Three-dimensional space radiation dose reconstruction method based on digital twinborn technology

Abstract

The invention discloses a three-dimensional radiation dose reconstruction method based on digital twinning, and the method comprises the steps: building a scintillation fiber dosimeter physical response model, and obtaining a linear relation between an optical signal and a dose; segmenting human organs based on a CT image, constructing a digital physical model by using a tissue equivalent material, and carrying out geometric modeling; calculating a dose field, and obtaining three-dimensional dose data; inserting an optical fiber dosimeter into a preset position of the phantom for measurement, and calculating an actual dose in combination with the linear relation; and the actually measured data is applied to an area with relatively large dose gradient change in the dose matrix obtained by calculation, so that three-dimensional dose distribution is reconstructed. According to the method, a closed loop is realized through real-time interaction of a physical model and a calculation model, the simulation time is shortened from several hours to a minute level through parallel calculation, the problem that three-dimensional dose distribution cannot be rapidly obtained through a traditional method is solved by combining optical fiber high-resolution measurement and total-space calculation, and the measurement deviation of a non-uniform medium is reduced through an energy segmentation correction strategy.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of radiotherapy, and particularly relates to a three-dimensional space radiation dose reconstruction method based on a digital twin technology. BACKGROUND

[0002] Proton radiotherapy is one of the important methods for treating malignant tumors. In the treatment process, the beam delivery system accurately delivers the main dose of the beam to the tumor target area, realizes killing of tumor tissues with the maximum irradiation dose, and at the same time, maximally avoids damage to the surrounding normal organ tissues. However, in the actual treatment process, due to the influence of factors such as physiological movement of patients, positioning accuracy, individual density difference and beam parameter change, there are differences between the transferred dose distributions in the tumor target area, and the uncertainty of the dose distribution is generated. At present, the quality assurance of clinical radiotherapy is mainly through pre-treatment plan verification and in-treatment beam state detection, but the actual irradiated dose in the patient's body cannot be quickly and accurately obtained. In recent years, the optical fiber-based dose measurement technology has gradually become a research hotspot in the field of radiotherapy dose verification due to its high spatial resolution, real-time measurement capability and anti-electromagnetic interference. However, how to quickly and accurately reconstruct the three-dimensional space dose distribution by using discrete optical fiber measurement values is still a challenge to be solved When dose verification is performed before treatment planning, the dose distribution in the human body cannot be directly measured, and the actual method adopted is to select a tissue equivalent material to form a measurement phantom, irradiate the phantom with a treatment beam, and measure the absorbed dose distribution in the phantom. Therefore, the phantom, as an important tool for QA, can replace the dose distribution in the human body by measuring the spatial dose distribution in the phantom, to a certain extent, solve the problem that the dose in the patient's body cannot be measured in real time. At the same time, the phantom can also be used for proton reference dose determination audit, end-to-end based dose determination audit, and can also be used to deeply understand the range accuracy in heterogeneous scenarios. In addition, when the phantom is used as part of independent audit, the dose information provided by the phantom can provide important understanding and understanding of any potential problems of the radiotherapy center, and realize the repeatability and consistency between the clinical trials and the complex treatment center. Therefore, most of the dose verification in the current proton therapy is carried out on the simulated irradiation phantom.

[0003] Traditional dose measurement methods such as ionization chamber and film dosimeter can provide accurate dose information, but they can only provide point or two-dimensional plane dose data, which is difficult to meet the needs of three-dimensional dose distribution reconstruction. In recent years, the dosimeter based on plastic scintillator (PSD) probe and optical fiber coupling is one of the types widely studied in the field of radiotherapy dosimetry, which is valued for its linear dose response, high spatial resolution and minimal quenching phenomenon in proton dose measurement. When plastic scintillator detector (PSD) is used in combination with plastic optical fiber, due to the water equivalent property, it will not interfere with the process of radiation energy deposition, thereby ensuring the accuracy of dose measurement. PSD has high spatial resolution and can be inserted, which can be used as a suitable method for real-time measurement of three-dimensional dose. It can not only read out the dose in real time, but also directly read out the dose when the phantom material is properly selected, which can realize high-precision three-dimensional dose reconstruction in the phantom, and will help the current clinical understanding of range uncertainty in heterogeneous space.

