OpenMC-based BNCT dose full-process automatic calculation output method

Abstract

The invention discloses an OpenMC-based BNCT dose full-process automatic calculation output method, and particularly relates to the technical field of boron neutron capture therapy dose calculation, and the method comprises the steps: reading a DICOM-CT image, automatically completing HU value segmentation, material assignment and voxel grid modeling, and generating an OpenMC geometry and material input card; setting a neutron / photon source, an operation mode and a counting grid, and executing Monte Carlo transport calculation; an ICRP flux-dose conversion coefficient library is built in, and dose calculation is automatically completed; the dose field is written into an Rt dose file, and the Rt dose file can be directly imported into mature medical image processing software for visual evaluation and plan optimization. According to the method, end-to-end automation from the CT image to the clinical available dose file is achieved, manual intervention is not needed, the BNCT dose calculation efficiency, precision and standardization level are remarkably improved, and a reliable tool is provided for BNCT dose simulation research.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of boron neutron capture therapy dose calculation, and particularly relates to a BNCT dose full-process automatic calculation output method based on an open-source Monte Carlo program OpenMC. BACKGROUND

[0002] BNCT is short for boron neutron capture therapy, which mainly utilizes the high-LET particles generated by the B(n, alpha) Li reaction to locally kill tumor cells, and the dose is determined by neutron flux, boron concentration and tissue macroscopic cross section, and accurate and efficient three-dimensional dose calculation is a core link for guaranteeing the safety and curative effect of BNCT. 10 B(n, alpha) 7 Li reaction, and accurate and efficient three-dimensional dose calculation is a core link for guaranteeing the safety and curative effect of BNCT.

[0003] At present, although there are solutions based on MCNP, PHITS and Geant4, there are still problems such as authorization restriction, interface closure, high script maintenance cost and poor compatibility with the DICOM radiotherapy standard, and OpenMC, as an open-source, modern, parallel computing and Python API supported Monte Carlo program, has the potential to build a BNCT automatic dose calculation and analysis platform, therefore, the application provides a BNCT dose full-process automatic calculation output method based on OpenMC. SUMMARY

[0004] The application aims to provide a BNCT dose full-process automatic calculation output method based on OpenMC, which realizes the automatic and high-fidelity conversion from patient CT images to clinically available dose files by constructing a complete and accurate boron neutron capture therapy dose calculation and standardized output system, so as to solve the above problems in the prior art.

[0005] In order to achieve the above-mentioned purpose, the application provides the following technical scheme: a BNCT dose full-process automatic calculation output method based on OpenMC, comprising the following steps:

[0006] S1, reading a patient CT image and analyzing the voxel-level Hounsfield unit HU value, dividing the full HU domain into 35 intervals, mapping each interval to a unique material number and corresponding element composition and density, and generating a Monte Carlo program OpenMC material card;

[0007] S2, assigning a specific material number 36 to the tumor region to ensure that the nuclear reaction cross section and energy deposition calculation of the tumor region in the Monte Carlo simulation are consistent with the benchmark research;

[0008] S3, directly converting the CT voxels into OpenMC rectangular grids in a 1:1 ratio without resampling, and establishing a three-dimensional grid geometry;

[0009] S4, setting a multi-group epithermal neutron surface source on one side surface of the grid, defining the incident intensity, energy and direction of the neutron source;

[0010] S5, superimposing a three-dimensional grid counter on the same grid, respectively counting the total neutron flux, fast neutron flux, thermal neutron flux and photon flux, and outputting the energy group flux matrix containing 498 energy groups, 204 group fast neutrons, 9 group thermal neutrons and 44 group photons;

[0011] S6, reading the OpenMC state point file through the flux-dose conversion script to obtain the flux of each energy group, and multiplying it by the preset dose conversion coefficient respectively to obtain the fast neutron dose, thermal neutron dose, photon dose and boron dose;

[0012] S7, according to the different damage effects of boron dose on tumor and normal tissue, combined with the CT region mask to perform regional dose weighting: the tumor area complex biological effect CBE=3.8, boron concentration 65ppm, normal brain tissue CBE=1.35, boron concentration 18ppm, and the boron dose of other regions is zero;

[0013] S8, superimposing the fast neutron dose, thermal neutron dose, photon dose and boron dose of the four types of physical dose to calculate the total equivalent dose, and then according to the voxel size (pixel spacing and layer thickness) and matrix dimension of the original CT image, the resolution of the dose matrix is improved to be completely consistent with the original CT image through the interpolation algorithm, and the spatial geometry is accurately aligned;

[0014] S9, calling the Rt dose file making script to write the five groups of three-dimensional dose matrix (total dose, fast neutron dose, thermal neutron dose, boron dose and photon dose) obtained in S8 into the RT dose file conforming to the DICOM radiotherapy standard.

