A radiation therapy system and a method for optimizing the treatment planning module of the radiation therapy system.

Abstract

To provide a radiation irradiation system and a control method therefor.SOLUTION: A radiation irradiation system and a control method therefor are provided. On the basis of weight proportions of elements in a human body and reaction intensities of the elements with a neutron and a photon, the element that has influence on simulation calculation results of the neutron and the photon in an application scenario of the radiation irradiation system is screened out, and during a simulation process, only the screened-out element is simulated, so that the calculation speed can be greatly improved, and the calculation time can be reduced.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] One aspect of the present invention relates to a radiation irradiation system, and another aspect of the present invention relates to a method for controlling a radiation irradiation system.

Background Art

[0002] With the development of atomic science, radiation therapies such as cobalt 60, linear accelerators, and electron beams have already become one of the main means of cancer treatment. However, conventional photon or electron therapies kill tumor cells due to the limitations of the physical conditions of the radiation itself, and at the same time, damage many normal tissues in the beam path. In addition, because the degree of sensitivity of tumor cells to radiation is different, in conventional radiation therapy, the treatment effect on malignant tumors with high radioresistance (for example, glioblastoma multiforme, melanoma) is not high.

[0003] In order to reduce radiation damage to normal tissues around tumors, the concept of targeted therapy in chemotherapy has been applied to radiation therapy. In addition, for tumor cells with high radioresistance, currently, radiation sources with high relative biological effectiveness (RBE), such as proton beam therapy, heavy particle therapy, and neutron capture therapy, are actively being developed. Among these, neutron capture therapy combines the above two concepts. For example, in Boron Neutron Capture Therapy (BNCT), boron-containing drugs specifically accumulate in tumor cells, and by combining with highly precise beam control, compared with conventional radiation specifically accumulate in tumor cells, and by combining with highly precise beam control, compared with conventional radiation ​​​​​​​​​Provide better cancer treatment options.

[0004] Boron neutron capture therapy utilizes the property that boron ( B)-containing drugs have a large capture cross-section for thermal neutrons, 10 B(n,α) 7 Li and through neutron capture and nuclear fission reactions, 4 two kinds of charged particles, He and 7 Li, are generated. Since the total flight path of the two kinds of particles corresponds to about the size of one cell, radiation damage to the living body can be suppressed at the cell level. The boron-containing drug selectively accumulates in tumor cells, and by combining it with an appropriate neutron beam source, it is possible to achieve the goal of partially killing tumor cells without causing significant damage to normal tissues.

[0005] To kill as many cancer cells as possible and reduce damage to normal cells, usually, before treating a patient, CT or PET imaging diagnosis is performed, and based on the imaging diagnosis results, information on the tissue materials of the human body is obtained. Based on the material information and the radiation source, a calculation model is established, the transport process of the radiation particles in the human body is simulated, and finally the dose distribution of the radiation particles in the human body is obtained, and the optimal plan for the dose distribution for the patient is selected as the treatment scheme for the patient.

[0006] Currently, the dose calculation module in the radiation treatment planning system is mainly obtained by simulating radiation particles using the Monte Carlo method. Conventional radiation therapy requires simulating the motion processes of photons and electrons, while radiation irradiation therapy requires simulating the motion processes of neutrons and photons. Currently, the Monte Carlo method is the most accurate method for dose calculation, but the calculation is time-consuming and consumes a large amount of memory.

[0007] ​​ Currently, general-purpose Monte Carlo programs such as MCNP and Geant4 are used for calculating radiation therapy. Programs are often used, and MCNP was initially used in reactor design calculations, Ge Ant4 was initially used for high-energy physics calculations and was designed in the early stages of the physics of radiation therapy. Since n is not considered, no special optimization is performed on the calculation of radiation therapy. Planning usually needs to be completed within a set timeframe; for example, radiation therapy, The radiation therapy planning system is required to be able to create a treatment scheme within one hour, and the dosimeter The calculation process accounts for the majority of the time spent creating the radiation therapy plan, therefore, the time spent creating the radiation therapy plan The dose calculation method needs to be optimized to reduce the risk. [Overview of the Initiative]

[0008] In order to overcome the shortcomings of the prior art, the radiation irradiation system according to the first aspect of the present invention is A therapeutic beam is generated, and the therapeutic beam is irradiated onto the target object to form the irradiated area. The dose is determined based on the irradiation device, the parameters of the therapeutic beam, and medical image data of the irradiated area. We perform simulation calculations and create a treatment plan, determining the weight percentage of elements in the human body and the middle Considering the reaction intensity with tensors and photons, neutrality is achieved in the application scenario of the radiation irradiation system. We screen for elements that have an influence on the simulation results of particles and photons. In the simulation process, only the screened elements are simulated. A treatment planning module and a control module that controls the irradiation of the beam irradiation device based on the treatment plan. Includes 'yuuru'.

[0009] In other examples, the elements screened by the treatment planning module were H, He, Li, Be, B, C, N, O, F, Na, Mg, Al, Si, P, S, Cl, Ar It is one or more elements selected from K and Ca.

[0010] More specifically, the elements screened by the treatment planning module were H, He , Li, Be, B, C, N, O, F, Na, Mg, Al, Si, P, S, Cl, Ar, K And Ca.

[0011] In another embodiment, the treatment planning module uses a database of the corresponding neutron cross-sections of each element. Remove the databases corresponding to temperatures other than 294K and 0K from the list.

[0012] The radiation irradiation system is preferably a neutron capture therapy system and a beam irradiation device. This includes a neutron generator, a beam shaping body and a treatment table, and the neutron generator is an accelerator and a t The accelerator, including the target, accelerates charged particles to generate a charged particle beam and targets. It interacts with the neutron beam to generate a neutron beam, and the beam shaping body uses the neutrons generated by the neutron generator. The beam can be adjusted to a predetermined beam quality, and the neutron beam generated by the neutron generator is The radiation passes through the orthopedic frame and is then delivered to the target object on the treatment table.

[0013] In a control method for a radiation irradiation system according to a second aspect of the present invention, The radiation irradiation systemThe radiation irradiation system includes a beam irradiation device that generates a therapeutic beam and irradiates a target body with the therapeutic beam to form an irradiated area; a treatment planning module that performs dose simulation calculations based on the parameters of the therapeutic beam and medical image data of the irradiated area, and creates a treatment plan; and a control module that controls the irradiation of the beam irradiation device based on the treatment plan. The control method for the radiation irradiation system includes a step of screening elements that have an influence on the simulation calculation results of neutrons and photons in the application scenario of the radiation irradiation system, by comprehensively considering the weight percentage of elements in the human body and the reaction intensity with neutrons and photons, and simulating only the screened elements during the simulation process.

[0014] In other examples, the elements screened by the treatment planning module were H, He, Li, Be, B, C, N, O, F, Na, Mg, Al, Si, P, S, Cl, Ar It is one or more elements selected from K and Ca.

[0015] More specifically, the elements screened by the treatment planning module were H, He , Li, Be, B, C, N, O, F, Na, Mg, Al, Si, P, S, Cl, Ar, K And Ca.

[0016] In another embodiment, the control method involves the treatment planning module performing the corresponding neutron phase analysis of each element. Remove databases corresponding to temperatures other than 294K and 0K from the area database. Includes further steps.

[0017] The radiation irradiation system is preferably a neutron capture therapy system and a beam irradiation device. This includes a neutron generator, a beam shaping body and a treatment table, and the neutron generator is an accelerator and a t The accelerator, including the target, accelerates charged particles to generate a charged particle beam and targets. It interacts with the neutron beam to generate a neutron beam, and the beam shaping body uses the neutrons generated by the neutron generator. The beam can be adjusted to a predetermined beam quality, and the neutron beam generated by the neutron generator is The radiation passes through the orthopedic frame and is then delivered to the target object on the treatment table.

[0018] The radiation irradiation system and control method according to embodiments of the present invention are, the weight of an element in the human body The ratio and the reaction intensity with neutrons and photons are combined to determine the application scenario of the radiation irradiation system. In this context, elements that have an influence on the simulation calculation results for neutrons and photons are specified in the script. The system is then screened, and only the screened elements are simulated during the simulation process. By doing so, the calculation speed of the treatment planning module is significantly improved, reducing the calculation time. It can be reduced.

[0019] A radiation irradiation system according to a third aspect of the present invention generates a therapeutic beam, and the therapeutic A beam irradiation device that irradiates an object to be irradiated with a therapeutic beam to form an irradiated area, and the therapeutic beam Dose simulation based on the parameters of the system and the medical image data of the irradiated area. A treatment planning module that performs calculations and creates a treatment plan, wherein different source particles Assign the simulation task to a different process or thread, and each process or thread After completing the calculation tasks of the treatment planning module, the final calculation results are obtained together. A control module that controls the irradiation of the beam irradiation device based on the treatment plan. , including.

