Beam dose control method, apparatus, and medium

Abstract

The application discloses a beam dose control method, device and medium, and relates to the technical field of radiotherapy; the method comprises the following steps: dynamically setting a dose rate range based on tumor site radiosensitivity grading, adjusting accelerator beam parameters to stabilize the dose rate; determining a single-pulse theoretical dose according to the dose rate and pulse width, obtaining a calibrated single-pulse dose through experimental calibration and correction, and establishing a mapping relationship between the single-pulse dose and accelerator control parameters; calculating the required pulse number according to the target total dose and the calibrated single-pulse dose, and generating an executable pulse sequence; and verifying the difference between the actual output cumulative dose and the preset dose in real time, and sending a feedback signal if the difference exceeds a preset threshold. The application realizes precise dose control under the condition of FLASH radiotherapy ultra-high dose rate, and improves the precision and stability of dose delivery through pulse sequence optimization and real-time feedback.

Description

[0001] This application is a divisional application of Chinese application No. 2025110146718, filed on July 23, 2025, entitled "Dual-mode beam and dose control system, method and apparatus for FLASH radiotherapy". Technical Field

[0002] This invention relates to the field of radiotherapy technology, and in particular to beam dose control methods, equipment and media. Background Technology

[0003] Proton and heavy ion therapy for tumors, with its unique depth-dose distribution characteristics of the Bragg peak, has become an internationally recognized advanced method of tumor radiotherapy. With the iteration of radiotherapy technology, FLASH radiotherapy has received widespread attention as a breakthrough technology. Its core feature is the delivery of ultra-high doses (usually greater than or equal to 40 Gy) in a very short time (usually less than 1 second), which significantly reduces the toxicity to normal tissues through the "FLASH effect".

[0004] However, FLASH radiotherapy presents entirely new challenges to dose control systems. Existing radiotherapy dose control technologies are primarily designed for conventional dose rates (fractionated radiotherapy) and cannot directly adapt to the ultra-high dose rate requirements of FLASH radiotherapy, mainly due to the following technical bottlenecks:

[0005] Traditional dose servo systems typically have response times in the hundreds of milliseconds, which cannot meet the millisecond-level dose cutoff requirements of FLASH radiotherapy and make it difficult to deliver ultra-high doses accurately in a very short time.

[0006] The fluctuations in the intensity and pulse width of pulsed beams (such as proton beams generated by synchrotron accelerators) are amplified by ultra-high dose rates, leading to a significant increase in dose error and affecting the safety and effectiveness of treatment.

[0007] Conventional ionization chambers are prone to saturation at ultra-high dose rates, making it impossible to achieve accurate real-time dose measurement and hindering FLASH beam output.

[0008] Existing dose control systems mostly use static parameter settings. Beam parameters (such as current intensity and pulse width) will drift during the actual beam output process. This drift will be amplified at ultra-high dose rates, causing the actual output dose to deviate from the preset value, making it difficult to guarantee dose consistency under long-term or short-pulse sequences. Summary of the Invention

[0009] The purpose of this invention is to provide a beam dose control method, device and medium that can adapt to the ultra-high dose rate and short time constraint of FLASH radiotherapy, so as to achieve pulse-level precise dose delivery and real-time feedback control.

[0010] The objective of this invention is achieved through the following technical solution:

[0011] In a first aspect, this application provides a beam dose control method, including:

[0012] Before establishing the mapping relationship between pulse parameters and single-pulse dose, the dose rate range is dynamically set based on the radiosensitivity classification of the tumor site; the accelerator beam parameters are adjusted to stabilize the dose rate within the preset dose rate range.

[0013] Based on the dose rate and pulse width, the theoretical dose of a single pulse is determined; the theoretical dose of a single pulse is corrected through experimental calibration, the calibrated dose of a single pulse is determined, and the mapping relationship between the dose of a single pulse and the accelerator control parameters is established.

[0014] Based on the target total dose and the calibrated single-pulse dose, calculate the required number of pulses and generate an executable pulse sequence;

[0015] The difference between the actual output cumulative dose and the preset dose is verified in real time; if the difference exceeds its preset threshold, a feedback signal is sent to the execution unit.

[0016] Preferably, determining the theoretical dose of a single pulse includes:

[0017] D_pulse_theory= D×PW_opt

[0018] Wherein, D_pulse_theory is the theoretical dose of a single pulse; D is the dose rate; PW_opt is the optimal pulse width; the optimal pulse width is selected by testing the coefficient of variation of the dose rate under different pulse widths and choosing the pulse width value with the smallest coefficient of variation.

[0019] Preferably, the experimental calibration adopts dual ionization chamber collaborative calibration: the main ionization chamber measures the total dose, the micro-dose ionization chamber measures the single-pulse dose, and the single-pulse theoretical dose is corrected to obtain the calibrated single-pulse dose.

[0020] Preferably, establishing the mapping relationship between single-pulse dose and accelerator control parameters includes: calibrating a two-dimensional lookup table of magnetic field strength, radio frequency voltage and single-pulse dose under constant pulse width, and generating a continuous function through polynomial fitting.

[0021] Preferably, generating the executable pulse sequence includes: calculating the number of pulses based on the target total dose and the calibrated single pulse dose, generating a sequence containing the width and current intensity of each pulse, and synchronizing the pulse emission interval with the accelerator radio frequency cycle.

[0022] Preferably, the pulse sequence employs a sequence optimization strategy based on intelligent prediction:

[0023] Based on the comparison results between the number of pulses and a preset threshold, the pulse sequence is divided into long sequences and short sequences;

[0024] When the pulse sequence is long, a verification pulse is inserted at a predetermined interval, and the pulse width is corrected according to the range drift. When the pulse sequence is short, a current intensity increment compensation strategy is adopted, and the beam attenuation trend is predicted based on the LSTM neural network. The dose rate attenuation is compensated by dynamically adjusting the current intensity.

[0025] Preferably, the real-time verification includes: verifying whether the single-pulse dose difference and the cumulative dose difference exceed their respective preset thresholds; if they exceed, feedback is provided and millisecond-level dose cutoff is achieved.

[0026] Secondly, this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the beam dose control method according to any one of the claims of this application.

[0027] Thirdly, this application provides a computer-readable storage medium that stores computer instructions, which, when read by a computer, execute the beam dose control steps described in any one of the claims of this application.

[0028] Compared with existing technologies, the beneficial effects of this invention include at least the following: By pre-calibrating the dose rate and selecting the optimal pulse width, the dose rate is stabilized within a preset range, laying the foundation for accurate calculation of single-pulse dose and effectively overcoming the problem of dose error amplification under ultra-high dose rates in traditional methods. Through dual-ionization chamber collaborative calibration, the theoretical single-pulse dose is corrected, and a mapping relationship with accelerator control parameters (such as magnetic field strength and radio frequency voltage) is established, enabling precise prediction and control of the single-pulse dose, solving the problem of insufficient accuracy of conventional calibration methods in FLASH mode. An automatic switching optimization strategy is implemented based on the number of pulses—for long sequences, range drift is monitored and subsequent pulse widths are corrected by inserting verification pulses; for short sequences, a current intensity increment strategy is adopted, and LSTM neural networks are used to predict beam attenuation trends for dynamic compensation. This adaptive mechanism significantly improves dose consistency under different pulse scale scenarios. By verifying the single-pulse dose difference and cumulative dose difference in real time, feedback is immediately provided and millisecond-level dose cutoff is achieved when the preset threshold is exceeded, forming a complete closed-loop control system that ensures dose safety during the extremely short irradiation process of FLASH radiotherapy. The pulse emission interval is synchronized with the accelerator radio frequency cycle, with the error controlled at the nanosecond level. This meets the high-precision timing requirements of FLASH radiotherapy, which has an irradiation time of less than 1 second, and avoids dose superposition errors caused by timing deviations. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of a dual-mode beam and dose control system for FLASH radiotherapy according to an embodiment of the present invention;

[0030] Figure 2 This is a schematic diagram of the dual working modes of an embodiment of the present invention;

[0031] Figure 3 This is a schematic diagram of the dose distribution test results according to an embodiment of the present invention;

[0032] Figure 4 This is a schematic diagram of the execution process of the dual-mode beam and dose control system for FLASH radiotherapy according to an embodiment of the present invention;

[0033] Figure 5 This is a schematic diagram of a dual-mode beam and dose control method for FLASH radiotherapy according to an embodiment of the present invention;

[0034] Figure 6 This is a schematic diagram of the beam dose control method according to an embodiment of the present invention;

[0035] Figure 7 This is a schematic diagram of the hardware connection between the cyclotron accelerator and the dual-mode beam and dose control system for FLASH radiotherapy according to an embodiment of the present invention. Detailed Implementation

[0036] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided to make the invention more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and therefore repeated descriptions of them will be omitted.

[0037] The terms used to express position and direction in this invention are illustrated with reference to the accompanying drawings, but changes can be made as needed, and all such changes are included within the scope of protection of this invention.

[0038] A cyclotron typically comprises five subsystems that, through the complex interactions between them, generate proton beams for clinical treatment or research. These five subsystems include an ion source subsystem, a radio frequency (RF) subsystem, a vacuum subsystem, a magnet subsystem, and an extraction subsystem. Visually, a cyclotron usually resembles a cylinder with two semi-circular cylindrical electrodes separated by a short distance. The RF subsystem uses this distance to create an alternating electric field, accelerating protons as they pass through. As the dipoles divide the cylinder into two separate parts, one electrode is shaped like the letter "D," and the other is shaped like an inverted "D"—hence the common name "D-electrode." A magnetic field exists within the D-electrode that guides particles (i.e., protons) from one D-electrode to the other, and protons also gain energy as they pass through this slit.

[0039] The protons are redirected and return along a curved path to the slit and the inverted D-shaped electrode. The protons pass through the slit and accelerate. They then enter the inverted D-shaped electrode and return along the curved path to the slit and the inverted D-shaped electrode again. This process repeats until the protons reach an extraction point near the outer edge of one of the D-shaped electrodes. At this point, the extraction subsystem guides the proton beam into downstream subsystems for either experimental purposes or to irradiate the patient.

[0040] Reference Figure 1 This invention proposes a dual-mode beam and dose control system for FLASH radiotherapy, the system comprising:

[0041] The mode switching module is used to drive an adjustable collimator located at the accelerator outlet or transport line through a drive mechanism to achieve dual-mode switching between clinical mode and FLASH mode, and to ensure the accuracy of the switching position through a redundant feedback mechanism.

[0042] The beam shaping module is used to optimize the Bragg peak depth dose distribution by using at least two materials with different physical properties based on the characteristics of proton and heavy ion Bragg peaks and by adjusting the thickness and / or quantity of the materials.

[0043] The dose control module is used to establish a mapping relationship between pulse parameters (current intensity, pulse width) and single-pulse dose, and calculate the required pulse sequence in combination with the target total dose to achieve precise dose control for FLASH radiotherapy.

[0044] Since the current intensity required for conventional treatment or experimentation in accelerators is much lower than that required in FLASH mode, higher current intensity places higher demands on various parameters and specifications of the accelerator subsystem, resulting in higher costs. This invention designs a collimator located at the accelerator exit or transport line to handle the switching between the two modes. It is used for beam shaping in conventional clinical mode, and can be completely offline in FLASH mode without collimating the beam, reducing beam loss and improving dose rate. The collimator is driven and controlled by a motor potentiometer. The position and voltage signal of the collimator are calibrated, and a redundant method of real-time position feedback using dual potentiometers is adopted. Its voltage value is displayed on the main control computer interface. The position is monitored in real time by dual redundant potentiometers at both ends of the limit switch to ensure the reliability of the position.

