain the parameter space of the irradiation plan, includes: The available energy parameter range, available incident direction parameter range, and beam dose adjustment range in the treatment resource information are subjected to boundary extraction processing to obtain the value boundary information of the optimization variables; Spatial mapping processing is performed on the target area contour and spatial position information in the target area geometry information to obtain the geometric constraint information of the target area in three-dimensional space; The target area desired dose distribution index and the dose constraint index of external tissues in the prescription requirements are quantified to obtain dose optimization target information; The boundary information, geometric constraint information, and dose optimization target information are subjected to variable definition and spatial combination processing to obtain an initial parameter space containing the beam pulse number variable, the pulse energy variable, the pulse incident direction variable, and the pulse dose contribution variable. The initial parameter space is subjected to physical realizability constraint verification to obtain the parameter space of the irradiation scheme.

3. The method according to claim 1, characterized in that, The process of extracting and processing information from treatment resource parameters and target area prescription parameters to obtain treatment resource information, target area geometric information, and prescription requirements includes: The treatment resource parameters are analyzed using equipment capability analysis to obtain the range of available electron beam energy parameters, the range of available incident direction parameters, and the range of beam dose adjustment. The medical image data in the target area prescription parameters are subjected to target area segmentation processing to obtain the target area outline and spatial location information; The clinical prescription data in the target area prescription parameters are subjected to dose target analysis to obtain the target area expected dose distribution index and the dose constraint index on external tissues of the target area. The target region contour and spatial location information are processed for organ at risk identification to obtain the organ at risk contour and its location information, and the organ at risk contour and its location information are used as the associated geometric data of the dose constraint index of the external tissue of the target region. The available electron beam energy parameter range, the available incident direction parameter range, the beam dose adjustment range, the target area contour and spatial location information, the target area expected dose distribution index, the dose constraint index for external tissues of the target area, and the contour and location information of the organs at risk are correlated and integrated to obtain treatment resource information, target area geometric information, and prescription requirements.

4. The method according to claim 1, characterized in that, The process of performing single-beam parameter optimization on each candidate incident direction in the parameter space yields a single-direction optimization result set covering multiple candidate incident angles, including: The candidate incident directions are subjected to direction traversal processing to obtain a single incident direction to be optimized; The beam parameters under the single incident direction are subjected to initial parameter setting processing to obtain an initial beam parameter combination, wherein the beam parameters include electron beam energy parameters, beam spot size parameters and source-target distance parameters; The initial beam parameter combination is subjected to a staged search process to obtain the optimal beam parameter combination; The target area dose deposition efficiency index is calculated and processed for the optimal beam parameter combination to obtain the single-direction optimization result under the single incident direction. The optimization results of each single direction corresponding to all candidate incident directions are aggregated to obtain a set of single-direction optimization results covering multiple candidate incident angles.

5. The method according to claim 1, characterized in that, The joint optimization processing of the single-direction optimization result set, considering both directional combination and dose contribution, to obtain the optimal combined irradiation scheme includes: The single-direction optimization results of each candidate incident direction in the single-direction optimization result set are processed to construct a direction combination space, thereby obtaining a direction combination search space; The dose allocation variable space is obtained by defining dose allocation variables for each direction combination in the direction combination search space. The optimal dose allocation scheme and the corresponding combined dose distribution index under each directional combination are obtained by jointly optimizing and solving the directional combination search space and the dose contribution allocation variable space. The optimal dose allocation scheme and the combined dose distribution index are comprehensively evaluated and sorted to obtain the best combined irradiation scheme.

6. The method according to claim 1, characterized in that, The three-dimensional dose distribution verification process for the optimal combined irradiation scheme yields the verification results, including: The optimal combination irradiation scheme is subjected to three-dimensional dose calculation processing to obtain the three-dimensional dose distribution data corresponding to the optimal combination irradiation scheme. The three-dimensional dose distribution data is compared with the target area expected dose distribution index in the prescription requirements on a voxel-by-voxel basis to obtain the target area dose distribution deviation data. The three-dimensional dose distribution data is compared with the dose constraint index for external tissues in the prescription requirements on a voxel-by-voxel basis to obtain dose distribution deviation data for external tissues. The dose distribution deviation data of the target area and the dose distribution deviation data of the tissues outside the target area are statistically processed to obtain the verification results.

7. An irradiation scheme generation device, characterized in that, include: The extraction module is used to extract and process information from treatment resource parameters and target area prescription parameters to obtain treatment resource information, target area geometric information and prescription requirements. The treatment resource information includes the range of available electron beam energy parameters, the range of available incident direction parameters and the range of beam dose adjustment. The target area geometric information includes the target area contour and spatial location information. The prescription requirements include the target area expected dose distribution index and the dose constraint index for external tissues of the target area. The processing module is used to perform parameter space construction processing on the treatment resource information, the target area geometry information and the prescription requirements to obtain the parameter space of the irradiation plan. The parameter space includes the beam pulse number parameter, the energy parameter of each pulse, the incident direction parameter of each pulse and the dose contribution parameter of each pulse. The processing module is also used to perform single-beam parameter optimization processing on the candidate incident directions in the parameter space to obtain a single-direction optimization result set covering multiple candidate incident angles. The single-beam parameter optimization is to search for the optimal single-beam parameter combination that makes the target area dose distribution meet the preset conditions in each candidate incident direction. The processing module is also used to perform joint optimization processing of direction combination and dose contribution on the single-direction optimization result set to obtain the best combined irradiation scheme. The joint optimization of direction combination and dose contribution is to select several incident directions from all candidate incident directions and assign corresponding dose contributions to each direction. The processing module is also used to perform three-dimensional dose distribution verification processing on the optimal combination irradiation scheme to obtain verification results; The determination module is used to judge and iteratively process the verification results and the optimal combination irradiation scheme to obtain the target irradiation scheme. In the case that the verification results do not meet the clinical acceptance criteria, the constraint boundary of the parameter space is adjusted according to the dose deviation information in the verification results, and the single beam parameter optimization processing, the joint optimization processing of direction combination and dose contribution, and the three-dimensional dose distribution verification processing are re-executed until the clinical acceptance criteria are met.

8. A computer device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.

Images