[0020] The radiation irradiation system is preferably a neutron capture therapy system, and the beam irradiation The apparatus includes a neutron generator, a beam shaping body, and a treatment table, the neutron generator being an accelerator The accelerator includes a device and a target, and accelerates charged particles to generate a charged particle beam, The beam shaping body interacts with the target to generate a neutron beam, and the neutron generation The neutron beam generated by the device can be adjusted to a predetermined beam quality, and the neutron generator The neutron beam generated by the process passes through the beam shaping body and reaches the irradiated object on the treatment table. It is irradiated.

[0021] In one embodiment, particles are processed by CPU process parallelism, thread parallelism, and GPU acceleration. The simulation showed that process parallelism and thread parallelism both utilize multi-core CPUs. Using this method to achieve parallel computing, the CPU process parallelism and thread parallelism computation process first, The system obtains the number of processes or threads and gets a value n, and then the system The first step is to uniformly divide the particles to be simulated into n groups, and then each thread Alternatively, one particle can be simulated independently in each process, and the counter The steps involve taking the data, and finally, the data obtained by the system in each process or thread. This includes the step of statistically analyzing the counts to obtain the final dose.

[0022] In one embodiment, GPU acceleration is achieved through parallel computing of the GPU's multi-core processor. The process of simulation calculation using GPU acceleration is as follows: First, the system generates random numbers, cross-sections The integrated data is transmitted from the CPU memory to the GPU graphics memory, and then, within the GPU... Each processor simulates, calculates, counts, and counts a single particle. The steps involve statistically counting the results into a global count, and then the system simulates Determine whether or not there are any particles that have not been simulated. If there are no particles that have not been simulated... The count results are transmitted from GPU memory to CPU memory, and the simulation is not performed. If there are missing particles, return to the previous step and continue until the simulation of all particles is complete. The step of continuing to simulate and count particles that have not been simulated. This includes,

[0023] In a control method for a radiation irradiation system according to a fourth aspect of the present invention, The radiation irradiation system The radiation irradiation system includes a beam irradiation device that generates a therapeutic beam and irradiates a target body with the therapeutic beam to form an irradiated area; a treatment planning module that performs dose simulation calculations based on the parameters of the therapeutic beam and medical image data of the irradiated area and creates a treatment plan; and a control module that controls the irradiation of the beam irradiation device based on the treatment plan. The control method for the radiation irradiation system includes assigning simulation tasks for different source particles to different processes or threads, and obtaining the final calculation results after each process or thread has completed its calculation tasks.

[0024] In one embodiment, particles are processed by CPU process parallelism, thread parallelism, and GPU acceleration. Let's run a simulation.

[0025] Preferably, both process parallelism and thread parallelism use a multi-core CPU. To enable parallel computing.

[0026] In one embodiment, the process-parallel and thread-parallel computation processes of the CPU are performed first by the system The system obtains the number of processes or threads and gets the value n, and then the system The first step is to uniformly divide the particles to be simulated into n groups, and then each thread... In each process, one particle is simulated independently, and the cow The steps to perform the test, and finally, the system obtained in each process or each thread This includes the step of statistically analyzing the counts to obtain the final dose.

[0027] Preferably, GPU acceleration is achieved by parallel computing of the GPU's multi-core processors. ru.

[0028] The process of simulation calculation using GPU acceleration involves the system first generating random numbers and cross-sectional data. Data is transmitted from CPU memory to GPU graphics memory, and then each program in the GPU... The sessa simulates, calculates, and counts a single particle, and the count results are then grouped. The steps involve statistically counting globally, and then the system is simulated. Determine whether or not there are any missing particles, and if there are no particles that have not been simulated, Cow The simulation results are transmitted from GPU memory to CPU memory, and unsimulated particles are... If there is a stain, return to the previous step and continue until the simulation of all particles is complete. The simulation continues to simulate unsimulated particles, and the counting step is... include.

[0029] The radiation irradiation system and control method according to embodiments of the present invention are for different source particle staining Assign the simulation task to a different process or thread, and each process or thread After completing the calculation tasks, the final calculation results are obtained all at once. This can significantly improve speed and reduce computation time.

[0030] A radiation irradiation system according to a fifth aspect of the present invention generates a therapeutic beam, and the therapeutic A beam irradiation device that irradiates an object to be irradiated with a therapeutic beam to form an irradiated area, and the therapeutic beam Based on the parameters of the device and the medical image data of the irradiated area, a dose simulation meter is used. A treatment planning module that performs calculations and creates a treatment plan, wherein the half-absorption thickness of the photon is 1. Treatment planning module that stops the simulation for photons if the value is below a predetermined value. A control module that controls the irradiation of the beam irradiation device based on the treatment plan, Includes.

[0031] The radiation irradiation system is preferably a neutron capture therapy system, and the beam irradiation The apparatus includes a neutron generator, a beam shaping body, and a treatment table, the neutron generator being an accelerator The accelerator includes a device and a target, and accelerates charged particles to generate a charged particle beam, The beam shaping body interacts with the target to generate a neutron beam, and the neutron generation The neutron beam generated by the device can be adjusted to a predetermined beam quality, and the neutron generator The neutron beam generated by the process passes through the beam shaping body and reaches the irradiated object on the treatment table. It is irradiated.

[0032] In one embodiment, the treatment planning module uses equation (1-1) to determine the half-absorption thickness t of the photon Calculate,

number

[0033] The first predetermined value is the size of a single cell, preferably 0.2m. When m is and the photon energy is less than or equal to the second predetermined value, the light corresponding to the photon energy The half-absorption thickness of the child is less than or equal to the first predetermined value, and the second predetermined value is 10 KeV.

[0034] In a control method for a radiation irradiation system according to a sixth aspect of the present invention, The radiation irradiation system The radiation irradiation system includes a beam irradiation device that generates a therapeutic beam and irradiates a target body with the therapeutic beam to form an irradiated area; a treatment planning module that performs dose simulation calculations based on the parameters of the therapeutic beam and medical image data of the irradiated area and creates a treatment plan; and a control module that controls the irradiation of the beam irradiation device based on the treatment plan. The control method for the radiation irradiation system includes the step of stopping the simulation for photons when the half-absorption thickness of the photons is less than or equal to a first predetermined value.

[0035] In one embodiment, the treatment planning module uses equation (1-1) to determine the half-absorption thickness t of the photon Calculate,

number

[0036] The first predetermined value is the size of a single cell.

[0037] Preferably, the first predetermined value is 0.2 mm.

[0038] If the photon energy is less than or equal to the second predetermined value, the half-absorption of the photon corresponding to the photon energy occurs. The yield thickness is less than or equal to the first predetermined value.

[0039] Preferably, the second predetermined value is 10 keV.

[0040] A radiation irradiation system and its control method according to an embodiment of the present invention, wherein the half-absorption thickness of the photons 1. If the value is less than or equal to a predetermined value, the simulation for the photon is stopped. Assuming accuracy is guaranteed, computation time can be reduced.

[0041] A radiation irradiation system according to a seventh aspect of the present invention generates a therapeutic beam, and A beam irradiation device that irradiates a target body with a beam to form the irradiated area, and parameters of the therapeutic beam. Based on medical image data of the irradiated area, dose simulation calculations are performed, and the treatment A treatment planning module that creates a treatment plan, simulating particles by reducing dispersion. A treatment planning module and a control module that controls the irradiation of the beam irradiation device based on the treatment plan. Includes joules.

[0042] Furthermore, variance reduction is implicit absorption, wait window games, and Russian roulette. The radiation irradiation system, including splitting, is a neutron capture therapy system. The beam irradiation device includes a neutron generator, a beam shaping body, and a treatment table, and the neutron generator This includes an accelerator and a target, the accelerator accelerating charged particles to generate a charged particle beam. Furthermore, it interacts with the target to generate neutron beams, and the beam shaping body is formed by a neutron generator. The generated neutron beam can be adjusted to a predetermined beam quality, and the neutrons generated by the neutron generator... The neutron beam passes through a beam shaping device and is irradiated onto the object to be irradiated on the treatment table.

[0043] In a control method for a radiation irradiation system according to the eighth aspect of the present invention, The radiation irradiation systemThe radiation irradiation system includes a beam irradiation device that generates a therapeutic beam and irradiates a target body with the therapeutic beam to form an irradiated area; a treatment planning module that performs dose simulation calculations based on the parameters of the therapeutic beam and medical image data of the irradiated area, and creates a treatment plan; and a control module that controls the irradiation of the beam irradiation device based on the treatment plan. The control method for the radiation irradiation system includes a step in which the treatment planning module simulates particles by dispersion reduction.