[0045] See attached document Figure 2 Dual working modes include:

[0046] Clinical mode: Potentiometer positioned at P2 (beam shaping position);

[0047] FLASH mode: Potentiometer set to P1 (completely remove beam current limitation);

[0048] Redundant monitoring: Dual potentiometers provide real-time feedback on position deviation to ensure safety and reliability.

[0049] In the area of ​​the accelerator exit beamline, a mechanical connection interface is designed for quick assembly and disassembly with conventional treatment heads. This bracket is designed to be compatible with conventional mechanical interfaces, allowing for quick installation and disassembly via a simple screw structure, thereby switching between FLASH and conventional mechanical modes.

[0050] From a security perspective, a combination of hardware and software is used to ensure security:

[0051] At the hardware level: the FLASH dosing module is connected to the clinical system only through a single fiber optic link, without much complex connection or interaction.

[0052] At the software level: The FLASH mode and clinical mode employ a dual-mode independent switching method. The FLASH mode operating interface is independently password protected, while the clinical mode automatically disables FLASH system configuration. Accelerator system parameters in the regular mode are stored in the regular mode's XML configuration file, while accelerator parameters in the FLASH mode are stored in the corresponding FLASH XML file; these modes are independent of each other. Standardized log display and export tools are provided at the FLASH software level, offering clues for data recording and machine fault troubleshooting during the process.

[0053] By switching between two collimators (clinical mode P2 and FLASH mode P1), beam shaping is achieved using the collimator in the conventional clinical mode, ensuring the accuracy of dose distribution in routine treatments. In FLASH mode, beam limitations are completely removed, reducing beam loss and significantly increasing the dose rate to meet the core requirement of "ultra-high dose rate" in FLASH radiotherapy. This rapid switching between the two modes allows the same device to meet the needs of both conventional radiotherapy and cutting-edge FLASH research, improving equipment utilization and clinical application flexibility.

[0054] Compared to designs that require separate high-performance accelerators for FLASH mode, this system achieves beam control through collimator switching, avoiding high performance requirements across the entire accelerator subsystem. In normal mode, it is not necessary to maintain ultra-high current intensity, reducing the manufacturing, maintenance costs, and operating losses of the accelerator. At the same time, the collimator's motor potentiometer drive and redundant feedback design achieve high-precision mode switching in a low-cost manner, further controlling equipment investment.

[0055] The collimator adopts a redundant design with real-time feedback of position using dual potentiometers, combined with dual redundant monitoring of limit switches. Through voltage signal calibration and visualization on the main control interface, the accuracy and reliability of position switching (P1 / P2) are ensured. The real-time deviation feedback mechanism of the dual potentiometers can detect positional abnormalities in a timely manner, avoid beam parameter abnormalities caused by mode switching errors, and improve treatment safety.

[0056] The quick-release mechanical interface in the accelerator exit beamline area is compatible with the mechanical connection of conventional treatment heads. The mechanical switching between FLASH mode and conventional mode can be completed through a simple screw structure, which greatly shortens the mode conversion time, reduces the operational complexity of medical staff, and meets the needs of rapid switching scenarios in clinical treatment.

[0057] At the hardware level, the FLASH dosing module is connected to the clinical system via a single-point fiber optic connection, reducing the risk of interference caused by complex interactions. At the software level, dual-mode independent switching, independent password protection, and parameter file isolation (Xml files are stored separately) are adopted to avoid parameter confusion and misoperation between modes. In addition, the standardized log tool of FLASH mode facilitates data traceability and troubleshooting, ensuring treatment safety in all aspects from operation process to data management.

[0058] In FLASH mode, the collimator limitation is removed to reduce beam loss and directly increase the dose rate. The beam shaping module uses graded modulation based on the Bragg peak characteristics, combined with the pulse sequence of the dose control module for precise calculation, to ensure the uniformity and accuracy of dose distribution for short-time, high-dose delivery in FLASH mode, reducing damage to normal tissues and improving treatment efficacy.

[0059] In summary, through its design of "precise switching, controllable cost, safety and reliability, and scenario adaptability," this system effectively solves the compatibility problem between FLASH radiotherapy and conventional modes, providing a practical and economical technical solution for the precision, efficiency, and cutting-edge research of clinical radiotherapy.

[0060] In one possible implementation, the beam shaping module includes a beam modulation unit, which implements graded beam modulation through the graded modulator, which includes a coarse adjustment layer and a fine adjustment layer.

[0061] The coarse adjustment layer comprises multiple first material components of the same or different thicknesses that are detachable. The first material components are modularly designed through a detachable connection mechanism to achieve stepped adjustment of the range. The materials of the first material components are selected from high atomic number materials, including but not limited to at least one of lead, tungsten, tungsten carbide, and high-density alloys.

[0062] The fine adjustment layer comprises multiple second material components of the same or different thicknesses and with an adjustable number of stacked layers. The second material components are used to achieve continuous fine adjustment of the range. The materials of the second material components are selected from low atomic number materials, including but not limited to at least one of polyethylene, polypropylene, polystyrene, and water-equivalent plastics.

[0063] The hierarchical modulation includes:

[0064] By combining different numbers of the first material component (N)i To achieve a wide range of range adjustment, the calibrated equivalent water thickness (WET) of each first material component is α. i (i=1,2,3,…);

[0065] Fine adjustment layer: by combining different numbers of second material components (M j To achieve short-range compensation, the calibration WET for each second material component is β. j (j=1,2,3,…);

[0066] Wherein, the total modulation range Ds of the coarse adjustment layer and the fine adjustment layer satisfies:

[0067] Ds=Σ(α i ×N i )+Σ(β j ×M j )+δ cal

[0068] α i For the calibrated equivalent water thickness of the i-th first material component, N i The quantity of the i-th first material component used;

[0069] β j M is the calibrated equivalent water thickness of the j-th second material component. j Let j be the number of stacked layers of the j-th second material component;

[0070] δ cal The scattering compensation amount is calibrated using Monte Carlo simulation.

[0071] In one possible implementation, the equivalent water thickness of the first material component is an integer multiple of 5-20 mm, the equivalent water thickness of the second material component is an integer multiple of 0.1-1 mm, and the equivalent water thickness is specified by the following formula:

[0072] WET=[(R water -R air ) / (R material -R air )]×t material

[0073] Where WET is the equivalent water thickness, R is the location of the Bragg peak, and t material R represents the physical thickness of the material. water R represents the Bragg peak range of the particle in water. air R represents the Bragg peak range of the particle in air. material The range of the Bragg peak of the particle in the material.

[0074] For example, the coarse adjustment layer consists of ≥4 types of carbon blocks with a WET value that is an integer multiple of 10 mm; the fine adjustment layer consists of ≥4 types of polyethylene sheets with a WET value that is an integer multiple of 1 mm.

[0075] In one possible implementation, the beam shaping module further includes a scattering suppression component, comprising:

[0076] The multi-leaf collimator, which is located close to the rear end of the modulator, has an aperture that is continuously adjustable in the range of 5–40 mm².

[0077] An ionization chamber support with a coaxiality error of ≤1mm with the collimator is used to support a transmission-type absolute dose ionization chamber.

[0078] In the beam shaping module, the result of optimizing the Bragg peak depth dose distribution satisfies:

[0079] The flatness of the broadened Bragg peak (SOBP) is ≤±3%, measured by a multi-layer ionization chamber array (0.1 mm step).

[0080] The distal dose drop gradient R80-R20 ≤ 2 mm was measured using Gafchromic film.

[0081] The scattering compensation amount δ cal The calibration methods include:

[0082] Establish a collimated aperture-lateral penumbra mapping table (Example: 10×10mm² aperture → penumbra 3.2±0.3mm);

[0083] Beam broadening effect under different material combinations was calculated based on Monte Carlo simulation.

[0084] The compensation amount is embedded in the range model so that the far-end dose drop gradient satisfies R80-R20≤2mm.

[0085] In one possible implementation, the dual-mode beam and dose control subsystem further includes:

[0086] Dedicated FLASH ionization chamber: Installed downstream of the beam shaping module, it integrates a temperature / pressure / humidity calibration PCB, a humidity control area, a metal high-voltage plane, fiber optic acquisition components, and a sealed window, adaptable to ultra-high dose rate beam acquisition; among them, the temperature, pressure, and humidity compensation PCB is used to calibrate the influence of environmental parameters on the dose in real time; the metal high-voltage plate is sealed and isolated from the fiber optic signal acquisition end.

[0087] Dosing unit: Connected to the FLASH ionization chamber via optical fiber, with a built-in signal amplifier to enable rapid dose measurement, data processing and control;

[0088] Supporting components: dedicated computer (for installing measurement and analysis software), fiber optic / connection cable (for component interconnection), Faraday cup / beam blocker (for collecting and blocking the beam).

[0089] Because conventional ionization chambers are prone to saturation and have slow response speeds, a dedicated FLASH ionization chamber is essential. The core components include a PCB circuit board for temperature, pressure, and humidity calibration, a humidity control area, a metal plane for the high-pressure area, fiber optic signal acquisition components, and two enclosed windows at the front and back. The combined dose distribution (Dose) control chassis is used to control and acquire Dose-related information.

[0090] The working principle of the above technical solution is as follows:

[0091] Traditional Bragg peak modulators have two major drawbacks: insufficient single-stage adjustment precision, resulting in a flatness of the spread Bragg peak (SOBP) >5%; and significant scattering effects, where high atomic number materials induce beam broadening, reducing the sharpness of the far-end dose drop (R80-R20 > 3mm).

[0092] When protons / heavy ions move through a medium, energy loss increases with depth, eventually forming an energy deposition peak (Bragg peak) at the end of the beam's range. To better utilize the unique depth-dose distribution characteristics of the Bragg peak for protons and heavy ions, the study and experimental regions are set within the Bragg peak area. The core of beam shaping is to adjust the beam's range in the medium so that the Bragg peak precisely covers the tumor target area while reducing dose deposition in normal tissues. A graded modulator (coarse adjustment layer + fine adjustment layer) achieves stepwise large-range range adjustment and continuous small-range compensation through the "stopping effect" of the material on the beam, ultimately matching the tumor depth requirements.

[0093] The coarse-tuning layer uses high atomic number materials (such as lead, tungsten, tungsten carbide, etc.), which have strong stopping power (particles lose energy quickly in the material) and a large equivalent water thickness (WET) (in integer multiples of 5-20 mm). When the beam passes through the coarse-tuning layer components, the high atomic number material will significantly shorten the Bragg peak range of the beam, and the increase or decrease in the number of components can achieve a large-range "step-like" adjustment of the range (for each additional component, the range adjustment is equal to its equivalent water thickness αi).

[0094] Based on the approximate depth of the tumor target area (e.g., 10-100mm), the number of first material components can be increased or decreased via a detachable connection mechanism. For example, if the range needs to be increased by 30mm, two components with αi=15mm can be combined (15×2=30mm) to achieve rapid, large-span coarse range adjustment, laying the foundation for subsequent fine-tuning.

[0095] The fine-tuning layer uses low atomic number materials (such as polyethylene, water-equivalent plastic, etc.), which have weak stopping power (slow particle energy loss) and small equivalent water thickness (0.1-1 mm integer multiples). When the beam passes through the fine-tuning layer assembly, the range change is small, and by continuously adjusting the number of stacked layers (such as increasing from 1 layer to 5 layers), "continuous" small-range compensation of the range can be achieved, making up for the insufficient accuracy caused by the stepped interval of the coarse-tuning layer.

[0096] After determining the approximate range in the coarse adjustment layer, precise compensation is achieved for the fine depth of the tumor target area (e.g., within ±2 mm) by increasing or decreasing the number of stacked layers of the second material component. For example, when the range deviation is 1.2 mm after coarse adjustment, 12 layers of the component with βj=0.1 mm (0.1×12=1.2 mm) can be stacked to ensure that the Bragg peak completely covers the target area.

