Method for calculating BNCT treatment dose and terminal

Abstract

This application provides a method and terminal for calculating BNCT treatment dose, comprising: acquiring a unified energy grid and storing macroscopic cross-sectional data, total macroscopic cross-sectional data, and principal cross-sectional data of each voxel at each energy point in the unified energy grid; determining a query energy point based on neutron energy to query the corresponding principal cross-sectional data from the unified energy grid, and performing virtual sampling of neutrons; querying the corresponding macroscopic cross-sectional data and total macroscopic cross-sectional data from the unified energy grid based on the neutron sampling point and the query energy point; obtaining the reaction weight and survival weight of neutrons at the sampling point based on the total macroscopic cross-sectional data and the principal cross-sectional data; calculating the dose component of the sampling point based on the reaction weight and the macroscopic cross-sectional data, and determining whether to perform the next virtual sampling of neutrons based on the survival weight, until all virtual sampling of neutrons is completed, and obtaining the BNCT treatment dose based on the dose component of each sampling point. This application calculates the BNCT treatment dose accurately and efficiently.

Description

Technical Field

[0001] This application belongs to the field of boron neutron technology and relates to a method and terminal for calculating BNCT therapeutic dose. Background Technology

[0002] Boron neutron capture therapy (BNCT) is a radiotherapy method used to treat tumors. It precisely generates alpha particles or lithium ions within tumor cells by combining boron-containing chemical agents with a neutron beam, thereby destroying the tumor cells. However, calculating the BNCT treatment dose has always been a technical challenge because the intensity and distribution of the neutron beam, as well as the distribution of the radiopharmaceutical in the body, are affected by many complex factors, leading to significant errors in dose calculation accuracy.

[0003] Currently, traditional BNCT treatment dose calculation methods, such as Monte Carlo simulations based on physical models, while providing some theoretical support, often fail to accurately assess the dose distribution received by patients during treatment in real-time due to the large computational load and high precision requirements in practical clinical applications. With the continuous development of BNCT technology, especially the upgrading of neutron sources and treatment equipment, existing dose calculation methods are struggling to meet the requirements for accurate, real-time, and efficient calculations. Therefore, how to accurately and efficiently calculate BNCT treatment doses is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] This application provides a method and terminal for calculating BNCT treatment dose, which solves the technical problem of how to accurately and efficiently calculate BNCT treatment dose.

[0005] In a first aspect, this application provides a method for calculating the BNCT treatment dose, the method comprising:

[0006] A unified energy grid is obtained, and macroscopic cross-sectional data, total macroscopic cross-sectional data, and principal cross-sectional data of each energy point are stored on the unified energy grid for each voxel.

[0007] The query energy point is determined based on the neutron energy to query the corresponding main cross-section data from the unified energy grid, and the neutrons are virtually sampled based on the main cross-section data;

[0008] Based on the neutron sampling points and the query energy points, the corresponding macroscopic cross-sectional data and the macroscopic total cross-sectional data are queried from the unified energy grid;

[0009] Based on the macroscopic total cross-sectional data and the main cross-sectional data, the reaction weight and survival weight of the neutron at the sampling point are obtained;

[0010] The dose component of the sampling point is calculated based on the reaction weight and the macroscopic cross-sectional data, and the survival weight is used to determine whether to perform another virtual sampling of neutrons until the virtual sampling of all neutrons is completed. Then, the BNCT treatment dose is obtained based on the dose component of each sampling point.

[0011] In some implementations of the first aspect, obtaining a unified energy grid includes:

[0012] The energy points corresponding to the microscopic cross-sectional data of each element are read from the original kernel database, and an index is created for all the energy points to generate a unified mesh;

[0013] The microscopic cross-sectional data of each element at each energy point are reconstructed to the unified grid and stored based on the energy point index in the unified grid to obtain the unified energy grid.

[0014] In some implementations of the first aspect, storing macroscopic cross-sectional data, total macroscopic cross-sectional data, and principal cross-sectional data of each voxel at each energy point on the unified energy grid includes:

[0015] Obtain a voxel model, and obtain the mass density and elemental mass fraction of each voxel in the voxel model;

[0016] Based on the mass density, the mass fraction of each element, and the micro-section data in the unified energy grid, the macro-section data of each voxel at each energy point and the total macro-section data are calculated and obtained.

[0017] The macroscopic total cross-sectional data of each voxel at each energy point is traversed, and the maximum value of the macroscopic total cross-sectional data at each energy point is taken as the corresponding main cross-sectional data;

[0018] The macroscopic cross-sectional data, the macroscopic total cross-sectional data, and the main cross-sectional data are stored based on the energy point index in the unified energy grid.

[0019] In some implementations of the first aspect, virtual sampling of neutrons based on the main cross-section data includes:

[0020] Based on the main cross-sectional data and random numbers, exponential distribution sampling is performed to obtain the sampling step size;

[0021] Based on the sampling step size, the neutron is moved along the current moving direction to the corresponding sampling point to complete one virtual sampling.

[0022] In some implementations of the first aspect, querying the corresponding macroscopic cross-sectional data and the macroscopic total cross-sectional data from the unified energy grid based on the neutron sampling point and the query energy point includes:

[0023] The current voxel corresponding to the sampling point is determined based on a preset position mapping relationship;

[0024] Based on the current voxel, query the macroscopic cross-sectional data and the macroscopic total cross-sectional data at the query energy point in the unified energy grid.