[0044] Furthermore, variance reduction is implicit absorption, wait window games, and Russian roulette. The steps include splitting and simulating particles by dispersion reduction. Step S1 involves acquiring one source particle and determining whether the particle will collide within the lattice cell. If it is determined that a collision will occur, steps S3 and S4 are executed in order, and if it is determined that no collision will occur, In this case, step S2 executes steps S5 and S6 in order, and steps S2 executes implicit absorption processing. S3 determines whether the weight is lower than the weight window, and if it is determined to be lower, Execute step S7, and if it is determined that it is not low, return to step S2, step S4, and Step S5 involves performing cyan roulette and splitting processes, and Russian roulette. The system determines whether or not to perform the process, and if it is determined to perform the Russian roulette process, then step S If step 7 is executed and it is determined that the Russian roulette process should not be performed, the process returns to step S2. Step S6, and a decision is made as to whether or not to destroy it. If it is decided to destroy it, then step S8. If it is determined not to execute and delete the item, the process returns to step S2 after executing step S9. Then, in step S7, it is determined whether or not the particle processing is complete, and if it is determined that it is complete, The flow is terminated, and if it is determined that it is not complete, it returns to step S1, step S8, and so on. It includes step S9 of dividing by the disappearance probability of

[0045] Furthermore, Russian roulette and splitting calculate and record the lattice space importance of each lattice, step S1, step S2 of obtaining particles, step S3 of performing weight window detection and operation on the particles, and the lattice space importance I before and after the particle crosses the lattice boundary And step S4 of calculating I, step S5 of comparing whether I is greater than I, if not, returning to step S3 after executing step S6, if so, executing step S7, step S6 of splitting the particles and reducing the weight of the particles, step S7 of judging whether to eliminate the particles, if it is judged to eliminate the particles, executing step S9, if it is judged not to eliminate the particles, returning to step S3 after executing step S8, step S8 of increasing the weight of the particles, step S9 of judging whether the simulation of the particles is completed, if it is judged that the simulation is not completed, returning to step S2, if it is judged that the simulation is completed, ending, and including

[0046] Furthermore, implicit absorption includes step S1 of obtaining particles, step S2 of judging whether the particles collide within the lattice cell, if it is judged that they collide, executing step S3, if it is judged that they do not collide, returning to step S1, step S3 of multiplying the weight by the scattering occurrence probability, step S4 of judging whether the weight of the particles is smaller than the lowest weight, if it is judged to be smaller, ​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​Execute step S5, and if it is determined that it is not small, return to step S2, step S4, and so on. The system determines whether or not to eliminate the particle, and if it determines that the particle should be eliminated, it performs step S7. If it is determined that the particles will not be eliminated, then step S6 is performed and then step S2 is performed. Go back, step S5, step S6 divide the weight by the annihilation probability, and then simulate the particle. It determines whether the process is complete or not, and if it determines that the simulation is not complete, step Returning to step S1, if it is determined that the simulation is complete, step S7 is executed to terminate the process. Includes.

[0047] Furthermore, the weight window game simulates particle motion in step S1 Then, it is determined whether the particle's weight is within the weight window range, and it is determined that it is within the range. If this occurs, return to step S1, and if it is determined that the result is not within the range, execute step S3. Step S2 determines whether the weight is larger than the weight window, and if it is determined to be larger If so, after executing step S4, return to step S1, and if it is determined that it is not large, Step S3 executes step S5, and splits the particles to reduce their weight. Step S4 is followed by a decision on whether or not to destroy it. If it is decided to destroy it, the process ends and it is destroyed. If it is decided not to allow it, the process returns to step S1 after executing step S6. The process includes step 5 and step S6, which increases the weight of the particles.

[0048] In one embodiment, the lattice space importance is obtained by solving the following adjoint transport equation. It was acquired,

number

[0049] More specifically, the particle has a lattice space importance of I n Import from the grid into the grid space The chest of drawers is n+1 It moves in the grid, I n+1 >I n If so, m=I n+1 / I n west, Split the particle into m particles, and reduce the weight of each particle to 1 / m of its original weight. I n+1 n Therefore, we can use the Russian roulette method on the particle and calculate P=I n+1 / I n Then, a random number x is sampled from 0 to 1, and if x is less than P, a particle is produced. If any particles remain, multiply their weight by 1 / P; otherwise, the particles are annihilated, and the particle simulation is performed. End the program.

[0050] In other examples, the particle weight has an upper limit of 10 and a lower limit of 0.25. If w is greater than 10, the integer part of w becomes w1, the decimal part becomes w2, and from 0 to 1... A random number x is sampled, and if x is less than w2, w = w1 + 1, and x becomes w2 If it is larger, set w=w1, then split one particle into w particles. The simulation is performed until the weight of each particle decreases to 1.

[0051] ​A radiation irradiation system and a control method thereof according to an embodiment of the present invention reduce particles by dispersion By performing simulations, we can reduce computation time while guaranteeing computational accuracy. It is possible.

[0052] A radiation irradiation system according to the ninth aspect of the present invention generates a therapeutic beam, and A beam irradiation device that irradiates a target body with a beam to form the irradiated area, and parameters of the therapeutic beam Based on the medical image data of the data and the irradiated area, dose simulation calculations are performed, and A treatment planning module for creating treatment plans, which uses a non-uniform grid for simulation A treatment planning module that performs calculations and a control module that controls the irradiation of the beam irradiation device based on the treatment plan. It includes a control module.

[0053] Furthermore, the treatment planning module uses the dimensions of the grid in critical areas to determine the dimensions of the grid in non-critical areas. Set it to a small value.

[0054] Furthermore, the treatment planning module reduces the grid size in the tumor region to 0.4 mm or less. Set it.

[0055] Furthermore, the treatment planning module sets the grid dimensions of the regions where blood, air, and bone are located to 1.6 Set to mm or larger.

[0056] Furthermore, the treatment planning module sets the grid dimensions of the area where normal muscle is located to 0.8 mm. Set it to be larger than 1.6mm and smaller than 1.6mm.

[0057] The radiation irradiation system is preferably a neutron capture therapy system and a beam irradiation device. This includes a neutron generator, a beam shaping body and a treatment table, and the neutron generator is an accelerator and a t The accelerator, including the target, accelerates charged particles to generate a charged particle beam and targets. It interacts with the neutron beam to generate a neutron beam, and the beam shaping body uses the neutrons generated by the neutron generator. The beam can be adjusted to a predetermined beam quality, and the neutron beam generated by the neutron generator is The radiation passes through the orthopedic frame and is then delivered to the target object on the treatment table.

[0058] In a control method for a radiation irradiation system according to a 10th aspect of the present invention, The radiation irradiation system The radiation irradiation system includes a beam irradiation device that generates a therapeutic beam and irradiates a target body with the therapeutic beam to form an irradiated area; a treatment planning module that performs dose simulation calculations based on the parameters of the therapeutic beam and medical image data of the irradiated area, and creates a treatment plan; and a control module that controls the irradiation of the beam irradiation device based on the treatment plan. The control method for the radiation irradiation system includes the step of the treatment planning module performing simulation calculations using a non-uniform grid.

[0059] The control method involves the treatment planning module controlling the dimensions of the grid in critical areas and the dimensions of the grid in non-critical areas. This includes further steps to set it to an even smaller size.

[0060] The control method involves the treatment planning module setting the grid dimensions of the region where the tumor is located to 0.4 mm or less. This includes further steps to be set below.

[0061] The control method involves the treatment planning module determining the dimensions of the grid in the regions where blood, air, and bone are located. Set to 1.6 mm or larger, and set the grid dimensions of the area where normal muscle is located to be larger than 0.8 mm. This further includes steps to set the value to less than 1.6 mm.

[0062] A radiation irradiation system and a control method according to an embodiment of the present invention use a non-uniform grid By performing simulation calculations, under the assumption that the computation time will not increase significantly, important areas This improves the computational accuracy of certain regions, satisfies the computational accuracy requirements for non-critical regions, and reduces computation time. It can be made to happen. [Brief explanation of the drawing]

[0063] [Figure 1] This is a schematic diagram of a module of a boron neutron capture therapy system according to an embodiment of the present invention. [Figure 2] This is a schematic diagram of a boron neutron capture therapy system according to an embodiment of the present invention. [Figure 3] This is a flowchart of process-parallel and thread-parallel computing in a CPU according to an embodiment of the present invention. [Figure 4] This is a flowchart of the simulation calculation using GPU acceleration in an embodiment of the present invention. [Figure 5] This is a flowchart for simulating particles using the dispersion reduction method in an embodiment of the present invention. [Figure 6] This is a flowchart of Russian roulette and splitting in an embodiment of the present invention. [Figure 7] This is a flowchart of implicit absorption in an embodiment of the present invention. [Figure 8] This is a flowchart of a wait window game in an embodiment of the present invention. [Modes for carrying out the invention]

[0064] The embodiments of the present invention will be described in more detail below with reference to the drawings, as demonstrated by those skilled in the art. If so, it can be carried out by referring to the text in the specification.