[0097] The selection of material thickness is mainly based on the equivalent water thickness of the particles in water, and the actual thickness of the material is defined by deducing from this. The equivalent water thickness (WET) is the core parameter for measuring the material’s blocking effect on the beam, and its calibration is based on the physical relationship of the Bragg peak range.

[0098] By utilizing the Bragg peak range difference of particles in water, air, and materials, the physical thickness of the material is transformed into an equivalent water thickness (WET). This is because the Rt of high atomic number materials... material <R water (Stronger ability to prevent), R of low atomic number materials materia Approaching R wate (With weaker blocking ability), it can accurately quantify the ability of different materials to adjust the beam range, providing a benchmark for the selection of components for coarse / fine adjustment.

[0099] Due to differences in materials and processing, the combination of range, thickness, and number of particles can be varied during testing to measure the dose rate of each group. The dose rate is measured using an absolute dose ionization chamber. The measurement position is calibrated using a position adjustment device and film cross-verification to confirm the accuracy of the measurement. After recording the material combination, equivalent water depth, and measured dose rate, the depth dose distribution of particles at the broadened Bragg peak can be observed. A corresponding FLASH broadened Bragg peak (SOBP) will be generated to meet the experimental requirements, ensuring that the region with the FLASH dose rate has a sufficiently wide Bragg peak.

[0100] In summary, the beam shaping module achieves precise control of beam depth dose distribution through the following steps:

[0101] WET calibration: Using physical formulas, the blocking effect of different materials is equivalent to water thickness, and a unified quantitative standard is established;

[0102] SOBP validation: Measurements using an ionization chamber and film ensure that the dose flatness of the plateau region and the distal drop gradient meet clinical requirements. SOBP is a "plateau region" formed by the superposition of multiple Bragg peaks; its width must cover the tumor target area. Flatness (dose fluctuation in the plateau region) and distal drop gradient (rapid dose decrease at the end of the plateau region) directly affect the dose received by normal tissues.

[0103] Flatness ≤ ±3%: Ensures uniform dose within the tumor target area and avoids excessively high / low local doses;

[0104] Distal drop gradient (R80-R20≤2mm): This allows the dose to drop rapidly outside the target area, protecting normal tissue.

[0105] The measurement parameters for broadened Bragg peak are shown in Table 1.

[0106] Table 1

[0107]

[0108] Scattering compensation: Record the correspondence between collimating aperture and penumbra to provide a basis for scattering compensation in clinical treatment; refer to Table 2 for the scattering compensation record table;

[0109] Table 2

[0110]

[0111] Ultimately, the beam's Bragg peak precisely covers the tumor target area, while SOBP optimization and scattering compensation protect normal tissues, adapting to the dose requirements of conventional radiotherapy and FLASH radiotherapy.

[0112] Among them, the adjustable collimator: through redundant control of dual potentiometers driven by a motor, it switches between clinical mode and FLASH mode to shape the beam cross section and optimize the dose rate;

[0113] Square support: Provides a mechanical interface to enable quick assembly and disassembly of treatment heads in different modes, ensuring stable beam transmission;

[0114] Ionization chamber: Real-time monitoring of beam dose parameters (intensity, distribution) and feedback to the control system to form a closed loop, ensuring treatment accuracy.

[0115] The adjustment effects of the coarse and fine adjustment layers are integrated through the total modulation range to ensure that the final range accurately matches the tumor depth. The coarse adjustment layer uses a large equivalent water thickness (5-20 mm) of high atomic number material to quickly match the approximate depth of the tumor target area, reducing adjustment steps. The fine adjustment layer uses a small equivalent water thickness (0.1-1 mm) of low atomic number material to compensate for the step interval of the coarse adjustment, achieving fine adjustment at the ±0.1 mm level. Whether in conventional clinical mode (requiring precise dose distribution) or FLASH mode (requiring range stability at high dose rates), graded modulation can be quickly adapted through component combination to ensure dynamic matching between beam range and tumor depth.

[0116] In one possible implementation, the physical thickness t of the first material component i With equivalent water thickness α i The following relationship must be satisfied:

[0117] α i =f1×t i Where f1 is the equivalent water thickness coefficient of the material, f1≥2.0 (corresponding to high atomic number materials), and the deviation of the f1 value of each first material component is ≤±0.5%;

[0118] The physical thickness t of the second material component j With equivalent water thickness β j The following relationship must be satisfied:

[0119] β j =f2×t j Where f2 is the equivalent water thickness coefficient of the material, 0.8≤f2≤1.2 (corresponding to low atomic number materials), and the deviation of f2 value for each second material component is ≤±0.3%.

[0120] The combination of the first material component and the second material component includes:

[0121] Layered arrangement: The first material component and the second material component are stacked sequentially along the beam direction;

[0122] Nested arrangement: The second material component is embedded in a preset groove of the first material component.

[0123] The working principle and effects of the above technical solution are as follows:

[0124] By quantifying material properties using the equivalent water thickness coefficient (f1 / f2) and optimizing the beam path using a stacked / nested arrangement, precise control of beam range and dose distribution can be achieved. The core logic consists of two steps:

[0125] High atomic number materials (first component): Due to their strong stopping power (rapid particle energy loss), a coefficient f1 ≥ 2.0 is needed to amplify the contribution of physical thickness to equivalent water thickness. For example, 1 mm thick tungsten (f1 = 2.5) has an equivalent water thickness αᵢ = 2.5 × 1 = 2.5 mm, achieving "thin material with large range adjustment," reducing the total material thickness along the beam path, and adapting to the "high dose rate requirement" of FLASH radiotherapy (the thinner the material, the smaller the beam loss, and the higher the dose rate).

[0126] Low atomic number materials (second component): Due to their weak stopping power (slow particle energy loss), the relationship between the physical thickness and the equivalent water thickness is finely adjusted using a coefficient of 0.8 ≤ f2 ≤ 1.2. For example, for 1 mm thick polyethylene (f2 = 1.0), the equivalent water thickness βⱼ = 1.0 × 1 = 1.0 mm, achieving "thick material with short range compensation" and avoiding insufficient mechanical strength due to the material being too thin during fine-tuning.

[0127] Stacked arrangement: The first component (high atomic number) and the second component (low atomic number) are stacked sequentially along the beam direction. The range is first significantly shortened by using high atomic number material (coarse adjustment), and then fine-tuned by using low atomic number material (fine adjustment). For example, when treating deep tumors (where long range is required), the stacked arrangement can quickly compress the beam range, adapting to the "deep dose coverage" requirements of conventional clinical models.

[0128] Nested arrangement: The second component (low atomic number) is embedded in a pre-set groove of the first component, allowing the thicknesses of the two materials to overlap in the beam direction, while also achieving a more uniform lateral distribution. For example, in FLASH mode, nested arrangement can reduce lateral beam scattering (because the low atomic number material fills the groove, optimizing the uniformity of the beam cross-section) and improve dose rate stability.

[0129] The large coefficient f1 (≥2.0) of high atomic number materials enables "thin materials to achieve long range adjustment", reducing the total thickness of materials in the beam path (e.g., 1 mm tungsten is equivalent to 2.5 mm of water, which is 1.5 mm less than the thickness of water directly), reducing beam loss and adapting to the high dose rate requirements of FLASH radiotherapy (dose rate is negatively correlated with beam loss).

[0130] The small coefficient f2 (0.8-1.2) of low atomic number materials enables "thick materials to achieve short range compensation", avoiding mechanical deformation caused by excessively thin materials during fine adjustment (such as polyethylene with a thickness ≥1mm, which has higher processability and stability), and ensuring fine adjustment accuracy.

[0131] The layered arrangement of "coarse adjustment followed by fine adjustment" is adapted to the "deep dose coverage" of conventional clinical models (e.g., if the tumor depth is 100mm, the range is first compressed to 90mm by layering high atomic number materials, and then fine-tuned to 100mm by using low atomic number materials).

[0132] Material coefficient deviation control (f1 deviation ≤ ±0.5%, f2 deviation ≤ ±0.3%) ensures that the equivalent water thickness of different batches of components is consistent, reducing the need for "component replacement calibration" in clinical treatment; the nested arrangement of mechanical structure compatibility allows the components to be adapted to the beam transport lines of different accelerators (only the groove size needs to be adjusted), improving the cross-platform versatility of the equipment.

[0133] The "thinning" of high atomic number materials (due to the large f1, the amount of material used can be reduced) reduces the procurement cost of components (e.g., the amount of tungsten used is reduced by 40%); the simple structure of stacked and nested arrangement facilitates rapid replacement of components in clinical settings (e.g., stacked components can be added or removed within 5 minutes), reducing equipment downtime.

[0134] In one possible implementation, the component selection of the coarse adjustment layer and the fine adjustment layer satisfies the following:

[0135] Max(α i ×N i ) / Σ(β j ×M j )∈[Cmin,Cmax];

[0136] The critical values ​​(Cmin, Cmax) for selecting the coarse adjustment layer and fine adjustment layer components are calibrated through the following simulation process:

[0137] (1) Constructing a dedicated simulation environment for FLASH radiotherapy

[0138] Based on a particle transport simulation program, an integrated dose-rate-dependent nonlinear scattering physics model is used to simulate the nonlinear behavior of scattering characteristics as dose rate dynamically changes when the ultra-high dose rate beam of FLASH radiotherapy interacts with materials.

[0139] A coupled heat-dose co-conduction module is used to simulate the real-time temperature changes of materials caused by short-term high-dose deposition and their dynamic impact on the beam blocking ability of materials.

[0140] (2) Perform scenario-based parameter scanning and indicator monitoring

[0141] The input parameters include typical beam characteristics and environmental parameters of FLASH radiotherapy, specifically including beam intensity, pulse duration, and initial environmental conditions of materials, to match the actual beam conditions in clinical FLASH studies.

[0142] Monitoring indicators include, but are not limited to: dose distribution uniformity of broadened Bragg peak (SOBP), the proportion of scattered dose to total dose, beam dose rate stability, and the drift of material equivalent water thickness with beam exposure time.

[0143] Cmin (Dominance Threshold of Coarse Adjustment Layer): When the ratio of the adjustment capability of the coarse adjustment layer to that of the fine adjustment layer is lower than Cmin, the monitoring results show that the uniformity of SOBP dose distribution exceeds the preset acceptable range, and the beam dose rate stability drops below the threshold, indicating that the coarse adjustment layer cannot effectively dominate the stepwise adjustment.

[0144] Cmax (Safety threshold of fine-tuning layer): When the above ratio is higher than Cmax, the monitoring results show that the proportion of scattered dose exceeds the safety threshold and the drift of the material equivalent water thickness exceeds the allowable range, indicating that the over-compensation of the fine-tuning layer leads to the loss of beam control.

[0145] The working principle of the above technical solution is as follows:

[0146] Using particle transport simulation programs, we consider the nonlinear behavior of scattering characteristics as the beam interacts with materials at ultra-high dose rates (≥40 Gy / s) in FLASH radiotherapy, where the scattering characteristics change dynamically with the dose rate. For example, ultra-high dose rates can cause changes in the microstructure of materials in a short period of time, and the scattering angle and probability no longer follow the conventional linear law. This complex change can be accurately simulated, providing a physical basis for subsequent regulation.

[0147] Depositing a large dose (e.g., 10-40 Gy in a short time) into a material will cause the material temperature to rise rapidly (potentially reaching tens or even hundreds of degrees Celsius). This temperature change will affect the material's ability to stop the beam (e.g., as the temperature rises, the stopping ability of some materials may decrease). The module simulates the dynamic correlation between temperature changes and stopping ability in real time, making the simulation closer to the real radiotherapy scenario.