[0015] Preferably, the specific material amount fraction setting of the tumor area in step S9 is: H: 10.2%, C: 14.3%, N: 2.2%, O: 72.1%, and other elements: 1.2%, and the material density is set to 1.06 g / cm 3 .

[0016] Preferably, in the step S9, the following sub-steps are specifically included:

[0017] S9-1, reading the DICOM sequence of the original CT to extract the examination instance UID, reference frame UID, image patient position, image patient orientation, pixel spacing and layer thickness;

[0018] S9-2, creating a new DICOM radiotherapy standard dose storage instance, the sequence instance adopts the StudyRoot generation rule, and the sequence instance UID is the same as the CT examination, but is independently numbered;

[0019] S9-3, multiply the floating-point dose array by 1x10 6 Convert the integer to a float, set the dose grid scaling factor to 1x10 -6

[0020] S9-4, write pixel data layer by layer, tag (0028, 0030) directly copy CT pixel spacing, tag (0020, 0032) copy CT image patient position, tag (3004, 000C) set dose unit to Gy, tag (3004, 000E) set dose type = physical dose;

[0021] S9-5, generate five independent Rt dose files for each group of dose, so as to realize the preservation of total dose, hydrogen dose, nitrogen dose, boron dose and photon dose.

[0022] Preferably, the step S9-3 adopts a high-precision storage scheme combining Uint32 integer type storage and dose grid scaling factor, which is used to store the equivalent dose value after relative biological effectiveness weighting. The calculated total relative biological effectiveness dose and other component dose array are multiplied by a preset scaling coefficient (1x10 6 ), mapped to 32-bit unsigned integer space for storage, and the corresponding dose grid scaling factor (1x10 -6 ) is set. When reading, inverse operation is performed through the scaling factor, so as to restore the integer pixel value to the original floating-point equivalent dose value without loss, and the theoretical resolution is as high as 1 μGy (1x10 -6 Gy).

[0023] Preferably, the step S9-4 adds (3004, 0005) dose accumulation type = PLAN in the DICOM header, and adds (3006, 0020) reference RT plan sequence, which is left blank for linking the plan file by subsequent medical image processing software.

[0024] Preferably, the five Rt dose files generated in the step S9-5 all retain the same reference frame UID, and the medical image processing software can simultaneously display the total dose, boron dose, fast neutron dose, thermal neutron dose and gamma dose under the same reference frame, and support separate window width and window level adjustment.

[0025] Preferably, the Rt dose file making script is realized based on pydicom, and the entry parameters include the original CT directory, the five NPY dose paths and the output directory.

[0026] Preferably, the dose conversion coefficient in the step S6 includes fast neutron dose conversion factor (204 group), thermal neutron dose conversion factor (9 group), photon dose conversion factor (44 group) and boron dose conversion factor (498 group). ​

[0027] Preferably, the CT region mask in the step S7 is obtained by target region delineation and preprocessing, for distinguishing tumor regions, normal brain tissue regions and other tissue regions.

[0028] In the above technical solution, the present application provides technical effects and advantages:

[0029] The present application realizes the dose simulation DICOM radiotherapy standard dose file generation of boron neutron capture therapy BNCT by using the open source Monte Carlo simulation program OpenMC, and constructs a complete and accurate BNCT dose calculation and standardized output system.

[0030] By dividing the Hounsfield unit HU value of the patient CT into 35 material intervals, accurate geometry modeling of 1:1 at the voxel level is realized, and resampling errors are avoided.

[0031] By setting a multi-group super-thermal neutron surface source and superimposing a multi-dimensional flux counter, the neutron and photon flux energy spectrum is comprehensively counted, the fast neutron dose, the thermal neutron dose, the photon dose, the boron dose and the total dose are calculated based on the flux-dose conversion coefficient and the region-specific boron concentration weighting, and finally the total dose, the boron dose, the fast neutron dose, the thermal neutron dose and the gamma dose file can be directly imported into the medical image processing software through the five-channel dose matrix fusion and DICOM radiotherapy standard packaging, supporting multi-modal visualization and independent window width and window level adjustment, thereby realizing high-precision modeling, multi-component dose separation and clinical standardized output of BNCT dose calculation, and providing comprehensive and reliable dosimetric guarantee for BNCT dose simulation. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0033] Figure 1 The overall flowchart of the present application is shown in the figure.