[0065] Neutron capture therapy has seen increased application in recent years as an effective means of treating cancer. Of these, boron neutron capture therapy became the most common. The neutrons used can be supplied by a nuclear reactor or an accelerator. Boron neutrons Capture therapy (Boron Neutron Capture Therapy, BNCT) Boron ( 10 B) The characteristic that the contained drug has a large trapping cross-section for thermal neutrons. Using 10 B(n,α) 7 Through Li neutron capture and nuclear fission reactions, 4 He and 7 Li and It generates two types of heavily charged particles, and these two types of heavily charged particles have an average energy of approximately 2.33 Me. V is high linear energy transfer (L It has the characteristics of being ET and having a short range, and the linear energy and range of the alpha particle are 150, respectively. KeV / μm, 8μm. 7 For Li heavy-charged particles, the voltages are 175 KeV / μm. Since the particle size is 5 μm and the combined range of the two types of particles is approximately the size of one cell, it is not effective against living organisms. Radiation damage is suppressed to the cellular level. Boron-containing drugs selectively accumulate in tumor cells, and By combining it with a precise neutron beam source, it is possible to target tumor cells without causing significant damage to normal tissue. This allows us to achieve the objective of precisely killing the cells.

[0066] In boron neutron capture therapy, the neutron source is involved in the nuclear reaction between the reactor or charged particles and the target. Regardless of the method, what is generated is a mixed radiation field, that is, the beam is low energy Including neutrons and photons ranging from low to high energies, boron neutron capture therapy for deep tumors. Regarding this, excluding epithermal neutrons, the higher the content of other radiation, the more likely it is to cause abnormalities in normal tissue. This also increases the proportion of selective dose deposition, thus causing these unnecessary dose depositions. It is necessary to reduce radiation exposure as much as possible. In addition to the quality factors of the air beam, neutron radiation is also a factor. To further understand dose distribution in the human body, embodiments of the present invention relate to human head tissue. The dose distribution is calculated using the artificial organ, and the quality elements of the beam in the artificial organ are considered. This will be used as a reference for tron ​​beam design.

[0067] The International Atomic Energy Agency (IAEA) has issued a warning regarding neutron sources used in clinical boron neutron capture therapy. They have issued five proposals regarding the quality elements of air beams, and these five proposals are based on different neutron sources. It can be used to compare the advantages and disadvantages of the neutron generation pathway and beam shaping. These can be used as reference when designing. The five proposals are as follows:

[0068] Epithermal neutron flux>1x10 9 n / cm 2 s Fast neutron contamination < 2 × 10⁻¹⁰ - 13 Gy-cm 2 / n Photon contamination < 2 × 10⁻¹⁰ -13 Gy-cm 2 / n The ratio of thermal neutrons to epithermal neutron flux. tron flux ratio < 0.05 The ratio of neutron current to flux. to flux ratio > 0.7 Note: The epithermal neutron energy range is 0.5 eV to 40 keV, and thermal neutron energy The region is less than 0.5 eV, and the fast neutron energy region is greater than 40 keV. .

[0069] As shown in Figure 1, the radiation irradiation system in this embodiment is a boron neutron capture therapy system. The stem 100 includes a neutron beam irradiation device 10, a treatment planning module 20, and a control module. Includes 30 ure. The neutron beam irradiation device 10 generates a therapeutic neutron beam N, and A therapeutic neutron beam N is irradiated onto the target object 200 to form the irradiated area. Treatment planning module The parameter 20 controls the therapeutic neutron beam N generated by the neutron beam irradiation device 10. Based on medical image data of the irradiated area, dose simulation calculations are performed, and the treatment A treatment plan is created, and based on the treatment plan, the neutron beam irradiation device 10 during irradiation therapy is subjected to The location of the irradiation site and the corresponding irradiation time are determined based on the treatment plan. After positioning the irradiated body 200, treatment can be started, and the control module 30 will The treatment plan corresponding to the current irradiated body 200 is retrieved from the treatment planning module 20, and The control module 30 controls the irradiation of the neutron beam irradiation device 10 based on the treatment plan. Other data information, for example, data from the neutron beam irradiation device 10, data from the irradiated object 200 You may receive messages such as "Ta".

[0070] As shown in Figure 2, in this embodiment, the neutron beam irradiation device 10 is a neutron generator The neutron generator 11 includes a beam shaping body 12, a collimator 13, and a treatment table 14. The system includes an accelerator 111 and a target T, wherein the accelerator 111 emits charged particles (e.g., protons, dendritic particles). Accelerating neutherium nuclei, etc., generates a charged particle beam P, such as a proton beam, and then the charged particle The beam P is irradiated onto the target T and interacts with the target T to form a neutron beam. N is generated, and the target T is preferably a metal target. Required neutron yield and energy, the energy and current of the accelerating charged particles that can be supplied, and the metal target Based on their physical and chemical properties, an appropriate nuclear reaction is selected and generally considered. Nuclear reactions are, 7 Li(p,n) 7 Be and 9 Be(p,n) 9 B and these 2 Both types of reactions are endothermic reactions. The two types of nuclear reactions have different energy thresholds. With voltages of 1.881 MeV and 2.055 MeV, these are ideal neutron sources for boron neutron capture therapy. Since it is an epithermal neutron at the keV energy level, theoretically, when the energy exceeds the threshold... By bombarding a metallic lithium target with slightly higher-energy protons, relatively low-energy protons can be generated. It can generate neutrons that can be used clinically without requiring much deceleration. It is possible to use two types of targets: metallic lithium (Li) and metallic beryllium (Be). This does not generate a sufficiently large neutron flux because the cross-section that interacts with protons at threshold energy is not high. To do this, protons with relatively high energy are generally selected to trigger nuclear reactions. The ideal target has a high neutron yield and the energy distribution of the generated neutrons is in the extrathermal zone. Radiation that is close to the tron ​​energy region (explained in detail below) and has strong penetrating properties is abundant. It does not occur frequently, is safe, inexpensive, easy to operate, and has properties such as high temperature resistance, but in practice In such cases, it is impossible to find a nuclear reaction that satisfies all requirements, and the embodiments of the present invention A target made of lithium metal is used. As is well known to those skilled in the art, The T may be manufactured from a metallic material other than lithium or beryllium, for example, tantalum. It is formed of (Ta) or tungsten (W), and the target T may be disc-shaped. It may also be in other solid forms, or a liquid material (liquid metal) may be used. Accelerator 11 1 is a linear accelerator, cyclotron, synchrotron, synchrocyclotron. In addition, the neutron generator 11 does not use an accelerator and a target, and may even be a nuclear reactor. i. The neutron source for boron neutron capture therapy is charged particles from a nuclear reactor or accelerator and the target and Regardless of whether it is due to nuclear reactions, what is actually produced is all a mixed radiation field, immediately The beam contains neutrons and photons ranging from low to high energies. Regarding boron neutron capture therapy, excluding epithermal neutrons, the higher the content of other types of radiation, the greater the risk. The proportion of non-selective dose deposition in normal tissue also increases, thus eliminating these unnecessary risks. It is necessary to reduce the radiation that causes the required dose as much as possible. Also, the normal assembly of the irradiated object Regarding the fabric, it is important to avoid excessive amounts of various types of radiation, which can similarly cause unnecessary dose deposition. It should be done.

[0071] The neutron beam N generated by the neutron generator 11 is then directed to the beam shaping body 12 and collimated. The beam passes through the 13 and is irradiated onto the 200 to be irradiated on the treatment table 14. The beam shaping body 12 is The beam quality of the neutron beam N generated by the neutron generator 11 can be adjusted, Meter 13 collects the neutron beam N and gives the neutron beam N high targeting ability during treatment. To make it clear, the present invention provides a beam without a collimator, where the beam is directed to the beam shaping body 12. After exiting the treatment area, the irradiated object 200 on the treatment table 14 may be directly irradiated.

[0072] The beam shaping body 12 further includes a reflector 121, a decelerator 122, a thermal neutron absorber 123, a radiation shield 124, and a beam outlet 125. Because the neutrons generated by the neutron generator 11 have a broad energy spectrum, and it is necessary to minimize the content of other types of neutrons and photons, other than epithermal neutrons that meet therapeutic needs, in order to avoid causing damage to the operator or the irradiated object, Neutron generator 11 The neutrons emitted from it Reducer 122 Through this process, the energy of fast neutrons (>40 keV) is adjusted to the epithermal neutron energy range (0.5 eV to 40 keV), and it is necessary to reduce epithermal neutrons (<0.5 eV) as much as possible. Reducer 122 The moderator is manufactured from a material with a large cross-section for interacting with fast neutrons and a small cross-section for interacting with epithermal neutrons. In a preferred embodiment, the moderator 122 is made of D2O, AlF3, Fluental,The reflector 121 is made of at least one of CaF2, Li2CO3, MgF2, and Al2O3, and surrounds the decelerator 122, and reflects neutrons that have passed through the decelerator 122 and diffused into the surroundings back to the neutron beam N to improve neutron utilization, and is made of a material with high neutron reflectivity, in a preferred embodiment, the reflector 121 is made of at least one of Pb or Ni, the decelerator 122 has a thermal neutron absorber 123 at its rear, and is made of a material with a large cross-section that interacts with thermal neutrons, in a preferred embodiment, the thermal neutron absorber 123 is made of Li-6, and passes through the decelerator 122 To absorb thermal neutrons and reduce the thermal neutron content in the neutron beam N, thereby avoiding excessive doses to superficial normal tissue during treatment, the thermal neutron absorber may be integrated with a speed reducer, and the speed reducer material may contain Li-6. The radiation shield 124 shields neutrons and photons leaking from parts other than the beam outlet 125, and the material of the radiation shield 124 includes at least one of a photon shielding material and a neutron shielding material. In a preferred embodiment, the material of the radiation shield 124 includes lead (Pb) as the photon shielding material and polyethylene (PE) as the neutron shielding material. The collimator 13 is installed behind the beam outlet 125, and the epithermal neutron beam exiting the collimator 13 irradiates the target body 200, passes through superficial normal tissue, and is then slowed down to thermal neutrons before reaching the tumor cells M. Beam shaping body 12 The structure may be anything else, as long as it obtains the epithermal neutron beam necessary for treatment. For the sake of clarity, if a collimator 13 is installed, the outlet of the collimator 13 can be considered as the beam outlet 125, which will be described later. In this embodiment, a radiation shielding device 15 is further installed between the irradiated object 200 and the beam outlet 125 to shield the radiation from the beam emitted from the beam outlet 125 to the normal tissue of the irradiated object. To make it easier to understand, the radiation shielding device 15 may not be installed.

