Multi-directional proton flash radiotherapy plan optimization method, system, device, and medium

Abstract

The present application belongs to the technical field of radiotherapy equipment, and particularly relates to a multi-directional proton flash radiotherapy plan optimization method, system, device and medium. The multi-directional proton flash radiotherapy device comprises a single proton source, a fast beam switching assembly, a multi-fixed gantry channel and a plan execution control assembly. The present application also provides a radiotherapy plan optimization method and system adapted to the multi-directional proton flash radiotherapy device. The method generates a high degree of freedom non-conformal energy modulator constraint reference plan under few constraints first, and then uses the reference plan as a continuous reference to gradually generate a final deliverable plan meeting strict physical constraints such as multi-directional fixed field, single energy within the field, and manufacturable modulator through a series of weakly coupled conversion steps such as point beam non-overlapping gridding, beam-by-beam dose responsibility separation, conformal energy modulator generation, energy compensation, and fixed modulator post-irradiation dose adjustment. The present application has good application prospects in the field of radiotherapy.

Description

Technical Field

[0001] This invention belongs to the field of radiotherapy equipment technology, specifically relating to a method, system, device, and medium for optimizing multidirectional proton flash radiotherapy planning. Background Technology

[0002] In recent years, ultra-high dose rate radiotherapy (FLASH radiotherapy) has attracted widespread attention due to its potential to reduce damage to normal tissues while maintaining tumor control in preclinical studies. Existing research typically considers an average dose rate of approximately 40 Gy / s or higher as one of the important physical conditions for the FLASH effect, and emphasizes that factors such as ultrashort irradiation time, single-pulse dose, instantaneous dose rate, beam pause time, and spatial dose rate distribution may collectively influence the FLASH effect. Therefore, radiotherapy planning systems for FLASH must not only meet conventional dose-volume constraints but also consider dose rate coverage, delivery time, and equipment deliverability. Among these, proton FLASH radiotherapy, due to the Bragg peak effect, is expected to achieve a superior dose distribution and has become a major research focus in FLASH radiotherapy.

[0003] Multi-angle irradiation is an important means to improve the conformity and concentration of dose within the tumor target area, enabling complete target coverage while minimizing impact on normal tissues. However, traditional proton therapy typically relies on rotating gantry to achieve multi-angle irradiation. Rotating gantry systems are bulky, heavy, and costly, and the mechanical rotation and angle switching increase the complexity of the treatment process and dose errors. The gantry rotation time is also significant, making it difficult to bring the overall treatment time window into the FLASH effect threshold. Furthermore, while traditional proton therapy allows a single treatment field to contain multiple energies, and traditional intensity-modulated proton therapy (IMRT) typically achieves tumor depth-direction dose coverage through multiple energy level switching, energy switching introduces additional time overhead, making it difficult to meet the short duration and high dose rate requirements of FLASH irradiation. If single-energy irradiation is used directly in traditional proton therapy equipment without energy switching, although energy switching time can be reduced, without additional range modulation or conformal compensation, a single-energy beam usually cannot simultaneously cover different depth regions within the tumor, easily leading to uneven dose distribution within the target area, localized cold zones, or increased dose in non-target areas, failing to meet clinical conformity requirements. In contrast, multi-angle single-energy irradiation, with its different incident angles and individual modulation capabilities, inherently offers advantages in dose distribution through multiple fixed single-energy fields, resulting in conformal and uniform dose distribution. Therefore, existing mechanical structures struggle to simultaneously support both FLASH irradiation and clinical conformal requirements. There is an urgent need to develop more efficient machine structures suitable for FLASH irradiation and to explore a new paradigm that achieves good tumor dose coverage without switching energy levels. Multi-field single-energy irradiation has emerged as a new direction for achieving good dose coverage in FLASH irradiation.

[0004] To avoid time delays caused by gantry movement, fixed multi-angle, multi-field radiotherapy equipment has been proposed, which is expected to achieve rapid multi-field proton irradiation without mechanical switching time, thus meeting the requirements for FLASH irradiation. However, this direction faces significant engineering design challenges. Currently, the development of multi-angle FLASH irradiation equipment mainly focuses on multi-directional X-ray or high-energy electron FLASH, and no proton multi-field radiotherapy equipment has been designed and implemented. Meanwhile, to achieve conformal irradiation under single-energy, single-field FLASH, Kang et al. (2022) proposed a structure in which a conformal energy modulator (CEM), a range shifter, and an aperture collimator are arranged sequentially along the radiation source to the patient's tumor center. This structure can achieve conformality under single-energy, single-field FLASH irradiation. This paradigm has been developed and registered as Conformal FLASH by IBA Corporation, and the team at KU Leuven (UC Louvain) has further carried out a patient-specific modulator project, demonstrating its clinical deliverability. Correspondingly, single-field FLASH inverse optimization systems based on patient dose requirements have matured, modulating CEM height and gradient, particle delivery at the scanning point, and energy through Monte Carlo dose calculations. However, these systems suffer from limitations such as irradiation of small target areas, poor dose conformity, local hot and cold spots, and high doses to organs at risk. To address these technical challenges, a feasible approach is to combine novel single-energy, single-field FLASH treatment paradigms with multi-angle FLASH devices to further advance the clinical translation of FLASH radiotherapy.

[0005] However, the computational difficulty of planning optimization methods in multi-angle proton FLASH scenarios has significantly increased due to the introduction of multi-angle and multi-field optimization objects. Specifically, this manifests in the following ways: First, existing studies mainly focus on single-beam schemes, paying more attention to the feasibility of specific gantry nozzles or specific cases, and have not yet directly covered the multi-beam joint planning requirements under fixed multi-directional devices. Second, existing methods often directly initiate optimization based on the patient's target dose distribution, resulting in strong coupling between variables such as CEM geometry, single-energy beam and scan point weights, energy selection, and point distribution. This leads to high optimization dimensionality, heavy computational burden, and insufficient interpretability of results. Third, in multi-directional iso-energy irradiation, each direction must independently form a manufacturable conformal energy modulator thickness map, while also jointly meeting the requirements of SBRT target coverage, hot spot limitation, and organ-at-risk constraints in terms of overall dose. Directly optimizing all directions, all beam spot weights, and all modulator thicknesses can easily lead to problems such as excessively long computation time, difficulty in repairing local cold spots, and unclear dose responsibility in different directions. Fourth, there is a potential trade-off between FLASH dose rate conditions and dose quality. Pursuing only dose rate or only physical dose is difficult to meet the requirements of clinical translation.

[0006] Currently, for multi-directional, iso-energy proton conformal FLASH scenarios, especially under conditions of fixed irradiation direction, rapid delivery, patient-specific modulator fabrication, and SBRT-level dose constraints, a faster, more stable, and interpretable planning optimization method suitable for engineering implementation is still needed. To date, no planning algorithm has been proposed that can meet these requirements and optimize for excellent dose distribution.

[0007] In summary, there is an urgent need in this field to propose a physical engineering design for a proton therapy system with a fixed irradiation field, and on this basis, to develop an efficient, interpretable, highly flexible, and widely adaptable optimization algorithm to meet the common needs of rapid FLASH irradiation and precise conformal treatment for patients, thereby promoting the reliable implementation and clinical translation of this system. Summary of the Invention

[0008] To address the problems of existing technologies, this invention provides a method, system, device, and medium for optimizing multidirectional proton flash radiotherapy planning.

[0009] A multidirectional proton flash radiotherapy device, comprising: A proton source, used to generate a proton beam; A beam transport assembly for transmitting the proton beam to several fixed gantry irradiation channels; A rapid beam switching assembly is used to quickly switch the specific fixed gantry irradiation channel into which the proton beam enters; Several fixed gantry irradiation channels are used to irradiate proton beams along several preset fixed irradiation directions, and the fixed gantry irradiation channels are arranged around the patient's treatment area; The fixed gantry irradiation channel includes a single-energy projection component and a beam shaping component. The single-energy projection component is used to project a proton beam into each fixed irradiation direction with a preset single-energy or energy-compensated single-energy proton beam. The beam shaping component is used to shape the proton beam in three-dimensional space so that its dose distribution matches the shape and depth of the tumor target area.

[0010] Preferred options also include: The plan execution control component is used to control the fast beam switching component, the monoenergetic irradiation component, and the beam shaping component to complete the irradiation of the multi-angle fixed field proton beam according to the plan parameters of the multi-directional fixed field proton FLASH plan. The generation process of the multi-directional fixed-field proton FLASH plan includes: Obtain patient plan input data; Based on the patient planning input data, a reference plan is generated under minimal constraints, including target volume dose coverage, target volume dose hotspots, shell dose constraints, organ at risk dose limits, extra-target dose hotspots, field fractional balance, and MU smoothness constraints. This reference plan serves as a continuous reference for subsequent physicalization plans. Based on the reference plan without conformal energy modulator constraints, a constrained plan space for the multi-directional fixed-field proton FLASH plan is constructed. The process of transforming a plan from few constraints to a constrained plan space includes the following steps: The multi-directional, multi-depth irradiation contributions in the reference plan are converted into a point structure that meets the requirement of lateral non-overlap. Using the point structure, the beam-by-beam reference dose in the reference plan, and the single-energy irradiation conditions as inputs, generate or correct the conformal energy modulator thickness map. After generating or correcting the conformal energy modulator thickness map, perform energy compensation or energy boost to adjust the MU fraction at the irradiation point; The aforementioned reference plan will be continuously referenced in each step of the plan conversion process; Within the constrained planning space, a physicalization plan is generated or modified so that the dose distribution of the physicalization plan continuously approximates the dose distribution of the reference plan. Monte Carlo dosing calculations were performed based on the final physicochemical plan; When the Monte Carlo dose meets the preset dosimetric acceptance criteria, the corresponding physical plan will be output as a multi-directional fixed-field proton FLASH plan.

[0011] Preferably, the beam shaping component comprises one or more of a conformal energy modulator, a range modulator, and an aperture collimator; the range modulator includes one or both of an initial range modulator and an equalizing range modulator. The reference plan, physicalization plan, or multi-directional fixed-field proton FLASH plan includes the following plan parameters: spatial orientation parameters for each fixed irradiation direction, single energy, compensated energy, conformal energy modulator thickness map, range modulator parameters, aperture collimator parameters, irradiation point coordinates, and MU fraction; the range modulator parameters include one or both of the initial range modulator parameters and the uniform energy range modulator parameters. The multi-directional proton flash radiotherapy device uses a single-energy proton beam globally. In a specific fixed irradiation direction, the single-energy proton beam in the field forms a conformal dose distribution through the thickness map of the conformal energy modulator, the initial range modulator parameters, the compensated energy, and the MU fraction adjustment. Finally, the global energy is synchronized through the uniform range displacement.

