A targeted isotope dosimetry method, apparatus, device, and medium

Abstract

The present application relates to the technical field of nuclear medicine, and discloses a kind of targeted isotope dose determination method, device, equipment and medium, obtain the PET three-dimensional image data and CT three-dimensional image data of target object;CT three-dimensional image data is mapped as non-uniform medium density distribution and material component index;After determining the activity value of each voxel in PET three-dimensional image data, a three-dimensional non-uniform source position probability weight model for Monte Carlo sampling is constructed;Call the decay skeleton diagram parameters in nuclear structure database, construct multi-particle composite source term;In the geometric space defined by non-uniform medium density distribution and material component index, generate three-dimensional dose rate distribution;Based on the preset pharmacokinetic parameters, three-dimensional dose rate distribution is processed, and the three-dimensional targeted isotope absorbed dose distribution of target object is obtained.The scheme realizes the leap from simple physical dose calculation to real biological effective dose evaluation, and improves the drug dose prediction precision.

Description

Technical Field

[0001] This invention belongs to the field of nuclear medicine technology, specifically relating to a method, apparatus, equipment, and medium for determining the dose of a targeted isotope. Background Technology

[0002] In targeted radioisotope therapy (such as , , In targeted therapy using monoclonal antibodies or peptides labeled with specific markers, accurate assessment of the absorbed dose to the tumor target area and organs at risk (OAR) is a core prerequisite for achieving precise and personalized treatment. However, current clinical and commercial treatment planning systems (TPS) heavily rely on the MIRD (Medical Internal Radiation Dose) paradigm or dose point kernel (DPK) convolutional algorithms (such as the Voxel S-value method) for calculating internal radiation dose.

[0003] The aforementioned conventional convolution kernel method suffers from a serious disconnect between theoretical dimensionality reduction and the physical mechanism: First, the convolution kernel algorithm is based on the assumption of translation invariance in an equivalent homogeneous medium of infinite water. However, when dealing with complex, non-homogeneous tissue interfaces in the human body (such as the lung-soft tissue-bone marrow interface), electrons (… The approximate path of continuous deceleration of particles and gamma ( The Compton scattering free path of X-rays undergoes severe nonlinear truncation, resulting in serious topological distortion in dose field calculations at tissue boundaries, which cannot meet the microdosimetric requirements for bone metastases or micrometastases.

[0004] Secondly, existing methods for constructing radioactive sources fail to achieve strong coupling with the decay scheme at the core of nuclear physics. They typically simplify the multiple energy spectra in cascade decays to a single average energy, neglecting the fact that... Particles (extremely high LET, range in the micrometer range) Particles (medium LET, millimeter range) and Rays (low LET, full-body cross-penetration) exhibit distinctly different differential cross sections and depositional topological features in the medium.

[0005] Finally, traditional static dose assessment completely decouples physical radiation transport from the biodistribution and metabolism of drugs, resulting in the derivation from the initial dose rate field to the cumulative total absorbed dose being merely a crude single-exponential decay scalar multiplication. Currently, the industry urgently needs a targeted isotope dose determination scheme. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention aims to provide a method, apparatus, device, and medium for determining the dose of a targeted isotope.

[0007] According to one aspect of this application, a method for determining a targeted isotope dose is disclosed, the method comprising: Acquire PET and CT 3D image data of the target object; The CT three-dimensional image data is mapped to a non-uniform medium density distribution and material composition index; The activity value of each voxel in the PET three-dimensional image data is determined, and a three-dimensional non-uniform source location probability weight model for Monte Carlo sampling is constructed based on the activity value of each voxel. By calling decay pattern parameters from the nuclear structure database, a decay pattern containing alpha particles and beta particles is constructed. - A multi-particle composite source term for particles and gamma rays, wherein the multi-particle composite source term integrates voxel activity weights and particle energy probability density distributions, and the voxel activity weights are obtained based on the normalization of the voxel activity distributions; Within the geometric space defined by the non-uniform medium density distribution and material composition index, for each sampled particle obtained from the non-uniform source location probability weight model, initial energy sampling is performed based on the particle energy probability density distribution corresponding to the sampled particle. The medium physical parameters corresponding to the current space are called according to the material composition index, the particle trajectory is tracked and energy deposition is accumulated to generate a three-dimensional dose rate distribution. Based on preset pharmacokinetic parameters, the three-dimensional dose rate distribution is processed to obtain the three-dimensional targeted isotope absorbed dose distribution of the target object.

[0008] The step of mapping the CT three-dimensional image data into a non-uniform medium density distribution and material composition index includes: Based on the mapping relationship between Huntsfield units and chemometrics school quasi-functionalities, the CT three-dimensional image data is converted voxel by voxel; The Huntsfield unit values ​​of each voxel in the CT three-dimensional image data are converted into electron density tensors through the first mapping relationship. Through the second mapping relationship, the Huntsfield unit values ​​of each voxel in the CT three-dimensional image data are converted into material composition index tensors. The material composition index tensors are used to distinguish between bone, soft tissue, lung and cavity regions. The non-uniform medium density distribution includes the electron density tensor. The first mapping relationship is as follows:

[0009] In the formula, In Huntsfield units, The conversion function, determined based on a chemometrics school lookup table, is used to map HU values ​​to the electron density of the corresponding tissue at a preset reference energy. , Preset reference energy, preset reference energy Determined based on the decay characteristics of the targeted isotope; The second mapping relationship is:

[0010] In the formula, Spatial location The index value at that location, Operators for finding the minimum parameter, For spatial location The Huntsfield unit value of the CT image corresponding to the voxel at that location. Let be the preset HU threshold for the i-th index, where i is the index independent variable.