[0025] In some implementations of the first aspect, obtaining the neutron's reaction weight and survival weight at the sampling point based on the macroscopic total cross-sectional data and the main cross-sectional data includes:

[0026] The ratio of the total macroscopic cross-sectional data to the main cross-sectional data is obtained, and the ratio is used as the physical reaction probability of neutrons.

[0027] Based on the physical reaction probability, the statistical weight of the neutron at the sampling point is deterministically decomposed to obtain the reaction weight and the survival weight respectively; wherein, the product of the physical reaction probability and the statistical weight is used as the reaction weight; and the difference between the statistical weight and the reaction weight is used as the survival weight.

[0028] In some implementations of the first aspect, determining whether to perform another virtual sampling of neutrons based on the survival weight includes:

[0029] If the survival weight is less than the preset survival threshold, then the neutron will not be virtual sampled again.

[0030] If the survival weight is not less than the preset survival threshold, then the neutrons are virtual sampled again, and the neutron statistical weight at the time of the next virtual sampling is obtained based on the survival weight and the neutron statistical weight.

[0031] In some implementations of the first aspect, calculating the dose component of the sampling point based on the response weight and the macroscopic cross-sectional data includes:

[0032] Obtain the ratio of each macroscopic cross-sectional data to the total macroscopic cross-sectional data;

[0033] Based on the ratios, the reaction weights are assigned to each physical reaction of the neutron, and multiplied by the unit energy release value of each physical reaction to calculate the dose component corresponding to each physical reaction at the sampling point.

[0034] In some implementations of the first aspect, obtaining the BNCT treatment dose based on the dose components of each sampling point includes:

[0035] The dose components corresponding to each physical reaction at each sampling point are added together to obtain the BNCT treatment dose; wherein, the physical reactions include boron capture reaction, nitrogen capture reaction, hydrogen scattering reaction, fast neutron reaction, and gamma reaction.

[0036] In a second aspect, this application also provides a terminal, including a processor and a memory, wherein the memory and the processor are communicatively connected; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to cause the terminal to perform the method as described in the first aspect.

[0037] As described above, the BNCT treatment dose calculation method and terminal described in this application have the following beneficial effects:

[0038] 1. This application performs unified kernel data preprocessing on all elements involved in BNCT treatment dose calculation and constructs a unified energy grid. This enables the cross-sectional information of each element to be obtained simultaneously with only one energy indexing operation at any neutron energy, avoiding repeated interpolation operations for multiple elements and multiple grids, and significantly reducing the energy cross-section lookup overhead.

[0039] 2. This application uniformly uses exponential distribution sampling based on the main cross section to control the flight distance of neutrons, and can determine the tissue or element type where the neutron is located simply by querying the location of the sampling point, so that neutrons can propagate continuously between different tissue boundaries, and effectively avoid the high-frequency boundary calculation overhead caused by the microstructure and complex organ geometry of the human body.

[0040] 3. This application uses continuous weighting to treat neutrons as simultaneously undergoing physical reaction and continued transmission at each sampling point, and reflects the relative contribution of the two physical processes through deterministic adjustment of the weights. This mathematically makes the originally discrete random collision process continuous, so that even in regions with extremely low acceptance probability, each sampling still makes an effective contribution to dose statistics, thereby completely eliminating the invalid sampling loops caused by virtual collisions near strong absorbers.

[0041] 4. This application supports the statistical needs of multiple dose components by calculating the continuous dose components of each sample, so that the dose contribution no longer depends on sparse physical collision events. Thus, it can still maintain stable and low variance dose estimation results in human soft tissue, air cavity or low density region, which is more suitable for the fine dose distribution requirements of BNCT.

[0042] 5. This application controls the number of neutrons transported to avoid a large number of low-weight neutrons occupying computational resources, and appropriately amplifies the weight of surviving neutrons to maintain statistical unbiasedness, thereby effectively balancing computational efficiency and statistical stability while ensuring computational accuracy.

[0043] 6. This application, through continuous weight allocation and continuous calculation of dose components, does not contain conditional jumps under ideal implementation, which is conducive to vectorized and parallelized calculations and has better scalability potential on multi-core CPU or accelerator platforms. Attached Figure Description

[0044] Figure 1 A flowchart illustrating the method for calculating the BNCT treatment dose according to an embodiment of this application is shown.

[0045] Figure 2 A schematic diagram of the process for storing cross-sectional data in a unified energy grid, as described in an embodiment of this application, is shown.

[0046] Figure 3 A schematic diagram of the process for obtaining the reaction weight and survival weight of the neutron at the sampling point according to an embodiment of this application is shown.

[0047] Figure 4 A schematic diagram of the process for calculating the dose component of the sampling point based on the reaction weight and the macroscopic cross-sectional data, as described in an embodiment of this application, is shown.

[0048] Figure 5 A schematic diagram illustrating the application flow of the BNCT treatment dose calculation method described in the embodiments of this application is shown.

[0049] Figure 6 A schematic diagram of the terminal described in an embodiment of this application is shown. Detailed Implementation

[0050] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0051] It should be noted that in the following description, reference is made to the accompanying drawings, which illustrate several embodiments of this application. It should be understood that other embodiments may also be used, and changes in mechanical composition, structure, electrical system, and operation may be made without departing from the spirit and scope of this application. The following detailed description should not be considered limiting, and the scope of the embodiments of this application is defined only by the claims of the published patent. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application. Spatial terms such as “upper,” “lower,” “left,” “right,” “below,” “below,” “lower part,” “above,” “upper part,” etc., may be used herein to illustrate the relationship between one element or feature shown in the figures and another element or feature.