[0012] Preferably, during the reference plan generation process, an objective function is established that includes penalties for insufficient target coverage, target hotspots, outer shell dose-limiting areas, organs at risk of dose, outside-target hotspots, field share balancing, and MU smoothing. in, This represents the radiation vector to be optimized in a less constrained plan. This represents the irradiance vector obtained by minimizing the above objective function under non-negativity constraints. This indicates a penalty for insufficient target coverage. Indicates the target area hotspot penalty item. This indicates a penalty for the target area's dose limitation region. This represents the dose penalty term for the r-th organ at risk. Indicates the penalty for hotspots outside the target area. This represents the dose fraction or exposure fraction balance term between fixed irradiation directions. This represents the smoothing term of the irradiation point MU. This indicates the weight of the fixed irradiation direction share balance term. Indicates the weight of the dose smoothing term; the total reference dose of the reference plan. Represented as: in, This represents the dose-effect matrix of the reference plan. Indicates the weight of exposure amount; The total reference dose of the reference plan is decomposed into beam-by-beam reference doses corresponding to each fixed irradiation direction: in, This represents the beam-by-beam reference dose for a fixed irradiation direction b. This represents the dose effect matrix for a fixed irradiation direction b. B represents the irradiation weight of a fixed irradiation direction b, where B represents the number of fixed irradiation directions. In the subsequent physicalization planning process, each fixed irradiation direction will be prioritized based on its corresponding... As a continuously approaching, beam-by-beam reference dose.

[0013] Preferably, the process of generating the final physicalization plan includes the following steps: Step 1: Repeat the weakly coupled conversion process, which includes the following steps: Step 1.1: Using the point structure, the beam-by-beam reference dose in the reference plan, and the single-energy irradiation conditions as inputs, generate or correct the conformal energy modulator thickness map; Step 1.2: Based on the generated or corrected conformal energy modulator thickness map, perform energy compensation or energy boosting on the corresponding fixed irradiation direction according to the target cold zone exploration results and the insufficient range. Step 1.3: Based on the generated or corrected conformal energy modulator thickness map, adjust the MU fraction at the irradiation point to obtain a candidate plan; Step 1.4: Evaluate the candidate plans obtained in Step 3 based on the Monte Carlo dose calculation results; The weakly coupled conversion process is executed in a maximum of three rounds. When a candidate plan in a certain round is better than the current best plan, the candidate plan is retained as the new best plan. When a candidate plan in a certain round deteriorates, it is rolled back to the current best plan. The energy compensation or energy boost includes calculating the range deficiency based on the water equivalent path length of the voxel in the cold zone of the target area along the corresponding fixed irradiation direction, and determining the energy increase based on the monotonic relationship between proton energy and range; when the far-end coverage margin is insufficient but a significant cold zone has not yet been formed, a preventive energy boost is performed on the corresponding fixed irradiation direction. Step 2: After obtaining the final optimal plan, a uniform range modulator is used to complete the global energy unification and obtain the final physical plan.

[0014] Preferably, the point structure is generated in the following manner: Construct a parent horizontal grid with a preset spacing, and set four non-overlapping child slots within each parent horizontal grid; Several depth irradiation contribution points located within the same parent horizontal grid in the reference plan are sorted according to MU share, dose contribution, target coverage importance, or organ-at-risk protection importance; the depth irradiation contribution point refers to a scan point in the reference plan located within a certain parent horizontal grid, corresponding to a target depth or water equivalent depth layer, and contributing dose to that depth region. The sorted depth illumination contribution points are assigned to different slots in the four non-overlapping sub-slots to avoid different depth contribution points occupying the same physical lateral position. When generating the point structure, an available sub-slot is selected for each depth illumination contribution point, and the cost of lateral movement is minimized. Let the depth irradiation contribution point be c, and its assigned sub-slot be s(c). Then the slot allocation target is: in, Indicates the contribution point of deep irradiation Assigned to sub-slot The cost; The cost function includes one or more of the following: lateral displacement cost, cross-parent lateral mesh penalty, and importance weight of depth illumination contribution points, and can be expressed as: Where, q= Indicates the contribution point of deep irradiation The importance of; Represents the horizontal coordinate of the candidate sub-slot; Indicates the contribution point of deep irradiation The original horizontal coordinates in the reference plan; Indicates the parent horizontal grid to which the candidate sub-slot belongs; Indicates the contribution point of deep irradiation The original parent horizontal grid; Indicates the penalty weight across the parent horizontal grid; This indicates the value to be taken when the candidate child slot is inconsistent with the original parent's horizontal grid. Indicator functions; When the number of depth illumination contribution points in the same parent horizontal grid exceeds four, the four highest-ranked depth illumination contribution points are retained first, and the unassigned, lower-ranked depth illumination contribution points are pruned, or the unassigned, lower-ranked depth illumination contribution points are migrated to the empty sub-slots in the adjacent parent horizontal grid. After generating the point structure, a reference association identifier is established for each irradiation point in the point structure. The information of the reference association identifier includes one or more of the following: original parent horizontal grid number, sub-slot number, fixed irradiation direction, target water equivalent depth, candidate depth layer number, original irradiation dose, normalized irradiation dose, importance evaluation value, beam-by-beam reference dose contribution number, and corresponding reference plan irradiation point number. The reference association identifier remains unchanged in subsequent steps of generating or correcting the conformal energy modulator thickness map and MU fraction adjustment.

[0015] Preferably, the adjustment of the MU share includes: After fixing the thickness map of the conformal energy modulator for each fixed irradiation direction, the MU fraction at the irradiation point is optimized. The MU adjustment aims to approximate the total reference dose of the reference plan without conformal energy modulator constraints with the dose synthesized from multiple fixed irradiation directions, and introduces MU constraint terms and field constraint terms into the MU adjustment target. Among them, the MU constraint term is used to limit the MU of a single irradiation point from deviating significantly from the reference plan, the non-overlapping point plan, or the plan of the previous cycle in the optimization cycle; The objective term considering the MU constraint is expressed as: ; The field constraint term is used to limit the total MU of each fixed irradiation direction from deviating significantly from the total MU of the corresponding fixed irradiation direction in the reference plan; The objective term considering the field constraint is expressed as: ; The MU adjustment target is expressed as: in: This represents the radiation weight obtained after optimization based on a non-overlapping point plan. Represents the irradiation weight to be optimized; This indicates that all exposure weights are non-negative; Represents the patient voxel weight matrix; This represents the dose effect matrix after fixing the modulator; This indicates an unlimited total reference dose; Indicates the first The weight of individual elements; This represents the reference value of the irradiation obtained after the modulator's master optimization. This represents the total number of radiation dose variables; Indicates the weight of the exposure maintenance term; Indicates the number of fixed irradiation directions; This indicates that the j-th irradiation point belongs to the j-th irradiation point. A fixed irradiation direction; Indicating an unrestricted reference plan Total irradiance from a fixed irradiation direction; This indicates the weight of the term that maintains the total amount of illumination in a fixed direction.

[0016] The present invention also provides a multi-directional proton flash radiotherapy planning optimization system, applied to the above-mentioned multi-directional proton flash radiotherapy device, comprising: The data input module is configured to acquire patient planned input data; The reference plan generation module is configured to generate a reference plan based on the patient plan input data, under minimal constraints including target volume dose coverage, target volume dose hotspots, shell dose limitations, organ at risk dose limitations, extra-target dose hotspots, field fractional balance, and MU smoothness constraints. The reference plan serves as a continuous reference for subsequent physicalization plans. The physicalization plan generation module is configured to construct a constrained plan space for a multi-directional fixed-field proton FLASH plan based on the reference plan without conformal energy modulator constraints; and to generate or modify physicalization plans within the constrained plan space so that the dose distribution of the physicalization plans continuously approximates the dose distribution of the reference plan; wherein the plan conversion process from less constrained conditions to a constrained plan space includes the following steps: The multi-directional, multi-depth irradiation contributions in the reference plan are converted into a point structure that meets the requirement of lateral non-overlap. Using the point structure, the beam-by-beam reference dose, and the single-energy irradiation conditions in the reference plan as inputs, a conformal energy modulator thickness map is generated or corrected. After generating or correcting the conformal energy modulator thickness map, energy compensation or energy boosting is performed to adjust the MU fraction of the irradiation point. Each step in the plan conversion process uses the reference plan as a continuous reference. The Monte Carlo dose calculation module is configured to perform Monte Carlo dose calculations based on the final physicalization plan; The output module is configured to output the corresponding physical plan as a multi-directional fixed-field proton FLASH plan when the Monte Carlo dose meets the preset dosimetric acceptance conditions.

[0017] The present invention also provides a method for optimizing multi-directional proton flash radiotherapy planning, applied to the above-mentioned multi-directional proton flash radiotherapy device, comprising the following steps: Obtain patient plan input data; Based on the patient planning input data, a reference plan is generated under minimal constraints, including target volume dose coverage, target volume dose hotspots, shell dose limitations, organ at risk dose limitations, extra-target dose hotspots, field fractional balance, and MU smoothness constraints. This reference plan serves as a continuous reference for subsequent physicalization plans. Based on the reference plan without conformal energy modulator constraints, a constrained plan space for the multi-directional fixed-field proton FLASH plan is constructed. The process of transforming a plan from few constraints to a constrained plan space includes the following steps: The multi-directional, multi-depth irradiation contributions in the reference plan are converted into a point structure that meets the requirement of lateral non-overlap. Using the point structure, the beam-by-beam reference dose in the reference plan, and the single-energy irradiation conditions as inputs, generate or correct the conformal energy modulator thickness map. After generating or correcting the conformal energy modulator thickness map, perform energy compensation or energy boost to adjust the MU fraction at the irradiation point; The aforementioned reference plan will be continuously referenced in each step of the plan conversion process; Within the constrained planning space, a physicalization plan is generated or modified so that the dose distribution of the physicalization plan continuously approximates the dose distribution of the reference plan. Monte Carlo dosing calculations were performed based on the final physicochemical plan; When the Monte Carlo dose meets the preset dosimetric acceptance criteria, the corresponding physical plan will be output as a multi-directional fixed-field proton FLASH plan.

[0018] The present invention also provides a computer-readable storage medium having stored thereon a computer program for implementing the above-described multidirectional proton flash radiotherapy planning optimization method.

[0019] This invention provides a novel multi-directional proton flash radiotherapy device compared to existing flash radiotherapy devices. The multi-directional proton flash radiotherapy device of this invention can perform multi-angle proton beam irradiation without gantry rotation and without energy switching, thus meeting the requirements of flash radiotherapy. Furthermore, due to the rapid switching between multi-angle irradiation, the multi-directional proton flash radiotherapy device of this invention has better conformability than existing devices.