[0011] Determining the activity value of each voxel in the PET 3D image data, and constructing a 3D non-uniform source location probability weighting model for Monte Carlo sampling based on the activity value of each voxel, includes: Extract the standard uptake values ​​of each voxel from the PET 3D image data; Obtain the total injection activity, weight, and preset cross-calibration coefficient of the target subject; Based on the total injected activity, the target subject's weight, the preset cross-calibration coefficient, and the standard intake value of each voxel, the activity value of each voxel is determined in conjunction with the first formula. The activity values ​​of all voxels are normalized in three-dimensional space to obtain the probability density function of the non-uniform source location in three-dimensional space. The probability density function of the three-dimensional non-uniform source location is used as the three-dimensional non-uniform source location probability weighting model for the Monte Carlo sampling. The first formula is: ; In the formula, Spatial location Location, Time The initial activity value at that time, This represents the standard intake value for voxels. Total injectable activity, For the target subject's weight, Preset cross-calibration coefficients; The probability density function of the non-uniform source location in three-dimensional space is: ; In the formula, Let be the probability density function of the non-uniform source location in three-dimensional space. The denominator is the total activity value of the three-dimensional space of the target object.

[0012] The decay profile parameters in the nuclear structure database are called to construct a structure containing alpha particles and beta particles. - Multi-particle composite source terms for particles and gamma rays include: Access a pre-defined nuclear structure database to extract decay pattern parameters of the target isotope and its daughter nuclei. These decay pattern parameters include at least: α decay energy, β decay energy, and so on. - Maximum decay energy, gamma-ray energy, branching ratio of each decay mode, alpha particle yield, and beta particle yield. - Particle production, gamma-ray production; Based on the decay framework parameters, α particles and β particles are constructed respectively. - Single-particle energy probability density distribution of particles and gamma rays , and ,in, Constructed based on discrete energy level transitions Based on Fermi theory Constructed based on characteristic spectral lines; Based on the aforementioned branch ratio, Output Output The output is obtained by weighting and combining the single-particle energy probability density distribution to generate the full-particle energy probability density distribution of the multi-particle composite source term. The multi-particle composite source term is constructed by fusing the full-particle energy probability density distribution with the voxel activity weights. The multi-particle composite source term is represented as follows: ; In the formula, For multi-particle composite source terms, it represents the spatial location. The energy is The particle emission probability density; Spatial location Voxel activity weights at locations; For the first Types of radiation patterns (including) particle, Particles and The output or branching ratio of (rays); For the first The single-particle energy probability density distribution of a certain type of particle.

[0013] Within the geometric space defined by the non-uniform medium density distribution and material composition index, for each sampled particle obtained from the non-uniform source location probability weight model, initial energy sampling is performed based on the particle energy probability density distribution corresponding to the sampled particle. Then, according to the material composition index, the corresponding medium physical parameters at the current space are called, particle trajectories are tracked, and energy deposition is accumulated to generate a three-dimensional dose rate distribution, including: Based on the particle energy probability density distribution corresponding to the sampled particle, the initial kinetic energy of the particle is determined by combining the inverse transformation sampling method. Based on the material composition index at the current space location, the corresponding medium physical parameters are retrieved and called from a pre-set physical parameter database. These medium physical parameters include at least mass density, electron density, atomic number, photoelectric effect cross section / Compton scattering cross section / pair effect cross section for gamma rays, and cross section for alpha particles and beta particles. - Data on the particle's stopping power; Solve the linear Boltzmann transport equation for gamma rays; Targeting alpha particles and beta particles - For charged particles, solve the Fokker-Planck collision evolution equations based on the continuous deceleration approximation; When a particle crosses a medium interface defined by different material composition indices, it is determined whether the remaining range of the current particle is less than the current voxel boundary distance; if so, the step size is reduced to the medium interface, and the material composition index and the corresponding medium physical parameters are updated after crossing the interface. The tracked particle energy deposition values ​​are accumulated into the corresponding voxel score array, and the particle flux and deposition energy in each voxel are statistically analyzed to generate the three-dimensional dose rate distribution. The linear Boltzmann transport equation is as follows: ; In the formula, Spatial location Particle emission energy and direction of movement Particle angular flux density at that location, Let E be the spatial position and E be the particle's emission energy. The unit vector in the direction of motion; The directional derivative characterizes the spatial migration and leakage terms of particles. This represents the overall macroscopic cross-section of the corresponding medium; It is a multi-particle composite source term distribution; The macroscopic differential scattering cross section represents the probability of a particle being scattered. The incident energy of the particle; The direction of motion of the incident particle. Spatial location Particle incident energy and direction of movement angular flux density of particles at that location; The Fock-Planck collision evolution equation is: ; In the formula, The stopping power of charged particles in the current medium characterizes the energy loss rate under the continuous deceleration approximation. It is the energy differential operator; The momentum partial differential term, used to characterize multiple Coulomb scattering, is used to describe the angular deflection of a charged particle's trajectory.

[0014] Based on preset pharmacokinetic parameters, the three-dimensional dose rate distribution is processed to obtain the three-dimensional targeted isotope absorbed dose distribution of the target object, including: Based on preset pharmacokinetic parameters, the biological metabolic constant and physical decay constant of the target isotope in the target voxel are obtained. Based on the aforementioned biological metabolic constants and physical decay constants, the effective decay constant of the target isotope within the target voxel is determined. Based on the effective decay constant, a time-activity retention functional for the target voxel is constructed. The physical dose rate tensor obtained from Monte Carlo simulation is integrated with the time activity retention functional of the target voxel in the time domain to obtain the three-dimensional target isotope absorbed dose distribution of the target object. The integral formula is as follows: ; In the formula, For the target voxel in space The time-activity retention functional (time integral activity) at a given location. For time The activity concentration of the target isotope within the target voxel; For integration time; ; In the formula, For the target object in space Three-dimensional targeted isotope cumulative absorbed dose distribution at the location; This is the physical dose rate tensor of the unit activity of the corresponding voxel generated by Monte Carlo simulation.

[0015] The process of tracking particle trajectories and accumulating energy deposition includes: During Monte Carlo particle transport, real-time monitoring of abrupt changes in energy deposition of charged particles at voxel boundaries is conducted. When a discontinuity in energy deposition is detected due to abrupt changes in the material composition index, a boundary energy deposition smoothing algorithm is activated to subdivide the particle step size across different medium interfaces.