[0052] Furthermore, as used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context indicates otherwise. It should be further understood that the terms “comprising,” “including,” indicate the presence of the stated feature, operation, element, component, item, kind, and / or group, but do not preclude the presence, occurrence, or addition of one or more other features, operations, elements, components, items, kinds, and / or groups. The terms “or” and “and / or” as used herein are to be interpreted inclusively, or mean any one or any combination thereof.

[0053] The Monte Carlo method is a widely used numerical simulation technique in neutron transport and dose calculation. Its core idea is to statistically obtain macroscopic physical quantities, such as flux, reaction rate, and dose distribution, by randomly sampling the motion and collision behavior of a large number of particles in a medium. However, in BNCT dose calculation, traditional Monte Carlo algorithms suffer from significant performance bottlenecks and computational efficiency limitations. BNCT dose calculation requires precise tracking of neutron behavior within the complex organ structures, tumor regions, and geometry of treatment devices in the human body. Traditional surface tracking (ST) methods require calculating the distance from the particle to the nearest geometric boundary within the current region during each particle flight and comparing it with the flight distance to determine whether the particle collides or crosses the boundary. For systems with numerous tiny geometric components, such as human organs, blood vessels, bones, and boron-containing tumor tissue, the overhead of such boundary intersection calculations is very high, becoming a major bottleneck in simulation efficiency.

[0054] Furthermore, the neutron cross-sections of different tissues and materials in BNCT treatment vary greatly. For example, boron-containing materials in tumors have high cross-sections, while surrounding soft tissues and coolants have low cross-sections. When dealing with regions with large cross-section differences and complex geometry, traditional Monte Carlo algorithms still require precise calculations of the comparison between each particle's flight distance and the boundary in low-cross-section regions, leading to a large amount of invalid computation and low efficiency. Simultaneously, traditional algorithms typically rely on track-length estimators (TLE) for dose or flux calculations, but at complex boundaries or at the junctions of high-cross-section materials, the statistical variance tends to increase, making it difficult to guarantee the stability and reliability of the calculation results. These factors collectively result in the low efficiency of traditional Monte Carlo methods in practical applications of BNCT dose calculations, especially under the requirement of high-precision dose distribution, where computation time and resource consumption are significant.

[0055] To improve the computational efficiency of Monte Carlo simulations under complex geometries, Woodcock Delta-Tracking introduces a global principal section. Virtual collisions allow particles to cross geometric boundaries without frequent intersection calculations, significantly reducing the computational overhead of boundary intersections. This method achieves high computational efficiency in certain large-scale neutron transport problems with complex micro-geometry components. However, when applied to BNCT dose calculations, this method also exhibits significant limitations and performance bottlenecks.

[0056] First, during BNCT treatment, the tumor tissue has a high boron content and a very large cross-section, while the surrounding soft tissue has a relatively low cross-section. In Delta-Tracking, the global principal cross-section must be the largest cross-section covering the entire geometric domain to ensure algorithm accuracy. This results in a very low neutron acceptance probability in the low cross-section region, with neutrons frequently experiencing virtual collisions. However, these virtual collisions do not change the neutron state or directly contribute to the dose count, leading to a large amount of redundant calculations and severely reducing simulation efficiency. This phenomenon is particularly pronounced in small-volume tumors containing localized strong absorbers or near shielding materials, and may even cause neutrons to enter an "infinite sampling loop."

[0057] Secondly, Delta-Tracking does not record the precise trajectory of neutrons crossing the boundary, therefore it is impossible to use a trajectory length flux estimator for counting, and only a collision flux estimator can be used. However, in BNCT dose calculation, there are regions with large volumes but extremely low cross-sections (such as air cavities or body cavity gaps), where the probability of neutrons physically colliding is extremely low, resulting in insufficient CFE statistical samples and excessively large variance in flux or dose estimation, affecting the accuracy of the results.

[0058] Finally, in deep-penetration calculation scenarios, the number of virtual collisions in Delta-Tracking increases with flight depth, leading to high computational resource consumption and making it difficult to combine with variance reduction techniques (such as weighted windows or importance sampling) for optimization. Since these techniques typically rely on clear boundary definitions and geometric path control, the spatial jump characteristics of Delta-Tracking actually limit its variance control capabilities. Therefore, although Delta-Tracking can reduce geometric intersection overhead to some extent, in BNCT dose calculation, the small volume of local high cross-sections, the dominance of low cross-section regions, and the problem of shielding deep penetration still limit its algorithm efficiency and counting accuracy.

[0059] To at least address the aforementioned technical problems, embodiments of this application provide a method and terminal for calculating BNCT treatment doses. This method can simultaneously achieve collaborative optimization across three levels: geometric processing, collision determination, and kernel data access in BNCT dose calculation. It can obtain accurate BNCT treatment dose results with high computational efficiency and low statistical variance under complex human geometry and high cross-sectional gradient conditions.

[0060] Figure 1 A flowchart illustrating the method for calculating the BNCT treatment dose according to an embodiment of this application is shown. Figure 1 As shown, the method for calculating the BNCT treatment dose provided in this application embodiment includes steps S1 to S5.

[0061] S1. Obtain a unified energy grid and store the macroscopic cross-sectional data, total macroscopic cross-sectional data, and principal cross-sectional data of each energy point for each voxel on the unified energy grid.