[0020] On the other hand, this invention provides a targeted radiotherapy planning optimization method and system. Through a planning generation paradigm of "from a less constrained reference plan to a constrained physicalized plan," it avoids strongly coupled free search from scratch under strict deliverable constraints. By separating dose responsibility for each beam, it ensures that the dose responsibility of each of the four fixed irradiation directions is retained within the reference plan. Through four-square non-overlapping point mapping, it allows multi-depth overlapping points in the high-degree-of-freedom reference plan to be converted into a point structure expressible by a conformal energy modulator. Through a reference association identifier preservation mechanism, it maintains the point relationships, field responsibilities, and MU allocation relationships with the reference plan during subsequent physicalization conversion. Through MU adjustment after CEM fixation and a weakly coupled conversion process, the final plan, while meeting single-energy irradiation and deliverable physical constraints, still approximates the dose quality of a less constrained reference plan. Finally, through Monte Carlo dosimetric acceptance, the output plan meets dosimetric requirements such as target coverage, hotspot control, organ-at-risk protection, and external hotspot limitation. Based on the above calculation method and system, this invention solves the problems of increased computational difficulty and reduced computational stability caused by the introduction of multi-directional irradiation.

[0021] In summary, this invention provides a multi-directional proton flash radiotherapy device and a radiotherapy planning optimization method adapted to the device. By decoupling the complex multivariate strong coupling optimization, the computational difficulty is significantly reduced. It can efficiently generate proton radiotherapy plans with good target coverage and controlled hotspots while meeting the physical conditions of FLASH ultrafast irradiation, and has great application prospects.

[0022] Obviously, based on the above description of the present invention, and according to common technical knowledge and conventional methods in the field, various other modifications, substitutions or alterations can be made without departing from the basic technical concept of the present invention.

[0023] The following detailed embodiments further illustrate the above-described content of the present invention. However, this should not be construed as limiting the scope of the present invention to the following examples. All technologies implemented based on the above-described content of the present invention fall within the scope of the present invention. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the structure of the multidirectional proton flash radiotherapy device according to Embodiment 1 of the present invention; Figure 2 This is a partial structural schematic diagram of the beam shaping component of the multi-directional proton flash radiotherapy device according to Embodiment 1 of the present invention; Figure 3 This is a flowchart illustrating Embodiment 2 of the present invention; Figure 4 This is an example diagram illustrating the process of optimizing a radiotherapy plan in Embodiment 2 of the present invention; Figure 5 This is a schematic diagram of a non-overlapping quadrangular grid structure under the constraint of a conformal energy modulator according to Embodiment 2 of the present invention.

[0025] Among them, 1-proton source, 2-quadrupole iron, 3-energy degrader, 4-collimator, 5-dipolar iron, 6-fast beam switching assembly, 7-vacuum window, 8-profile detector, 9-X-direction scanning magnet, 10-Y-direction scanning magnet, 11-position detector, 12-dose detector, 13-beam shaping assembly, 14-specific angle field, 15-beams at various angles, 16-uniform energy range modulator, 17-conformal energy modulator, 18-aperture collimator, 19-target area isocenter, 20-primary range modulator.

[0026] 13-X represents beam shaping components for different firing fields. Except for the front aperture collimator, their actual component parameters are not the same. Further, 13-1 represents a 90° fixed firing field beam shaping component, 13-2 represents a 45° fixed firing field beam shaping component, 13-3 represents a 0° fixed firing field beam shaping component, and 13-4 represents a -45° fixed firing field beam shaping component.

[0027] 14-X represents the field number used to distinguish different fixed field angles. Further, 14-1 represents a 90° fixed field, 14-2 represents a 45° fixed field, 14-3 represents a 0° fixed field, and 14-4 represents a -45° fixed field. In embodiments of the present invention, the reference standard for angles is as follows: a vertical field facing the patient is defined as a 0° field, 45° refers to a 45° leftward rotation relative to the patient's midline, and -45° refers to a 45° leftward rotation relative to the patient's midline. The left-right mapping of -90° and 90° is similar.

[0028] 15-X represents the beam. Further, 15-1 represents a 90° fixed beam, 15-2 represents a 45° fixed beam, 15-3 represents a 0° fixed beam, and 15-4 represents a -45° fixed beam.

[0029] 16-X represents a uniform range modulator. Further, 16-1 represents a 90° fixed field uniform range modulator, 16-2 represents a 45° fixed field uniform range modulator, and 16-3 represents a 0° fixed field uniform range modulator. In this embodiment, since the -45° field has the longest range, the automatic optimization method of the radiotherapy planning optimization method considers a uniform range modulator thickness of 0 at this angle to be the optimal optimization scheme. Therefore, no uniform range modulator is set in the -45° fixed field in the figure. In other embodiments, a uniform range modulator may be set in the direction of the -45° fixed field based on the results of the optimization scheme.

[0030] 20 represents the primary range modulator. In this embodiment, the three fixed radiation fields other than the -45° fixed radiation field are automatically optimized by the radiotherapy planning optimization method, which determines that a primary range modulator thickness of 0 is the optimal optimization scheme. Therefore, no primary range modulator is set in the other three directions in the figure. In other embodiments, primary range modulators may also be set in the directions of these fixed radiation fields according to the results of the optimization scheme. Detailed Implementation

[0031] The present invention will be further described below with reference to embodiments. It should be understood that the following embodiments are only for illustrating the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. Data acquisition, data transmission, data storage, dose calculation, hardware connection, electromagnetic control, beam monitoring, patient positioning, and treatment safety interlocks, etc., not specifically described, can be implemented using techniques already disclosed or conventional in the art. In the following embodiments, the conformal energy modulator is abbreviated as CEM, and the irradiation dose at the irradiation point is abbreviated as MU.

[0032] Example 1: Multidirectional Proton Flash Radiotherapy Device This embodiment provides a multidirectional proton flash radiotherapy device, the overall structure of which is shown in the schematic diagram below. Figure 1 As shown.

[0033] The specific components include: Proton source 1 is used to generate a proton beam. In this embodiment, due to the setting of the fast beam switching component 6, the number of proton sources 1 can be set to one, that is, a single proton source, thereby reducing the cost and the overall complexity of the equipment. A beam transport assembly for transmitting the proton beam to several fixed gantry irradiation channels; The rapid beam switching assembly 6 is used to rapidly switch the specific fixed gantry irradiation channel into which the proton beam enters. Several fixed gantry irradiation channels are used to irradiate proton beams along several preset fixed irradiation directions, and the fixed gantry irradiation channels are arranged around the patient's treatment area; The plan execution control component is used to control the fast beam switching component, monoenergetic irradiation component, and beam shaping component to complete the irradiation of the multi-angle fixed-field proton beam according to the plan parameters of the multi-directional fixed-field proton FLASH plan.

[0034] Specifically: In this embodiment, the beam transmission assembly includes structures such as a quadrupole 2, a de-energizer 3, a collimator 4, and a dipole 5 arranged along the proton beam transmission direction.

[0035] In this embodiment, the fast beam switching component 6 is selected from Kicker magnets or other electromagnetic deflection components.

[0036] The fixed gantry irradiation channel includes a single-energy projection component and a beam shaping component 13. The single-energy projection component is used to project a proton beam with a preset single energy or an energy-compensated single-energy proton beam in each fixed irradiation direction. The beam shaping component is used to shape the proton beam in three-dimensional space so that its dose distribution matches the shape and depth of the tumor target area. In this embodiment, as an example, the number of fixed irradiation directions (i.e., the number of fixed gantry irradiation channels) is preferably four, corresponding to four fixed fields. In other embodiments, three, five, six or more fixed irradiation directions can be set according to the gantry structure and patient treatment needs. The specific number should be selected by comprehensively considering the treatment plan, device cost, and treatment room space.

[0037] Figure 1 In this context, θ represents the rotation angle of the multi-directional fixed treatment gantry around the Z-axis (vertical direction). The multi-directional proton flash radiotherapy device can rotate around the Z-axis by a preset angle according to the distribution of scattered radiation generated by the de-energizer, the spatial arrangement of the treatment room, or the patient's positioning requirements, in order to reduce the impact of the de-energizer radiation on subsequent beam transmission, dose monitoring, or treatment area.

[0038] In this embodiment, the monoenergetic irradiation component includes a vacuum window 7, a profile detector 8, an X-direction scanning magnet 9, a Y-direction scanning magnet 10, a position detector 11, and a dose detector 12 arranged along the proton beam transmission direction.

[0039] In this embodiment, the specific beam shaping component 13 is configured as follows: Figure 2 As shown, the system includes one or more of the following: a conformal energy modulator 17, a range modulator (including a primary range modulator 20 and an equalizing range modulator 16), and an aperture collimator 18. The conformal energy modulator 17 is used to modulate the water equivalent range of the proton beam at different lateral positions according to the patient's anatomy and the target depth distribution in the fixed irradiation direction; the range modulator is used to correct the overall range of the proton beam and to unify the final global energy; the aperture collimator 18 is used to limit the lateral range of the beam and assist in controlling the dose outside the target area.

[0040] Figure 1 and Figure 2 In the diagram, 13-X represents beam shaping components for different radiation fields. Based on the optimization results of the radiotherapy planning optimization method, their actual component parameters are not the same, except for the front aperture collimator. Further, this embodiment includes: a 90° fixed radiation field beam shaping component 13-1, a 45° fixed radiation field beam shaping component 13-2, a 0° fixed radiation field beam shaping component 13-3, and a -45° fixed radiation field beam shaping component 13-4.

[0041] 14-X represents the field number used to distinguish different fixed field angles. Further, in this embodiment, the fixed field angles include: 90° fixed field 14-1, 45° fixed field 14-2, 0° fixed field 14-3, and -45° fixed field 14-4. In this embodiment, the reference standard for the angle is as follows: a vertical field facing the patient is defined as a 0° field, 45° refers to a 45° leftward rotation relative to the patient's midline, and -45° refers to a 45° leftward rotation relative to the patient's midline. The left-right mapping of -90° and 90° is similar.

[0042] 15-X represents a beam, which further includes: a 90° fixed field beam 15-1, a 45° fixed field beam 15-2, a 0° fixed field beam 15-3, and a -45° fixed field beam 15-4.

[0043] 16-X represents the uniform range modulator, further comprising: a 90° fixed field uniform range modulator 16-1, a 45° fixed field uniform range modulator 16-2, and a 0° fixed field uniform range modulator 16-3. In this embodiment, since the -45° field has the longest range, the algorithm automatically optimizes and considers a uniform range modulator thickness of 0 at this angle to be the optimal solution. Therefore, no uniform range modulator is set in the -45° fixed field in the figure. In other embodiments, a uniform range modulator may be set in the direction of the -45° fixed field based on the optimization results. Furthermore, in this embodiment, the algorithm automatically optimizes and considers a primary range modulator thickness of 0 in the other three fixed fields besides the -45° fixed field to be the optimal solution. Therefore, no primary range modulator is set in the other three directions in the figure. In other embodiments, primary range modulators may be set in the directions of these fixed fields based on the optimization results.