[0034] Figure 2 The element model of the present application is shown in the figure.

[0035] Figure 3 The dose distribution diagram of the present application is shown in the figure. DETAILED DESCRIPTION

[0036] In order to make those skilled in the art better understand the technical solutions of the present application, the present application will be further described in detail with reference to the drawings.

[0037] The application provides a BNCT dose full-process automatic calculation output method based on OpenMC, as shown in the formula. Figures 1-2

[0038] Specifically, in step one, first, the CT image of the patient is read, and then the HU value at the voxel level is analyzed, the lower limit -1000 and the upper limit +1000 of the HU value are taken as the value range, the full HU domain is divided into 35 intervals, different material information is set for each HU interval, including material density and element composition percentage, that is, each interval is mapped to a unique material number and corresponding element composition and density, so as to generate an OpenMC material card.

[0039] In step two, the tumor region is assigned a specific material number 36, so as to ensure that the nuclear reaction cross section and energy deposition calculation of the tumor region in the Monte Carlo simulation are consistent with the benchmark research.

[0040] In step three, the CT voxel is directly converted into an OpenMC rectangular grid at a ratio of 1:1, without resampling, and a three-dimensional grid geometry is established.

[0041] In step four, a multi-group superhot neutron surface source is set on one side surface of the grid, and the incident intensity, energy and direction of the neutron source are defined.

[0042] As can be seen from the above, the technical process starts from the fine processing of the CT Hounsfield unit (HU) value, the value range of -1000 to +1000 is divided into 35 material intervals, and the material density and element composition are accurately set for each interval, laying an accurate physical foundation for the Monte Carlo simulation, and a three-dimensional geometric model completely consistent with the patient's anatomical structure is established through the 1:1 direct conversion of the CT voxel to the grid, without resampling, thereby eliminating the geometric error from the source.

[0043] In step five, a three-dimensional grid counter of 64*64*79 is superimposed on the same grid, and the total neutron flux, fast neutron flux, thermal neutron flux and photon flux are counted respectively, and the energy group flux matrix containing 498 energy groups, 204 group fast neutrons, 9 group thermal neutrons and 44 group photons is output.

[0044] ​In step six, the OpenMC state point file is read by the flux-dose conversion script to obtain the flux of each energy group, and multiplied by the preset dose conversion coefficient respectively to obtain the fast neutron dose, thermal neutron dose, photon dose and boron dose. Further, the dose conversion coefficient includes the fast neutron dose conversion factor (204 groups), the thermal neutron dose conversion factor (9 groups), the photon dose conversion factor (44 groups) and the boron dose conversion factor (498 groups).

[0045] In step seven, according to the different damage effects of boron dose on tumor and normal tissue, combined with the CT region mask to perform regional dose weighting: the tumor area composite biological effect CBE = 3.8, boron concentration 65ppm, normal brain tissue CBE = 1.35, boron concentration 18ppm, and other regions boron dose is zero. Further, the CT region mask is obtained by preprocessing, which is used to distinguish the tumor area, normal brain tissue area and other tissue area.

[0046] From the above, in the physical simulation stage, the flow is set by the multi-group super-thermal neutron surface source with reasonable energy spectrum and angle distribution, and the multi-dimensional grid counter covering multiple particle types, and the flux spectrum information of neutrons and photons is comprehensively obtained. In the subsequent dose calculation link, not only the preset conversion coefficient is used to convert the flux into physical dose, but also the key innovation based on biological effect is introduced. According to the CT region mask, different composite biological effect coefficients (CBE) and boron concentrations are used for weighting the boron dose of tumor and normal brain tissue area respectively, so that the calculation result is more in line with the treatment concept of BNCT.

[0047] In step eight, the fast neutron dose, thermal neutron dose, photon dose and boron dose are superimposed to calculate the total equivalent dose. Then, according to the voxel size (such as pixel spacing and layer thickness) and matrix dimension of the original CT image, the resolution of the dose matrix is improved to be completely consistent with the original CT image through interpolation algorithm, so as to ensure the accurate alignment of the two in spatial geometry;