[0073] The irradiated body 200 is subjected to the administration or injection of a boron (B-10)-containing drug, followed by the application of boron. The boron-containing drug selectively accumulates in tumor cells M, and then the boron (B-10)-containing drug is attacked by thermal neutrons. By utilizing the characteristic of having a high capture cross-sectional area, 10 B(n,α) 7 Li neutron Through capture and nuclear fission reactions, 4 He and 7 It produces two types of heavy charged particles, Li. The charged particles have an average energy of approximately 2.33 MeV and offer high linear energy transfer (Li It has the characteristics of near energy transfer (LET) and short flight range. The linear energy transfer and range of the alpha particle are 150 keV / μm and 8 μm, respectively. 7 L The linear energy transfer and range of the heavily charged particles are 175 keV / μm and 5 μm, respectively, and there are two types. Because the total range of the particles is equivalent to the size of a single cell, radiation damage to living organisms occurs at the cellular level. The goal is to partially kill tumor cells, assuming that the damage is contained and does not cause significant harm to normal tissue. The target can be achieved.

[0074] The boron neutron capture therapy system 100 is entirely housed within a concrete structure. Specifically, the boron neutron capture therapy system 100 includes an irradiation chamber 101 and charged particles The beam generation chamber 102 is further included, and the irradiated body 200 on the treatment table 14 is in the irradiation chamber 101. The patient receives treatment by irradiation with a neutron beam N, and the charged particle beam generation chamber 102 is an accelerator 1 The beam shaping body 12, which at least partially accommodates 11, connects the irradiation chamber 101 to the charged particle beam It is at least partially contained within the partition wall 103 separating it from the generation chamber 102. The partition wall 103 completely separates the irradiation chamber 101 from the charged particle beam generation chamber 102. It may also be a partial separation between the irradiation chamber 101 and the charged particle beam generation chamber 102. It may be present, and the irradiation chamber 101 and the charged particle beam generation chamber 102 are in communication. Target T There may be one or more of them, and the charged particle beam P selectively targets one or more of them. It acts on one or more target T, either on that target T or on multiple target T simultaneously. A therapeutic neutron beam N is generated. Depending on the number of targets T, the beam shaping body 12, The collimator 13 and the treatment table 14 may be one or more, and the multiple treatment tables may be the same It may be installed in the irradiation room, or an individual irradiation room may be installed at each treatment table. Irradiation Room 1 01 and the charged particle beam generation chamber 102 are enclosed by a concrete wall W (including a partition wall 103). It is a confined space, and the concrete structure is the operating over of the Boron Neutron Capture Therapy System 100. It can shield neutrons and other radiation that leak out over time.

[0075] In order to kill cancer cells to the greatest extent possible while reducing radiation damage to normal tissue, In setting up the image module 20, the accuracy of the dose distribution of epithermal neutrons and photons is extremely important. Yes. In application scenarios of radiation irradiation systems, dose calculations are performed using photon and neutron cross-section data. It is loaded into the database, and because the neutron cross-section database is very large, radiation irradiation The installation of the software that creates the system's treatment plan occupies a large amount of memory space. Furthermore, the memory occupied when performing dose calculation simulations using the said software. The space required is large, and the computation time is long. The following embodiment of the present invention is a series of treatment planning modules This provides a method for optimizing the code, reducing the amount of memory used and decreasing the dose calculation time. This fulfills the requirement to quickly create a radiation therapy planning scheme. (See diagram below) Each optimization method will be explained. [Examples]

[0076] The database is optimized in the treatment planning module 20.

[0077] The human body contains a total of over 60 elements, of which over 20 are essential. It is essential and plays a crucial role in maintaining the normal physiological functions of the human body. Elements with high abundance include carbon, hydrogen, oxygen, nitrogen, phosphorus, chlorine, sodium, and magnesium. Potassium and calcium, along with carbon, hydrogen, oxygen, and nitrogen, constitute the organic matter in the human body. It is the primary element, making up 96% of the total body weight, and the remaining 0.01% is greater than the majority of the body. The elements are calcium, potassium, phosphorus, sulfur, chlorine, magnesium, and sodium. These are the components of the human body, and their respective percentages of total body weight are 1.5%, 0.35%, and 1%, respectively. The concentrations are 0.25%, 0.15%, 0.05%, and 0.15%. Therefore, carbon, hydrogen, acid Contains elemental, nitrogen, calcium, potassium, phosphorus, sulfur, chlorine, magnesium, and sodium. The amount already reaches 99.45% of the total body weight, and Monte Carlo has measured over 100 elements. The simulation and calculation system is stored, and each time, the simulation is performed using Monte Carlo. When performing simulations and calculations, since each element is simulated and calculated, The time required is long. Also, the calculation and simulation system that handles many elements is stored in memory. Because of this, the Monte Carlo database is very large. However, radiation exposure In the application scenarios of the system, Monte Carlo is used to analyze particles such as photons and neutrons within the human body. In contrast, when performing simulation calculations, the percentage of the total weight of the human body is large (0.01 Accurate calculation results can be obtained simply by performing simulation calculations for elements in the human body (greater than %). It is possible to obtain this, and simulation calculations for some trace elements take a long time. The calculation results are used to simulate elements that make up more than one ten-thousandth of the total weight of the human body. The difference is within 0.001% compared to the result obtained by performing a calculation.

[0078] Furthermore, certain elements, such as Fe, occupy a large weight proportion in the human body, but neutrons and photons Because it does not react with neutrons and photons, or the reaction intensity with neutrons and photons is not high, such elemental stains The influence of simulation calculation results on the overall calculation results may be ignored. For example, for a certain element, For example, although boron occupies a small weight percentage in the human body, it is used in application scenarios for radiation irradiation systems. In this case, a large amount of boron-containing drug was injected into the human body, and the reaction with neutrons was intense, and the final calculation result It has a significant impact on the results.

[0079] The weight percentage of an element in the human body and its reaction intensity with neutrons and photons are combined to determine the radiation exposure. Shadow on the neutron and photon simulation calculation results in the system's application scenarios. Elements with resonance (H, He, Li, Be, B, C, N, O, F, Na, Mg, Al, S) Screening (one or more elements selected from i, P, S, Cl, Ar, K, and Ca) The system then performs a simulation, and during the simulation process, only the screened elements are simulated. By doing so, the simulation calculation time is reduced while guaranteeing calculation accuracy. This can improve the execution efficiency of Monte Carlo, and also screen these The database capacity corresponding to the elements is only 2% to 3% of the original database capacity. The memory requirements for magnetic disks will decrease significantly, leading to lower manufacturing costs.

[0080] On the other hand, in the neutron cross-section database corresponding to each element, Monte Carlo system For selection, there are many temperatures, for example, 0K, 1200K, 2500K, 250K, Cross-sectional area data corresponding to 294K, 600K, and 900K is stored, but radiation irradiation In the system's application scenario, simulation calculations are performed only on the human body model. The normal body temperature is 310K to 315K. Based on the Monte Carlo principle... And, after the temperature is entered, the system will use the many temperatures that have been stored in memory. The system automatically matches the temperature closest to the target temperature and uses that temperature as a simulation parameter. To perform simulation calculations, after the body temperature is input, the system will then... The temperature of 294K stored in the database is matched, and 294K is used as a parameter. Then the simulation calculation is performed. Also, the temperature used when calculating the Doppler effect is , 0K. In other words, in the application scenario of the radiation irradiation system, the database is stored Of the temperatures listed, only 294K and 0K are used; the other temperatures are for radiation irradiation. Because it is superfluous for the application scenario of the radiation system, it is not suitable for temperatures other than 294K and 0K. Assuming that the corresponding database is removed and the simulation calculation structure is not affected, The database size is reduced to about one-third of the original size, lowering the database's execution costs. .