[0044] In this embodiment, the number, deflection angle, magnetic field strength, rise time, and fall time of specific components such as the Kicker magnet, quadrupole 2, and dipole 5, as well as the spatial arrangement of the fixed gantry channel and the specific structure of the beam transmission pipeline, can be determined by the hardware implementer based on the single proton source type, treatment room space, number of fixed fields, and beam energy range. The implementation of this invention does not depend on a specific component size or a specific fixed gantry mechanical structure. As long as the system can provide multiple fixed irradiation directions, intra-field monoenergetic irradiation capability, and patient-specific beam shaping capability, it falls within the scope of this invention.

[0045] In this embodiment, as a specific example, the system preferably has four fixed gantry irradiation channels, each corresponding to a fixed irradiation direction. The irradiation plan for each fixed irradiation direction is generated by the multi-directional proton flash radiotherapy planning optimization method described in Embodiment 2 below. The final output multi-directional fixed-field proton FLASH plan includes spatial direction parameters. Each fixed irradiation direction includes a single energy, a compensated energy, a CEM thickness map, aperture collimator parameters, a set of range modulator parameters (including primary range modulator parameters and equalizing range modulator parameters), a set of irradiation point coordinates, and the corresponding MU fraction. The planning execution control component controls the switching timing of the fast beam switching component 6 (Kicker magnet or electromagnetic deflection component) according to these planning parameters, so that the proton beam enters different fixed gantry irradiation channels sequentially or in a preset order, and controls the single-energy irradiation component and beam shaping component 13 to complete multi-angle fixed-field single-energy irradiation.

[0046] Example 2: Method and System for Optimizing Multidirectional Proton Flash Radiotherapy Planning In Example 1, a novel multi-directional proton flash radiotherapy device was provided. However, the introduction of the multi-directional irradiation structure significantly increased the difficulty and stability of radiotherapy planning calculations. To address this issue, this example provides a radiotherapy planning optimization method and system applicable to the multi-directional proton flash radiotherapy device described in Example 1.

[0047] The multi-directional proton flash radiotherapy planning optimization system can be integrated into existing computing devices and communicate with the multi-directional proton flash radiotherapy device described in Example 1. The computing devices can be selected from PCs, servers, edge computing devices, etc.; or, it can be integrated into the multi-directional proton flash radiotherapy device described in Example 1 to form an embedded device.

[0048] The multi-directional proton flash radiotherapy planning optimization system includes: The data input module is configured to acquire patient planned input data; The reference plan generation module is configured to generate a reference plan based on the patient plan input data, under minimal constraints including target volume dose coverage, target volume dose hotspots, shell dose constraints, organ at risk dose limits, extra-target dose hotspots, field fractional balance, and MU smoothness constraints. The reference plan serves as a continuous reference for subsequent physicalization plans. The physicalization plan generation module is configured to construct a constrained plan space for a multi-directional fixed-field proton FLASH plan based on the reference plan without conformal energy modulator constraints. The process of transforming a plan from few constraints to a constrained plan space includes the following steps: The multi-directional, multi-depth irradiation contributions in the reference plan are converted into a point structure that meets the requirement of lateral non-overlap. Using the point structure, the beam-by-beam reference dose in the reference plan, and the single-energy irradiation conditions as inputs, a conformal energy modulator thickness map is generated or corrected. In this invention, the single-energy irradiation condition is an energy parameter. For a given fixed irradiation direction, the maximum energy from the original multi-energy layer in the less constrained reference plan corresponding to that direction is selected, or a larger energy that can cover the target range requirement in that direction is artificially set as the single energy for that fixed irradiation direction. All irradiation points within this fixed irradiation direction are irradiated based on this single energy, and multi-energy layer switching is not performed within the same fixed irradiation direction.

[0049] After generating or correcting the conformal energy modulator thickness map, energy compensation or energy boosting is performed based on single-energy irradiation conditions to adjust the MU fraction at the irradiation point; The aforementioned reference plan will be continuously referenced in each step of the plan conversion process; Within the constrained planning space, a physicalization plan is generated or modified so that the dose distribution of the physicalization plan continuously approximates the dose distribution of the reference plan. The Monte Carlo dose calculation module is configured to perform Monte Carlo dose calculations based on the final physicalization plan; The output module is configured to output the corresponding physical plan as a multi-directional fixed-field proton FLASH plan when the Monte Carlo dose meets the preset dosimetric acceptance conditions.

[0050] Compared to traditional single-field irradiation, multi-field irradiation systems introduce multiple angles as new variables. Traditional FLASH component optimization algorithms cannot logically respond to and optimize each field to achieve a good dose distribution, making it difficult for any algorithm to efficiently produce optimized results through combination. To address this issue, in this embodiment, the core idea of ​​the method for optimizing multi-directional proton flash radiotherapy plans using the aforementioned system is not to simultaneously optimize the point beam coordinates, MU, energy compensation, and conformal energy modulator thickness map of all fixed irradiation directions under strict deliverable constraints. Instead, it first generates a reference plan with higher dosimetric freedom under fewer constraints, and then uses this reference plan as a continuous reference to sequentially complete operations such as four-square non-overlapping point mapping, conformal energy modulator thickness map generation or correction, energy compensation, MU adjustment after modulator fixation, and Monte Carlo dosimetric acceptance. Through this weakly coupled transformation process, the system avoids simultaneously and freely optimizing all fields, irradiation points, energy, MU, and CEM thickness maps under strict physical delivery constraints, thereby reducing optimization difficulty and improving plan stability.

[0051] Specifically, the flowchart of this embodiment is shown below. Figure 3 As shown, it includes the following steps: Step S1, Obtain patient plan input data: First, patient planning input data is acquired. In this embodiment, the patient planning input data includes: patient CT images, patient body contour, planned target volume (PTV) or tumor target volume (GTV), target volume outer shell dose limitation area, one or more organs at risk, body regions, hot spot control area outside the target volume, prescribed dose, target coverage target, target volume hot spot limitation, organ at risk dose limitation, hot spot limitation outside the target volume, and fixed irradiation direction set.

[0052] The set of fixed irradiation directions is denoted as: in, These represent four fixed illumination directions.

[0053] The patient voxel set involved in dose calculation or dose evaluation is denoted as: Where N represents the number of voxels. The set of voxels for the planned target region or tumor target region is denoted as: The set of voxels representing the r-th organ at risk is denoted as: The dose-limiting region of the target area shell is denoted as Body regions are recorded as The hotspot control area outside the target area is denoted as Prescription dosage is recorded as follows: The upper limit of the target hotspot dose is denoted as , No. The upper limit of dose for each organ at risk is denoted as The upper limit of the hot spot dose outside the target area is denoted as .

[0054] Step S2, generate a reference plan with fewer constraints: Based on the patient planning input data, including target dose coverage Target area dose hotspot Shell dose constraint Dosage limitations that endanger organs Dose hotspots outside the target area Balance of field share and MU smoothness constraints Under fewer constraints, a reference plan is generated, which serves as a continuous reference for subsequent physicalization plans.

[0055] Specifically, the necessary parameters for constructing fewer constraints are first calculated based on the patient's planned input data: In this embodiment, the system uses four fixed angles by default: in, This indicates the transverse radiation field. Because the patient's tumor may be skewed, a transverse radiation field that is too far away would produce unnecessary doses to the body. Therefore, the transverse radiation field is selectively applied based on whether the tumor is skewed to the left or right side of the patient. or This is to ensure that the transverse radiation field is closer to the side of the tumor than to the skin. This corresponds to the actual positioning requirement of the patient being head-in and feet-in, which is equivalent here. Therefore, the radiation field is divided into four types based on its physical distance from the tumor: distal oblique field, vertical field, proximal oblique field, and transverse field. After generating the initial four field angles, this method further performs lateral adaptive determination of the transverse radiation field.

[0056] As an example, in a specific radiotherapy case, the requested angles were 0°, 45°, -45°, and -90°. After lateral field determination, the actual effective angles became 0°, 45°, -45°, and 90°. That is, the fourth lateral field was reversed from -90° to +90°, entering from the patient's left side to bring the superficial lateral beam closer to the target area, reducing unnecessary deep penetration paths. After the lateral field was reversed, the system simultaneously rearranged the roles of the four fields. Beams with an absolute angle of 90° were marked as superficial lateral beams; oblique beams with the same sign and an angle of 45° were marked as superficial oblique beams; oblique beams with a different sign and an angle of 45° were marked as deep oblique beams; and the 0-degree beam was defined as a vertical field.