[0148] The actual beam characteristics of FLASH radiotherapy (the beam intensity may be several times higher than that of conventional radiotherapy, and the pulse duration may be as short as sub-milliseconds) and environmental parameters (initial material temperature, air pressure, etc.) are entered to ensure that the simulation conditions are consistent with those of clinical FLASH studies, so that the subsequent calibrated critical values ​​can directly serve clinical practice.

[0149] Expanded Bragg peak (SOBP) is key to accurate tumor coverage in FLASH radiotherapy. Poor homogeneity can lead to uneven dose distribution within the tumor, affecting efficacy and increasing the risk of damage to normal tissues. Strict monitoring is required (e.g., homogeneity deviation ≤ ±3%).

[0150] If the scattered dose is too high (e.g., exceeding 10% of the total dose), it will cause unnecessary dose deposition in normal tissues around the tumor. Its proportion must be controlled within a safe range (e.g., ≤5%).

[0151] FLASH radiotherapy relies on a stable high dose rate to ensure efficacy. Large dose rate fluctuations (such as exceeding ±2%) will cause the actual total dose to deviate from the target, so it is necessary to maintain stability (such as fluctuations ≤ ±1%).

[0152] The equivalent water thickness of the material will change due to temperature and dosage, which directly affects the accuracy of beam range control. The drift amplitude needs to be limited (e.g., ≤0.1mm).

[0153] Dynamically determine the critical value, when Max(α) i ×N i ) / Σ(β j ×M j When the dose distribution is less than Cmin, the coarse adjustment layer has weak adjustment capabilities and cannot dominate the step-by-step wide-range range control. Simulations show that the uniformity of SOBP dose distribution deteriorates (e.g., deviation > ±5%), and the stability of beam dose rate falls below the threshold (e.g., fluctuation > ±3%). At this time, it is necessary to add coarse adjustment layer components or adjust materials to strengthen the dominant role of coarse adjustment.

[0154] When Max(α) i ×N i ) / Σ(β j ×M j When the value is greater than Cmax, the fine-tuning layer overcompensates, the proportion of scattered dose in the simulation exceeds the standard (e.g., >7%), the material equivalent water thickness drifts out of control (e.g., >0.2mm), and the beam control is chaotic. It is necessary to reduce the intervention of the fine-tuning layer or optimize the material combination so that the fine-tuning can return to the positioning of "precise compensation in a small range".

[0155] The effects of the above technical solution are as follows:

[0156] This system precisely adapts to the physical characteristics of FLASH radiotherapy using a dedicated simulation environment (nonlinear scattering model + thermal-dose coordination module). It accurately reproduces the complex physical processes of beam-material interaction under the "ultra-high dose rate, short-time high-dose deposition" scenario of FLASH radiotherapy, solving the problem of conventional simulations neglecting the dynamic effects of dose rate and temperature, and making critical value calibration more closely aligned with actual clinical needs. It improves the safety and effectiveness of beam modulation by monitoring indicators such as the proportion of scattered dose and the drift of equivalent water thickness, providing early warning of the risks of insufficient coarse adjustment and excessive fine adjustment. This avoids treatment safety issues caused by uncontrolled beam modulation (such as excessively high scattered dose damaging normal tissue, or dose rate fluctuations leading to total dose deviation), ensuring patient safety. It clearly defines the division of labor between the coarse adjustment layer ("leading stepwise adjustment") and the fine adjustment layer ("responsible for precise compensation"), allowing beam range modulation to efficiently cover the tumor (coarse adjustment quickly matches the approximate depth) and finely correct deviations (fine adjustment compensates for minor errors), improving the effectiveness of FLASH radiotherapy in killing tumors.

[0157] In one possible implementation, δ cal Calibration was achieved through a combination of collimation aperture and material combinations.

[0158]

[0159] The area of ​​the collimating hole. The reference area is determined by the collimator design parameters of the radiotherapy equipment.

[0160] μ is the linear scattering coefficient of the material (cm²). -1 (Obtained through Monte Carlo simulation of Geant4, positively correlated with material atomic number and beam energy).

[0161] The total thickness of the fine-tuning layer (cm, thickness of low atomic number material stacking);

[0162] The number of the first material components (to match the needs of the coarse adjustment layer step adjustment);

[0163] δflash is a compensation feature specific to FLASH radiotherapy (due to nonlinear scattering caused by ultra-high dose rates, δflash = ...). D^0.5, (where D is the nonlinear scattering coefficient and D is the dose rate, in Gy / s).

[0164] b1, b2 The scattering coefficients fitted by Monte Carlo simulation and experiment are (b1∈[0.1,0.5], b2∈[0.01,0.1]). ∈[0.005,0.02], which is suitable for beam energy and material composition.

[0165] The working principle of the above technical solution is as follows:

[0166] The collimating aperture area determines the initial cross-sectional size of the beam; its ratio to the reference area reflects the change in the beam's "initial scattering space." Material scattering attenuation term. This describes the attenuation effect of the fine-tuning layer material on beam scattering. The thicker the material and the larger the scattering coefficient, the more significantly beam scattering is suppressed, and the compensation amount needs to be adjusted accordingly. The more components in the coarse-tuning layer, the more complex the beam scattering in the coarse-tuning stage becomes, and this linear term needs to compensate for the scattering deviation introduced by the coarse-tuning.

[0167] Due to the ultra-high dose rate of FLASH radiotherapy (e.g., ≥40Gy / s), the interaction between the beam and the material exhibits nonlinear scattering (the linear scattering model of conventional radiotherapy fails). The FLASH radiotherapy-specific compensation item is designed to compensate for the nonlinear scattering deviation under ultra-high dose rates, reflecting the special characteristics of the FLASH radiotherapy scenario.

[0168] The above technical solution achieves the following effects: it quantifies the nonlinear scattering caused by the ultra-high dose rate of FLASH radiotherapy, solving the problem of conventional simulations neglecting the nonlinear effects of dose rate. At an ultra-high dose rate of 60 Gy / s, it can accurately compensate for dose deviations caused by nonlinear scattering, ensuring the accuracy of FLASH radiotherapy doses. By coupling multiple physical parameters such as collimation aperture area, material scattering coefficient, and number of coarse adjustment components, it covers the entire process of beam scattering influence from "collimation and shaping → coarse adjustment scattering → fine adjustment attenuation," achieving "full-chain scattering compensation" and making beam control more suitable for the complex physical environment of FLASH radiotherapy.

[0169] In one possible implementation, the dose control module includes:

[0170] The dose rate precalibration unit is used to dynamically set the dose rate range based on the radiosensitivity classification of the tumor site before establishing the mapping relationship between pulse parameters and single pulse dose; and to adjust the accelerator beam parameters to stabilize the dose rate within the preset dose rate range.

[0171] The single-pulse dose modeling unit is used to calculate the theoretical single-pulse dose by combining the pulse width and dose rate, and to correct the theoretical single-pulse dose through experimental calibration, thereby establishing a mapping relationship between the calibrated single-pulse dose and the accelerator control parameters.

[0172] D_pulse_theory=D×PW_opt

[0173] D_pulse_theory is the theoretical single-pulse dose; D is the dose rate; PW_opt is the optimal pulse width;

[0174] Among them, by testing different pulse widths, the coefficient of variation of the dose rate under each pulse width is calculated based on the monitoring data, and the pulse width value with the smallest coefficient of variation is selected as the optimal pulse width;

[0175] Through laboratory calibration (dual ionization chamber co-calibration (total dose measured in the main ionization chamber, single-pulse dose measured in the micro-dose ionization chamber)), the theoretical single-pulse dose is corrected, the calibrated single-pulse dose D_pulse_cal is determined, and its mapping relationship with accelerator control parameters (such as magnetic field strength B and radio frequency voltage VRF) is established; D_pulse_cal = f(B, VRF); f() is the mapping relationship function; for example, this mapping relationship can be specifically represented by a continuous function generated by polynomial fitting;

[0176] A pulse sequence generation unit is used to receive the optimal pulse width and the calibrated single-pulse dose, calculate the required number of pulses based on the target total dose, and generate an executable pulse sequence; the executable pulse sequence includes pulse width and current intensity; Map={Pulse1:(PW1,I1),Pulse2:(PW2,I2),...,PulseNp :(PW Np ,I Np )};

[0177] Np = round(D_total / D_pulse_cal)

[0178] round() is for automatic rounding; D_total is the target total dose;

[0179] The verification unit is equipped with a fiber-optic coupled dose monitoring system (sampling frequency ≥ 1 kHz) and performs the following operations: monitors the cumulative dose online through a transmission-type absolute dose ionization chamber, verifies in real time the difference between the actual output cumulative dose and the preset dose, as well as the single-pulse dose difference, and sends a feedback signal to the execution unit if any difference exceeds its preset threshold.

[0180] For example: Single-pulse dose difference |D_pulse_meas - D_pulse_cal| / D_pulse_cal ≤ 3%

[0181] D_pulse_meas is the mean dose of the pulse sequence;

[0182] (D_measured-D_total) / D_total∣×100%<5;

[0183] D_measured is the real-time cumulative dose; D_total is the preset total dose.

[0184] The execution unit is used to execute the pulse sequence, loading the optimized pulse sequence Map into the accelerator control system; it executes Np pulses strictly according to Map to achieve millisecond-level dose cutoff accuracy. The pulse emission interval is synchronized with the accelerator radio frequency period (error ≤ 10 ns).

[0185] The working principle of the above technical solution is as follows:

[0186] Based on the radiosensitivity of tumor sites (e.g., intracranial tumors are more sensitive to dose), the dose rate range is dynamically set (e.g., ≥40 Gy / s in FLASH mode, 2-10 Gy / s in conventional mode). By adjusting the core parameters of the accelerator beam (magnetic field strength B, radio frequency voltage VRF), the output dose rate is stabilized within a preset range. For example, for trunk tumors, the dose rate is locked at 50±2 Gy / s to provide a stable benchmark for subsequent single-pulse dose calculations and avoid initial errors caused by dose rate fluctuations.

[0187] Different pulse widths (e.g., 0.5-2ms) were tested, and dose rate data for each pulse width was collected using a fiber-optic coupled monitoring system. The coefficient of variation (CV = standard deviation / mean) was calculated, and the pulse width with the smallest CV was selected as the optimal pulse width (PW_opt). For example, when PW = 1ms, the dose rate coefficient of variation is 1.2% (less than 2.5% for 0.8ms and 1.8% for 1.5ms), so PW_opt = 1ms, ensuring the stability of the single-pulse dose.

[0188] The theoretical single-pulse dose is calculated based on the optimal pulse width (D_pulse_theory=D×PW_opt, where D is the pre-calibrated dose rate); then, through dual ionization chamber co-calibration, the total dose is measured in the main ionization chamber, and the single-pulse dose is measured in the micro-dose ionization chamber (high time resolution). The calibrated single-pulse dose (D_pulse_cal) is obtained by correcting the theoretical value.

[0189] For example, a pulse with a theoretical value of 0.05 Gy is calibrated and corrected to 0.048 Gy, which is closer to the actual output.

[0190] The D_pulse_cal function is associated with the accelerator control parameters (B, VRF) to form a mapping function D_pulse_cal = f(B, VRF). For example, when B = 0.5T and VRF = 5MV, D_pulse_cal = 0.048Gy. Subsequently, the single-pulse dose can be directly adjusted by adjusting B or VRF to achieve parameterized and precise control.