[0081] Table 1 shows the results for the Monte Carlo system at temperatures of 294K, 310K, and 330K, respectively. This is the result of neutron dose simulation calculations performed on the same human body model.

[0082] [Table 1]

[0083] As can be seen from the data in Table 1, the final calculation of neutron dose due to small temperature fluctuations is as follows. The impact on the results can be ignored, and in application scenarios of radiation irradiation systems, the system Simply by holding the database corresponding to the stored 294K, the usage request can be fulfilled. This is possible, and the error in the final calculation result is within an acceptable range. [Examples]

[0084] During the particle simulation process, interrupt the process and interrupt the particle simulation time. Reduce.

[0085] In application scenarios of radiation irradiation systems, dose calculations include neutron dose calculations and photon dose calculations. This includes calculations. In the process of motion of a photon within the body of the irradiated object, the photon energy and half of the photon energy are involved. The absorption thickness gradually decreases, but the absorption cross-section of the photon is related to the photon energy and the half-absorption thickness of the photon. It increases rapidly as the decrease in . When the half-absorption thickness of a photon is smaller than the size of a single cell. The probability that a photon will be absorbed into the cell in which it is located is very high, and in this case, the photon It directly assumes that all energy is deposited within the cell where it is located, and the simulation is Do not continue the simulation, but continue with the dose distribution and photon simulation calculated after this processing. The error between the calculated dose distribution and the actual distribution is within 0.1%.

[0086] Human body cells are 2-200 μm in size, and typically, when applying a radiation irradiation system... Therefore, if the minimum dimension of the model's lattice is 0.8 mm, the half-absorption thickness of the photon is the minimum dimension of the lattice. If it is less than one-quarter of that value, the photon simulation can be stopped.

[0087] Specifically, the half-absorption thickness t of the photon is calculated using equation (1-1),

number

[0088] The following calculations will use the human skeleton as an example to determine the half-absorption thickness in skeletons with different photon energy values. ru.

[0089] [Table 2]

[0090] As can be seen from Table 2, in the framework, when the photon energy is less than 10 keV, The half-absorption thickness of the photon is less than 0.2 mm, meaning the photon energy is less than 10 keV. In that case, there is a very high probability that the photon will be absorbed within the lattice where it is currently located. In this case, the dose obtained by stopping the photon simulation and the photon simulation further... The difference between the obtained dose and the measured dose is within an acceptable range.

[0091] The half-absorption thickness of a photon in cellular tissue is determined by the material through which the photon passes and its energy. Therefore, it is possible to set cutoff energies corresponding to various materials, and photon energy If the energy is below this cutoff energy, the photon simulation will stop. This can be done by performing simulation calculations on a human body model. Obtain photon doses corresponding to different photon cutoff energies.

[0092] [Table 3]

[0093] Table 3 As can be seen, when the photon energy is 10 keV or less, the error between the photon dose calculated after stopping the photon simulation calculation and the dose distribution obtained by continuing the photon simulation calculation is within 0.1%.

[0094] From the above, in the process of the photon simulation calculation, the photon energy is determined to be a certain cutoff. If it is less than the tooff energy, the half-absorption thickness of the photon is smaller than the size of a single cell. In this case, stop the photon simulation if it does not affect the calculated dose distribution. This can reduce the calculation time for photon dose.

[0095] In other embodiments, a first predetermined value and 2. Set a predetermined value, and if the half-absorption thickness of the photon is less than or equal to the first predetermined value, or if the photon E If the energy is less than or equal to the second predetermined value, the photon simulation is stopped, and the first predetermined The value may be greater than the size of a single cell, or it may be less than the size of a single cell. The second predetermined value may be 12KeV, 16KeV, or the like. [Examples]

[0096] Use parallel processing technology to accelerate the dose calculation module.

[0097] The treatment planning module of the radiation irradiation system uses the Monte Carlo method to select different source particles. The simulation process is completely independent, meaning that the simulation of different particles is... The order of particles does not affect the calculation results, and based on these calculation characteristics, the sequence of different source particles Assign the simulation task to a different process or thread, and each process or thread After completing the calculation tasks, the final calculation results are obtained all at once. It provides two types of parallelism: CPU process parallelism and thread parallelism, as well as GPU acceleration. It is possible.

[0098] As shown in Figure 3, the so-called CPU process parallelism and thread parallelism are... First, the system obtains the number of processes or threads and obtains the value n, and then The system uniformly divides the particles to be simulated into n groups, and each Simulate one particle independently in each thread or process, The first step is to count, and finally, the system retrieves the result in each process or thread. The steps include statistically analyzing the counts obtained to obtain the final dose. Assign it to a process or thread and simulate and calculate it. The simulation and calculation time can be reduced by simulating all particles using a single process or thread. Reduce the time required to perform the calculation to 1 / n (process) or 1 / 2n. .

[0099] Both process parallelism and thread parallelism utilize multi-core CPUs to perform parallel computations. In other words, GPU acceleration is achieved through parallel computing of the GPU's multi-core processors. The effects of seth and thread parallelism are limited by the CPU's resource count, and in general standalone applications... In the case of the 1, there are 4 cores, and in this way, process parallelism and thread parallelism are speed It can improve performance by up to 4x (processes) or 8x (threads), with one GPU handling multiple processes. The processor is integrated, and logically, the computing speed can be improved by several times.

[0100] As shown in FIG. 4, in the process of simulation calculation by so-called GPU acceleration, first, the system transmits random numbers, cross-sectional area data, etc. from the CPU memory to the GPU graphic memory and then each processor in the GPU simulates, calculates, and counts a single particle, and steps of statistically counting the count results into the global count, and then the system judges whether there are particles that have not been simulated, and if there are no particles that have not been simulated, transmits the count results from the GPU memory to the CPU memory, and if there are particles that have not been simulated, returns to the previous step, and continues to simulate and count the particles that have not been simulated until the simulation of all particles is completed, including steps of simulating and counting.

[0101] Because the GPU has many processors, the simulation calculation by GPU acceleration can significantly improve the computing speed and reduce the calculation time. Also, the CPU uses an Intel Xeon processor with 2.27GHz priced at about 6000 yuan while the GPU uses an NVIDIA Tesla C2050 priced at about 9000 yuan. Each GPU has a total of 448 processors, and these 448 processors are equivalent to 448 CPUs. The calculation time of the CPU is 50 to 70 times that of the GPU calculation time. Therefore, compared with the CPU, performing simulation calculation using the GPU has great advantages in terms of price and time taken.

Example

[0102] We use the Monte Carlo method of decreasing variance to accelerate and converge the process.

[0103] The Monte Carlo method has rich variance reduction techniques to improve the computation speed of programs. Variance reduction methods usable in treatment planning systems include implicit absorption, weight window games, This includes Russian roulette and splitting, and Russian roulette and splitting The ting consists of the Russian roulette method and the splitting method. Russian roulette method of dispersion reduction, splitting method, lattice space importance, Interpret implicit absorption and the wait window game.

[0104] Regarding the Russian roulette method, the weight of the particles is usually set with an upper limit of 10 and a lower limit This is set to 0.25. The particle weight w is smaller than a certain predetermined value (for example, 0.25). If it decreases to a certain value, sample a random number x from 0 to 1, and if x is less than w, the grain If the child survives, the particle's weight returns to 1, and x is greater than or equal to w, the particle is annihilated, and the particle's End simulation.

[0105] Regarding the splitting method, if the particle weight w is greater than a certain value (for example, 10) Combine the integer part of w and the decimal part and set it to w1, then sample a random number x from 0 to 1. If x is less than w², then w = w1 + 1, and if x is greater than w², then w = w First, set the weight to 1, then split one particle into w particles, and reduce the weight of each particle to 1. The simulation will continue until the result is reached.

[0106] Regarding lattice space importance, the dose contribution from particles in different regions of the model is There are differences, and spatial importance represents the contribution of dose by particles, and the simulation Before proceeding, the grid space input of each grid should be determined based on parameters such as tumor location and model characteristics. Calculate the portance. For a particle, the lattice space importance is I n From the grid to the grid space Portance is I n+1 Move to the grid, I n+1 >I n If so, m=I n+1 / I n to The particle is split into m particles, and the weight of each particle is reduced to the original 1 / m. I n+1 n Therefore, we can use the Russian roulette method on the particle and calculate P=I n+1 / I n Then, a random number x is sampled from 0 to 1, and if x is less than P, a particle is produced. If any particles remain, multiply their weight by 1 / P; otherwise, the particles are annihilated, and the particle simulation is performed. End the program.

[0107] Regarding the calculation method for lattice space importance, lattice space importance is calculated by calculating the adjoint bundle. The adjoint bundle is obtained by solving the adjoint transport equation below. ,

number

[0108] Regarding implicit absorption, when a particle interacts with a substance, the particle can only scatter and is not absorbed. Each time it collides, the weight of the particle is multiplied by P / P 散乱 / P 総 When the weight of the particle drops to a certain value (for example 0.25), the particle is processed using the Russian roulette method.