[0057] Then, patient geometry initialization is performed. The patient's planned input data also includes the target area's three-dimensional bounding box parameters, which include the left-right length, front-back length, and head-to-toe length. The system calculates the target area's cross-sectional diagonal based on the left-right and front-back lengths. : in, Indicates the length in the left and right directions. This indicates the forward and backward length. In this embodiment, the reference cross-sectional dimension is set to 60.0 mm. When the diagonal of the target area's cross-section does not exceed 60.0 mm, no additional allowance is added. When the diagonal of the target area's cross-section exceeds 60.0 mm, an additional allowance is added for every 20.0 mm exceeding this limit, with each allowance corresponding to an additional 5.0 mm edge allowance, rounded up to the nearest 5.0 mm barrel width. This additional allowance can be expressed as: in, Indicates additional edge allowance. This indicates the number of size increments corresponding to the excess amount, i.e., how many additional aperture allowance increments are needed. In this embodiment, each 20.0 mm excess amount corresponds to one increment. This indicates the amount by which the lateral dimension of the target area exceeds the reference lateral dimension, expressed in millimeters; specifically, it is the diagonal length of the target area within the lateral plane. The portion exceeding the reference lateral dimension of 60.0 mm. When When not exceeding 60.0 mm, Take 0. This function rounds up to the nearest integer, rounding the value within the parentheses to the nearest integer. For example, ceil(0.1) = 1, ceil(1.0) = 1, and ceil(1.2) = 2. express Increase by one level every 20 mm, with each level being 5 mm. For example... In this embodiment, the basic target area margin is 0.0 mm, and the basic aperture edge gap is 7.0 mm. Therefore, the final target area margin is... and the final aperture edge gap They are respectively: beam working diameter Inner diameter of the aperture and aperture outer diameter They are respectively: in, This refers to the transverse wall thickness of the aperture. If any predicted dimension exceeds the corresponding upper limit, the system reports the size as too large; if none exceed the upper limit, the system reports the status as normal and applies the largest aperture size to all four fixed irradiation directions. The aforementioned aperture setting is a geometric initialization parameter, its function being to consistently apply the same set of patient geometric boundaries to the reference plan generation and subsequent physicalization plan generation stages, rather than serving as a subsequent optimization variable. Under minimal constraints—without introducing conformal energy modulator geometric constraints, single-energy physical delivery constraints, or lateral non-overlapping point constraints—the system generates a minimally constrained plan without conformal energy modulator constraints. Since this plan serves as a reference for subsequent optimization, it can also be called a reference plan. During the generation of this minimally constrained plan, an objective function is established that includes penalties for insufficient target coverage, target hotspots, outer shell dose-limited regions, organs at risk of dose, outside-target hotspots, field share balancing, and MU smoothing. in, This represents the radiation vector to be optimized in a less constrained plan. This represents the irradiance vector obtained by minimizing the above objective function under non-negativity constraints. This indicates a penalty for insufficient target coverage. Indicates the target area hotspot penalty item. This indicates a penalty for the target area's dose limitation region. This represents the dose penalty term for the r-th organ at risk. Indicates the penalty for hotspots outside the target area. This represents the dose fraction or exposure fraction balance term between fixed irradiation directions. This represents the smoothing term of the irradiation point MU. This indicates the weight of the fixed irradiation direction share balance term. The terms represent the weights of the radiation smoothing term; they are used to adjust the relative importance of the beam fraction equalization requirement and the radiation spatial smoothing requirement in the objective function, respectively. The target area cold spot penalty can be written as: The target area hotspot penalty can be written as: Punishment for endangering organs can be written as: The penalty for hotspots outside the target area can be written as: in, . Let represent the planned dose value at the i-th voxel, i∈T, and let represent that the i-th voxel belongs to the target region voxel set, i∈OAR. r This indicates that the i-th voxel belongs to the r-th set of voxels in the endangered organ, i∈H. outside This indicates that the i-th voxel belongs to the hotspot evaluation region outside the target area. This represents the target area cold spot penalty weight, used to control the target area dose from falling below the prescribed dose D. pres The intensity of the penalty at that time; the larger the value, the more the optimization tends to increase the dose in the under-dose region of the target area. This represents the target hotspot penalty weight, used to control the target dose from exceeding the target hotspot limit. The intensity of the penalty at that time; the larger the value, the more the optimization tends to suppress excessively high doses within the target area. This represents the dose over-limit penalty weight for the r-th organ at risk, used to control the dose to that organ from exceeding its dose limit. The severity of punishment at different times; different weights can be set for different organs at risk. This represents the penalty weight for hotspots outside the target area, used to control dose exceeding the limit within the hotspot evaluation area outside the target area. The intensity of the penalty at that time; the larger the value, the more the optimization tends to reduce the unexpected high dose outside the target area.

[0058] The total reference dose of the low-constraint plan Represented as: in, This represents the dose-effect matrix of the less-constrained plan. Indicates the weight of exposure amount; The total reference dose of the reference plan is decomposed into beam-by-beam reference doses corresponding to each fixed irradiation direction: in, This represents the beam-by-beam reference dose for a fixed irradiation direction b. This represents the dose effect matrix for a fixed irradiation direction b. This represents the dose weight of fixed irradiation direction b, where B represents the number of fixed irradiation directions, and the subscript 0 indicates the dose vector to be optimized in a less constrained plan. Through this step, each of the four fixed irradiation directions obtains its own dose responsibility. During subsequent generation or correction of the conformal energy modulator thickness map, each fixed irradiation direction preferentially approximates its corresponding beam-by-beam reference dose. As an example, to avoid dose responsibility collapse among the four fixed irradiation directions, this embodiment sets the total dose share of the distal oblique field and vertical field to 40%, defining it as the deep irradiation group, and sets the total dose share of the proximal oblique field and lateral field to 60%, defining it as the shallow irradiation group, and makes the irradiation doses of each field within the group as similar as possible.

[0059] Through the above-described beam-by-beam dose responsibility separation, each fixed irradiation direction preferentially approaches its corresponding beam during the subsequent physicochemical conversion process. This avoids strong coupling and free optimization of conformal energy modulator thickness maps, energy compensation, and MU fractions for four fixed irradiation directions at the same stage. To prevent a single field from bearing almost the entire dose during the reference planning stage, this invention sets field fraction constraints. .

[0060] As a specific example, in this embodiment, the threshold for endangering organ weights... The settings are shown in Table 1, with weights for the cold zone item. Set to 2.0, weight of hot items Set to 0.8, weight of background hotspot items Set to 0.35, and set the mask inflation to 3 voxels.

[0061] Table 1. Organ-at-risk dose limits applied in this embodiment. Preferred Target D in the Reference Planning Phase 95 Reaching 30 Gy and making target D max Less than 39 Gy; considering that subsequent CEM physicalization may cause dose fluctuations, the system sets a gating target: Regulation D 95 Greater than 27 Gy Regulation D max A non-strict threshold of less than 42 Gy is used to prevent the reference plan or intermediate candidate plan from being prematurely eliminated due to slight fluctuations. These preferred targets can be set as constraints in the reference plan optimization process.

[0062] Step S3, Generate the physicalization plan: Based on the reference plan without conformal energy modulator constraints, a constrained plan space for a multi-directional fixed-field proton FLASH plan is constructed. Within this constrained plan space, physicalization plans are generated or modified to ensure that the dose distribution of these physicalization plans continuously approximates the dose distribution of the reference plan.

[0063] Specifically, the process of transforming a plan from few constraints to a constrained plan space includes the following steps: 1. Generate a non-overlapping square grid structure: The multi-directional, multi-depth irradiation contributions in the reference plan are converted into a point structure that meets the requirement of lateral non-overlapping (i.e., a four-square non-overlapping point structure).

[0064] The point structure is generated in the following manner: A parent horizontal grid is constructed with a preset spacing. In a preferred embodiment, the spacing of the parent horizontal grid is 8 mm. Four non-overlapping sub-slots are set within each parent horizontal grid. In a preferred embodiment, the sub-slots are 4 mm lateral offset sub-slots, so that each point occupies a square grid size with a 4 mm boundary.

[0065] The four child slots corresponding to the p-th parent horizontal grid are represented as follows: The coordinates of the sub-slot are represented as follows: in, The coordinates of the parent's horizontal grid center. For the first The sub-slots are offset laterally relative to the center of the parent horizontal grid.

[0066] Several depth irradiation contribution points located within the same parent horizontal grid in the reference plan are sorted according to MU share, dose contribution, target coverage importance, or organ-at-risk protection importance. A depth irradiation contribution point refers to a scan point in the reference plan located within a parent horizontal grid, corresponding to a target depth or water-equivalent depth layer, and contributing dose to that depth region. In other words, when the same horizontal position corresponds to irradiation contributions at multiple different depths, i.e., when scan points overlap, each depth layer scan point can be considered a depth irradiation contribution point. Since depth irradiation contribution points within this region contribute dose at different depths within that horizontal position, sorting, filtering, moving, and deleting are necessary to ensure that the dose of non-overlapping plans remains as constant as possible.

[0067] In this embodiment, eight candidate depth layers are first generated along the depth direction. These eight candidate depth layers are generated from the 5th, 15th, 25th, 40th, 55th, 70th, 85th, and 95th percentiles of the equivalent water depth distribution in the target area under a fixed irradiation direction, and are denoted as: in, Indicates the first The equivalent depth distribution of water in the target area under a fixed irradiation direction. Percentile.

[0068] In this embodiment, multiple depth illumination contribution points located within the same parent horizontal grid in the reference plan are represented as follows: in, Indicates the horizontal grid number of the parent element. Indicates depth or spheric number. Indicates the equivalent depth of the target water. This indicates the MU or radiation dose at that contribution point. This indicates the importance assessment value of the contribution point. The importance assessment value can be determined based on MU, dose contribution, target coverage contribution, or organ-at-risk protection contribution.

[0069] The sorted depth illumination contribution points are assigned to different slots in the four non-overlapping sub-slots to avoid different depth contribution points occupying the same physical lateral position. When the number of depth illumination contribution points in the same parent lateral grid exceeds four, the four sorted depth illumination contribution points are retained first, and the unassigned sorted depth illumination contribution points are pruned, or the unassigned sorted depth illumination contribution points are migrated to the empty sub-slots in the adjacent parent lateral grid.

[0070] When generating the point structure, an available sub-slot is selected for each depth illumination contribution point, and the cost of lateral movement is minimized. Let the depth irradiation contribution point be c, and its assigned sub-slot be s(c). Then the slot allocation target is: in, Indicates the contribution point of deep irradiation Assigned to sub-slot The cost; The cost function includes one or more of the following: lateral displacement cost, cross-parent lateral mesh penalty, and importance weight of depth illumination contribution points, and can be expressed as: Where, q= , Indicates the contribution point of deep irradiation The importance of; Represents the horizontal coordinate of the candidate sub-slot; Indicates the contribution point of deep irradiation The original horizontal coordinates in a less constrained plan; Indicates the parent horizontal grid to which the candidate sub-slot belongs; Indicates the contribution point of deep irradiation The original parent horizontal grid; Indicates the penalty weight across the parent horizontal grid; This indicates the value to be taken when the candidate child slot is inconsistent with the original parent's horizontal grid. The indicator function. Wherein, The importance rating value is used to indicate the importance of the c-th depth irradiation contribution point in the process of point allocation, retention, migration, or thickness map correction. The larger the value, the more important the irradiation point is for maintaining the reference dose or target coverage, and the more likely optimization will retain the point or reduce its offset.

[0071] After generating the point structure, a reference association identifier is established for each irradiation point in the point structure. The information of the reference association identifier includes one or more of the following: original parent horizontal grid number, sub-slot number, fixed irradiation direction, target water equivalent depth, candidate depth layer number, original irradiation dose, normalized irradiation dose, importance evaluation value, beam-by-beam reference dose contribution number, and corresponding reference plan irradiation point number.

[0072] In subsequent steps of generating or revising the conformal energy modulator thickness map, energy compensation, and MU share adjustment, the reference association identifier remains unchanged. That is, based on the reference association identifier, the point relationships, field dose responsibilities, and MU allocation relationships between the constrained physicalization plan and the less constrained plan without conformal energy modulator constraints are maintained. This ensures that the subsequent plan conversion process does not involve a free search for irradiation points, field responsibilities, and MU shares again, but rather performs physical corrections while preserving the structural relationships of the less constrained plan.

[0073] 2. Generate or correct the thickness map of the conformal energy modulator: Using the point structure (i.e., the four-square non-overlapping point structure), the beam-by-beam reference dose and single-energy irradiation conditions in the less constrained plan as inputs, a conformal energy modulator thickness map is generated or corrected.