[0191] Based on the target total dose (D_total, e.g., 4 Gy) and the calibrated single-pulse dose (D_pulse_cal = 0.048 Gy), calculate the required number of pulses: Np = round(D_total / D_pulse_cal) = round(4 / 0.048) = 83 pulses. Generate a sequence containing the width (PW) and current intensity (I) of each pulse: Map = {Pulse1:(PW_opt,I1),Pulse2:(PW_opt,I2),...,Pulse83:(PW_opt,I83)}

[0192] The verification unit is equipped with a fiber-coupled dose monitoring system (sampling frequency ≥ 1 kHz) and performs two core verifications in real time:

[0193] Single-pulse dose difference verification: Calculate |D_pulse_meas - D_pulse_cal| / D_pulse_cal (where D_pulse_meas is the measured average single-pulse dose), which must be ≤3%. For example, if D_pulse_cal = 0.048 Gy and the measured average is 0.049 Gy, the deviation is 2.1% (meets the requirement); if the deviation reaches 4%, a feedback signal is triggered.

[0194] Cumulative dose difference verification: Calculate |D_measured-D_total| / D_total×100% (where D_measured is the real-time cumulative dose). |D_measured−D_total| / D_total×100% must be less than, for example, 5%. For example, with a target total dose of 4 Gy, when the cumulative dose reaches 3.8 Gy, the deviation is 5%, triggering an alert; if it reaches 3.7 Gy (deviation 7.5%), a cutoff signal is immediately sent.

[0195] The execution unit loads the pulse sequence map into the accelerator control system and executes Np pulses strictly according to the sequence:

[0196] Timing synchronization: The pulse emission interval is synchronized with the accelerator radio frequency cycle (error ≤ 10ns) to avoid dose superposition errors caused by timing deviation;

[0197] Rapid cutoff: After receiving the feedback signal from the verification unit, millisecond-level dose cutoff is achieved (response time ≤ 0.5ms). For example, when the single pulse deviation exceeds the limit, the current pulse is terminated within 0.3ms to prevent overdose.

[0198] Redundancy guarantee: After truncation, a redundant pulse sequence is activated to ensure that the total dose can still be accurately achieved (if the original sequence is interrupted, the redundant sequence will supplement the remaining 0.2 Gy dose).

[0199] The effects of the above technical solution are as follows:

[0200] FLASH radiotherapy is a radiotherapy technique that delivers ultra-high doses (≥40Gy) within an extremely short time (typically <1s). Its core advantage lies in its ability to significantly reduce toxicity to normal tissues (FLASH effect). However, the coupling of extremely short time and high dose presents a severe challenge to dose control. This protocol meets the high-precision requirements of "short-time, high-dose deposition" in FLASH radiotherapy through optimal pulse width screening (minimum coefficient of variation) and dual ionization chamber calibration. The cumulative dose deviation is <5%, and combined with real-time verified dynamic early warning, it avoids insufficient or excessive total dose due to equipment drift, ensuring tumor killing effect. The fiber optic coupling system with a sampling frequency of ≥1kHz can capture instantaneous dose fluctuations at ultra-high dose rates (≥40Gy / s) in FLASH mode, solving the problem of "insufficient sampling" in conventional monitoring systems. Millisecond-level dose cutoff (≤0.5ms) and radiofrequency cycle synchronization (error ≤10ns) adapt to the stringent timing requirements of FLASH radiotherapy with irradiation times of less than 1s, avoiding the risk of dose runaway at high dose rates.

[0201] Dual-dimensional verification safeguard: Dual monitoring of single-pulse dose difference and cumulative dose difference ensures dose safety from both "instantaneous" and "overall" levels, reducing damage to normal tissues caused by dose deviation;

[0202] Based on the dynamic dose rate setting of tumor radiosensitivity, it can adapt to the radiotherapy needs of tumors in different locations (such as intracranial and trunk) and achieve personalized dose control; after truncation, redundant pulse sequences are activated to ensure that the treatment process is not interrupted, thereby improving the fault tolerance of the equipment and the success rate of clinical treatment.

[0203] The specific execution effects of the parameters are shown in Table 3:

[0204] Table 3

[0205]

[0206] After 10 repeated verifications, the dose output error was less than 3%, and the stability reached 97.2%.

[0207] In one possible implementation, the pre-calibration of the dose rate includes:

[0208] Based on the radiosensitivity of the target tumor site, the dose rate D range is dynamically set (e.g., 40-60 Gy / s for intracranial tumors, 60-100 Gy / s for trunk tumors); the accelerator beam parameters are adjusted to stabilize the dose rate D within the preset range.

[0209] Optimize pulse current intensity (I) and pulse width (PW) to ensure that the single-pulse dose stability meets the preset stability threshold.

[0210] The optimized pulse current intensity (I) and pulse width (PW) include:

[0211] Fixed PW adjustment I, so that the inter-pulse dose fluctuation is less than the first fluctuation threshold (e.g., 3%).

[0212] Adjust PW with a fixed I to make the pulse width accuracy error less than the first error threshold (e.g., 0.1ms).

[0213] In one possible implementation, the mapping relationship D_pulse_cal=f(B, VRF) is established as follows:

[0214] Under constant PW, calibrate the (B, VRF) combination and the two-dimensional lookup table of D_pulse_cal;

[0215] Generating continuous functions through polynomial fitting:

[0216]

[0217] The coefficient Au is determined by the least squares method; u = 0, 1, 2, 3, 4.

[0218] The effects of the above technical solution are as follows:

[0219] The dose rate is dynamically set based on the tumor location (intracranial / trunk) to balance tumor killing and normal tissue protection. For example, 40-60 Gy / s is used for intracranial tumors to reduce nerve damage, while 60-100 Gy / s is used for trunk tumors to enhance tumor killing, improving the personalization of radiotherapy. Through dual-dimensional optimization of I and PW, the single-pulse dose fluctuation is reduced to <3% and the pulse width error to <0.1 ms, meeting the stringent requirements of FLASH radiotherapy for "short-time, high-dose deposition" (conventional radiotherapy allows for 5% fluctuation), ensuring dose accuracy. Lookup tables make associations "visible," allowing clinicians to quickly match parameters (e.g., queries can be completed within 200 ms); continuous polynomial functions support dose prediction with arbitrary parameter combinations, overcoming the limitations of lookup tables and adapting to complex radiotherapy plans. Pre-calibration and optimization processes reduce the time spent repeatedly debugging B and VRF dose finding in clinical practice (e.g., from 30 minutes to 5 minutes), improving the startup efficiency of radiotherapy equipment and indirectly reducing medical costs.

[0220] In one possible implementation, the pulse sequence generation unit is configured to execute a sequence optimization strategy based on intelligent prediction:

[0221] Based on the number of pulses, sequences are divided into long sequences and short sequences (e.g., if Np > 100, it is a long sequence; if Np < 100, it is a short sequence), and appropriate optimization strategies are adopted for each, including:

[0222] Long sequences are distributed using a uniform distribution strategy. = ; = ; This represents the average width of the long sequence of pulses; This represents the average value of the current intensity of a long sequence of pulses.

[0223] Short sequences employ an incremental flow intensity compensation strategy:

[0224]

[0225]

[0226] in, The first pulse in the pulse sequence The current intensity of each pulse, PW represents the current intensity of the first pulse in the pulse sequence. The first pulse in the pulse sequence The pulse width of each pulse; The upper limit of the current intensity allowed by the accelerator hardware (determined through pre-calibration, e.g., <200nA, to avoid excessive single-pulse dose); constraints: ≤ (Always limit the current intensity within the hardware safety threshold);

[0227] The pulse sequence includes a preset emergency redundant pulse, whose parameters can be dynamically canceled and activated only when the preceding pulse dose is detected to be insufficient.

[0228] In one possible implementation, the pulse sequence generation unit is configured to execute a sequence optimization strategy based on intelligent prediction:

[0229] Based on the number of pulses, sequences are divided into long sequences and short sequences (e.g., if Np > 100, it is a long sequence; if Np is less than or equal to 100, it is a short sequence), and appropriate optimization strategies are applied to each:

[0230] The long sequence employs uniform distribution combined with a drift compensation strategy, periodically correcting the parameters of subsequent pulses in advance based on the beam parameter drift trend of the preceding pulse; for example, inserting one verification pulse every h pulses (e.g., 10 pulses) to correct the pulse width of subsequent pulses.

[0231] The revised pulse width formula:

[0232] PWg'= +ΔPW×(gh×k)

[0233] Wherein, PWg' is the nth pulse in the pulse sequence. The pulse width after pulse correction; ΔPW=k_drift×ΔR (ΔR is the range drift measured by the verification pulse, k_drift is the pulse width correction coefficient, which is experimentally calibrated to k_drift∈[0.1,0.3]mm / Gy); k=floor((g-1) / h) (k is the sequence number of the verification pulse, k=0,1,2,..., representing the (k+1)th group of h pulses);

[0234] Constraint: PWg'≤PW_max (PW_max is the maximum pulse width allowed by the hardware, to avoid excessive single-pulse dose);

[0235] The short sequence employs a strategy of increasing current intensity combined with prediction and compensation. It uses a neural network to predict the attenuation trend of beam parameters and dynamically adjusts the current intensity to compensate for dose rate attenuation.

[0236]

[0237]

[0238] in, The first pulse in the pulse sequence The current intensity of each pulse, The current intensity of the first pulse in the pulse sequence; The upper limit of the current intensity allowed by the accelerator hardware (determined through pre-calibration, e.g., <200nA, to avoid excessive single-pulse dose); constraints: ≤ (Always limit the current intensity within the hardware safety threshold); The dynamic compensation coefficients are determined through a dual-drive approach of LSTM neural network and physical constraints. The dose rate drift is calculated in real time via temperature / vacuum monitoring.

[0239] Determining dynamic compensation coefficients through a dual-drive approach of LSTM neural network and physical constraints includes:

[0240] Based on historical pulse sequence data (beam parameters, dose rate, temperature, vacuum level), the beam attenuation trend of the current pulse is predicted, and the initial compensation coefficient is output. _init;

[0241] The initial compensation coefficients are corrected by applying physical constraints (such as dose rate stability thresholds) to obtain the final compensation coefficients. .

[0242] = _init×(1+λ (Ca-C_threshold))

[0243] Where Ca is the actual value of the constraint index (e.g., the standard deviation of the current dose rate σ_current), C_threshold is the constraint threshold (e.g., 2%), and λ is the constraint strength coefficient (λ∈[0.1,0.3], calibrated experimentally to balance prediction bias and constraint strength).

[0244] The effects of the above technical solution are as follows:

[0245] In long sequences (such as whole bone marrow irradiation requiring 500+ pulses), minute fluctuations in beam parameters (current intensity, pulse width) can accumulate and amplify, causing the total dose deviation to exceed the clinical threshold. By employing a uniform distribution strategy, all pulse parameters are forced to be consistent, transforming the "accumulated risk of fluctuations" into a "fixed deviation" (which can be compensated for through pre-calibration), reducing the total dose error in long sequences, and meeting the dose accuracy requirements of "large fractionation, long sequence" FLASH radiotherapy (e.g., error <0.4Gy for a single 20Gy irradiation).

[0246] One verification pulse is inserted every hour to measure range drift in real time and correct subsequent pulse widths, thereby improving the range drift compensation rate of subsequent pulses.

[0247] In short-sequence ablation (e.g., 10-30 pulses for intracranial metastases), accelerator hardware (e.g., radiofrequency source, magnet) can experience parameter drift due to instantaneous high load (e.g., current intensity attenuation of 2-5% / pulse), leading to insufficient dose at the end of the pulse. A current intensity increment strategy is employed, linearly increasing the current intensity with the pulse sequence to compensate for hardware attenuation. An LSTM neural network learns from historical pulse data (current intensity, dose rate, temperature, vacuum level) to predict the beam attenuation trend of the current pulse, outputting a dynamic compensation coefficient. Physical constraints are then used to correct this, reducing the dose deviation at the end of the short-sequence ablation and meeting the clinical needs for "precise ablation of small tumors."