[0062] In some embodiments, obtaining a unified energy grid includes: reading the energy points corresponding to the micro-section data of each element from the original nuclear database, and indexing all the energy points to generate a unified grid; reconstructing the micro-section data of each element at each of the energy points to the unified grid, and storing it based on the energy point index in the unified grid to obtain the unified energy grid.

[0063] For example, for all human tissue elements and boron-10 that may be involved in dose calculation, all raw energy points corresponding to the microscopic cross-sectional data of each element are read from the raw nuclear database (ENDF, etc.). These raw energy points are then merged, deduplicated, and sorted to form a joint energy point set covering the entire energy range. A unique integer index (e.g., 1, 2, ..., K) is assigned to each energy point to generate a unified grid. Subsequently, the raw microscopic cross-sectional data of each element is interpolated or mapped to each energy point in the unified grid, thus obtaining the energy point in the grid. The microscopic cross-sectional data of the i-th element is And it is stored in the form of energy point index to obtain a unified energy grid.

[0064] Among them, microscopic cross-sectional data refers to the probability area describing a specific reaction (such as elastic scattering, boron neutron capture reaction, nitrogen capture reaction, hydrogen scattering reaction, etc.) between a single neutron and a single target nucleus.

[0065] Furthermore, it is also necessary to store other cross-sectional data of each energy point in a unified energy grid. Figure 2 A schematic diagram illustrating the process of storing cross-sectional data in a unified energy grid according to an embodiment of this application is shown. Figure 2 As shown, storing macroscopic cross-sectional data, total macroscopic cross-sectional data, and main cross-sectional data of each energy point for each voxel on the unified energy grid includes steps S11 to S14.

[0066] S11. Obtain the voxel model and obtain the mass density and elemental mass fraction of each voxel in the voxel model.

[0067] In some embodiments, a regularly arranged voxel model is first generated using three-dimensional image data obtained from CT scans. Each voxel in the voxel model... For a cubic spatial unit of fixed size, let the voxel size be... Furthermore, the Hounsfield Unit (HU) values ​​in the CT images are converted into tissue mass density. Information composed of elements, and in each voxel Assign corresponding material properties, including the mass fractions of hydrogen, carbon, nitrogen, oxygen, and boron-10. .

[0068] S12. Based on the mass density, mass fraction of each element and micro-section data in the unified energy grid, calculate and obtain the macro-section data and total macro-section data of each voxel at each energy point.

[0069] In some embodiments, microsectional data are obtained from the original nuclear database and interpolated onto a unified energy grid. and the known mass density of each voxel. and element mass fraction For voxels Each energy point on the unified energy grid The macroscopic total cross section at the location is calculated, and the specific calculation is shown in Equation (1).

[0070]

[0071] in, voxels At the energy point The total macroscopic cross section at voxel r refers to the total probability that a neutron at the spatial position corresponding to voxel r will undergo any type of reaction with the medium (composed of a large number of target atomic nuclei and extranuclear electrons). It is Avogadro's constant; Let be the atomic weight of the i-th element; For the i-th element at energy point The total microscopic cross-section at that location. Among them, The i-th element at the energy point Microsection at the location Additive acquisition refers to the total probability area of ​​a single neutron reacting with a single target nucleus in any type of reaction.

[0072] Furthermore, regarding voxels Each energy point on the unified energy grid The macroscopic section at the location is calculated, and the specific calculation is shown in Equation (2).

[0073]

[0074] in, voxels At the energy point The macroscopic section at voxel r refers to the probability that a neutron reacts with the medium at the spatial position corresponding to voxel r. It is Avogadro's constant; Let be the atomic weight of the i-th element; For the i-th element at energy point The microscopic cross section of reaction x at the location.

[0075] S13. Traverse the macroscopic total cross-sectional data of each voxel at each energy point, and take the maximum value of the macroscopic total cross-sectional data at each energy point as the corresponding main cross-sectional data.

[0076] In some embodiments, for energy points in a unified energy grid The macroscopic total cross section of each voxel at the energy point is traversed, and the maximum value is taken as the main cross section data corresponding to the energy point. Specifically, it is shown in equation (3).

[0077]

[0078] In some embodiments, during BNCT treatment, the principal cross-sectional data are often determined by the tumor voxel with the highest boron concentration.

[0079] S14. Store macroscopic cross-sectional data, macroscopic total cross-sectional data, and main cross-sectional data based on the energy point index in the unified energy grid.

[0080] In some embodiments, macroscopic cross-sectional data, macroscopic total cross-sectional data, and main cross-sectional data are stored according to energy point indices in a unified energy grid. Furthermore, in terms of data storage structure, these cross-sectional data are arranged according to energy point indices and stored in the form of contiguous arrays or structured data blocks, so that different elements have corresponding cross-sectional data at the same energy point index position, eliminating the problem of energy grid inconsistencies between nuclides.

[0081] S2. Determine the query energy point based on the neutron energy to query the corresponding main cross section data from the unified energy grid, and perform virtual sampling of neutrons based on the main cross section data.

[0082] In some embodiments, neutrons are initialized to obtain neutron energy, initial neutron position, and initial neutron statistical weights, and a Monte Carlo transport simulation begins. During neutron transport, an energy indexing and positioning operation is performed on the unified energy grid based on the neutron energy to determine the corresponding query energy point. The corresponding master section data is then retrieved from the unified energy grid based on the query energy point, and neutrons are virtually sampled based on the master section data.