[0074] Each fixed irradiation direction employs an in-field single-energy proton beam, without multi-layer switching within the same fixed irradiation direction. The single-energy proton beam forms a conformal dose distribution through the combined effects of a conformal energy modulator thickness map, a range modulator, compensation energy, and MU fraction adjustment.

[0075] Inputs used to generate or correct conformal energy modulator thickness maps include site structure, beam-by-beam reference dose, single-energy irradiation conditions, reference association identifier, candidate depth layers, and irradiation point importance evaluation values. Conformal energy modulator thickness boundary and aperture geometry parameters ( and One or more of the following.

[0076] Algorithms for generating or correcting thickness maps of conformal energy modulators can be implemented based on existing algorithms. They can be implemented using existing CEM generation methods, CEM solving functions in open-source or commercial treatment planning systems, or other algorithms capable of generating or correcting thickness maps.

[0077] In this embodiment, in one specific implementation, the initial input parameters for the conformal energy modulator thickness map solving algorithm can be formed based on the target water equivalent depth at each non-overlapping irradiation point and the single-energy range of the fixed irradiation direction. Let the first... The current single energy is [value] for a fixed irradiation direction. The range function is The equivalent depth of the target water at the q-th non-overlapping point is The initial water equivalent thickness at that point can be denoted as: When manufacturing boundaries need to be considered, it can be written as: in, This indicates the minimum water-equivalent thickness that CEM can manufacture. This indicates the maximum water-equivalent thickness that CEM can produce. This represents a truncation function that restricts x to [ , The cutoff function within the interval, i.e., when x < Time to take When x> Time to take Otherwise, take x itself.

[0078] The initial physical height can be written as: in, It represents the relative blocking ability of CEM material, also known as the relative water equivalent conversion factor, used to convert the water equivalent thickness into the material's physical thickness.

[0079] In this embodiment, regarding CEM, for the continuous modulator surface, the first... The equivalent thickness diagram of the modulator water in each irradiation field is denoted as: The material physics height diagram is as follows: in, This represents the b-th illumination field on the horizontal coordinate. The equivalent thickness of the modulator water at the location; This represents the actual height of the modulator at the same location; Represents the horizontal coordinate under the eye view.

[0080] In this step, a physically reasonable initial modulator is first derived using the range difference. This allows for a relatively close dose distribution in the early stages of optimization, reducing unnecessary computational burden. This stage establishes the initial configuration of all physical components and determines their physical location, laying the foundation for the next step of optimization and Monte Carlo simulation.

[0081] Next, each field undergoes an independent optimization process from a minimally constrained plan without CEM to one with CEM. Since the four fields were previously split, this step can be computed in parallel, improving computer efficiency. Within each fixed irradiation field, the modulator thickness map is optimized to ensure that the single-beam dose after adding the modulator is as close as possible to the reference dose of that field. In this step, only the CEM thickness map is modified; other parameters are not altered.

[0082] For the first Each irradiation field, with the dose after adding the modulator, is denoted as . The superscript T indicates that the dose is produced by the modulator thickness map. The result after modulation is not a matrix transpose.

[0083] The reference dose for this irradiation field is The optimization objective is written as: in: Representing the Equivalent thickness map of modulator water in each irradiation field; Representing the The irradiation weight of each irradiation field; Represents the first value calculated under the current modulator and current irradiance. Individual irradiation field dose; Representing the Reference dose for each irradiation field; Representing the A voxel weight matrix for each irradiation field is used to enhance the importance of key voxels such as the target area, organs at risk, and hotspot areas. The symbol represents the squared error calculation. In this optimization objective, the modulator is required to reproduce the dose responsibility of the b-th irradiation field in an unconstrained plan. Since the reference dose is set as the single-field dose distribution of a less constrained plan, the final optimization focus is... Instead of the originally planned dose distribution .

[0084] To update the modulator thickness, the effect of dose error on the modulator thickness needs to be calculated. Let the gradient of the objective function with respect to dose be: in: Representing the The first illumination field Dosage residual gradient of individual elements; Representing the The objective function for modulator optimization in each illumination field; Representing the Current dose of individual factor.

[0085] Assume the effect of modulator thickness on voxel dosage is: in: This represents the modulator position in the b-th illumination field. The thickness variation on the first Effect of individual body dosage; Represents the position of the modulator The water equivalent thickness.

[0086] The direction of modulator thickness update is: in: Represents the position of the modulator The update direction is determined by the location. Therefore, if thickening at a location improves the dose error, the update proceeds along the thickening direction; if thinning at a location improves the dose error, the update proceeds along the thinning direction.

[0087] Each round of modulator thickness update can be written as: in: This represents the equivalent thickness of the modulator water at the k-th iteration; This represents the equivalent thickness of the modulator water after the (k+1)th iteration; This represents the update direction at the k-th iteration; This represents the step size of the k-th iteration; This represents the minimum water equivalent thickness allowed by the modulator; This represents the maximum water equivalent thickness allowed by the modulator; This represents a projection or clipping operation, indicating that the updated thickness cannot be lower than the minimum value or exceed the maximum value, and needs to be accurately reproduced because...

[0088] The optimal step size can be determined by the quantiles of the update direction: in: This represents the maximum allowable thickness variation scale in the k-th iteration; It is the 90th percentile of the absolute value of the update direction. This design avoids excessive overall updates caused by a few outlier gradient points. In addition, the compensation is dynamic, performing coarse adjustments when the result is far from the optimization result and fine adjustments when it is close to the optimization result.

[0089] If the dose target deteriorates after a certain update, the modulator thickness and dose from the previous round are rolled back, and the step size is reduced before retrying. In this example, the primary optimization of the compensated energy modulator adopts a beam-by-beam reference dose matching engine in the style of an open proton program system, and the primary downstream modulator source is a complete water equivalent thickness map. The optimization is divided into 6 preprocessing iterations, 6 coarse-stage iterations, 12 intermediate-stage iterations, and 18 fine-stage iterations; the corresponding maximum step sizes are 8.0 mm, 4.0 mm, and 2.0 mm, respectively, and the minimum step size limit is 0.10 mm. The objective function mode is mask shape matching, with a high dose threshold of 50% of the reference dose and a low dose threshold of 20%. The minimum step size limit is 0.10 mm, the absolute convergence tolerance is 0.002, and the acceptable objective function value is 0.002.

[0090] 3. Perform energy compensation: After generating or correcting the CEM thickness map, dose calculations are performed on the currently optimized plan, and cold zones or areas with insufficient range in the target area are identified. As an example, in this embodiment, cold zone determination uses a cold zone dose ratio of 0.90 and a reference dose fraction of 0.45, so the cold zone set can be represented as: Hotspot identification uses a hotspot ratio of 1.03 and a reference dose fraction of 0.70. The hotspot set can be represented as: in, This indicates the dose of voxel numbered i. This represents the reference dose at the i-th voxel of the less constrained reference plan.

[0091] When performing spatial positioning of cold and hot zones, the center weighting scale is set to 10 mm.

[0092] For the For a fixed irradiation direction, calculate the maximum water equivalent path length of the voxel in the cold region along that direction: in, Indicates the first Individual elements along the first The equivalent path length of water in a fixed irradiation direction. If the energy of this field under the current single-energy irradiation conditions... Corresponding range If the range is less than the range required for the cold zone, the insufficient range is expressed as: in, This indicates the safety margin of the firing range.

[0093] During energy compensation, the system calculates the range deficiency based on the equivalent path length of water along the corresponding fixed irradiation direction of the voxel in the cold zone of the target area, and determines the energy increase based on the monotonic relationship between proton energy and range. Let the first... The current energy is for a fixed irradiation direction. (i.e., the energy of the field under single-energy illumination conditions), the range function is The range obtained from cold zone exploration is insufficient. Then the increase in energy can be approximated as: in, This represents the energy safety boundary. The energy after compensation. for: To prevent the CEM thickness from exceeding the manufacturing range after energy increase, the following must be met: in, This represents the equivalent depth of water near the target area along the b-th fixed irradiation direction. Indicates the maximum permissible water equivalent thickness of the modulator. This indicates a safety margin for modulator thickness. This constraint is used to ensure energy... Even after the upgrade, the CEM still has enough thickness to pull the excessive range back into the target area.

[0094] In this embodiment, after the cold zone energy is repaired, a Monte Carlo test is performed. If it still does not reach an acceptable range, the next round of repair is performed, and this cycle is executed for a maximum of 3 rounds. The weight of organs at risk in the cycle is multiplied by 1.50. Energy priority repair is based on the candidate energy increase given by the irradiation field; the conventional maximum energy increase is 10 MeV; the aggressive coverage mode allows an automatic maximum increase of at least 24 MeV, with an additional overshoot of 4 MeV. When the distal coverage margin is insufficient but no obvious cold zone has been formed, the system can perform a preventive energy boost on the corresponding fixed irradiation direction. In a preferred embodiment, the preventive energy boost is as follows: in, The energy after compensation is, This indicates the compensated energy after preventative boosting. Energy compensation or boosting triggers CEM thickness map reconstruction or correction, ensuring the boosted range is still reshaped to the target depth region via the CEM thickness map, preventing dose increases in non-target areas caused by simple energy boosting. In some embodiments, the system can also synchronize global energy via a uniform energy range modulator, ensuring the final four fixed irradiation directions meet monoenergetic irradiation and system delivery requirements.

[0095] 4. Adjust MU after fixing CEM: Generate or correct the conformal energy modulator thickness map and determine the irradiation energy of the field. Then, the MU fraction of the irradiation point is adjusted.

[0096] The adjustment of the MU share includes: After fixing the thickness map of the conformal energy modulator for each fixed irradiation direction, the MU fraction at the irradiation point is optimized. The MU adjustment aims to approximate the total reference dose of the constrained plan without conformal energy modulator constraints with the dose synthesized from multiple fixed irradiation directions as the MU adjustment target, and introduces MU constraint terms and field constraint terms into the MU adjustment target. Among them, the MU constraint term is used to limit the MU of a single irradiation point from deviating significantly from the plan of the previous cycle in the less constrained plan, the non-overlapping point plan, or the optimization cycle. The objective term considering the MU constraint is expressed as: ; The field constraint term is used to limit the total MU of each fixed irradiation direction from deviating significantly from the total MU of the corresponding fixed irradiation direction in the reference plan; The objective term considering the field constraint is expressed as: ; The MU adjustment target is expressed as: in: This represents the radiation weight obtained after optimization based on a non-overlapping point plan. Represents the irradiation weight to be optimized; This indicates that all exposure weights are non-negative; Represents the patient voxel weight matrix; This represents the dose effect matrix after fixing the modulator; This indicates an unlimited total reference dose; Indicates the first The weight of individual elements; This represents the reference value of the irradiation obtained after the modulator's master optimization. This represents the total number of radiation dose variables; Indicates the weight of the exposure maintenance term; Indicates the number of fixed irradiation directions; This indicates that the j-th irradiation point belongs to the j-th irradiation point. A fixed irradiation direction; Indicating an unrestricted reference plan Total irradiance from a fixed irradiation direction; This indicates the weight of the term that maintains the total amount of illumination in a fixed direction.