[0109] Regarding the weight window game, when the weight of the particle is greater than a certain value, for example, greater than 10 , the splitting method is performed. When the weight of the particle is less than a certain value, for example, less than 0.25 , the Russian roulette method is performed. The two values here are set based on software performance .

[0110] Generally speaking, when a particle moves from a location with a small lattice space importance to a location with a large lattice space importance, the weight of the particle improves, and the particle does not disappear. When the weight decreases , for the case where the weight of the particle decreases, when the particle moves from a location with a large lattice space importance to a location with a small lattice space importance, the weight of the particle decreases, and splitting or implicit absorption processing is performed on the particle. Due to the large absorption cross-section of thermal neutrons in the human body (mainly elements such as N, B, etc ), in order to accelerate the convergence speed of the dose calculation of the lattice at locations far from the neutron source, particles should move to these lattices through implicit absorption. When a particle leaves the treatment area , continuing the simulation of the particle has little significance for the dose calculation in the treatment area. In this case, continuing the calculation will waste computational resources. Therefore, by setting the lattice space importance , the probability of such a situation can be reduced. When the weight of the particle is too small , the contribution of the particle to the count is very small, and the simulation Continuing in this case is not very meaningful for the dose calculation in the treatment area. In this case, continuing the calculation will waste computational resources. Therefore, by setting the lattice space importance , the probability of such a situation can be reduced. When the weight of the particle is too small , the contribution of the particle to the count is very small, and the simulation Continuing this will waste computational resources, and if the particle weights are too large, a single count will Because it can cause things to get too big and lead to the wrong results, weight win The Dough should be set to bring the particle weights within an appropriate range. Therefore, in combination with implicit absorption... In addition, the grid space importance and wait window game can be set rationally. Therefore, it is necessary to accelerate the convergence speed of the Monte Carlo program.

[0111] As shown in Figure 5, the flow for simulating particles using the dispersion reduction method is as follows: Step S1 involves acquiring a child particle, and then determining whether the particle will collide within the lattice cell. If a collision occurs, If a decision is made, steps S3 and S4 are executed in order, and if it is determined that there is no collision, step Step S2 executes S5 and S6 in order, step S3 performs implicit absorption processing, and the weights Determine whether the value is below the wait window, and if it is determined to be below, execute step S7. If it is determined that the value is not low, return to step S2, then step S4, and Russian Roulette Step S5 involves performing a splitting process and whether or not to perform a Russian roulette process. The system determines whether to perform Russian roulette, and if it determines to perform Russian roulette, it executes step S7. If it is decided not to perform the cyan roulette process, return to step S2, step S6 and Then, a decision is made as to whether or not to destroy it, and if it is decided to destroy it, step S8 is executed and it is destroyed. If it is determined not to allow it, the process returns to step S2 after executing step S9. 7. Determine whether the particle processing is complete or not. If it is determined to be complete, terminate the flow. If it is determined that the process cannot be completed, return to step S1, then to step S8, and the weights are determined by the probability of extinction. Includes the division step S9.

[0112] As shown in Figure 6, the flow of Russian roulette and splitting is as follows: Step S1: Calculate and record the child space importance, and step S1: Obtain the particle. Step S2 is a step in which weight window detection and manipulation are performed on the particles, and the particles are Importance of the grid space before and after crossing the subboundary I n and I n+1 Step S4 to calculate and , I n I n+1 We compared whether it was greater or less, and if it was not greater, we performed step S6. Later, return to step S3, and if large, perform step S7, step S5, and the particle Step S6 involves splitting the particles and reducing their weight, and then eliminating the particles. The system determines whether or not to destroy the particle, and if it determines to destroy the particle, it executes step S9 to destroy the particle. If it is decided not to allow it, the process returns to step S3 after executing step S8. Step 7 involves increasing the weight of the particle, and step S8 involves determining whether the particle simulation is complete or not. The system determines whether the simulation is complete, and if it is determined that the simulation is not complete, it returns to step S2. Step S9 includes terminating if it is determined that the simulation is complete.

[0113] As shown in Figure 7, the implicit absorption flow involves step S1 acquiring a particle and the particle being in the lattice. The system determines whether a collision will occur within the cell, and if a collision is determined, it executes step S3. If it is determined that there is no collision, return to step S1, then step S2, and assign the scattering probability to the weight. Multiply by step S3 and determine whether the particle weight is smaller than the lowest weight. If it is smaller, If a determination is made, perform step S5; if it is determined that it is not small, return to step S2. Step S4 is performed to determine whether or not to eliminate the particle, and if it is determined that the particle should be eliminated... If step S7 is executed and it is determined that the particles should not be eliminated, then step S6 is executed. Later, return to step S2, step S5, and step S6 divide the weight by the probability of annihilation, and the particle Determine whether the simulation is complete or not, and if it is determined that the simulation is not complete. If the process is interrupted, return to step S1, and if it is determined that the simulation is complete, terminate. Step S7 includes,

[0114] As shown in Figure 8, the flow of a weight window game simulates particle motion. Step S1 involves determining whether the particle weight is within the weight window range. If it is determined to be within range, return to step S1; if it is determined to be outside range, Step S2 executes step S3 and determines whether the weight is larger than the weight window. If it is determined to be large, the process returns to step S1 after executing step S4, and if it is large If it is determined that step S5 is performed, step S3 is performed and the particles are split. Step S4 involves reducing the weight and deciding whether or not to eliminate it, and it was decided to eliminate it. If it is determined that the process will not be terminated and deleted, then step S1 will be executed after step S6. The process includes step S5, which returns to the previous step, and step S6, which increases the weight of the particle.

[0115] To compare the acceleration effect on dose calculation using the dispersion reduction method, we will compare the case when using implicit absorption and We will compare the computation time and standard deviation of the results with and without implicit absorption. For this, calculating 10 million particles and using the implicit absorption method, the computation time is 158 The calculation is performed in 4 seconds, with a standard deviation of 11% for dose calculation, and without using the implicit absorption method. The time is 1258s, the standard deviation of the dose calculation is 14.5%, and the standard deviation is implicitly absorbed. To reduce the standard deviation to the same value as when using the method, simulation The number of particles being tested was increased to 16 million, and the simulation calculation time was increased to 2045 seconds. This requires that the computation time increases by 461 seconds compared to when using the implicit absorption method, in other words Furthermore, using the implicit absorption method reduces computation time by approximately 20%. [Examples]

[0116] The simulation calculation will be performed using a non-uniform rectangular grid.

[0117] Traditional calculations of human body models often use a uniform grid, and in some cases, precision in subdomains is required. Because the number of grid cells must be increased for denser calculations, the computation time becomes longer. The row memory will increase exponentially. The obtained materials are often uniform in their lattice structure, and the materials in many of the connection areas are the same. For example, air, blood, etc., and for some of these areas, fine doses Since it is not necessary to calculate the distribution precisely, a coarse grid can be used to substitute for a fine grid, and tumor To achieve more precise calculations for important locations such as these, it is necessary to divide the grid into finer sections. There is a need. A non-uniform grid is used to calculate the dose, improving calculation accuracy, and during calculation... If the amount of memory used during execution does not increase significantly, and the computation time does not increase significantly, the total amount of critical memory will be calculated accordingly. To improve computational accuracy and reduce computation time while satisfying the computational accuracy requirements for non-critical areas. It is possible.

[0118] Below, simulation calculations were performed with different grid dimensions, and the results for different grid dimensions were as follows: We will compare the impact on simulation time and calculation accuracy.

[0119] [Table 4]

[0120] In this experiment, the mixed grating used was located in the direction of neutron incidence, with the front 5 cm section being the most prominent part. The first part is composed of a 0.4mm grid, and the middle 5cm section is composed of a 0.8mm grid. The remaining part consists of a 1.6 mm grid, and as can be seen from Table 3, a non-uniform grid is used. By using this method, the total number of grids is reduced to 2 / 5 of the original number of 0.4 mm grids, and the grid is not required The required memory was reduced to 1 / 3 of the original, and the calculation time was reduced to 34.7% of the original, 0 Based on the results obtained by calculation using a 0.4 mm grid, calculation using a mixed grid The neutron error in the calculated results is less than 0.1%, and the photon error is 0.2%. Less than %.

[0121] The configuration patterns of the mixed grating are not limited to those listed above, but may also apply to the specific irradiated object. It can be set according to the specific situation. In general, the dimensions of the grid are determined by the importance of the area. This is determined by, for example, the grid in the area where the tumor is located has small dimensions, with grids of 4 mm or less. The grid in the region where blood, air, and bone are located is large in size, and is 1.6 mm or larger. The dimensions of the grid in the area where other normal tissues such as muscle are located, set above, are greater than 0.8 mm. It will be set to be significantly smaller than 1.6mm.