[0016] According to another aspect of this application, a targeted isotope dose determination device is also disclosed, the device comprising: The 3D image data acquisition module is used to acquire PET 3D image data and CT 3D image data of the target object; The mapping module is used to map the CT three-dimensional image data into a non-uniform medium density distribution and material composition index; A three-dimensional non-uniform source location probability weight model construction module is used to determine the activity value of each voxel in the PET three-dimensional image data, and construct a three-dimensional non-uniform source location probability weight model for Monte Carlo sampling based on the activity value of each voxel. The multi-particle composite source term invocation module is used to invoke decay pattern parameters from the nuclear structure database to construct a composite source term containing alpha particles and beta particles. - A multi-particle composite source term for particles and gamma rays, wherein the multi-particle composite source term integrates voxel activity weights and particle energy probability density distributions, and the voxel activity weights are obtained based on the normalization of the voxel activity distributions; The three-dimensional dose rate distribution generation module is used to generate a three-dimensional dose rate distribution for each sampled particle sampled from the non-uniform source location probability weight model within the geometric space defined by the non-uniform medium density distribution and material composition index. It performs initial energy sampling based on the particle energy probability density distribution corresponding to the sampled particle, calls the medium physical parameters corresponding to the current space according to the material composition index, tracks the particle trajectory and accumulates energy deposition. The absorbed dose distribution determination module is used to process the three-dimensional dose rate distribution based on preset pharmacokinetic parameters to obtain the three-dimensional targeted isotope absorbed dose distribution of the target object.

[0017] According to another aspect of this application, an electronic device is also disclosed, the electronic device including a memory and at least one processor, the memory storing instructions; the at least one processor invokes the instructions in the memory to cause the electronic device to perform the steps of the targeted isotope dose determination method as described in any of the preceding claims.

[0018] According to another aspect of this application, a computer-readable storage medium is also disclosed, wherein instructions are stored on the computer-readable storage medium, characterized in that, when executed by a processor, the instructions implement the various steps of the targeted isotope dose determination method as described in any of the preceding claims.

[0019] This invention includes, but is not limited to, the following beneficial effects: This solution maps CT image data to a non-uniform medium density distribution and material composition index, and combines it with PET image data to construct a three-dimensional non-uniform source location probability weight model, thereby realizing the restoration of the target object's internal anatomical structure and the true spatial distribution of radiopharmaceuticals. Furthermore, it utilizes decay pattern parameters from a nuclear structure database to construct a model including alpha particles, beta particles, and other decay patterns. - This method employs a multi-particle composite source term for particles and gamma rays, and performs dual-mode particle transport calculations in heterogeneous media within a defined geometric space. It dynamically invokes corresponding medium physical parameters based on material composition indices to track particle trajectories and accumulate energy deposition, thereby improving the physical realism of dose calculations. A first mapping function converts HU values ​​into electron density tensors, enhancing the accuracy of the electron density distribution relied upon for energy deposition calculations in particle transport. Furthermore, this reference energy is adaptively set based on the decay characteristics of the target isotope, enhancing the method's universality and specificity. Secondly, a second mapping relationship converts HU values ​​into material composition index tensors. A preset HU threshold is used to classify voxels by tissue type (e.g., bone, soft tissue, lungs, and cavities), providing material boundary conditions for subsequent simulations of particle physical behavior in different media. This dual-mapping strategy not only preserves the anatomical details of CT images but also endows them with physically computable properties, enabling Monte Carlo simulations to realistically reflect particle transport processes in heterogeneous environments within the target body, thus improving the physical fidelity and clinical applicability of dose calculations. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0021] Figure 1 This is a flowchart of the targeted isotope dose determination method according to an embodiment of this application; Figure 2 This is a structural block diagram of the targeted isotope dose determination device according to an embodiment of this application; Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application; In the diagram, 300 is the electronic device, 310 is the processor, 320 is the memory, 330 is the storage medium, 340 is the power supply, 350 is the wired or wireless network interface, 360 is the input / output interface, 331 is the operating system, 332 is the data, and 333 is the application program. Detailed Implementation

[0022] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] For ease of understanding, the specific process of the embodiments of the present invention will be described below. Figure 1 This is a flowchart of the targeted isotope dose determination method according to an embodiment of this application. (See attached document.) Figure 1 It includes the following steps: S101. Obtain PET 3D image data and CT 3D image data of the target object.

[0024] Specifically, PET (Positron Emission Tomography) and CT (Computed Tomography) 3D image data of the target object can be acquired within the same scanning time period. The CT 3D image data is reconstructed from the attenuated signals acquired after the X-ray tube rotates around the target object, possessing excellent spatial and density resolution. It is used to accurately delineate the anatomical structures within the target object (such as bones, soft tissues, lungs, and hollow organs), and stores the linear attenuation coefficients of each voxel in the form of a Hounsfield Unit (HU) matrix, providing a physical basis for subsequent construction of non-uniform medium density distribution and material composition indexing. The PET 3D image data is reconstructed by detecting the positron annihilation photon pairs emitted by the radioactive tracer injected into the target object, reflecting the dynamic distribution characteristics of the radiopharmaceutical within the organism. After attenuation and scattering correction, the functional metabolic activity of each voxel is characterized in the form of a Standard Uptake Value (SUV) matrix, providing quantitative activity distribution information for subsequent construction of a 3D non-uniform source location probability weight model. To ensure spatial consistency of multimodal data, PET and CT data can be registered and fused. Rigid or nonlinear transformations can be used to eliminate image misalignment caused by differences in scanning position or organ peristalsis, generating dual-modal three-dimensional volume data that are perfectly aligned in the spatial coordinate system. This ensures accurate mapping of anatomical structures and functional distributions in subsequent Monte Carlo simulations, laying a solid data foundation for high-precision dose calculation.

[0025] S102. Map CT three-dimensional image data into a non-uniform medium density distribution and material composition index.