[0097] 5. Perform weakly coupled transformation iterations The final physicalization plan was generated through the following steps: Step 1: Repeat the weakly coupled conversion process, which includes the following steps: Step 1.1: Following the same method as the previous steps, take the point structure, the beam-by-beam reference dose in the reference plan, and the single-energy irradiation conditions as inputs to generate or correct the conformal energy modulator thickness map; Step 1.2: Following the same method as the previous steps, based on the generated or corrected conformal energy modulator thickness map, and according to the target area cold zone exploration results and the insufficient range, perform energy compensation or energy enhancement on the corresponding fixed irradiation direction. Step 1.3: Following the same method as the previous steps, adjust the MU fraction of the irradiation point based on the generated or corrected conformal energy modulator thickness map to obtain a candidate plan; Step 1.4: Evaluate the target coverage, target hotspots, organ-at-risk doses, extra-target hotspots, doses in the outer shell dose-limited region, and doses in the body region based on the Monte Carlo dose calculation results obtained in Step 3. or ratio; The weakly coupled conversion process is executed in a maximum of three rounds. When a candidate plan in a certain round is better than the current best plan, the candidate plan is retained as the new best plan. When a candidate plan in a certain round deteriorates, it is rolled back to the current best plan. The energy compensation or energy boost includes calculating the range deficiency based on the water equivalent path length of the voxel in the cold zone of the target area along the corresponding fixed irradiation direction, and determining the energy increase based on the monotonic relationship between proton energy and range; when the far-end coverage margin is insufficient but a significant cold zone has not yet been formed, a preventive energy boost is performed on the corresponding fixed irradiation direction. Step 2: After obtaining the final optimal plan, a uniform range modulator is used to complete the global energy unification and obtain the final physical plan.

[0098] In this embodiment, the default maximum number of iterations for the less constrained reference plan is 50; the default maximum number of iterations for generating or correcting the conformal energy modulator thickness map is 60; the default maximum number of iterations for the energy compensation stage is 30; the default maximum number of iterations for recalculating the total dose residual conformal energy modulator candidate is 30; the default maximum number of iterations for Monte Carlo true small beam correction is 80; and the default maximum number of iterations for fine-tuning the per-beam irradiation dose after fixing the conformal energy modulator is 60.

[0099] L-BFGS-B algorithm objective function change convergence threshold Set as Projection gradient convergence threshold Set as The maximum number of line searches in each major iteration Set it to 30.

[0100] Step S4, Monte Carlo Dosage Calculation and Final Acceptance: Monte Carlo dose calculations are performed based on the final physicalization plan. When the Monte Carlo dose meets the preset dosimetric acceptance criteria, the corresponding physicalization plan is output as a multi-directional fixed-field proton FLASH plan.

[0101] In this embodiment, after the weak coupling conversion process is completed, the system adds an equal-energy range modulator with the same logic as the energy modulator to unify the global energy and perform Monte Carlo dose calculation based on the final fixed irradiation direction, final single energy, final conformal energy modulator thickness map, final compensated energy, final irradiation point coordinates, and final MU fraction.

[0102] The average range modulator is calculated as follows: A uniform maximum energy delivery strategy was employed before the final Monte Carlo dose calculation. First, the values ​​of each beam were read. Compensated energy, and all beams are taken. The maximum value of the compensated energy is used as the global delivery energy; if the user configures a higher unified energy... If the two values ​​are greater, then the larger value is taken. Subsequently, for each beam, the difference between the water range corresponding to the unified global energy and the water range corresponding to the original anchoring energy of the beam is calculated, and this difference is used as the water equivalent thickness of the beam's range modulator.

[0103] The thickness of the average range modulator for this firing field is: in, To unify overall energy; This is an optional feature, i.e., a unified global energy for four illumination fields configured manually; The energy after compensation for the b-th field; The required increase in the thickness of the uniform range modulator; For energy Corresponding underwater range; The water equivalent thickness of the modulator for the b-th firing field range; The physical thickness of the range modulator; The relative stopping power of the range modulator material. Calculation. or The ratio, and its use as a dosimetric gating indicator: in, This indicates the dose corresponding to 1% of the volume of the high-dose region. It should be noted that the final acceptance in this embodiment is primarily based on Monte Carlo dosimetry indicators, and the voxel FLASH dose rate threshold is not used as a necessary acceptance criterion for the output plan.

[0104] The Monte Carlo dosimetric index (candidate plan comprehensive penalty score) is calculated using the following formula: in, This represents the overall penalty score for the candidate plan; a lower score is better. This indicates that PTVD95 is valid; Indicates valid PTV Dmax; Indicates target D95; Indicates the lower limit of the gate D95; Indicates the upper limit of the target Dmax; Indicates the upper limit of the gate Dmax; This indicates that hotspots outside the target area have exceeded the limit; Indicates the volume of hotspots outside the target area; This indicates that the amount of drugs used to endanger organs has exceeded the limit.

[0105] A candidate plan (the plan obtained after a round of optimization) is considered better mainly in two cases: D95 should be increased by at least 0.15 Gy, while Dmax should deteriorate by no more than 0.75 Gy, and hot spots and OAR outside the target area should not deteriorate significantly. Alternatively, the overall penalty score improves by at least 0.10, while D95 decreases by no more than 0.10 Gy, and hotspots, OARs, and out-of-target hotspots remain under control.

[0106] If the final candidate plan does not meet the dosimetric acceptance requirements, the system returns the current best candidate plan, or triggers the regeneration or correction of the conformal energy modulator thickness map, energy compensation, or MU adjustment based on the reason for the failure.

[0107] In this embodiment, the final output plan includes directional markers and spatial direction parameters for four fixed irradiation directions, the final single energy corresponding to each fixed irradiation direction, the compensated energy or range modulator (including primary range modulator and average energy range modulator) parameters corresponding to each fixed irradiation direction, the CEM thickness map corresponding to each fixed irradiation direction, the irradiation point coordinates corresponding to each fixed irradiation direction, the MU fraction corresponding to each irradiation point, the final Monte Carlo three-dimensional dose distribution, target area coverage results, target area hotspot results, organ-at-risk dose results, target area hotspot results, target area dose-limited region dose results, and body region dose results. and result.

[0108] For a specific application example, Tables 2 and 3 below show the changes in various optimization target indicators during the radiotherapy planning optimization process according to the method of this embodiment.

[0109] Table 2 shows the changes in the target area and D95 ​​during the optimization process of Embodiment 2 of the present invention. Table 3 shows the changes in the average dose to organs at risk during the optimization process of Example 2 of the present invention. Table 4 shows the changes in the maximum dose to the target area and organs at risk during the optimization process of Example 2 of the present invention. As can be seen from the above data, this embodiment provides a corresponding radiotherapy planning optimization method for the novel multi-directional proton flash radiotherapy device provided by the present invention, which can optimize the radiotherapy plan for specific optimization objectives.

Claims

1. A multidirectional proton flash radiotherapy device, characterized in that, include: A proton source, used to generate a proton beam; A beam transport assembly for transmitting the proton beam to several fixed gantry irradiation channels; A rapid beam switching assembly is used to quickly switch the specific fixed gantry irradiation channel into which the proton beam enters; Several fixed gantry irradiation channels are used to irradiate proton beams along several preset fixed irradiation directions, and the fixed gantry irradiation channels are arranged around the patient's treatment area; The fixed gantry irradiation channel includes a single-energy projection component and a beam shaping component. The single-energy projection component is used to project a proton beam into each fixed irradiation direction with a preset single-energy or energy-compensated single-energy proton beam. The beam shaping component is used to shape the proton beam in three-dimensional space so that its dose distribution matches the shape and depth of the tumor target area.

2. The multidirectional proton flash radiotherapy device according to claim 1, characterized in that, Also includes: The plan execution control component is used to control the fast beam switching component, the monoenergetic irradiation component, and the beam shaping component to complete the irradiation of the multi-angle fixed field proton beam according to the plan parameters of the multi-directional fixed field proton FLASH plan. The generation process of the multi-directional fixed-field proton FLASH plan includes: Obtain patient plan input data; Based on the patient planning input data, a reference plan is generated under minimal constraints, including target volume dose coverage, target volume dose hotspots, shell dose constraints, organ at risk dose limits, extra-target dose hotspots, field fractional balance, and MU smoothness constraints. This reference plan serves as a continuous reference for subsequent physicalization plans. Based on the reference plan without conformal energy modulator constraints, a constrained plan space for the multi-directional fixed-field proton FLASH plan is constructed. The process of transforming a plan from few constraints to a constrained plan space includes the following steps: The multi-directional, multi-depth irradiation contributions in the reference plan are converted into a point structure that meets the requirement of lateral non-overlap. Using the point structure, the beam-by-beam reference dose in the reference plan, and the single-energy irradiation conditions as inputs, generate or correct the conformal energy modulator thickness map. After generating or correcting the conformal energy modulator thickness map, perform energy compensation or energy boost to adjust the MU fraction at the irradiation point; The aforementioned reference plan will be continuously referenced in each step of the plan conversion process; Within the constrained planning space, a physicalization plan is generated or modified so that the dose distribution of the physicalization plan continuously approximates the dose distribution of the reference plan. Monte Carlo dosing calculations were performed based on the final physicochemical plan; When the Monte Carlo dose meets the preset dosimetric acceptance criteria, the corresponding physical plan will be output as a multi-directional fixed-field proton FLASH plan.

3. The multi-directional proton flash radiotherapy device according to claim 2, characterized in that, The beam shaping assembly comprises one or more of a conformal energy modulator, a range modulator, and an aperture collimator; the range modulator includes one or both of an initial range modulator and an equalizing range modulator. The reference plan, physicalization plan, or multi-directional fixed-field proton FLASH plan includes the following plan parameters: spatial orientation parameters for each fixed irradiation direction, single energy, compensated energy, conformal energy modulator thickness map, range modulator parameters, aperture collimator parameters, irradiation point coordinates, and MU fraction; the range modulator parameters include one or both of the initial range modulator parameters and the uniform energy range modulator parameters. The multi-directional proton flash radiotherapy device uses a single-energy proton beam globally. In a specific fixed irradiation direction, the single-energy proton beam in the field forms a conformal dose distribution through the thickness map of the conformal energy modulator, the initial range modulator parameters, the compensated energy, and the MU fraction adjustment. Finally, the global energy is synchronized through the uniform range displacement.