[0004] MC dose calculation has been introduced into proton therapy, but as a statistical method, the total number of particles in simulation determines the accuracy of dose calculation. In order to achieve sufficient statistical uncertainty, a large number of particle beams are usually required and a long calculation time is needed. The time required to calculate the dose distribution of a patient using the conventional MC method is from several hours to several tens of hours, which seriously hinders the application of the MC method in clinical dose assurance. In order to break through the technical bottleneck of traditional proton dose calculation methods in terms of calculation efficiency and real-time monitoring ability, a GPU-accelerated proton dose distribution calculation method is proposed, and a scintillating optical fiber dose measurement system is integrated to build a dose verification closed loop, and finally realize real-time reconstruction of three-dimensional space radiation dose of phantom QA. SUMMARY

[0005] The purpose of the present application is to provide a three-dimensional space radiation dose reconstruction method based on digital twin technology to solve the problem of insufficient calculation efficiency and real-time monitoring ability in the prior art MC dose calculation to achieve sufficient statistical uncertainty, which usually requires a large number of particle beams and a long calculation time.

[0006] To solve the above technical problems, the technical scheme adopted by the present application is: A three-dimensional space radiation dose reconstruction method based on digital twin technology, comprising the following steps: Step S1, a scintillating optical fiber dosimeter physical response model is established to obtain the linear scale relationship between optical signal intensity and radiation dose; Step S2, based on CT image data, the three-dimensional coordinates of human organs or tissues are obtained by segmentation, and geometric modeling is performed, and a simulated human digital physical model is constructed by using tissue equivalent materials; Step S3, using GPU accelerated Monte Carlo simulation code to calculate the dose field, obtain the three-dimensional space dose data; Step S4, insert the optical fiber dosimeter into the preset position of the digital physical model of the simulated human body, and measure the dose value of the position with large gradient change difference in real time; Step S5, based on the GPU calculation data and the measured value of the optical fiber, the measured dose of the difference point is fed back to the dose data calculated by the GPU in real time, and the three-dimensional space radiation dose distribution of the model is inversely calculated through interactive adjustment.

[0007] According to the above technical scheme, in step S1, the physical response model of the scintillating optical fiber dosimeter is established, and the linear scale relationship between the optical signal intensity and the radiation dose is obtained, which is specifically: The linear scale relationship between the optical signal intensity and the radiation dose is established by verifying the radiation response characteristics of the scintillating optical fiber dosimeter in the non-uniform medium by the Monte Carlo method, and the linear scale relationship is specifically: Wherein, I is the number of scintillating photons collected by the detector, D is the real-time dose of the scintillating optical fiber, and k, b are fitting parameters.

[0008] According to the above technical scheme, in step S1, the scintillating optical fiber dosimeter is in the energy range of 70MeV-230MeV.

[0009] According to the above technical scheme, in step S2, the human organs or tissues are automatically segmented based on the CT image data, and the three-dimensional coordinates thereof are obtained, which are specifically: The organ segmentation of medical CT image converts the pixel unit of the phantom CT image into the voxel unit, and based on the three-dimensional coordinate data of the voxel unit, the three-dimensional geometric structure of the digital physical model of the simulated human body is constructed by the volume rendering algorithm.

[0010] According to the above technical scheme, in step S2, the tissue equivalent material includes the density and elements of soft tissue, fat and bone, which are directly filled into the corresponding organs or tissues.

[0011] According to the above technical scheme, in step S3, the pre-processing step of GPU accelerated calculation includes: configuring the proton beam flow parameters by reading the treatment plan file, constructing a three-dimensional geometric model based on the voxel coordinate file output by the MOOSE software, and filling the corresponding tissue equivalent material parameters into each voxel unit of the geometric model.

[0012] According to the above technical scheme, in step S3, the simulation process executed by the GPU calculation unit includes: Step S301, simulate the energy deposition, multiple coulomb scattering and non-elastic nuclear reaction process of protons through parallel computing threads; Step S302, the random transport path of the secondary particle generated by the proton nuclear reaction is tracked in real time; Step S303, the memory allocation of the integral voxel is managed by using a hash table data structure, and only the memory space of the voxel unit interacting with the particle is allocated.

[0013] According to the above technical scheme, in step S4, the optical fiber dosimeter is inserted into a predetermined position of the phantom, such as: the isocenter position of the proton beam, the tissue interface dose gradient region of the muscle and the pancreas, the pancreas and the liver, and the center position of the target treatment region and its adjacent high dose gradient but low non-uniformity region.