[0026] In one embodiment, step S102 may specifically include: Step 2.1: Based on the mapping relationship between Huntsfield units and chemometric quasi-functional, voxel-by-voxel conversion is performed on the CT three-dimensional image data.

[0027] Specifically, the system first reads the registered and aligned CT 3D image data from step S101 and extracts the Huntsfield unit value HU(r) corresponding to each spatial voxel. Since the HU value is essentially a relative measure of the degree of X-ray absorption in tissue and is not directly equivalent to the electron density or material composition required for particle transport calculations, a stoichiometric calibrated and theoretically fitted quasi-functional is introduced as a conversion bridge. This functional is constructed based on measurement data of tissue equivalent phantoms with various known chemical compositions (such as polystyrene, polymethyl methacrylate, and hydroxyapatite) under standard CT scanning conditions, establishing a nonlinear mapping relationship between the HU value and physical density and atomic number. Through this functional, the system performs point-by-point conversion on each voxel, thereby transforming the original CT grayscale values ​​into a parameter field with clear physical meaning, providing the basic input for subsequent dose calculations.

[0028] Step 2.2: Convert the Huntsfield unit values ​​of each voxel in the CT three-dimensional image data into electronic density tensors through the first mapping relationship.

[0029] Step 2.3: Through the second mapping relationship, convert the Huntsfield unit values ​​of each voxel in the CT three-dimensional image data into material composition index tensors. The material composition index tensors are used to distinguish between bone, soft tissue, lung and cavity regions. The non-uniform medium density distribution includes the electron density tensor.

[0030] The first mapping relationship is as follows: ; In the formula, In Huntsfield units, The conversion function, determined based on a chemometrics school lookup table, is used to map HU values ​​to the electron density of the corresponding tissue at a preset reference energy. , Preset reference energy, preset reference energy Determined based on the decay characteristics of the targeted isotope.

[0031] The second mapping relationship is: ; In the formula, Spatial location The index value at that location, Operators for finding the minimum parameter, For spatial location The Huntsfield unit value of the CT image corresponding to the voxel at that location. Let be the preset HU threshold for the i-th index, where i is the index independent variable.

[0032] Specifically, in the subsequent Monte Carlo simulations, This will serve as an identifier for interface boundary conditions. For example, =1 represents the skeleton. =2 represents soft tissue. =3 represents the lungs. =4 represents a cavity. It tells the simulation program which medium the particle is in, so that it can call the corresponding physical parameters (such as density and stopping power) to perform transport calculations.

[0033] In this formula, argmin is used to iterate through all preset HU thresholds and find the index i corresponding to the threshold that is closest to the current voxel HU value. For example, if there are 5 preset thresholds (corresponding to 5 materials), argmin i It will return the number with the smallest difference among these 5 indices (0 to 4 or 1 to 5), thus completing the automatic classification. The system uses pre-defined HU (Heat) thresholds to distinguish different tissues. For example, the system might pre-define: less than -800 for air, -800 to -400 for lungs, -400 to 100 for soft tissue, and greater than 100 for bone. The formula achieves automated tissue segmentation by finding the closest threshold.

[0034] Understandably, through mapping, the system discretizes the continuous HU value domain into several material categories with clear physical meanings, such as air, lungs, soft tissue, and bone, and assigns each voxel an integer index value, forming a material composition index tensor. This index tensor is used to quickly retrieve the physical parameters of the corresponding material in Monte Carlo simulations, such as atomic number, mass density, and stopping power, and supports intelligent judgment and step size control of particle transport at heterogeneous interfaces. Simultaneously, this index tensor, together with the electron density tensor obtained in step 2.2, constitutes a complete characterization of the density distribution of the non-uniform medium, providing a structured and parameterized physical medium input for subsequent dose calculations.

[0035] S103. Determine the activity value of each voxel in the PET three-dimensional image data, and construct a three-dimensional non-uniform source location probability weight model for Monte Carlo sampling based on the activity value of each voxel.

[0036] Specifically, in one example, step S103 may include the following steps: Step 3.1: Extract the standard uptake values ​​of each voxel from the PET 3D image data.

[0037] Specifically, the PET 3D image data, already registered and aligned with CT in step S101, is first read, and the Standard Uptake Value (SUV) corresponding to each spatial voxel is extracted. The SUV value is a quantitative parameter obtained by performing attenuation correction, scattering correction, normalization correction, and possible partial volume effect compensation on the raw PET count data. Its physical meaning is the standardized expression of the concentration of radiotracer per unit mass of tissue relative to the injected dose, usually expressed in g / mL or %ID / g. In medical image analysis, the SUV value is widely used to characterize tissue metabolic activity, especially in oncology as an important indicator for differentiating between benign and malignant tissues.

[0038] Step 3.2: Obtain the total injection activity, weight, and preset cross-calibration coefficient of the target subject.

[0039] Total activity (TURP) refers to the total radioactivity of the radiopharmaceutical injected intravenously into the target subject during clinical procedures. It is usually expressed in MBq or mCi and must be accurately recorded at the time of injection. The target subject's weight is used to standardize the SUV value to eliminate the influence of individual body size differences on the activity distribution, and is usually expressed in kg. The preset cross-calibration coefficient is a correction factor used to convert the SUV value (dimensionless or relative unit) into an absolute activity value (such as Bq or MBq). Its value can be preset or dynamically adjusted according to the pharmacokinetic model of the tracer used, the scanning protocol, and the equipment calibration status.

[0040] Step 3.3: Based on the total injection activity, the target subject's weight, the preset cross-calibration coefficient, and the standard intake value of each voxel, and in conjunction with the first formula, determine the activity value of each voxel.

[0041] The first formula is: ; In the formula, Spatial location Location, Time The initial activity value at that time, This represents the standard intake value for voxels. Total injectable activity, For the target subject's weight, These are the preset cross-calibration coefficients.