[0248] See attached document Figure 4 In one possible implementation, the execution process of the dual-mode beam and dose control system for FLASH radiotherapy includes:

[0249] S1. Responding to the FLASH radiotherapy research command, initiate the mode switching process and trigger the drive mechanism to remove the conventional mode treatment head;

[0250] S2. Install the FLASH kit to the accelerator outlet / transport line adapter position. Use the laser light calibration function to detect and correct the physical installation position of the FLASH kit to ensure that the positional accuracy meets the preset requirements (e.g., ≤±0.1mm).

[0251] S3. Controls the TC (Terminal Controller) main computer to switch from clinical mode to FLASH mode, executes a dual-mode independent switching mechanism, automatically disables clinical mode configuration, and loads FLASH mode exclusive permissions and control logic;

[0252] S4. Security Status Verification:

[0253] Verify the collimator status and confirm that the collimator is fully pulled out of the accelerator beam channel in normal mode.

[0254] Verify the compatibility of the FLASH kit with the accelerator's physical connections, electrical signals, and dose control logic;

[0255] S5.FLASH parameter loading and beam output:

[0256] Load the pre-configured FLASH mode beam parameters (including dose rate, pulse width, current intensity, etc.), drive the beam shaping module and dose control module to work together, and output the FLASH mode beam;

[0257] S6. Dosage Verification and Trial Execution:

[0258] The beam dose is collected and verified using a dedicated FLASH ionization chamber and dosimeter box. Once the dose accuracy is confirmed to meet the preset threshold (e.g., error ≤ 5%), the FLASH radiotherapy research trial is initiated.

[0259] S7. Switch back to mode:

[0260] After the FLASH research trial ended, the trigger mechanism removed the FLASH suite, redeployed the conventional mode treatment head, controlled the TC main control computer to switch back to clinical mode, and restored the clinical mode configuration and permissions.

[0261] See attached document Figure 5 This application also provides a dual-mode beam and dose control method for FLASH radiotherapy, characterized in that the method includes:

[0262] The adjustable collimator located at the accelerator outlet or transport line is driven by a drive mechanism to achieve dual-mode switching between clinical mode and FLASH mode, and the switching position accuracy is ensured through a redundant feedback mechanism.

[0263] Based on the characteristics of the proton and heavy ion Bragg peak, at least two materials with different physical properties are used. By adjusting the material thickness and / or quantity, the beam range and dose distribution can be modulated in a graded manner to optimize the Bragg peak depth dose distribution.

[0264] By establishing a mapping relationship between pulse parameters and single-pulse dose, and combining this with the target total dose to calculate the required pulse sequence, precise dose control for FLASH radiotherapy can be achieved.

[0265] In one possible implementation, beam hierarchical modulation is achieved through a hierarchical modulator, which includes a coarse tuning layer and a fine tuning layer.

[0266] The coarse adjustment layer comprises multiple first material components of the same or different thicknesses that are detachable. The first material components are modularly designed through a detachable connection mechanism to achieve step-wise adjustment of the range. The material of the first material components is selected from high atomic number materials.

[0267] The fine adjustment layer comprises multiple second material components of the same or different thicknesses and with an adjustable number of stacked layers. The second material components are used to achieve continuous fine adjustment of the range. The material of the second material components is selected from low atomic number materials.

[0268] Wherein, the total modulation range Ds of the coarse adjustment layer and the fine adjustment layer satisfies:

[0269] Ds=Σ(α i ×N i )+Σ(β j ×M j )+δ cal ,

[0270] α i For the calibrated equivalent water thickness of the i-th first material component, N i The quantity of the i-th first material component used;

[0271] β j M is the calibrated equivalent water thickness of the j-th second material component. j Let j be the number of stacked layers of the j-th second material component;

[0272] δ cal The scattering compensation amount is calibrated using Monte Carlo simulation.

[0273] In one possible implementation, the equivalent water thickness of the first material component is an integer multiple of 5-20 mm, the equivalent water thickness of the second material component is an integer multiple of 0.1-1 mm, and the equivalent water thickness is specified by the following formula:

[0274] WET=[(R water -R air ) / (R material -R air )]×t material

[0275] Where WET is the equivalent water thickness, R is the location of the Bragg peak, and t material R represents the physical thickness of the material. water R represents the Bragg peak range of the particle in water. air R represents the Bragg peak range of the particle in air. material The range of the Bragg peak of the particle in the material.

[0276] In one possible implementation, the physical thickness t of the first material component i With equivalent water thickness α i The following relationship must be satisfied:

[0277] α i =f1×t i Where f1 is the equivalent water thickness coefficient of the material, f1≥2.0 (corresponding to high atomic number materials), and the deviation of the f1 value of each first material component is ≤±0.5%;

[0278] The physical thickness t of the second material component j With equivalent water thickness β j The following relationship must be satisfied:

[0279] β j =f2×t j Where f2 is the equivalent water thickness coefficient of the material, 0.8≤f2≤1.2 (corresponding to low atomic number materials), and the deviation of f2 value for each second material component is ≤±0.3%.

[0280] The combination of the first material component and the second material component includes:

[0281] Layered arrangement: The first material component and the second material component are stacked sequentially along the beam direction;

[0282] Nested arrangement: The second material component is embedded in a preset groove of the first material component.

[0283] In one possible implementation, the component selection of the coarse adjustment layer and the fine adjustment layer satisfies the following:

[0284] Max(α i ×N i ) / Σ(β j ×M j )∈[Cmin,Cmax];

[0285] The critical values ​​(Cmin, Cmax) for selecting the coarse adjustment layer and fine adjustment layer components are calibrated through the following simulation process:

[0286] (1) Constructing a dedicated simulation environment for FLASH radiotherapy

[0287] Based on a particle transport simulation program, an integrated dose-rate-dependent nonlinear scattering physics model is used to simulate the nonlinear behavior of scattering characteristics as dose rate dynamically changes when the ultra-high dose rate beam of FLASH radiotherapy interacts with materials.

[0288] A coupled heat-dose co-conduction module is used to simulate the real-time temperature changes of materials caused by short-term high-dose deposition and their dynamic impact on the beam blocking ability of materials.

[0289] (2) Perform scenario-based parameter scanning and indicator monitoring

[0290] The input parameters include typical beam characteristics and environmental parameters of FLASH radiotherapy, specifically including beam intensity, pulse duration, and initial environmental conditions of materials, to match the actual beam conditions in clinical FLASH studies.

[0291] Monitoring indicators include, but are not limited to: dose distribution uniformity of broadened Bragg peak (SOBP), the proportion of scattered dose to total dose, beam dose rate stability, and the drift of material equivalent water thickness with beam exposure time.

[0292] Cmin (Dominance Threshold of Coarse Adjustment Layer): When the ratio of the adjustment capability of the coarse adjustment layer to that of the fine adjustment layer is lower than Cmin, the monitoring results show that the uniformity of SOBP dose distribution exceeds the preset acceptable range, and the beam dose rate stability drops below the threshold, indicating that the coarse adjustment layer cannot effectively dominate the stepwise adjustment.

[0293] Cmax (Safety threshold of fine-tuning layer): When the above ratio is higher than Cmax, the monitoring results show that the proportion of scattered dose exceeds the safety threshold and the drift of the material equivalent water thickness exceeds the allowable range, indicating that the over-compensation of the fine-tuning layer leads to the loss of beam control.

[0294] In one possible implementation, δ cal Calibration was achieved through a combination of collimation aperture and material combinations.

[0295]

[0296] The area of ​​the collimating hole. The reference area is determined by the collimator design parameters of the radiotherapy equipment.

[0297] μ is the linear scattering coefficient of the material (cm²). -1 (Obtained through Monte Carlo simulation of Geant4, positively correlated with material atomic number and beam energy).

[0298] The total thickness of the fine-tuning layer (cm, thickness of low atomic number material stacking);

[0299] The number of the first material components (to match the needs of the coarse adjustment layer step adjustment);

[0300] δflash is a compensation feature specific to FLASH radiotherapy (due to nonlinear scattering caused by ultra-high dose rates, δflash = ...). D^0.5, (where D is the nonlinear scattering coefficient and D is the dose rate, in Gy / s).

[0301] b1, b2 The scattering coefficients fitted by Monte Carlo simulation and experiment are (b1∈[0.1,0.5], b2∈[0.01,0.1]). ∈[0.005,0.02], which is suitable for beam energy and material composition.

[0302] In one possible implementation, the precise dose control of FLASH radiotherapy is achieved by establishing a mapping relationship between pulse parameters and single-pulse dose, and calculating the required pulse sequence in conjunction with the target total dose, including:

[0303] Before establishing the mapping relationship between pulse parameters and single-pulse dose, the dose rate range is dynamically set based on the radiosensitivity classification of the tumor site; the accelerator beam parameters are adjusted to stabilize the dose rate within the preset dose rate range.

[0304] Based on the dose rate and pulse width, the theoretical dose of a single pulse is determined; the theoretical dose of a single pulse is corrected through experimental calibration, the calibrated dose of a single pulse is determined, and the mapping relationship between the dose of a single pulse and the accelerator control parameters is established.

[0305] D_pulse_theory=D×PW_opt

[0306] D_pulse_theory is the theoretical single-pulse dose; D is the dose rate; PW_opt is the optimal pulse width;

[0307] Among them, by testing different pulse widths, the coefficient of variation of the dose rate under each pulse width is calculated based on the monitoring data, and the pulse width value with the smallest coefficient of variation is selected as the optimal pulse width;

[0308] Through laboratory calibration (dual ionization chamber co-calibration (total dose measured in the main ionization chamber, single-pulse dose measured in the micro-dose ionization chamber)), the theoretical single-pulse dose is corrected, the calibrated single-pulse dose D_pulse_cal is determined, and its mapping relationship with accelerator control parameters (such as magnetic field strength B and radio frequency voltage VRF) is established; D_pulse_cal=f(B,VRF); f() is the mapping relationship function;

[0309] Based on the target total dose and the calibrated single-pulse dose, the required number of pulses is calculated, and an executable pulse sequence is generated; the executable pulse sequence includes pulse width and current intensity; Map={Pulse1:(PW1,I1),Pulse2:(PW2,I2),...,Pulse Np :(PW Np ,I Np )};

[0310] Np = round(D_total / D_pulse_cal)

[0311] round() is for automatic rounding; D_total is the target total dose;

[0312] The difference between the actual output cumulative dose and the preset dose is verified in real time; if the difference exceeds its preset threshold, a feedback signal is sent to the execution unit.

[0313] The optimized pulse sequence Map is loaded into the accelerator control system; Np pulses are executed strictly according to Map to achieve millisecond-level dose cutoff accuracy. The pulse emission interval is synchronized with the accelerator RF period (error ≤ 10 ns).

[0314] In one possible implementation, the pulse sequence generation unit is configured to execute a sequence optimization strategy based on intelligent prediction:

[0315] Based on the number of pulses, sequences are divided into long sequences and short sequences (e.g., if Np > 100, it is a long sequence; if Np < 100, it is a short sequence), and appropriate optimization strategies are adopted for each, including:

[0316] Long sequences are distributed using a uniform distribution strategy. = ; = ; This represents the average width of the long sequence of pulses; This represents the average value of the current intensity of a long sequence of pulses.

[0317] Short sequences employ an incremental flow intensity compensation strategy:

[0318]

[0319]

[0320] in, The first pulse in the pulse sequence The current intensity of each pulse, PW represents the current intensity of the first pulse in the pulse sequence. The first pulse in the pulse sequence The pulse width of each pulse; The upper limit of the current intensity allowed by the accelerator hardware (determined through pre-calibration, e.g., <200nA, to avoid excessive single-pulse dose); constraints: ≤ (Always limit the current intensity within the hardware safety threshold);

[0321] The pulse sequence includes a preset emergency redundant pulse, whose parameters can be dynamically canceled and activated only when the preceding pulse dose is detected to be insufficient.