4. The multidirectional proton flash radiotherapy device according to claim 2, characterized in that, During the reference plan generation process, an objective function is established that includes penalties for insufficient target coverage, target hotspots, outer shell dose-limiting areas, organ-at-risk doses, outside-target hotspots, field-share balancing, and MU smoothing. in, This represents the radiation vector to be optimized in a less constrained plan. This represents the irradiance vector obtained by minimizing the above objective function under non-negativity constraints. This indicates a penalty for insufficient target coverage. Indicates the target area hotspot penalty item. This indicates a penalty for the target area's dose limitation region. This represents the dose penalty term for the r-th organ at risk. Indicates the penalty for hotspots outside the target area. This represents the dose fraction or exposure fraction balance term between fixed irradiation directions. This represents the smoothing term of the irradiation point MU. This indicates the weight of the fixed irradiation direction share balance term. Indicates the weight of the dose smoothing term; the total reference dose of the reference plan. Represented as: in, This represents the dose-effect matrix of the reference plan. Indicates the weight of exposure amount; The total reference dose of the reference plan is decomposed into beam-by-beam reference doses corresponding to each fixed irradiation direction: in, This represents the beam-by-beam reference dose for a fixed irradiation direction b. This represents the dose effect matrix for a fixed irradiation direction b. B represents the irradiation weight of a fixed irradiation direction b, where B represents the number of fixed irradiation directions. In the subsequent physicalization planning process, each fixed irradiation direction will be prioritized based on its corresponding... As a continuously approaching, beam-by-beam reference dose.

5. The multidirectional proton flash radiotherapy device according to claim 2, characterized in that, The final physicalization plan was generated through the following steps: Step 1: Repeat the weakly coupled conversion process, which includes the following steps: Step 1.1: Using the point structure, the beam-by-beam reference dose in the reference plan, and the single-energy irradiation conditions as inputs, generate or correct the conformal energy modulator thickness map; Step 1.2: Based on the generated or corrected conformal energy modulator thickness map, perform energy compensation or energy boosting on the corresponding fixed irradiation direction according to the target cold zone exploration results and the insufficient range. Step 1.3: Based on the generated or corrected conformal energy modulator thickness map, adjust the MU fraction at the irradiation point to obtain a candidate plan; Step 1.4: Evaluate the candidate plans obtained in Step 3 based on the Monte Carlo dose calculation results; The weakly coupled conversion process is executed in a maximum of three rounds. When a candidate plan in a certain round is better than the current best plan, the candidate plan is retained as the new best plan. When a candidate plan in a certain round deteriorates, it is rolled back to the current best plan. The energy compensation or energy boost includes calculating the range deficiency based on the water equivalent path length of the voxel in the cold zone of the target area along the corresponding fixed irradiation direction, and determining the energy increase based on the monotonic relationship between proton energy and range; when the far-end coverage margin is insufficient but a significant cold zone has not yet been formed, a preventive energy boost is performed on the corresponding fixed irradiation direction. Step 2: After obtaining the final optimal plan, a uniform range modulator is used to complete the global energy unification and obtain the final physical plan.

6. The multidirectional proton flash radiotherapy device according to claim 5, characterized in that, The point structure is generated in the following manner: Construct a parent horizontal grid with a preset spacing, and set four non-overlapping child slots within each parent horizontal grid; Several depth irradiation contribution points located within the same parent horizontal grid in the reference plan are sorted according to MU share, dose contribution, target coverage importance, or organ-at-risk protection importance; the depth irradiation contribution point refers to a scan point in the reference plan located within a certain parent horizontal grid, corresponding to a target depth or water equivalent depth layer, and contributing dose to that depth region. The sorted depth illumination contribution points are assigned to different slots in the four non-overlapping sub-slots to avoid different depth contribution points occupying the same physical lateral position. When generating the point structure, an available sub-slot is selected for each depth illumination contribution point, and the cost of lateral movement is minimized. Let the depth irradiation contribution point be c, and its assigned sub-slot be s(c). Then the slot allocation target is: in, Indicates the contribution point of deep irradiation Assigned to sub-slot The cost; The cost function includes one or more of the following: lateral displacement cost, cross-parent lateral mesh penalty, and importance weight of depth illumination contribution points, and can be expressed as: Where, q= Indicates the contribution point of deep irradiation The importance of; Represents the horizontal coordinate of the candidate sub-slot; Indicates the contribution point of deep irradiation The original horizontal coordinates in the reference plan; Indicates the parent horizontal grid to which the candidate sub-slot belongs; Indicates the contribution point of deep irradiation The original parent horizontal grid; Indicates the penalty weight across the parent horizontal grid; This indicates the value to be taken when the candidate child slot is inconsistent with the original parent's horizontal grid. Indicator functions; When the number of depth illumination contribution points in the same parent horizontal grid exceeds four, the four highest-ranked depth illumination contribution points are retained first, and the unassigned, lower-ranked depth illumination contribution points are pruned, or the unassigned, lower-ranked depth illumination contribution points are migrated to the empty sub-slots in the adjacent parent horizontal grid. After generating the point structure, a reference association identifier is established for each irradiation point in the point structure. The information of the reference association identifier includes one or more of the following: original parent horizontal grid number, sub-slot number, fixed irradiation direction, target water equivalent depth, candidate depth layer number, original irradiation dose, normalized irradiation dose, importance evaluation value, beam-by-beam reference dose contribution number, and corresponding reference plan irradiation point number. The reference association identifier remains unchanged in subsequent steps of generating or correcting the conformal energy modulator thickness map and MU fraction adjustment.

7. The multidirectional proton flash radiotherapy device according to claim 5, characterized in that, The adjustment of the MU share includes: After fixing the thickness map of the conformal energy modulator for each fixed irradiation direction, the MU fraction at the irradiation point is optimized. The MU adjustment aims to approximate the total reference dose of the reference plan without conformal energy modulator constraints with the dose synthesized from multiple fixed irradiation directions, and introduces MU constraint terms and field constraint terms into the MU adjustment target. Among them, the MU constraint term is used to limit the MU of a single irradiation point from deviating significantly from the reference plan, the non-overlapping point plan, or the plan of the previous cycle in the optimization cycle; The objective term considering the MU constraint is expressed as: ; The field constraint term is used to limit the total MU of each fixed irradiation direction from deviating significantly from the total MU of the corresponding fixed irradiation direction in the reference plan; The objective term considering the field constraint is expressed as: ; The MU adjustment target is expressed as: in: This represents the radiation weight obtained after optimization based on a non-overlapping point plan. Represents the irradiation weight to be optimized; This indicates that all exposure weights are non-negative; Represents the patient voxel weight matrix; This represents the dose effect matrix after fixing the modulator; This indicates an unlimited total reference dose; Indicates the first The weight of individual elements; This represents the reference value of the irradiation obtained after the modulator's master optimization. This represents the total number of radiation dose variables; Indicates the weight of the exposure maintenance term; Indicates the number of fixed irradiation directions; This indicates that the j-th irradiation point belongs to the j-th irradiation point. A fixed irradiation direction; Indicating an unrestricted reference plan Total irradiance from a fixed irradiation direction; This indicates the weight of the term that maintains the total amount of illumination in a fixed direction.

8. A multi-directional proton flash radiotherapy planning optimization system, characterized in that, The multidirectional proton flash radiotherapy device according to any one of claims 1-7 is characterized in that it comprises: The data input module is configured to acquire patient planned input data; The reference plan generation module is configured to generate a reference plan based on the patient plan input data, under minimal constraints including target volume dose coverage, target volume dose hotspots, shell dose limitations, organ at risk dose limitations, extra-target dose hotspots, field fractional balance, and MU smoothness constraints. The reference plan serves as a continuous reference for subsequent physicalization plans. The physicalization plan generation module is configured to construct a constrained plan space for a multi-directional fixed-field proton FLASH plan based on the reference plan without conformal energy modulator constraints; and to generate or modify physicalization plans within the constrained plan space so that the dose distribution of the physicalization plans continuously approximates the dose distribution of the reference plan; wherein the plan conversion process from less constrained conditions to a constrained plan space includes the following steps: The multi-directional, multi-depth irradiation contributions in the reference plan are converted into a point structure that meets the requirement of lateral non-overlap. Using the point structure, the beam-by-beam reference dose, and the single-energy irradiation conditions in the reference plan as inputs, a conformal energy modulator thickness map is generated or corrected. After generating or correcting the conformal energy modulator thickness map, energy compensation or energy boosting is performed to adjust the MU fraction of the irradiation point. Each step in the plan conversion process uses the reference plan as a continuous reference. The Monte Carlo dose calculation module is configured to perform Monte Carlo dose calculations based on the final physicalization plan; The output module is configured to output the corresponding physical plan as a multi-directional fixed-field proton FLASH plan when the Monte Carlo dose meets the preset dosimetric acceptance conditions.

9. A method for optimizing multi-directional proton flash radiotherapy planning, characterized in that, The multidirectional proton flash radiotherapy device according to any one of claims 1-7 is characterized by comprising the following steps: Obtain patient plan input data; Based on the patient planning input data, a reference plan is generated under minimal constraints, including target volume dose coverage, target volume dose hotspots, shell dose limitations, organ at risk dose limitations, extra-target dose hotspots, field fractional balance, and MU smoothness constraints. This reference plan serves as a continuous reference for subsequent physicalization plans. Based on the reference plan without conformal energy modulator constraints, a constrained plan space for the multi-directional fixed-field proton FLASH plan is constructed. The process of transforming a plan from few constraints to a constrained plan space includes the following steps: The multi-directional, multi-depth irradiation contributions in the reference plan are converted into a point structure that meets the requirement of lateral non-overlap. Using the point structure, the beam-by-beam reference dose in the reference plan, and the single-energy irradiation conditions as inputs, generate or correct the conformal energy modulator thickness map. After generating or correcting the conformal energy modulator thickness map, perform energy compensation or energy boost to adjust the MU fraction at the irradiation point; The aforementioned reference plan will be continuously referenced in each step of the plan conversion process; Within the constrained planning space, a physicalization plan is generated or modified so that the dose distribution of the physicalization plan continuously approximates the dose distribution of the reference plan. Monte Carlo dosing calculations were performed based on the final physicochemical plan; When the Monte Carlo dose meets the preset dosimetric acceptance criteria, the corresponding physical plan will be output as a multi-directional fixed-field proton FLASH plan.

10. A computer-readable storage medium, characterized in that, It stores a computer program for implementing the multidirectional proton flash radiotherapy planning optimization method as described in claim 9.

Images