[0083] In some embodiments, virtual sampling of neutrons based on the main cross-section data includes: performing exponential distribution sampling based on the main cross-section data and random numbers to obtain the sampling step size; and moving the neutron along the current moving direction to the corresponding sampling point based on the sampling step size to complete one virtual sampling.

[0084] The calculation process of the sampling step size is shown in equation (4).

[0085]

[0086] in, It is a random number. To query the principal section data corresponding to energy point E, where s is the sampling step size, yes The probability density function follows an exponential distribution. Through equation (4), the non-uniform medium in space can be mathematically equivalent to a "pseudo-uniform" medium, so that the statistical distribution of neutron flight distance remains consistent throughout the entire space and the neutron energy remains unchanged. When a neutron moves along the current direction of movement by a sampling step, it is considered to have completed one virtual sampling, regardless of whether it crosses the voxel boundary in the middle, thereby realizing continuous propagation across voxels.

[0087] S3. Based on the neutron sampling points and query energy points, query the corresponding macroscopic cross-sectional data and macroscopic total cross-sectional data from the unified energy grid.

[0088] In some embodiments, the current voxel corresponding to the sampling point is determined based on a preset position mapping relationship, and the macroscopic cross-sectional data and the macroscopic total cross-sectional data at the query energy point are queried in the unified energy grid based on the current voxel.

[0089] For example, let the position vector of the i-th sampling point of the neutron be... Then, the corresponding current voxel can be determined through a preset position mapping relationship. And based on the current voxels The corresponding macroscopic cross-sectional data and the total macroscopic cross-sectional data are obtained directly from the unified energy grid by querying the energy points. The preset position mapping relationship is shown in equation (5).

[0090]

[0091] Among them, the voxel size is , These are the origin coordinates of the voxel mesh. The offset of the sampling point relative to the origin is calculated, divided by the voxel size, and then rounded down. The voxel to which the sampling point belongs is obtained. index This positional mapping relationship ensures that all points within the same voxel have the same index, and based on this index, the macroscopic total cross-section at the corresponding energy point and the macroscopic cross-sectional data of various reactions can be directly accessed, thereby effectively reducing the number of instructions executed and memory access latency.

[0092] S4. Obtain the reaction weight and survival weight of neutrons at the sampling points based on the macroscopic total cross-sectional data and the main cross-sectional data.

[0093] Figure 3 A schematic diagram illustrating the process of obtaining the reaction weight and survival weight of the neutron at the sampling point according to an embodiment of this application is shown. Figure 3 As shown, obtaining the reaction weight and survival weight of the neutron at the sampling point based on the macroscopic total cross-section data and the main cross-section data includes steps S41 and S42.

[0094] S41. Obtain the ratio of the total macroscopic cross-sectional data to the main cross-sectional data, and use the ratio as the physical reaction probability of neutrons.

[0095] In some embodiments, the total macroscopic cross-sectional data obtained based on the sampling point and the query energy point reflects the intensity of any physical interaction between neutrons of the corresponding energy at the sampling point. This ratio is calculated by comparing it with the previously obtained main cross-sectional data of the query energy point, and the ratio is used as the probability that a neutron will have a real physical interaction at the sampling point, i.e., the probability of a neutron's physical reaction. This probability is numerically equal to the ratio of the local macroscopic total cross-section to the main cross-section, and its physical meaning corresponds to the proportion of interaction between the "real medium" and the "pseudo-homogeneous medium" at the sampling point, as shown in Equation (6).

[0096]

[0097] in, To query the main cross-section data corresponding to energy point E, This represents the macroscopic total cross-sectional data of the current voxel at the query energy point E corresponding to the i-th sampling point.

[0098] S42. Based on the physical reaction probability, the statistical weight of neutrons at the sampling points is decomposed deterministically to obtain the reaction weight and survival weight respectively.

[0099] In some embodiments, the statistical weight of the neutron at the sampling point is deterministically decomposed based on the physical reaction probability to obtain the reaction weight and the survival weight respectively. After obtaining the physical reaction probability of the neutron at the sampling point, the probability is no longer sampled randomly to determine whether a virtual collision has occurred. Instead, the statistical weight of the neutron at the sampling point is deterministically decomposed according to the physical reaction probability to obtain the reaction weight and the survival weight respectively.

[0100] In some embodiments, the reaction weight represents the weight share “consumed and used for physical scoring” at the sampling point, and its value is obtained by the product of the physical reaction probability and the statistical weight, as shown in Equation (7).

[0101]

[0102] in, For reaction weights, The neutron statistical weights are from the previous sampling.

[0103] In some embodiments, the difference between the statistical weight and the response weight is used as the survival weight.

[0104] Among them, the survival weight is considered to be the effective weight for the neutron to continue to propagate forward after the sampling point is processed, and is considered to be the transmission component that has not interacted. Its value is obtained by the difference between the statistical weight and the reaction weight, as shown in Equation (8).

[0105]

[0106] in, Survival weight, The neutron statistical weights are from the previous sampling.

[0107] Therefore, each sampling can deterministically decompose the statistical weights at the sampling point without involving any random variables or conditional judgments. This means it can be repeated and stably executed at each main cross-section step sampling point. In this way, the neutron propagation process in space is transformed into a series of processes with "gradually decaying weights accompanied by continuous physical scoring," no longer relying on discrete random collision events. This deterministic weight evolution mechanism completely eliminates conditional branch judgments based on random numbers at the algorithm level, ensuring that each particle trajectory remains consistent along the execution path. Consequently, in SIMD vectorized computation or GPU massively parallel computing environments, it can significantly reduce branch divergence between threads, improve computational resource utilization, and effectively enhance the overall execution efficiency of Monte Carlo simulation in BNCT dose calculation.