[0122] The algorithm used for dose calculation is the Monte Carlo algorithm. The R algorithm has the advantage of high computational accuracy, but the disadvantages are its slow convergence speed and computation time. Because of its length, optimizing computational efficiency is the most important part. Optimization methods can improve computational efficiency to varying degrees and reduce computation time.

[0123] The above description of exemplary specific embodiments of the present invention will help those skilled in the art to understand the present invention. While this facilitates the process, it is clear that the present invention is not limited to the scope of specific embodiments and is not limited to those skilled in the art. The spirit and scope of the invention are limited and determined in the attached claims, with various variations being the extent of the invention. If these changes are present within the context, they are evident and all fall within the scope of the claims of this invention.

Claims

1. A beam irradiation device that generates a therapeutic beam and irradiates the object to be irradiated with the therapeutic beam to form an irradiated area, A treatment planning module that performs dose simulation calculations and creates a treatment plan based on the parameters of the treatment beam and the medical image data of the irradiated area, comprising a treatment planning module that simulates particles by dispersion reduction, Includes a control module that controls the irradiation of the beam irradiation device based on the treatment plan, The aforementioned variance reduction includes implicit absorption, weight window games, Russian roulette, and splitting. The step of simulating particles with reduced dispersion is, Step S1 involves acquiring one source particle, Step S2 determines whether or not a particle will collide within the lattice cell. If it is determined that a collision will occur, steps S3 and S4 are executed in order. If it is determined that a collision will not occur, steps S5 and S6 are executed in order. Step S3 involves implicit absorption processing, The system determines whether the weight is lower than the weight window. If it is lower, it executes step S7; otherwise, it returns to step S2, and so on. Step S5 involves performing Russian roulette and splitting processing, The system determines whether or not to perform a Russian roulette game. If it decides to perform a Russian roulette game, it executes step S7. If it decides not to perform a Russian roulette game, it returns to step S2, and so on. Step S7 involves determining whether or not to destroy the data. If it is determined to destroy the data, step S8 is executed. If it is determined not to destroy the data, step S9 is executed, and then the process returns to step S2. The process determines whether the particle processing is complete or not. If it is determined to be complete, the flow ends; if it is determined not to be complete, the process returns to step S1, and so on. Step S9 involves dividing the weight by the probability of extinction, Features including, Radiation irradiation system.

2. The aforementioned variance reduction includes implicit absorption, weight window games, Russian roulette, and splitting. The aforementioned radiation irradiation system is a neutron capture therapy system, The beam irradiation apparatus includes a neutron generator, a beam shaping body, and a treatment table. The neutron generator includes an accelerator and a target, The accelerator accelerates charged particles to generate a charged particle beam, and interacts with the target to generate a neutron beam. The beam shaping body is capable of adjusting the neutron beam generated by the neutron generator to a predetermined beam quality. The neutron beam generated by the neutron generator is characterized in that it passes through the beam shaping body and is irradiated onto the object to be irradiated on the treatment table. The radiation irradiation system according to claim 1.

3. A method for optimizing the treatment planning module of a radiation irradiation system, The aforementioned radiation irradiation system is A beam irradiation device that generates a therapeutic beam and irradiates the object to be irradiated with the therapeutic beam to form an irradiated area, A treatment planning module that performs dose simulation calculations based on the parameters of the treatment beam and the medical image data of the irradiated area, and creates a treatment plan, Includes, The method for optimizing the treatment planning module of the radiation irradiation system includes the step of the treatment planning module simulating particles by dispersion reduction, The aforementioned variance reduction includes implicit absorption, weight window games, Russian roulette, and splitting. The step of simulating particles with reduced dispersion is, Step S1 involves acquiring one source particle, Step S2 determines whether or not a particle will collide within the lattice cell. If it is determined that a collision will occur, steps S3 and S4 are executed in order. If it is determined that a collision will not occur, steps S5 and S6 are executed in order. Step S3 involves implicit absorption processing, The system determines whether the weight is lower than the weight window. If it is lower, it executes step S7; otherwise, it returns to step S2, and so on. Step S5 involves performing Russian roulette and splitting processing, The system determines whether or not to perform a Russian roulette game. If it decides to perform a Russian roulette game, it executes step S7. If it decides not to perform a Russian roulette game, it returns to step S2, and so on. Step S7 involves determining whether or not to destroy the data. If it is determined to destroy the data, step S8 is executed. If it is determined not to destroy the data, step S9 is executed, and then the process returns to step S2. The process determines whether the particle processing is complete or not. If it is determined to be complete, the flow ends; if it is determined not to be complete, the process returns to step S1, and so on. Step S9 involves dividing the weight by the probability of extinction, Features including, A method for optimizing the treatment planning module of a radiation therapy system.

4. The aforementioned Russian roulette and splitting are, Step S1 involves calculating and recording the grid space importance of each grid, Step S2 for obtaining particles, Step S3 involves performing weight window detection and manipulation on the particles, The importance of lattice space I before and after a particle crosses a lattice boundary. ｎ and I ｎ＋１ Step S4 calculates, I ｎ I ｎ＋１ The system compares whether it is greater or less. If it is not greater, it performs step S6 and returns to step S3. If it is greater, it performs step S7, and so on. Step S6 involves splitting the particles and reducing their weight, Step S7 involves determining whether or not to eliminate the particle, executing step S9 if it is determined to eliminate the particle, and then returning to step S3 after executing step S8 if it is determined not to eliminate the particle. Step S8 involves increasing the weight of the particles, Step S9 determines whether the particle simulation is complete or not. If it is determined that the simulation is not complete, the process returns to step S2. If it is determined that the simulation is complete, the process terminates. Features including, A method for optimizing the treatment planning module of a radiation irradiation system according to claim 3.

5. The aforementioned implicit absorption is, Step S1 for obtaining particles, The process determines whether or not a particle will collide within the lattice cell. If a collision is determined, step S3 is executed; if no collision is determined, the process returns to step S1, and so on. Step S3 involves multiplying the weight by the probability of scattering occurring, The system determines whether the particle weight is less than the lowest weight. If it is determined to be less, it executes step S5. If it is determined not to be less than the lowest weight, it returns to step S2, and so on. Step S5 involves determining whether or not to eliminate the particle, executing step S7 if it is determined to eliminate the particle, and then returning to step S2 after executing step S6 if it is determined not to eliminate the particle. Step S6 involves dividing the weight by the probability of extinction, Step S7 determines whether the particle simulation is complete or not. If it is determined that the simulation is not complete, the process returns to step S1. If it is determined that the simulation is complete, the process terminates. Features including, A method for optimizing the treatment planning module of a radiation irradiation system according to claim 3.

6. Wait window games are Step S1 simulates particle motion, Step S2 determines whether the particle's weight is within the weight window range. If it is within the range, the process returns to step S1; if it is not within the range, step S3 is executed. Step S3 determines whether the weight is greater than or equal to the weight window. If it is determined to be greater, step S4 is executed, and then the process returns to step S1. If it is determined not to be greater, step S5 is executed. Step S4 involves splitting the particles to reduce their weight, Step S5 involves determining whether or not to destroy the data. If it is determined to destroy the data, the process ends. If it is determined not to destroy the data, step S6 is executed, followed by a return to step S1. Step S6 involves increasing the weight of the particles, Features including, A method for optimizing the treatment planning module of a radiation irradiation system according to claim 3.

7. The lattice space importance is obtained by solving the adjoint transport equation below, [Math 1] In the formula, Φ ＊ This is an adjoint bundle, S ＊ is the associated source, v is the velocity of the particle motion, Ω is the direction of the particle motion, Σ ｔ Σ is the reaction cross-section where a particle and matter collide. ｓ The characteristic is that is the scattering cross-section, r is the position where the particle is located, E is the energy of the particle, and t is time. A method for optimizing the treatment planning module of a radiation irradiation system according to claim 4.

8. The particle has a lattice space importance of I ｎ and moves from a lattice with a lattice space importance of I ｎ＋１ to a lattice with a lattice space importance of I ｎ＋１ If I ｎ > I ｎ＋１ then set m = I ｎ / I ｎ＋１ Split the particle into m particles, reduce the weight of each particle to 1 / m of the original weight, and if I ｎ < I ｎ＋１ then perform the Russian roulette method on the particle, set P = I ｎ / I Sample a random number x from 0 to 1. If x is less than P, the particle survives and its weight is multiplied by 1 / P; otherwise, the particle is eliminated and the simulation of the particle ends. This is characterized by A method for optimizing the treatment planning module of a radiation irradiation system according to claim 7.

9. The particle weight has an upper limit of 10 and a lower limit of 0.

25. If the particle weight w is greater than 10, the integer part of w is set to w1 and the decimal part to w2. A random number x is sampled from 0 to 1. If x is less than w2, w = w1 + 1; if x is greater than w2, w = w1. Then, the single particle is split into w particles, and the simulation is performed until the weight of each particle decreases to 1. A method for optimizing a treatment planning module of a radiation irradiation system according to any one of claims 3 to 8.

Images