[0042] For example, the Standardized Intake Value (SUV) can include the maximum standardized intake value, the average standardized intake value, and the peak standardized intake value. The Standardized Intake Value (SUV) can be a semi-quantitative indicator calculated by delineating a region of interest (ROI) and combining it with the clinical parameters of the target subject. For example, injecting a radioactive tracer (such as...) into the target subject... 18 Following F-FDG, after a specific imaging time (e.g., 60 minutes), a PET / CT scan is performed. Subsequently, attenuation correction (AC) is applied to the PET image using density information from the CT image to eliminate the absorption of radiation by tissues, resulting in a radioactivity map. Further, on the PET image, a region is delineated by the physician or software to assess the lesion's uptake. Common measurement methods include: 2D planar ROI: A two-dimensional region is manually or semi-automatically drawn on a single axial slice. 3D volumetric ROI (VOI): The entire lesion's three-dimensional metabolic volume is automatically or semi-automatically delineated based on algorithms, providing a more comprehensive reflection of the lesion's true metabolic status. Further, within the selected ROI / VOI, the system extracts the following data for calculation: SUV max (Maximum Standard Intake Value): The highest radioactivity concentration value of a single pixel within the ROI.

[0043] SUV mean (Average Standard Uptake): The average of the radioactivity concentration of all pixels within the ROI. It reflects the overall uptake, but is greatly affected by the size and location of the ROI.

[0044] SUV peak (Peak Standard Intake): In a fixed small volume (e.g., 1 cm³) 3 The highest average value within the spherical region.

[0045] After determining the standard uptake value for each voxel, the initial activity value is obtained by substituting it into the first formula.

[0046] Step 3.4: Normalize the activity values ​​of all voxels in three-dimensional space to obtain the probability density function of the non-uniform source location in three-dimensional space.

[0047] The probability density function of the non-uniform source location in three-dimensional space is: ; In the formula, The denominator is the total activity value of the three-dimensional space of the target object.

[0048] Step 3.5: Use the probability density function of the three-dimensional non-uniform source location as the three-dimensional non-uniform source location probability weighting model for Monte Carlo sampling.

[0049] Specifically, this step uses the probability density function obtained above as the sampling weight model for generating source particle positions in the Monte Carlo simulation. During Monte Carlo transport, the system no longer employs uniform random sampling or simple grid partitioning. Instead, it implements weighted random sampling (such as accept-rejection, importance sampling, or stratified sampling) in three-dimensional space based on the magnitude of the probability density function. For example, in high-activity regions (such as the tumor core), the probability density function value is larger, and the sampling probability is correspondingly higher; while in low-activity or background regions, the probability density function value is smaller, and the sampling probability is lower. This adaptive sampling strategy based on source strength distribution can reduce the number of invalid samples, improve computational efficiency, and effectively reflect the true spatial distribution characteristics of radiopharmaceuticals in vivo.

[0050] S104. Call the decay pattern parameters from the nuclear structure database to construct a model containing alpha particles and beta particles. - Multi-particle composite source term of particles and gamma rays.

[0051] Among them, the multi-particle composite source term integrates the voxel activity weight and the particle energy probability density distribution, and the voxel activity weight is obtained based on the normalization of the voxel activity distribution.

[0052] Specifically, in one example, step S104 may include the following steps: Step 4.1: Access the pre-set nuclear structure database and extract the decay profile parameters of the target isotope and its daughter nuclei. The decay profile parameters should include at least the α decay energy. β - Maximum decay energy γ-ray energy and the branching ratio of each decay mode. Alpha particle production β - Particle production gamma ray production .

[0053] Step 4.2: Based on the decay framework parameters, construct α particles and β particles respectively. - Single-particle energy probability density distribution of particles and gamma rays , and , among which, among which, Constructed based on discrete energy level transitions Based on Fermi theory Constructed based on characteristic spectral lines.

[0054] For example, for an alpha particle, because its energy is discrete during decay (originating from internuclear energy level transitions), the single-particle energy probability density distribution... The discrete energy level transition is used for construction, that is, a Dirac delta function or a narrow Gaussian peak is set at the energy of the j-th α decay level, with the weight being the branch ratio b₁ of the corresponding energy level. This can be expressed as: ; In the formula, Let be the single-particle energy probability density distribution of the alpha particle. The decay branching ratio corresponding to the j-th energy level. For Dirac delta function, Let be the energy value corresponding to the j-th α decay level; For β - The particle, whose energy spectrum is continuously distributed and follows Fermi theory, has P_β(E) constructed as a continuous probability density function from 0 to E_β_max, typically including shape factor corrections (such as Coulomb field effects), and its form can be simplified to: ; in, For β - The single-particle energy probability density distribution of a particle, where C is the normalization constant. For β - The maximum energy of decay, F(Z, E) is the Fermi function, and Z is the atomic number of the daughter nucleus; For gamma rays, their energy exhibits discrete characteristics (originating from the de-excitation of nuclear excited states). P_γ(E) is constructed based on a characteristic spectral line model, which sets a narrow peak (such as a Lorentzian or Gaussian line) at E_γ,k, with the weight being the branching ratio b corresponding to the gamma transition. k It can be represented in the following form: ; in, The natural width of the γ transition. Let ∑ be the single-particle energy probability density distribution of gamma rays. k For the summation of the energy levels of the k-th γ transition, b k The branch ratio for the k-th transition. It is a Lorentz line type. Let be the center energy of the k-th γ transition. Step 4.3: Based on branch ratio Output Output Output The energy probability density distribution of single particles is weighted and combined to generate the full particle energy probability density distribution of the multi-particle composite source term. .

[0055] Step 4.4: Distribute the energy probability density of all particles By integrating with voxel activity weights, a multi-particle composite source term is constructed.

[0056] The multi-particle composite source term is represented as: ; In the formula, For multi-particle composite source terms, it represents the spatial location. The energy is The particle emission probability density; Spatial location Voxel activity weights at locations; For the first Types of radiation patterns (including) particle, Particles and The output or branching ratio of (rays); For the first The single-particle energy probability density distribution of a certain type of particle.