[0014] According to the above technical scheme, in step S5, the dose difference point is adjusted interactively by the measured dose data, specifically: by comparing the measured value of the optical fiber dosimeter with the GPU calculation value, the correction factor for different tissues and energy segments in the dose gradient region is determined, wherein, for the full energy segment of the fat tissue, a negative deviation correction factor is used; for the 74-100MeV energy segment of the bone tissue, a positive deviation correction factor is used; for the 150-230MeV energy segment, a negative deviation correction factor is used.

[0015] According to the above technical scheme, in step S5, the reconstruction result of the three-dimensional space radiation dose distribution can be visualized in the following way: Generating dose slice graphs of different depths; Based on the three-dimensional dose data calculated by the GPU, a dose distribution cloud chart in the phantom is generated.

[0016] Compared with the prior art, the present application has the following beneficial effects: In the present application, through real-time interaction of the physical and computational models of the phantom, a closed loop of dose measurement, calculation and correction is realized, and the reconstruction accuracy is improved. GPU parallel computing shortens the Monte Carlo simulation time from several hours to minutes, meeting the real-time needs of clinical practice. Combined with high-resolution measurement of optical fibers and full-space calculation of GPUs, the problem of being unable to obtain three-dimensional dose distribution in traditional methods is solved; through energy segment correction strategy, the dose measurement deviation in non-uniform medium is significantly reduced, and the method is suitable for dose verification of complex anatomical structures. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The flowchart of the three-dimensional space dose reconstruction method of the proton beam therapy of the present application; Figure 2 The schematic diagram of the dose deposition difference rate of the scintillating optical fiber in the water and tissue equivalent material of the present application; Figure 3 The schematic diagram of the chest, abdomen and part of the human body organs constructed by the three-dimensional coordinates of the segmented voxels of the organs or tissues of the present application; Figure 4This is a schematic diagram of the human spine, ribs, pelvis, etc., constructed using the three-dimensional coordinates of voxels segmented from organs or tissues according to the present invention. Figure 5 This is a monoenergetic one-dimensional dose distribution map obtained by GPU computation according to the present invention; Figure 6 This is a flowchart of the GPU-accelerated method for rapid reconstruction of three-dimensional spatial radiation according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 like Figure 1 As shown, a method for reconstructing three-dimensional spatial radiation dose based on digital twin technology includes the following steps: Step S1: Establish a physical response model of the scintillation fiber optic dosimeter to obtain a linear scale relationship between optical signal intensity and radiation dose; Step S2: Based on CT images, human organs or tissues are automatically segmented using multi-organ segmentation software to obtain their three-dimensional coordinates. A simulated human digital physical model is constructed using tissue equivalent materials and input into Monte Carlo simulation software for geometric modeling. Step S3: Calculate the dose field using GPU-accelerated Monte Carlo simulation code to obtain three-dimensional spatial dose data; Step S4: Insert the fiber optic dosimeter into the predetermined position of the phantom, measure the dose value at that position in real time, and calculate the actual dose based on the linear scale relationship. Step S5: Based on GPU-computed data and fiber optic measured values, the measured dose difference points are fed back to the dose data calculated by the GPU in real time. Through interactive adjustment, the three-dimensional spatial radiation dose distribution of the model is quickly retrieved.

[0020] In this invention, a closed loop of dose measurement, calculation, and correction is achieved through real-time interaction between the physical model and the computational model of the phantom, improving reconstruction accuracy. GPU parallel computing reduces Monte Carlo simulation time from hours to minutes, meeting real-time clinical needs. Combining high-resolution fiber optic measurement with GPU full-space computation solves the problem of traditional methods being unable to quickly obtain three-dimensional dose distribution; through an energy segmentation correction strategy, dose measurement deviation in non-uniform media is significantly reduced, making it suitable for dose verification of complex anatomical structures.

[0021] Example 2 The embodiment is a further refinement of embodiment one. The application provides a GPU accelerated proton therapy three-dimensional space dose reconstruction method based on digital twin technology, comprising the following steps: establishing a coupling physical response model of the multi-parameter scintillating optical fiber dosimeter; in order to accurately analyze the influence of the dosimeter measurement accuracy and the radiation field disturbance on the dosimeter measurement value, a correction scheme for the optical fiber dosimeter is established; the micro dosimeter is placed in the phantom for dose measurement, and the correction factor obtained by the correction scheme is coupled with the three-dimensional space dose reconstruction algorithm of the GPU program in the digital phantom to reconstruct the dose distribution in the phantom.