[0108] Furthermore, by using a unified energy grid, a stable energy index and cross-section access process can be maintained even with an increase in elements. This ensures that kernel data access will not become a computational bottleneck during high-frequency sampling and weight decomposition, providing efficient and reliable data support for BNCT dose calculation under complex voxel geometry conditions.

[0109] S5. Calculate the dose component of the sampling point based on the reaction weight and macroscopic cross-sectional data, and determine whether to perform the next virtual sampling of neutrons based on the survival weight, until the virtual sampling of all neutrons is completed, and obtain the BNCT treatment dose based on the dose component of each sampling point.

[0110] In some embodiments, this application no longer limits dose counting to the actual location where a real physical collision event of a neutron occurs, but instead regards the physical interaction corresponding to each sampling point in each step sampling process based on the principal cross section as a statistically significant continuous process, so that the dose can be continuously accumulated along the particle propagation path. Figure 4 A schematic diagram illustrating the process of calculating the dose component of the sampling point based on the reaction weight and the macroscopic cross-sectional data, as described in an embodiment of this application, is shown. Figure 4 As shown, calculating the dose component of the sampling point based on the reaction weight and the macroscopic cross-sectional data includes steps S51 and S52.

[0111] S51. Obtain the ratio of each macroscopic section data to the total macroscopic section data.

[0112] S52. Based on each ratio, the reaction weights are assigned to each physical reaction of the neutron, and multiplied by the unit energy release value of each physical reaction to calculate the dose component corresponding to each physical reaction at the sampling point.

[0113] In some embodiments, after a neutron completes a flight step sampling based on the main cross-section and arrives at a new sampling point, the probability of a physical reaction occurring at that location is assigned in a weighted manner. Based on this, by utilizing the reaction weights and combining macroscopic cross-sectional data from different reaction channels, the corresponding dose components can be continuously statistically analyzed to obtain the final BNCT treatment dose, without waiting for the actual physical reaction of the neutron to occur during the random sampling process.

[0114] In some embodiments, each sampling point is considered a potential location where various physical reactions have occurred. Under this mechanism, the contribution of each sampling point to each therapeutic dose component is determined by three factors: first, the reaction weight assigned to the particle at that sampling point; second, the proportion of the macroscopic cross-sectional data corresponding to each reaction relative to the main cross-section at that sampling point location and neutron energy; and third, the energy release corresponding to a single physical reaction. Thus, effective dose samples can be stably generated in each sampling step, whether in regions with frequent reactions or in surrounding normal tissue regions where reactions occur infrequently. This statistically equivalently achieves an average count of a large number of possible physical reaction events without having to wait for these events to occur explicitly one by one during random transport.

[0115] In some embodiments, the physical reactions include boron capture reactions, nitrogen capture reactions, hydrogen scattering reactions, fast neutron reactions, and gamma reactions.

[0116] Taking the boron neutron capture reaction as an example, each sampling point is considered as a potential boron neutron capture reaction site. The boron capture dose corresponding to the boron neutron capture reaction at each sampling point is calculated as shown in Equation (9).

[0117]

[0118] in, This represents the boron capture dose statistically calculated at the sampling point corresponding to the i-th sampling. The response weights assigned to the i-th sample are For a neutron with energy E, the sample point corresponding to the i-th sample is... Macroscopic cross section of the reaction This represents the energy released in a single boron neutron capture reaction.

[0119] Similarly, the statistical process for nitrogen capture dose and hydrogen scattering dose also employs the same calculation procedure. For example, at each sampling point, based on the proportion of the macroscopic cross-sectional data of the corresponding reaction channel in the main cross-section, the reaction weights are proportionally allocated to the nitrogen capture reaction and the hydrogen elastic scattering reaction, and then multiplied by the unit energy release value of the corresponding reaction, thereby continuously including the corresponding energy deposition into their respective dose components. This approach avoids the problem of scarce statistical samples caused by the low probability of certain reactions occurring.

[0120] In some embodiments, the nitrogen capture dose is calculated as shown in equation (10).

[0121]

[0122] in, Let be the nitrogen capture dose statistically calculated at the sampling point corresponding to the i-th sampling. The response weights assigned to the i-th sample are This represents the macroscopic cross-section of the nitrogen capture reaction that occurs at the sampling point corresponding to the i-th sampling, where the neutron has energy E. This refers to the energy released in a single nitrogen capture reaction.

[0123] In some embodiments, the hydrogen scattering dose is calculated as shown in equation (11).

[0124]

[0125] in, Let be the hydrogen scattering dose statistically calculated at the sampling point corresponding to the i-th sampling. The response weights assigned to the i-th sample are Let E be the macroscopic cross section of the hydrogen scattering reaction that occurs at the sampling point corresponding to the i-th sampling of a neutron with energy E. This refers to the energy released in a single hydrogen scattering reaction.

[0126] Similarly, the energy deposition caused by fast neutrons and gamma rays can be included in the dose component corresponding to each sampling point through the same weighting and continuous counting method.