[0057] S105. Within the geometric space defined by the non-uniform medium density distribution and material composition index, for each sampled particle obtained from the non-uniform source location probability weight model, initial energy sampling is performed based on the particle energy probability density distribution corresponding to the sampled particle. The medium physical parameters at the current space are called according to the material composition index, the particle trajectory is tracked and energy deposition is accumulated to generate a three-dimensional dose rate distribution.

[0058] Specifically, in one example, step S105 may include the following steps: Step 5.1: Based on the particle energy probability density distribution corresponding to the sampled particle, determine the initial kinetic energy of the particle using the inverse transformation sampling method.

[0059] For example, the emission position of a particle can be extracted at a spatial location based on the source term weights. Then, based on the probability density distribution of the total particle energy, the initial kinetic energy of the particle can be determined using inverse transform sampling. The core of this method lies in obtaining the cumulative distribution function by integrating the probability density function.

[0060] Step 5.2: Based on the material composition index at the current space location, retrieve and call the medium physical parameters corresponding to the current medium from the preset physical parameter database. The medium physical parameters include at least the mass density, electron density, atomic number, photoelectric effect cross section / Compton scattering cross section / electron pair effect cross section for γ rays, and stopping power data for α particles and β− particles.

[0061] Step 5.3: Solve the linear Boltzmann transport equation for γ-rays.

[0062] Step 5.4: Targeting alpha particles and beta particles - For charged particles, solve the Fokker-Planck collision evolution equations based on the continuous deceleration approximation.

[0063] Step 5.5: When a particle crosses the interface of a medium defined by different material composition indices, determine whether the remaining range of the current particle is less than the current voxel boundary distance; if so, reduce the step size to the medium interface and update the material composition index and the corresponding medium physical parameters after crossing the interface.

[0064] Step 5.6: Accumulate the tracked particle energy deposition values ​​into the corresponding voxel score array, and calculate the particle flux and deposition energy in each voxel to generate a three-dimensional dose rate distribution.

[0065] The linear Boltzmann transport equation is as follows: ; In the formula, Spatial location Particle emission energy and direction of movement Particle angular flux density at that location, Let E be the spatial position and E be the particle's emission energy. The unit vector in the direction of motion; The directional derivative characterizes the spatial migration and leakage terms of particles. The macroscopic total cross-section of the corresponding medium represents its spatial location. At energy E, the total probability of any interaction between a particle and a medium (including absorption, scattering, etc.); The distribution of multi-particle composite source terms represents the spatial location. Particle emission energy E, direction of motion The number of particles generated per unit time; The macroscopic differential scattering cross section represents the probability of a particle being scattered. The incident energy of the particle; The direction of motion of the incident particle. Spatial location Particle incident energy and direction of movement angular flux density of particles at that location; The Fock-Planck collision evolution equation is: ; In the formula, The stopping power of charged particles in the current medium characterizes the energy loss rate under the continuous deceleration approximation. It is the energy differential operator; The momentum partial differential term, used to characterize multiple Coulomb scattering, is used to describe the angular deflection of a charged particle's trajectory.

[0066] For example, tracking particle trajectories and accumulating energy deposition includes: During Monte Carlo particle transport, real-time monitoring of abrupt changes in energy deposition of charged particles at voxel boundaries is conducted. When a discontinuity in energy deposition is detected due to abrupt changes in the material composition index, a boundary energy deposition smoothing algorithm is activated to subdivide the particle step size across different medium interfaces.

[0067] S106. Based on preset pharmacokinetic parameters, the three-dimensional dose rate distribution is processed to obtain the three-dimensional targeted isotope absorbed dose distribution of the target object.

[0068] Specifically, in one example, step S106 may include the following steps: Step 6.1: Based on the preset pharmacokinetic parameters, obtain the biological metabolic constant and physical decay constant of the target isotope in the target voxel.

[0069] Specifically, the system first reads the parameters of a preset pharmacokinetic model, which is usually established based on clinical studies or population pharmacokinetic data to describe the absorption, distribution, metabolism, and excretion (ADME) process of radiopharmaceuticals in the target body. The system then extracts the biometabolic constants of the target isotope within the target voxel (i.e., the three-dimensional units with specific material composition and physical density divided by CT images in step S102), such as: the absorption rate constant ka (characterizing the rate at which the drug enters the bloodstream from the injection site); the distribution volume Vd (characterizing the apparent volume of the drug distributed in the body); the clearance rate constant ke (characterizing the rate at which the drug is eliminated from the body); and possible tissue-specific retention constants (such as the uptake half-life T1 / 2,tumor of tumor tissue). Simultaneously, the system obtains the physical decay constant λphys of the target isotope from a nuclear structure database or a preset decay model, defined as λphys = ln(2) / T1 / 2,phys, where T1 / 2,phys is the physical half-life of the isotope (e.g., the uptake half-life of the tumor tissue). 177 Lu's T1 / 2, phys ≈ 6.65 days. These parameters together form the basic inputs of the pharmacokinetic-physical decay coupling model.

[0070] Step 6.2: Based on biological metabolic constants and physical decay constants, determine the effective decay constant of the target isotope within the target voxel.

[0071] Specifically, biological metabolic processes can be coupled with physical decay processes to calculate the effective decay constant of the target isotope within the target voxel.

[0072] Step 6.3: Construct the time-activity retention functional of the target voxel based on the effective decay constant.

[0073] Step 6.4: Integrate the physical dose rate tensor obtained from the Monte Carlo simulation with the target voxel time-activity retention functional in the time domain to obtain the three-dimensional target isotope absorbed dose distribution of the target object. The integral formula is as follows: ; In the formula, For the target voxel in space The time-activity retention functional (time integral activity) at a given location. For time The activity concentration of the target isotope within the target voxel; For integration time; ; In the formula, For the target object in space Three-dimensional targeted isotope cumulative absorbed dose distribution at the location; This is the physical dose rate tensor of the unit activity of the corresponding voxel generated by Monte Carlo simulation.