[0022] The application provides a scintillating optical fiber dosimeter radiation dose response model based on a Monte Carlo method, which comprehensively considers the collection efficiency, dose dependence, water equivalence and other factors of the scintillation crystal and optical fiber material. In order to realize efficient radiation dose measurement, the application selects a BCF-60 plastic scintillation crystal with a diameter of 1mm produced by Saint Gobain as the optical fiber scintillation probe, and adopts Eska Premier 1mm single-core optical fiber (Mitsubishi Chemical Corporation, Chiyoda-ku) as the transmission optical fiber. The combination of the two materials establishes the radiation response model of the optical fiber dosimeter. The measurement value of the optical fiber dosimeter is the light signal intensity I, and there is a linear scale relationship between the signal intensity and the radiation dose D. The relationship can be expressed as: Wherein I is the number of scintillation photons collected by the detector, D is the real-time dose of the scintillating optical fiber, and k and b are fitting parameters.

[0023] Firstly, the dose response model of the optical fiber dosimeter in different media is established through pre-simulation measurement. The radiation response model of the optical fiber dosimeter for human tissue equivalent materials is constructed, and simulation verification is carried out based on the Monte Carlo method, the influence of different tissues on the response of the dosimeter is quantified, and the prediction accuracy of the model in non-uniform media is discussed, and the clinical feasibility is evaluated. The tissue equivalent material parameters provided by ICRU No. 44 report and NST are adopted, including density, element composition, etc., and representative human tissue equivalent materials are selected for radiation dosimetry calculation, including soft tissue, fat, bone and the like.

[0024] The tissue equivalence of the scintillating optical fiber dosimeter is verified, and simulation verification is carried out in the energy range of 70MeV-230MeV.

[0025] From Figure 2It can be seen that the difference rate of dose deposition between the scintillating fiber and water is close to 0% at all energies, indicating that the scintillating fiber dosimeter has the best water equivalence, which meets the expected design goal. In the low energy range (74-100 MeV), the scintillating fiber has a small dose deposition deviation in soft tissue (within ±2%), and the equivalence is good; but as the energy increases (150-230 MeV), the deviation gradually increases to ±4%, which can be corrected by a simple correction factor. In fat, there is a negative deviation (-4% to -6%) at all energies, and the higher the energy, the more significant the deviation. This is due to the low density and low stopping power of fat, which causes protons to penetrate deeper and reduce dose deposition at 10 mm. At low energy (74-100 MeV), the scintillating fiber shows a significant positive deviation (+6%-+4%) in bone due to the high density of bone, which causes protons to quickly deposit energy; but as the energy increases (150-230 MeV), the deviation gradually decreases or even turns negative (-2%), which may be related to the penetration enhancement of high-energy protons and the shift of Bragg peak position. The equivalence of bone is highly dependent on energy and needs to be adjusted accordingly. In addition, within this energy range, the difference rate between the scintillating fiber and water is basically constant as the energy changes, and the dose difference of other tissues after correction is significantly reduced and can be ignored, so the scintillating fiber dosimeter shows good energy independence and does not need to be calibrated for energy. Therefore, the scintillating fiber dosimeter used in this paper has good tissue equivalence in the energy range of 70 MeV-230 MeV and can be used for absolute dose measurement in proton therapy.

[0026] Establish a digital model of the simulated human body: In general, CT images are easy to obtain and have rich anatomical features, making them the best choice for creating personalized body models. The body modeling process for human organs and tissues using medical image processing technology includes organ segmentation and three-dimensional reconstruction. Automatic organ segmentation is a key step in building patient models. By using multi-organ segmentation software (MOOSE) to automatically segment existing CT images. Import the captured CT images into the software, which can specifically segment the corresponding tissues or organs according to the selected model, and obtain the three-dimensional coordinates of the pixels of the corresponding organs or tissues. Use volume rendering method to regard the pixels as voxel units, and use these three-dimensional coordinates to establish a simulated human physical model. For example Figure 3 With Figure 4 Part of the human organs and bones constructed using the three-dimensional coordinates of the segmented organ or tissue voxels.

[0027] Simultaneously, the data was input into Monte Carlo simulation software for geometric modeling. Using tissue equivalent material parameters provided by ICRU Report No. 44 and NST, including density and elemental composition, representative human tissue equivalent materials were selected for radiation dosimetry calculations, including soft tissue, fat, and bone. Unlike traditional material allocation methods based on Hu values, this invention can directly use different tissue equivalent materials to fill corresponding organs or tissues.