[0127] Furthermore, after all sampling is completed, the dose components corresponding to each physical response at each sampling point are summed to obtain the BNCT treatment dose. Since the entire dose counting process does not depend on whether a real physical collision event occurred at the sampling point, but rather on continuous accumulation based on probability weights, it can maintain both high sample utilization and low statistical variance in BNCT scenarios where the spatial scale is highly non-uniform and the response cross-section varies across multiple orders of magnitude. This is particularly important for accurately assessing the dose difference between tumor regions and normal tissues.

[0128] Furthermore, as the Monte Carlo simulation continues, neutrons are continuously transported within the voxel model, and their statistical weights are redistributed at each sampling point. With the accumulation of sampling steps, the neutron statistical weights show a continuous decreasing trend, potentially dropping to very low values ​​in the later stages of transport. At this point, if the complete transport simulation and dose statistics are continued for all low-weight particles during the calculation, although their statistical contribution to the final dose result is already very limited, they still require computational resources similar to those of high-weight particles, resulting in unnecessary computational overhead.

[0129] Therefore, in some embodiments, after each sampling is completed, it is determined whether to perform another virtual sampling of neutrons based on the survival weight, thereby controlling the number of neutrons. Specifically, if the survival weight is less than the preset survival threshold, the neutrons are not subjected to another virtual sampling; if the survival weight is not less than the preset survival threshold, the neutrons are subjected to another virtual sampling, and the neutron statistical weight for the next virtual sampling is obtained based on the survival weight and the neutron statistical weight.

[0130] In some embodiments, the statistical weight of the neutron in the current sampling round is divided by the survival weight to amplify the statistical weight, which is then used as the neutron statistical weight in the next virtual sampling. This allows the neutron to statistically compensate for the contribution loss that may be caused by neutrons at the end of the sampling. In this way, the total number of particles that need to be tracked can be effectively reduced without changing the overall statistical expectation, the overall calculation time can be shortened, and the convergence speed and computational efficiency of BNCT dose calculation can be further improved.

[0131] In some embodiments, the preset survival threshold can be set according to the requirements of statistical accuracy and computational efficiency of the specific computational task, so as to distinguish between neutron trajectories that still have statistical significance and neutron trajectories whose contribution is negligible. This application does not limit the specific value.

[0132] Figure 5 A schematic diagram illustrating the application flow of the BNCT treatment dose calculation method described in the embodiments of this application is shown. Figure 5As shown, when calculating the BNCT treatment dose, a voxel model is first generated using 3D image data obtained from CT scans. Based on the mass density and elemental information of each voxel in the voxel model, combined with energy point information from the original nuclear database, a unified energy grid is constructed. This allows for direct querying of the cross-sectional data corresponding to the neutron energy through energy point index lookup. Next, the neutrons are initialized and transport simulation begins. During the transport simulation, multiple virtual samples of the neutrons are performed using the sampling step size obtained from the main cross-sectional data. Statistical weights are assigned at each sampling point to obtain the corresponding reaction weight and survival weight. Furthermore, dose component statistics are performed based on the reaction weight, and the neutron sampling is terminated based on the survival weight. If the survival weight is less than a preset survival threshold, the neutron is not sampled again; otherwise, sampling continues until all neutrons have been sampled, at which point the final BNCT treatment dose is output.

[0133] Therefore, this application achieves systematic optimization of complex voxel geometry, local strong absorbers, and statistical problems involving multiple dose components in BNCT dose calculation. This application not only eliminates invalid sampling loops in high boron concentration regions but also significantly reduces nuclear data retrieval overhead, thereby greatly improving the overall efficiency and convergence speed of BNCT dose calculation while ensuring computational accuracy.

[0134] The scope of protection for the BNCT treatment dose calculation method of this application is not limited to the execution order of the steps listed in this embodiment. Any solution implemented by adding, subtracting, or replacing steps in the prior art based on the principles of this application is included within the scope of protection of this application.

[0135] It should also be understood that the module division in the embodiments of this application is illustrative and only represents a logical functional division; in actual implementation, there may be other division methods. Furthermore, the functional modules in the various embodiments of this application can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0136] This application also provides a terminal. Figure 6 A schematic diagram of the structure of the terminal described in an embodiment of this application is shown, as follows: Figure 6As shown, the terminal 700 in this embodiment includes at least one processor 701, a memory 702, at least one network interface 704, and a user interface 706. Furthermore, the various components in the terminal 700 are coupled together via a bus system 705. It is understood that the bus system 705 is used to implement communication between these components. In addition to a data bus, the bus system 705 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 6 In this context, all buses are labeled as bus systems. The user interface 706 may include a monitor, keyboard, mouse, trackball, clicker, buttons, touchpad, or touchscreen, etc.

[0137] It is understood that memory 702 can be volatile memory or non-volatile memory, or both. This application does not specifically limit it. In the embodiments of this application, memory 702 is used to store various types of data to support the operation of terminal 700. Examples of this data include: any executable program for operation on terminal 700, such as operating system 7021 and application 7022; operating system 7021 includes various system programs, such as framework layer, core library layer, driver layer, etc., used to implement various basic services and handle hardware-based tasks. Application 7022 can include various applications, such as media player, browser, etc. The method for calculating the BNCT treatment dose provided in the embodiments of this application can be included in application 7022.

[0138] The BNCT treatment dosage calculation method disclosed in the above embodiments of this application can be applied to, or implemented by, processor 701. Processor 701 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed through integrated logic circuits in the hardware of processor 701 or through software instructions. Processor 701 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 701 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor 701 may be a microprocessor or any conventional processor, etc.