[0322] In one possible implementation, the pulse sequence generation unit is configured to execute a sequence optimization strategy based on intelligent prediction:

[0323] Based on the number of pulses, sequences are divided into long sequences and short sequences (e.g., if Np > 100, it is a long sequence; if Np is less than or equal to 100, it is a short sequence), and appropriate optimization strategies are applied to each:

[0324] The long sequence employs uniform distribution combined with a drift compensation strategy, periodically correcting the parameters of subsequent pulses in advance based on the beam parameter drift trend of the preceding pulse; for example, inserting one verification pulse every h pulses (e.g., 10 pulses) to correct the pulse width of subsequent pulses.

[0325] The revised pulse width formula:

[0326] PWg'= +ΔPW×(gh×k)

[0327] Wherein, PWg' is the nth pulse in the pulse sequence. The pulse width after pulse correction; ΔPW=k_drift×ΔR (ΔR is the range drift measured by the verification pulse, k_drift is the pulse width correction coefficient, which is experimentally calibrated to k_drift∈[0.1,0.3]mm / Gy); k=floor((g-1) / h) (k is the sequence number of the verification pulse, k=0,1,2,..., representing the (k+1)th group of h pulses);

[0328] Constraint: PWg'≤PW_max (PW_max is the maximum pulse width allowed by the hardware, to avoid excessive single-pulse dose);

[0329] The short sequence employs a strategy of increasing current intensity combined with prediction and compensation. It uses a neural network to predict the attenuation trend of beam parameters and dynamically adjusts the current intensity to compensate for dose rate attenuation.

[0330]

[0331]

[0332] in, The first pulse in the pulse sequence The current intensity of each pulse, The current intensity of the first pulse in the pulse sequence; The upper limit of the current intensity allowed by the accelerator hardware (determined through pre-calibration, e.g., <200nA, to avoid excessive single-pulse dose); constraints: ≤ (Always limit the current intensity within the hardware safety threshold); The dynamic compensation coefficients are determined through a dual-drive approach of LSTM neural network and physical constraints. The dose rate drift is calculated in real time via temperature / vacuum monitoring.

[0333] Determining dynamic compensation coefficients through a dual-drive approach of LSTM neural network and physical constraints includes:

[0334] Based on historical pulse sequence data (beam parameters, dose rate, temperature, vacuum level), the beam attenuation trend of the current pulse is predicted, and the initial compensation coefficient is output. _init;

[0335] The initial compensation coefficients are corrected by applying physical constraints (such as dose rate stability thresholds) to obtain the final compensation coefficients. .

[0336] = _init×(1+λ (Ca-C_threshold))

[0337] Where Ca is the actual value of the constraint index (e.g., the standard deviation of the current dose rate σ_current), C_threshold is the constraint threshold (e.g., 2%), and λ is the constraint strength coefficient (λ∈[0.1,0.3], calibrated experimentally to balance prediction bias and constraint strength).

[0338] The working principle and effect of the above technical solution are the same as those of the dual-mode beam and dose control system for FLASH radiotherapy in the embodiments of this application, and will not be repeated here.

[0339] This application embodiment also provides a graded modulator for realizing graded modulation of beam range and dose distribution, including a coarse adjustment layer and a fine adjustment layer.

[0340] The coarse adjustment layer comprises multiple first material components of the same or different thicknesses that are detachable. The first material components are modularly designed through a detachable connection mechanism to achieve step-wise adjustment of the range. The material of the first material components is selected from high atomic number materials.

[0341] The fine adjustment layer comprises multiple second material components of the same or different thicknesses and with an adjustable number of stacked layers. The second material components are used to achieve continuous fine adjustment of the range. The material of the second material components is selected from low atomic number materials.

[0342] Wherein, the total modulation range Ds of the coarse adjustment layer and the fine adjustment layer satisfies:

[0343] Ds=Σ(α i ×N i )+Σ(β j ×M j )+δ cal ,

[0344] α i For the calibrated equivalent water thickness of the i-th first material component, N i The quantity of the i-th first material component used;

[0345] β j M is the calibrated equivalent water thickness of the j-th second material component. j Let j be the number of stacked layers of the j-th second material component;

[0346] δ cal The scattering compensation amount is calibrated using Monte Carlo simulation.

[0347] This application also provides a beam hierarchical modulation method for achieving hierarchical modulation of beam range and dose distribution. The method is implemented by a hierarchical modulator, including a coarse adjustment layer and a fine adjustment layer.

[0348] The coarse adjustment layer comprises multiple first material components of the same or different thicknesses that are detachable. The first material components are modularly designed through a detachable connection mechanism to achieve step-wise adjustment of the range. The material of the first material components is selected from high atomic number materials.

[0349] The fine adjustment layer comprises multiple second material components of the same or different thicknesses and with an adjustable number of stacked layers. The second material components are used to achieve continuous fine adjustment of the range. The material of the second material components is selected from low atomic number materials.

[0350] Wherein, the total modulation range Ds of the coarse adjustment layer and the fine adjustment layer satisfies:

[0351] Ds=Σ(α i ×N i )+Σ(β j ×M j )+δ cal ,

[0352] α i For the calibrated equivalent water thickness of the i-th first material component, N i The quantity of the i-th first material component used;

[0353] β j M is the calibrated equivalent water thickness of the j-th second material component. j Let j be the number of stacked layers of the j-th second material component;

[0354] δ cal The scattering compensation amount is calibrated using Monte Carlo simulation.

[0355] In one possible implementation, the equivalent water thickness of the first material component is an integer multiple of 5-20 mm, the equivalent water thickness of the second material component is an integer multiple of 0.1-1 mm, and the equivalent water thickness is specified by the following formula:

[0356] WET=[(R water -R air ) / (R material -R air )]×t material

[0357] Where WET is the equivalent water thickness, R is the location of the Bragg peak, and t material R represents the physical thickness of the material. water R represents the Bragg peak range of the particle in water. airR represents the Bragg peak range of the particle in air. material The range of the Bragg peak of the particle in the material.

[0358] In one possible implementation, the physical thickness t of the first material component i With equivalent water thickness α i The following relationship must be satisfied:

[0359] α i =f1×t i Where f1 is the equivalent water thickness coefficient of the material, f1≥2.0 (corresponding to high atomic number materials), and the deviation of the f1 value of each first material component is ≤±0.5%;

[0360] The physical thickness t of the second material component j With equivalent water thickness β j The following relationship must be satisfied:

[0361] β j =f2×t j Where f2 is the equivalent water thickness coefficient of the material, 0.8≤f2≤1.2 (corresponding to low atomic number materials), and the deviation of f2 value for each second material component is ≤±0.3%.

[0362] The combination of the first material component and the second material component includes:

[0363] Layered arrangement: The first material component and the second material component are stacked sequentially along the beam direction;

[0364] Nested arrangement: The second material component is embedded in a preset groove of the first material component.

[0365] In one possible implementation, the component selection of the coarse adjustment layer and the fine adjustment layer satisfies the following:

[0366] Max(α i ×N i ) / Σ(β j ×M j )∈[Cmin,Cmax];

[0367] The critical values ​​(Cmin, Cmax) for selecting the coarse adjustment layer and fine adjustment layer components are calibrated through the following simulation process:

[0368] (1) Constructing a dedicated simulation environment for FLASH radiotherapy

[0369] Based on a particle transport simulation program, an integrated dose-rate-dependent nonlinear scattering physics model is used to simulate the nonlinear behavior of scattering characteristics as dose rate dynamically changes when the ultra-high dose rate beam of FLASH radiotherapy interacts with materials.

[0370] A coupled heat-dose co-conduction module is used to simulate the real-time temperature changes of materials caused by short-term high-dose deposition and their dynamic impact on the beam blocking ability of materials.

[0371] (2) Perform scenario-based parameter scanning and indicator monitoring

[0372] The input parameters include typical beam characteristics and environmental parameters of FLASH radiotherapy, specifically including beam intensity, pulse duration, and initial environmental conditions of materials, to match the actual beam conditions in clinical FLASH studies.

[0373] Monitoring indicators include, but are not limited to: dose distribution uniformity of broadened Bragg peak (SOBP), the proportion of scattered dose to total dose, beam dose rate stability, and the drift of material equivalent water thickness with beam exposure time.

[0374] Cmin (Dominance Threshold of Coarse Adjustment Layer): When the ratio of the adjustment capability of the coarse adjustment layer to that of the fine adjustment layer is lower than Cmin, the monitoring results show that the uniformity of SOBP dose distribution exceeds the preset acceptable range, and the beam dose rate stability drops below the threshold, indicating that the coarse adjustment layer cannot effectively dominate the stepwise adjustment.

[0375] Cmax (Safety threshold of fine-tuning layer): When the above ratio is higher than Cmax, the monitoring results show that the proportion of scattered dose exceeds the safety threshold and the drift of the material equivalent water thickness exceeds the allowable range, indicating that the over-compensation of the fine-tuning layer leads to the loss of beam control.

[0376] In one possible implementation, δ cal Calibration was achieved through a combination of collimation aperture and material combinations.

[0377]

[0378] The area of ​​the collimating hole. The reference area is determined by the collimator design parameters of the radiotherapy equipment.

[0379] μ is the linear scattering coefficient of the material (cm²). -1 (Obtained through Monte Carlo simulation of Geant4, positively correlated with material atomic number and beam energy).

[0380] The total thickness of the fine-tuning layer (cm, thickness of low atomic number material stacking);

[0381] The number of the first material components (to match the needs of the coarse adjustment layer step adjustment);

[0382] δflash is a compensation feature specific to FLASH radiotherapy (due to nonlinear scattering caused by ultra-high dose rates, δflash = ...). D^0.5, (where D is the nonlinear scattering coefficient and D is the dose rate, in Gy / s).

[0383] b1, b2 The scattering coefficients fitted by Monte Carlo simulation and experiment are (b1∈[0.1,0.5], b2∈[0.01,0.1]). ∈[0.005,0.02], which is suitable for beam energy and material composition.

[0384] The working principle and effect of the above technical solution are the same as those of the beam shaping module of the dual-mode beam and dose control system for FLASH radiotherapy in the embodiments of this application, and will not be repeated here.

[0385] See attached document Figure 6 This application also provides a beam dose control method, including:

[0386] Before establishing the mapping relationship between pulse parameters and single-pulse dose, the dose rate range is dynamically set based on the radiosensitivity classification of the tumor site; the accelerator beam parameters are adjusted to stabilize the dose rate within the preset dose rate range.

[0387] Based on the dose rate and pulse width, the theoretical dose of a single pulse is determined; the theoretical dose of a single pulse is corrected through experimental calibration, the calibrated dose of a single pulse is determined, and the mapping relationship between the dose of a single pulse and the accelerator control parameters is established.

[0388] Based on the target total dose and the calibrated single-pulse dose, calculate the required number of pulses and generate an executable pulse sequence;

[0389] The difference between the actual output cumulative dose and the preset dose is verified in real time; if the difference exceeds its preset threshold, a feedback signal is sent to the execution unit.

[0390] Before establishing the mapping relationship between pulse parameters and single-pulse dose, the dose rate range is dynamically set based on the radiosensitivity classification of the tumor site; the accelerator beam parameters are adjusted to stabilize the dose rate within the preset dose rate range.

[0391] Based on the dose rate and pulse width, the theoretical dose of a single pulse is determined; the theoretical dose of a single pulse is corrected through experimental calibration, the calibrated dose of a single pulse is determined, and the mapping relationship between the dose of a single pulse and the accelerator control parameters is established.