[0028] This invention utilizes GPU-accelerated Monte Carlo simulation to calculate dose: The dose field is calculated using MOQUI, a GPU-accelerated proton dose calculation Monte Carlo code. The parallel computing power of the GPU significantly improves the speed of particle transport simulation, making it particularly suitable for complex, non-uniform media environments. Combining GPU-accelerated computation with a dose inversion method based on fiber optic dosimeter data, a three-dimensional dose distribution inversion of the phantom is achieved through efficient simulation of particle transport and dose deposition processes, encompassing data acquisition, physical modeling, and mathematical inversion methods.

[0029] To minimize the problems caused by an excessive number of optical fibers, this invention arranges a smaller number of optical fiber dosimeters at the isocenter of the geometric model, based on the beam direction and dose distribution. For example... Figure 5 The beam is shown passing through areas with large dose gradients, such as the muscle-pancreas interface and the pancreas-liver interface, resulting in high sensitivity to measurement location. Therefore, optical fibers are inserted near the interfaces of different tissues for dose verification. An optical fiber is also inserted near the target area to avoid locations with excessively high non-uniformity in the high dose gradient, thereby reducing measurement errors. In addition, an optical fiber is also placed at the center of the target area to ensure the representativeness of the measurement.

[0030] like Figure 6 As shown, this invention proposes a GPU-accelerated three-dimensional dose calculation method, which is used to evaluate the performance of fiber optic dosimeters in radiation therapy and to reconstruct three-dimensional dose in a phantom. The entire calculation process can be divided into three main parts: preprocessing, GPU computation and data processing, and result reconstruction and correction.

[0031] First, in the preprocessing stage, the CPU is responsible for configuring the beam parameters according to the set requirements and constructing a 3D model based on the voxel coordinate file segmented from MOOSE. During this process, corresponding tissue equivalent material data also needs to be filled in to ensure that the model can realistically reflect the characteristics of human tissue. This process provides the necessary input data for subsequent calculations and ensures the accuracy of the simulation.

[0032] After these preprocessing steps are completed, the computational task is handed over to the GPU. The GPU first copies the data from the CPU to its own memory and then performs calculations in MOQUI using a simplified physical model that includes processes such as electromagnetics, ionization, and elastic collisions between protons and oxygen.

[0033] The transport history of each incident proton can be considered an independent event, making it suitable for parallel processing by GPUs. By decomposing the particle history into multiple subtasks, GPUs can simultaneously track the paths of tens of thousands of protons, significantly reducing computation time. In simulating the interaction between a proton beam and human tissue, GPUs can process the following key steps in parallel: 1. Proton transport path simulation: GPU threads are used to perform parallel calculations of proton energy deposition, multiple Coulomb scattering and inelastic nuclear reactions, reducing unnecessary complex physical processes to lower the computational burden; 2. Secondary particle generation and tracking: Generate high-quality, high-speed pseudo-random numbers to simulate random processes such as particle free path, scattering angle, and interaction type, and process the random transport of secondary particles (such as neutrons and alpha particles) generated by nuclear reactions in real time. 3. Memory management optimization: MOQUI uses a hash table as the data structure for its integration system. In traditional data structures, each voxel needs to be allocated to an array for integration, i.e., an array structure. However, in a hash table, only voxels that interact with particles occupy memory in the data structure, and the data can be searched efficiently to determine the address. This allows more elements to be stored in the GPU, greatly improving memory utilization efficiency.

[0034] Finally, after data processing, the GPU downloads the calculation results back to the CPU to reconstruct the three-dimensional dose distribution map in the phantom. Based on this, the dose at the insertion point of the fiber optic dosimeter can be read in real time according to a pre-established linear dose calibration formula. This method not only allows for real-time dose detection during measurement but also enables correction of dose difference points after measurement, ensuring the accuracy of dose measurement. Finally, dose slice maps and dose distribution cloud maps at different depths are generated based on the three-dimensional dose data.