[0139] In an exemplary embodiment, terminal 700 may be used by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), or complex programmable logic devices (CPLDs) to execute the aforementioned method.

[0140] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented using computer program-related hardware. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0141] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0142] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method for calculating the dosage of BNCT treatment, characterized in that, The method includes: A unified energy grid is obtained, and macroscopic cross-sectional data, total macroscopic cross-sectional data, and principal cross-sectional data of each energy point are stored on the unified energy grid for each voxel. The query energy point is determined based on the neutron energy to query the corresponding main cross-section data from the unified energy grid, and the neutrons are virtually sampled based on the main cross-section data; Based on the neutron sampling points and the query energy points, the corresponding macroscopic cross-sectional data and the macroscopic total cross-sectional data are queried from the unified energy grid; Based on the macroscopic total cross-sectional data and the main cross-sectional data, the reaction weight and survival weight of the neutron at the sampling point are obtained; The dose component of the sampling point is calculated based on the reaction weight and the macroscopic cross-sectional data, and the survival weight is used to determine whether to perform another virtual sampling of neutrons until the virtual sampling of all neutrons is completed. Then, the BNCT treatment dose is obtained based on the dose component of each sampling point.

2. The method for calculating the BNCT treatment dose according to claim 1, characterized in that, Obtaining a unified energy grid includes: The energy points corresponding to the microscopic cross-sectional data of each element are read from the original kernel database, and an index is created for all the energy points to generate a unified mesh; The microscopic cross-sectional data of each element at each energy point are reconstructed to the unified grid and stored based on the energy point index in the unified grid to obtain the unified energy grid.

3. The method for calculating the BNCT treatment dose according to claim 1, characterized in that, The unified energy grid stores macroscopic cross-sectional data, total macroscopic cross-sectional data, and principal cross-sectional data of each energy point for each voxel, including: Obtain a voxel model, and obtain the mass density and elemental mass fraction of each voxel in the voxel model; Based on the mass density, the mass fraction of each element, and the micro-section data in the unified energy grid, the macro-section data of each voxel at each energy point and the total macro-section data are calculated and obtained. The macroscopic total cross-sectional data of each voxel at each energy point is traversed, and the maximum value of the macroscopic total cross-sectional data at each energy point is taken as the corresponding main cross-sectional data; The macroscopic cross-sectional data, the macroscopic total cross-sectional data, and the main cross-sectional data are stored based on the energy point index in the unified energy grid.

4. The method for calculating the BNCT treatment dose according to claim 1, characterized in that, Virtual sampling of neutrons based on the main cross-section data includes: Based on the main cross-sectional data and random numbers, exponential distribution sampling is performed to obtain the sampling step size; Based on the sampling step size, the neutron is moved along the current moving direction to the corresponding sampling point to complete one virtual sampling.

5. The method for calculating the BNCT treatment dose according to claim 1, characterized in that, Based on the neutron sampling points and the query energy points, the corresponding macroscopic cross-sectional data and the total macroscopic cross-sectional data are retrieved from the unified energy grid, including: The current voxel corresponding to the sampling point is determined based on a preset position mapping relationship; Based on the current voxel, query the macroscopic cross-sectional data and the macroscopic total cross-sectional data at the query energy point in the unified energy grid.

6. The method for calculating the BNCT treatment dose according to claim 1, characterized in that, The reaction weight and survival weight of the neutron at the sampling point are obtained based on the macroscopic total cross-sectional data and the main cross-sectional data, including: The ratio of the total macroscopic cross-sectional data to the main cross-sectional data is obtained, and the ratio is used as the physical reaction probability of neutrons. Based on the physical reaction probability, the statistical weight of the neutron at the sampling point is deterministically decomposed to obtain the reaction weight and the survival weight respectively; wherein, the product of the physical reaction probability and the statistical weight is used as the reaction weight; and the difference between the statistical weight and the reaction weight is used as the survival weight.

7. The method for calculating the BNCT treatment dose according to claim 1, characterized in that, Determining whether to perform another virtual sampling of neutrons based on the survival weight includes: If the survival weight is less than the preset survival threshold, then the neutron will not be virtual sampled again. If the survival weight is not less than the preset survival threshold, then the neutrons are virtual sampled again, and the neutron statistical weight at the time of the next virtual sampling is obtained based on the survival weight and the neutron statistical weight.

8. The method for calculating the BNCT treatment dose according to claim 1, characterized in that, Calculating the dose components at the sampling points based on the response weights and the macroscopic cross-sectional data includes: Obtain the ratio of each macroscopic cross-sectional data to the total macroscopic cross-sectional data; Based on the ratios, the reaction weights are assigned to each physical reaction of the neutron, and multiplied by the unit energy release value of each physical reaction to calculate the dose component corresponding to each physical reaction at the sampling point.

9. The method for calculating the BNCT treatment dose according to claim 8, characterized in that, The BNCT treatment dose is obtained based on the dose components of each of the aforementioned sampling points, including: The dose components corresponding to each physical reaction at each sampling point are added together to obtain the BNCT treatment dose; wherein, the physical reactions include boron capture reaction, nitrogen capture reaction, hydrogen scattering reaction, fast neutron reaction, and gamma reaction.

10. A terminal, characterized in that, It includes a processor and a memory, with the memory and the processor having a communicative connection; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to cause the terminal to perform the method as described in any one of claims 1 to 9.

Images