[0048] In step nine, first, the DICOM sequence of the original CT is read to extract key metadata such as examination instance UID, reference frame UID, image patient position, image patient orientation, pixel spacing and layer thickness, to ensure that the dose file and the original image are completely consistent in geometry and reference coordinate system. Then, a new DICOM radiotherapy standard dose storage instance is created, the UID adopts the StudyRoot generation rule, the sequence instance UID belongs to the same examination as the CT, but is independently numbered to maintain data correlation and traceability; the high-precision storage scheme of Uint32 integer type combined with dose grid scaling factor is used to store the equivalent dose value weighted by relative biological effect, and the total relative biological effect dose and other component dose arrays calculated are multiplied by a preset scaling coefficient (1×10 6The dose grid is mapped to a 32-bit unsigned integer space for storage, and the corresponding dose grid scaling factor (1×10) is set. -6 During reading, the inverse operation is performed using the scaling factor, which allows the integer pixel value to be losslessly restored to the original floating-point equivalent dose value, with a theoretical resolution of up to 1 μGy (i.e., 1 × 10⁻⁶). -6 Next, pixel data is written layer by layer, and key labels are copied and set in the DICOM header file. Specifically, the pixel spacing of the CT (labels 0028, 0030) and the patient position in the image (labels 0020, 0032) are directly copied to maintain spatial consistency. The dose unit label (3004, 000C) is set to Gy (DoseUnits is Gy) and the dose type label (3004, 000E) is set to PHYSICAL (dose type = physical dose) to comply with clinical dose storage specifications. At the same time, (3004, 0005) dose accumulation type = PLAN and (3006, 0020) reference RT plan sequence are added (left blank for subsequent TPS linking to the plan file) to further improve the clinical usability of the dose file. Finally, independent Rt dose files are generated for each of the five dose matrices. All files retain the same UID, and independent Rt dose files are generated for each of the five dose groups. This allows for the preservation of total dose, hydrogen dose, nitrogen dose, boron dose, and photon dose, enabling medical image processing software to simultaneously display total dose, boron dose, fast neutron dose, thermal neutron dose, and gamma dose within the same reference frame. Individual window width and window level adjustments are also supported, facilitating subsequent retrieval, visualization, and analysis in dose analysis. Furthermore, the Rt dose generation script is implemented based on pydicom, with input parameters including the original CT directory, the five NPY dose paths, and the output directory. This means that the generated dose files can be directly imported into mature medical image processing software for visualization evaluation and treatment plan optimization.

[0049] Therefore, the final output of the process fully reflects its clinical practical value. By resampling the multi-component dose matrix to the same resolution as the original CT and generating a dose file that fully conforms to DICOM radiotherapy standards, it successfully achieves seamless integration with mature medical image processing software. The high-precision storage scheme combining Uint32 integer storage and dose grid scaling factor ensures 1×10⁻⁶ accuracy. -6 While achieving high resolution, storage efficiency has been optimized. More importantly, the five independent dose files generated by the process (total dose, fast neutron dose, thermal neutron dose, boron dose, and photon dose) all retain the same reference frame UID, providing a powerful tool for the accurate evaluation and optimization of BNCT treatment plans.

[0050] This technical process establishes a complete and accurate system for calculating and standardizing boron neutron capture therapy (BNCT) doses. Its core lies in achieving automated and high-fidelity conversion from patient CT images to clinically usable dose files.

[0051] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for automatic calculation and output of BNCT dose based on OpenMC, characterized in that, Includes the following steps: S1. Read the patient's CT image and analyze the voxel-level Heinz unit HU value. Divide the entire HU domain into 35 intervals. Each interval is mapped to a unique material number and the corresponding elemental composition and density. Generate the Monte Carlo program OpenMC material card. S2. Assign a specific material number of 36 to the tumor region to ensure that the nuclear reaction cross section and energy deposition calculation of the tumor region in the Monte Carlo simulation are consistent with the benchmark study. S3. Convert CT voxels directly into OpenMC rectangular grids at a 1:1 ratio, without resampling, and establish three-dimensional grid geometry; S4. Set up multiple groups of superheated neutron surface sources on one side of the grid and define the incident intensity, energy and direction of the neutron sources; S5. Superimpose a three-dimensional grid counter on the same grid, and count the total neutron flux, fast neutron flux, thermal neutron flux and photon flux respectively, and output the energy group flux matrix containing 498 energy groups, 204 fast neutron groups, 9 thermal neutron groups and 44 photon groups. S6. Read the OpenMC state point file through the flux-dose conversion script, obtain the flux of each energy group, and multiply it by the preset dose conversion coefficient to obtain the fast neutron dose, thermal neutron dose, photon dose and boron dose. S7. Based on the different damaging effects of boron dose on tumors and normal tissues, regional dose weighting is performed in combination with CT area mask: the composite biological effect (CBE) of the tumor area is 3.8, and the boron concentration is 65 ppm; the CBE of normal brain tissue is 1.35, and the boron concentration is 18 ppm; the boron dose in other areas is set to zero. S8. The four types of physical doses—fast neutron dose, thermal neutron dose, photon dose, and boron dose—are superimposed to calculate the total equivalent dose. Then, based on the voxel size and matrix dimension of the original CT image (including pixel spacing and slice thickness), the resolution of the dose matrix is ​​improved to be completely consistent with the original CT image through an interpolation algorithm to ensure that the two are precisely aligned in spatial geometry. S9. Call the Rt script to write the five sets of three-dimensional dose matrices obtained in S8 into the RT dose file that conforms to the DICOM radiotherapy standard. The five sets of three-dimensional dose matrices include total dose, fast neutron dose, thermal neutron dose, boron dose, and photon dose.