[0035] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0036] Finally, it should be noted that the above only describes the preferred embodiments of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art will appreciate that the technical solutions described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalent ones. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for reconstructing three-dimensional spatial radiation dose based on digital twin technology, characterized in that: Includes the following steps: Step S1: Establish a physical response model of the scintillation fiber optic dosimeter to obtain a linear scale relationship between optical signal intensity and radiation dose; Step S2: Based on CT image data, obtain the three-dimensional coordinates of human organs or tissues by segmenting them, perform geometric modeling, and construct a simulated human digital physical model using tissue equivalent materials. Step S3: Calculate the dose field using GPU-accelerated Monte Carlo simulation code to obtain three-dimensional spatial dose data of the simulated human digital model. Step S4: Insert the fiber optic dosimeter into the preset position of the simulated human digital physical model and measure the dose value at the position where the fiber optic dose gradient change is large in real time. Step S5: Based on GPU-computed data and fiber optic measured values, the measured dose at the difference points is fed back to the dose data calculated by the GPU in real time. Through interactive adjustment, the three-dimensional spatial radiation dose distribution of the model is retrieved.

2. The method for reconstructing three-dimensional spatial radiation dose based on digital twin technology according to claim 1, characterized in that: In step S1, a physical response model of the scintillation fiber optic dosimeter is established to obtain the linear scale relationship between optical signal intensity and radiation dose, specifically as follows: The radiation response characteristics of the scintillation fiber optic dosimeter in a non-uniform medium were verified using the Monte Carlo method, and a linear scaling relationship between optical signal intensity and radiation dose was established. Specifically, the linear scaling relationship is as follows: Where I is the number of scintillation photons collected by the detector, D is the real-time dose of the scintillation fiber, and k and b are fitting parameters.

3. The method for reconstructing three-dimensional spatial radiation dose based on digital twin technology according to claim 2, characterized in that: In step S1, the scintillation fiber optic dosimeter operates within an energy range of 70 MeV to 230 MeV.

4. The method for reconstructing three-dimensional spatial radiation dose based on digital twin technology according to claim 1, characterized in that: In step S2, the automatic segmentation of human organs or tissues based on CT image data and the acquisition of their three-dimensional coordinates are specifically as follows: Organ segmentation of medical CT images converts pixel units of phantom CT images into voxel units. Based on the three-dimensional coordinate data of voxel units, a three-dimensional geometric structure of a simulated human digital physical model is constructed through a volume rendering algorithm.

5. The method for reconstructing three-dimensional spatial radiation dose based on digital twin technology according to claim 4, characterized in that: In step S2, the tissue equivalent material includes the density and elements of soft tissue, fat, and bone, and is directly filled into the corresponding organ or tissue.

6. The method for reconstructing three-dimensional spatial radiation dose based on digital twin technology according to claim 1, characterized in that: In step S3, the preprocessing steps for GPU-accelerated computation include: configuring proton beam parameters by reading the treatment plan file, constructing a three-dimensional geometric model based on the voxel coordinate file output by the MOOSE software, and filling each voxel unit of the geometric model with the corresponding tissue equivalent material parameters.

7. A method for reconstructing three-dimensional spatial radiation dose based on digital twin technology according to claim 1 or 6, characterized in that: In step S3, the simulation process performed by the GPU computing unit includes: Step S301: Simulate the energy deposition, multiple Coulomb scattering, and inelastic nuclear reaction processes of protons using parallel computing threads; Step S302: Real-time tracking of the random transport paths of secondary particles generated by proton nuclear reactions; Step S303: A hash table data structure is used to manage the memory allocation of integral voxels, and memory space is allocated only for voxel units that interact with particles.

8. The method for reconstructing three-dimensional spatial radiation dose based on digital twin technology according to claim 1, characterized in that: In step S4, the fiber optic dosimeter is inserted into the predetermined position of the phantom, including but not limited to: the isocenter position of the proton beam, the dose gradient region at the interface of tissues such as muscle and pancreas, pancreas and liver, and the center position of the target treatment area and its adjacent high dose gradient but low non-uniformity region.

9. A method for reconstructing three-dimensional spatial radiation dose based on digital twin technology according to claim 1, characterized in that: In step S5, the interactive adjustment of dose difference points based on measured dose data specifically involves: by comparing the measured values ​​of the fiber optic dosimeter with the GPU calculated values, determining the correction factors for different tissues and energy ranges in the dose gradient region. Specifically, a negative deviation correction factor is used for the full energy range of adipose tissue; a positive deviation correction factor is used for the 74-100MeV energy range of bone tissue; and a negative deviation correction factor is used for the 150-230MeV energy range.

10. A three-dimensional spatial radiation dose reconstruction method based on digital twin technology according to claim 1, characterized in that: In step S5, the reconstruction results of the three-dimensional spatial radiation dose distribution are visualized in the following way: Generate dose slices at different depths; Based on the three-dimensional dose data calculated by the GPU, a dose distribution cloud map is generated within the phantom.

Images