[0074] For ease of understanding, the following is a specific example of this solution: This section delves into the underlying algorithmic technology, focusing on the core concept of comprehensively managing the three-dimensional activity distribution of PET, reconstructing decay framework diagrams from multiple particle sources, particle transport in heterogeneous media, and integrating pharmacokinetics, and breaking down the process. This embodiment relies on a heterogeneous server with extremely high concurrent tensor computing capabilities (such as the CUDA architecture).

[0075] 1. Initial conditions and boundary modeling of multiphysics non-uniform mesh established. Before launching the joint dose simulation system, the absolute physical boundary (CT) and activity topological boundary (PET) of the target object must be established in a unified world coordinate system.

[0076] Anatomical geometry and physical section boundary establishment: Reading the 3D CT image of the target object, and through a rigorous mapping from the Hounsfield Unit (HU) to a stoichiometric calibration functional, generating two high-resolution 3D discrete tensors: the tissue electron density tensor. and material composition index tensor In subsequent Monte Carlo transport, these two tensors constitute the particle attenuation and scattering response cross section. The absolute physical boundary.

[0077] Three-dimensional activity topological boundary established: reading the injected targeted radiopharmaceutical (e.g.) ) Specific time point Three-dimensional PET images. Standard uptake values ​​for each voxel were extracted. Based on total injected activity target subject weight and cross-calibration coefficients Establish the absolute initial activity value in three-dimensional space: ; The initial activity value directly defines the non-uniform spatial probability density function of the radioactive source in the three-dimensional target manifold. : ; 2. Core Mathematics and Partial Differential Derivation of Microscopic Transport and Pharmacokinetic Multi-Field Coupling Unlike external radiation therapy, internal radiation dosimetry is a typical isotropic endogenous mixed radiation field, and its partial differential operators contain extremely strong energy, particle type and time coupling correlations.

[0078] (I) Derivation of the multi-particle generation operator based on decay framework diagram Using nuclear structure evaluation data (ENSDF), a structure containing nuclear structure evaluation data (ENSDF) was established for the target isotope. The generation matrix of a type of radiative particle. Define the first... Types of radiation (such as) Continuous spectrum Characteristic peaks, The branching ratio of the characteristic spectrum is The energy probability density distribution functional is .

[0079] The Monte Carlo full-phase space composite source term generation operator is defined as: ; in, For the generation operator of composite source terms, for A random emission direction vector uniformly distributed within the solid angle. Let E be the spatial location and E be the particle energy. For particle type, The Dirac function characterizes the type of particle. Spatial location Voxel activity weights at the location For summation, Let i be the branching ratio of the i-th type of particle. Let be the single-particle energy probability density distribution of the i-th type of particle. Let i be the i-th particle type.

[0080] (II) Derivation of Boltzmann transport and dose rate tensor calculus of mixed particles When the aforementioned source terms are fed into the heterogeneous three-dimensional lattice constructed by the CT of the target object, for gamma ( Solving the steady-state linear Boltzmann transport equation (LBTE) for photons and secondary neutral particles: ; In the formula, The particle angular flux number represents the position in space. Energy E, direction of motion The number of particles passing through a unit area per unit time. For streamline curvature; This represents the overall macroscopic cross-section of the corresponding medium; For scattering source terms; Let E' be the macroscopic differential scattering cross section, representing the energy E′ and the direction of scattering. Particles, scattered with energy E, direction The probability of; for photon source term; for , For charged particles that undergo extremely high-frequency Coulomb collisions with orbital electrons, solve the Fokker-Planck transport partial differential approximation based on the Continuous Deceleration Approximation (CSDA) and multiple Coulomb scattering: ; in, For spatial convection, For spatial gradient operators, This is the energy loss term. Stopping power refers to the energy lost per unit path length when a charged particle (such as an electron or an alpha particle) travels through a medium. For the partial derivative of energy, For scattering source terms; For differential scattering cross section, For electron source; In the Boltzmann transport equations This usually appears as an energy loss term. For charged particles, due to the extremely high frequency of Coulomb collisions with the orbital electrons of the medium atoms, the particle loses energy for every very short distance it travels. This continuous energy loss can be expressed as... This method is used for quantification. It is rigorously derived from the fundamental theory of quantum mechanics and the interaction between charged particles and matter. Its core source is the Bethe-Bloch formula, which is currently available and will not be elaborated upon here.

[0081] Tracking step length in each Monte Carlo Internally, the energy is accumulated through collisional ionization and bremsstrahlung deposition. After traversing the evolution of tens of millions of particles, the energy is determined based on the local mass of the voxels. Normalized output static three-dimensional absolute physical dose rate tensor (Unit: Gy / s or Gy / MBq). This tensor exhibits extremely high-resolution dose jump distributions at microscopic interfaces (such as the junction of trabeculae and red bone marrow), completely overturning the smoothness fallacy of convolution kernels. For spatial location The mass of a single voxel at that location. Spatial location The concentration of substances at that location, The volume is the volume of a voxel.

[0082] (III) Derivation of the pharmacokinetic (PK) time differential equation and the spatiotemporal functional of total dose Internal irradiation must take into account the drug's metabolic and excretory processes within the target body. A pharmacokinetic compartmental partial differential model is constructed for each specific voxel. The physical decay constant is assumed to be... The biological metabolic constant is (Controlled by blood supply in the tumor microenvironment).

[0083] Intravoxel-targeted drug activity concentration The evolutionary ordinary differential equation is: ; In the formula, This represents the rate of change of drug activity concentration over time. For the item to be cleared, among which, The decay constant corresponding to the physical half-life. The biological clearance constant, For input items, representing time t and spatial location The rate of drug activity upon newly introduced voxels. It is generally assumed that a clearance phase occurs after injection (ignoring the uptake phase). Alternatively, a double-exponential fitting method can be used (voxel time-activity retention functional). Defined as: ; In the formula, To quickly eliminate the exponential decay term of the component, Let be the local rapid biological clearance constant, and t be the time. For the exponential decay term of the slow-clearing component, The relative weights of the slow-clearing components, This represents the localized, slow biological clearance constant.