[0392] D_pulse_theory=D×PW_opt

[0393] D_pulse_theory is the theoretical single-pulse dose; D is the dose rate; PW_opt is the optimal pulse width;

[0394] Among them, by testing different pulse widths, the coefficient of variation of the dose rate under each pulse width is calculated based on the monitoring data, and the pulse width value with the smallest coefficient of variation is selected as the optimal pulse width;

[0395] Through laboratory calibration (dual ionization chamber co-calibration (total dose measured in the main ionization chamber, single-pulse dose measured in the micro-dose ionization chamber)), the theoretical single-pulse dose is corrected, the calibrated single-pulse dose D_pulse_cal is determined, and its mapping relationship with accelerator control parameters (such as magnetic field strength B and radio frequency voltage VRF) is established; D_pulse_cal=f(B,VRF); f() is the mapping relationship function;

[0396] Based on the target total dose and the calibrated single-pulse dose, the required number of pulses is calculated, and an executable pulse sequence is generated; the executable pulse sequence includes pulse width and current intensity; Map={Pulse1:(PW1,I1),Pulse2:(PW2,I2),...,Pulse Np :(PW Np ,I Np )};

[0397] Np = round(D_total / D_pulse_cal)

[0398] round() is for automatic rounding; D_total is the target total dose;

[0399] The difference between the actual output cumulative dose and the preset dose is verified in real time; if the difference exceeds its preset threshold, a feedback signal is sent to the execution unit.

[0400] The optimized pulse sequence Map is loaded into the accelerator control system; Np pulses are executed strictly according to Map to achieve millisecond-level dose cutoff accuracy. The pulse emission interval is synchronized with the accelerator RF period (error ≤ 10 ns).

[0401] In one possible implementation, the pulse sequence generation unit is configured to execute a sequence optimization strategy based on intelligent prediction:

[0402] Based on the number of pulses, sequences are divided into long sequences and short sequences (e.g., if Np > 100, it is a long sequence; if Np < 100, it is a short sequence), and appropriate optimization strategies are adopted for each, including:

[0403] Long sequences are distributed using a uniform distribution strategy. = ; = ; This represents the average width of the long sequence of pulses; This represents the average value of the current intensity of a long sequence of pulses.

[0404] Short sequences employ an incremental flow intensity compensation strategy:

[0405]

[0406]

[0407] in, The first pulse in the pulse sequence The current intensity of each pulse, PW represents the current intensity of the first pulse in the pulse sequence. The first pulse in the pulse sequence The pulse width of each pulse; The upper limit of the current intensity allowed by the accelerator hardware (determined through pre-calibration, e.g., <200nA, to avoid excessive single-pulse dose); constraints: ≤ (Always limit the current intensity within the hardware safety threshold);

[0408] The pulse sequence includes a preset emergency redundant pulse, whose parameters can be dynamically canceled and activated only when the preceding pulse dose is detected to be insufficient.

[0409] In one possible implementation, the pulse sequence generation unit is configured to execute a sequence optimization strategy based on intelligent prediction:

[0410] Based on the number of pulses, sequences are divided into long sequences and short sequences (e.g., if Np > 100, it is a long sequence; if Np is less than or equal to 100, it is a short sequence), and appropriate optimization strategies are applied to each:

[0411] The long sequence employs uniform distribution combined with a drift compensation strategy, periodically correcting the parameters of subsequent pulses in advance based on the beam parameter drift trend of the preceding pulse; for example, inserting one verification pulse every h pulses (e.g., 10 pulses) to correct the pulse width of subsequent pulses.

[0412] The revised pulse width formula:

[0413] PWg'= +ΔPW×(gh×k)

[0414] Wherein, PWg' is the nth pulse in the pulse sequence. The pulse width after pulse correction; ΔPW=k_drift×ΔR (ΔR is the range drift measured by the verification pulse, k_drift is the pulse width correction coefficient, which is experimentally calibrated to k_drift∈[0.1,0.3]mm / Gy); k=floor((g-1) / h) (k is the sequence number of the verification pulse, k=0,1,2,..., representing the (k+1)th group of h pulses);

[0415] Constraint: PWg'≤PW_max (PW_max is the maximum pulse width allowed by the hardware, to avoid excessive single-pulse dose);

[0416] The short sequence employs a strategy of increasing current intensity combined with prediction and compensation. It uses a neural network to predict the attenuation trend of beam parameters and dynamically adjusts the current intensity to compensate for dose rate attenuation.

[0417]

[0418]

[0419] in, The first pulse in the pulse sequence The current intensity of each pulse, The current intensity of the first pulse in the pulse sequence; The upper limit of the current intensity allowed by the accelerator hardware (determined through pre-calibration, e.g., <200nA, to avoid excessive single-pulse dose); constraints: ≤ (Always limit the current intensity within the hardware safety threshold); The dynamic compensation coefficients are determined through a dual-drive approach of LSTM neural network and physical constraints. The dose rate drift is calculated in real time via temperature / vacuum monitoring.

[0420] Determining dynamic compensation coefficients through a dual-drive approach of LSTM neural network and physical constraints includes:

[0421] Based on historical pulse sequence data (beam parameters, dose rate, temperature, vacuum level), the beam attenuation trend of the current pulse is predicted, and the initial compensation coefficient is output. _init;

[0422] The initial compensation coefficients are corrected by applying physical constraints (such as dose rate stability thresholds) to obtain the final compensation coefficients. .

[0423] = _init×(1+λ (Ca-C_threshold))

[0424] Where Ca is the actual value of the constraint index (e.g., the standard deviation of the current dose rate σ_current), C_threshold is the constraint threshold (e.g., 2%), and λ is the constraint strength coefficient (λ∈[0.1,0.3], calibrated experimentally to balance prediction bias and constraint strength).

[0425] The working principle and effect of the above technical solution are the same as those of the dose control module of the dual-mode beam and dose regulation system for FLASH radiotherapy in the embodiments of this application, and will not be repeated here.

[0426] See attached document Figure 7 This application provides a radiotherapy system, including:

[0427] A cyclotron is used as a beam generating device. Through the synergistic action of an ion source subsystem, a radio frequency subsystem, a vacuum subsystem, a magnet subsystem, and an extraction subsystem, it generates a proton beam for clinical radiotherapy or FLASH research. The magnet subsystem guides the acceleration and deflection of the proton beam through D-shaped electrodes and a magnetic field. The radio frequency subsystem continuously energizes the proton beam through an alternating electric field until the proton beam reaches the output of the extraction subsystem.

[0428] The dual-mode beam and dose control system for FLASH radiotherapy described in any embodiment of this application is connected to the beam output terminal of the cyclotron accelerator and includes:

[0429] Mode switching module: used to drive the adjustable collimator (located at the accelerator outlet or transport line) to achieve dual switching between clinical mode and FLASH mode, and ensures the accuracy of the switching position through a redundant feedback mechanism;

[0430] Beam shaping module: Based on the characteristics of proton and heavy ion Bragg peaks, it uses at least two materials with different physical properties to hierarchically adjust the beam range and dose distribution, and optimize the Bragg peak depth dose.

[0431] Dose control module: Used to establish the mapping relationship between pulse parameters (current intensity, pulse width) and single pulse dose, and generate pulse sequence in combination with the target total dose to achieve precise dose control of FLASH radiotherapy;

[0432] The cyclotron provides the initial proton beam for the dual-mode beam and dose control system. The dual-mode control system adapts to the conventional dose requirements of clinical radiotherapy and the ultra-high dose rate requirements of FLASH radiotherapy through mode switching, beam shaping, and dose control, thus synergistically realizing the functions of precision radiotherapy for tumors and FLASH research.

[0433] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described in this application or the functions of the system described in this application.

[0434] This application also provides a computer-readable storage medium for storing a computer program. When the computer program is executed, it implements the steps of any of the methods described in this application. The specific implementation method is consistent with the implementation method and the technical effect achieved in the above method embodiments, and some contents will not be repeated.

[0435] In this application, a readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. The program product can take the form of any combination of one or more readable media. A readable medium can be a readable signal medium or a readable storage medium. A readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0436] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium capable of sending, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, or any suitable combination thereof. Program code for performing operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar programming languages. The program code may be executed entirely on a user computing device, partially on an associated device, as a standalone software package, partially on a user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing devices can be connected to user computing devices via any type of network, including local area networks (LANs) or wide area networks (WANs), or they can be connected to external computing devices (e.g., via the Internet using an Internet service provider).

[0437] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the invention without departing from the principles and spirit of the invention, and all such changes should fall within the protection scope of the claims of the present invention.

Claims

1. A beam dose control method, characterized in that, include: Before establishing the mapping relationship between pulse parameters and single-pulse dose, the dose rate range is dynamically set based on the radiosensitivity classification of the tumor site. Adjust the accelerator beam parameters to stabilize the dose rate within the preset dose rate range; The theoretical dose per pulse is determined based on the dose rate and pulse width. The theoretical dose of a single pulse was corrected by experimental calibration, the calibrated dose of the single pulse was determined, and the mapping relationship between the dose of the single pulse and the accelerator control parameters was established. Based on the target total dose and the calibrated single-pulse dose, calculate the required number of pulses and generate an executable pulse sequence; Real-time verification of the difference between the actual cumulative output dose and the preset dose; If the difference exceeds its preset threshold, a feedback signal is sent to the execution unit.

2. The beam dose control method according to claim 1, characterized in that, The determination of the theoretical dose for a single pulse includes: D_pulse_theory= D×PW_opt Wherein, D_pulse_theory is the theoretical dose of a single pulse; D is the dose rate; PW_opt is the optimal pulse width; the optimal pulse width is selected by testing the coefficient of variation of the dose rate under different pulse widths and choosing the pulse width value with the smallest coefficient of variation.

3. The beam dose control method according to claim 1, characterized in that, The experimental calibration employed a dual-ionization chamber synergistic calibration: the main ionization chamber measured the total dose, while the micro-dose ionization chamber measured the single-pulse dose, and the calibrated single-pulse dose was obtained by correcting the theoretical single-pulse dose.

4. The beam dose control method according to claim 1, characterized in that, The process of establishing the mapping relationship between single-pulse dose and accelerator control parameters includes: calibrating a two-dimensional lookup table of magnetic field strength, radio frequency voltage and single-pulse dose under constant pulse width, and generating a continuous function through polynomial fitting.

5. The beam dose control method according to claim 1, characterized in that, The generation of the executable pulse sequence includes: calculating the number of pulses based on the target total dose and the calibrated single pulse dose, generating a sequence containing the pulse width and current intensity of each pulse, and synchronizing the pulse emission interval with the accelerator radio frequency cycle.

6. The beam dose control method according to claim 5, characterized in that, The pulse sequence employs a sequence optimization strategy based on intelligent prediction: Based on the comparison results between the number of pulses and a preset threshold, the pulse sequence is divided into long sequences and short sequences; When the pulse sequence is long, a verification pulse is inserted at a predetermined interval, and the pulse width is corrected according to the range drift. When the pulse sequence is short, a current intensity increment compensation strategy is adopted, and the beam attenuation trend is predicted based on the LSTM neural network. The dose rate attenuation is compensated by dynamically adjusting the current intensity.

7. The beam dose control method according to claim 1, characterized in that, The real-time verification includes: verifying whether the single-pulse dose difference and cumulative dose difference exceed their respective preset thresholds; if they exceed, feedback is provided and millisecond-level dose cutoff is achieved.

8. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program performing the steps of the method as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The storage medium stores computer instructions, and when the computer reads the computer instructions, the computer executes the steps of the method as described in any one of claims 1-7.

Images