2. The method for automatic calculation and output of BNCT dose based on OpenMC according to claim 1, characterized in that, In step S9, the tumor region is assigned a specific material fraction as follows: H: 10.2%, C: 14.3%, N: 2.2%, O: 72.1%, other elements: 1.2%, and the material density is set to 1.06 g / cm³. 3 .

3. The method for automatic calculation and output of BNCT dose based on OpenMC according to claim 1, characterized in that, Step S9 specifically includes the following sub-steps: S9-1. Read the original DICOM sequence of the CT scan and extract the examination instance UID, reference frame UID, patient position in the image, patient orientation in the image, pixel spacing and slice thickness. S9-2. Create a new DICOM radiotherapy standard dose storage instance. The sequence instance is generated using the StudyRoot generation rule. The sequence instance UID belongs to the same examination as CT, but is numbered independently. S9-3, Multiply the floating-point dose array by 1×10 6 Convert to an integer and set the dose grid scaling factor to 1×10. -6 ; S9-4. Write pixel data layer by layer. Labels (0028, 0030) directly copy the CT pixel spacing. Labels (0020, 0032) copy the patient position in the CT image. Labels (3004, 000C) set the dose unit to Gy. Labels (3004, 000E) set the dose type to physical dose. S9-5. Generate independent Rt files for each of the five dose groups, thereby enabling the storage of total dose, hydrogen dose, nitrogen dose, boron dose, and photon dose.

4. The method for automatic calculation and output of BNCT dose based on OpenMC according to claim 3, characterized in that, Step S9-3 employs a high-precision storage scheme combining Uint32 integer storage with a dose grid scaling factor to store the equivalent dose value weighted by relative biological effects. The calculated total relative biological effect dose and other component dose arrays are multiplied by a preset scaling factor of 1×10. 6 The data is mapped to a 32-bit unsigned integer space for storage, and the corresponding dose grid scaling factor is set to 1×10. -6 During reading, the inverse operation is performed using the scaling factor, which allows the integer pixel value to be losslessly restored to the original floating-point equivalent dose value, with a theoretical resolution of up to 1 μGy, or 1 × 10⁻⁶. -6 Gy.

5. The method for automatic calculation and output of BNCT dose based on OpenMC according to claim 3, characterized in that, In step S9-4, add (3004, 0005) dose accumulation type = PLAN to the DICOM header, and add (3006, 0020) reference RT plan sequence, leaving it blank for subsequent medical image processing software to link the plan file.

6. The method for automatic calculation and output of BNCT dose based on OpenMC according to claim 3, characterized in that, The five-channel Rt files generated in step S9-5 all retain the same reference frame UID. The medical image processing software can simultaneously display the total dose, boron dose, fast neutron dose, thermal neutron dose, and gamma dose under the same reference frame, and supports individual window width and window level adjustment.

7. The method for automatic calculation and output of BNCT dose based on OpenMC according to claim 1, characterized in that, The Rt dose generation script is implemented based on pydicom, and the input parameters include the original CT directory, five NPY dose paths, and the output directory.

8. The method for automatic calculation and output of BNCT dose based on OpenMC according to claim 1, characterized in that, The dose conversion coefficients in step S6 include the fast neutron dose conversion factor, the thermal neutron dose conversion factor, the photon dose conversion factor, and the boron dose conversion factor.

9. The method for automatic calculation and output of BNCT dose based on OpenMC according to claim 1, characterized in that, The CT region mask in step S7 is obtained through target delineation and preprocessing, and is used to distinguish between tumor regions and normal brain tissue regions.

Images