[0084] The dose rate tensor solved by physical calculus The above-mentioned biological metabolic retention functional in the time domain By performing a strongly coupled Riemann integration, the final cumulative total absorbed dose tensor for targeted isotope release is obtained: ; 3. Execution flow of low-level simulation and multimodal coupled dimensionality reduction control The system implements extremely stringent feedback and concurrency control in the underlying computing power scheduling and multi-field coupling. The specific steps are as follows: Step 1: Concurrent initialization of multi-particle emission threads based on hash space mapping: The CT-derived physical parameter matrix and the PET-derived activity sampling matrix are loaded into the GPU's global memory. The sampling is constructed using either the alias method or inverse transform sampling. A discrete voxel emission sampling index tree with complexity of L2. The underlying emitter kernel is based on... Generate particle coordinates and randomly determine the particle state of this thread based on ENSDF data (e.g., 60% probability of generating 0.5 MeV particles). Rays, with a 40% probability of generating an additional 0.2 MeV. (Associated clusters). Tens of millions of parallel computing threads (CUDA Threads) synchronously enter the physical transport cycle.

[0085] Step 2: Adaptive step-size transport and scoring with voxel boundary cross detection: When traversing within the grid, the underlying algorithm invokes octree-based spatial voxel ray-tracing logic. This is especially relevant for extremely short-range ray tracing. When particles or low-energy secondary electrons cross interfaces with strong density gradients (such as the tracheal wall-lung tissue interface), the system automatically performs geometric truncation stepping, using independent material sections on both sides of the interface. and the ability to stop The continuous energy loss is calculated by precisely storing the energy in the voxel score array using the underlying atomic primitive `atomicAdd()`. This process completely avoids the energy leakage of traditional convolution kernels at heterogeneous interfaces.

[0086] Step 3: Inverse fitting of parameters at multiple time points and output of the final dose extreme value: In actual clinical manifolds, in order to obtain non-uniform biological metabolic constants... The system needs to read a series of SPECT / PET follow-up data from different time points (e.g., 4h, 24h, 96h) of the target object. Voxel-wise curve fitting is performed using the Levenberg-Marquardt nonlinear least squares method, and the data is then reconstructed in reverse. The feature parameter array.

[0087] Finally, the static output of Monte Carlo will be... The tensor and the fitted Time-Integrated Activity (TIA) matrix are subjected to tensor dot product based on the Hadamard Product. The system output is a perfect fusion of the extreme precision of multi-particle transport in underlying nuclear physics and the time evolution of macroscopic physiological pharmacokinetics (DICOM-RT Dose format), resulting in an ultimate three-dimensional dose field. This patented system breaks through the bottleneck of convolution kernel methods and establishes a technological barrier for the next generation of precise radioisotope internal radiotherapy (TRT) dose planning.

[0088] Furthermore, Figure 2 This is a structural block diagram of a targeted isotope dose determination device according to an embodiment of the application, such as... Figure 2 As shown, the device includes: The 3D image data acquisition module is used to acquire PET 3D image data and CT 3D image data of the target object; The mapping module is used to map CT 3D image data into non-uniform medium density distribution and material composition index; The module for constructing a three-dimensional non-uniform source location probability weight model is used to determine the activity value of each voxel in PET three-dimensional image data. Based on the activity value of each voxel, a three-dimensional non-uniform source location probability weight model for Monte Carlo sampling is constructed. The multi-particle composite source term invocation module is used to invoke decay pattern parameters from the nuclear structure database to construct a composite source term containing alpha particles and beta particles. - The multi-particle composite source term for particles and gamma rays, wherein the multi-particle composite source term integrates voxel activity weights and particle energy probability density distributions, and the voxel activity weights are obtained based on the normalization of the voxel activity distribution. The three-dimensional dose rate distribution generation module is used to generate a three-dimensional dose rate distribution for each sampled particle obtained from the probability weight model of the non-uniform source location within the geometric space defined by the non-uniform medium density distribution and material composition index. It performs initial energy sampling based on the particle energy probability density distribution corresponding to the sampled particle, calls the medium physical parameters corresponding to the current space according to the material composition index, tracks the particle trajectory and accumulates energy deposition. The absorbed dose distribution determination module is used to process the three-dimensional dose rate distribution based on preset pharmacokinetic parameters to obtain the three-dimensional targeted isotope absorbed dose distribution of the target object.

[0089] The application of the relevant modules of the system in this example can be found in the above introduction to the principles of the method, and will not be repeated here.

[0090] above Figure 2 The targeted isotope dose determination device in this embodiment of the invention is described in detail from the perspective of modular functional entities. The electronic device in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0091] Figure 3 This is a schematic diagram of the structure of an electronic device 300 provided in an embodiment of the present invention. The electronic device 300 can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) for storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the electronic device 300. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the electronic device 300.

[0092] Electronic device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, MacOSX, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The illustrated electronic device structure does not constitute a limitation on electronic devices and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0093] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of any of the above-described methods for determining the dose of a targeted isotope.

[0094] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0095] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0096] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining the dosage of a targeted isotope, characterized in that, The method includes: Acquire PET and CT 3D image data of the target object; The CT three-dimensional image data is mapped to a non-uniform medium density distribution and material composition index; The activity value of each voxel in the PET three-dimensional image data is determined, and a three-dimensional non-uniform source location probability weight model for Monte Carlo sampling is constructed based on the activity value of each voxel. By calling decay pattern parameters from the nuclear structure database, a decay pattern containing alpha particles and beta particles is constructed. - A multi-particle composite source term for particles and gamma rays, wherein the multi-particle composite source term integrates voxel activity weights and particle energy probability density distributions, and the voxel activity weights are obtained based on the normalization of the voxel activity distributions; Within the geometric space defined by the non-uniform medium density distribution and material composition index, for each sampled particle obtained from the non-uniform source location probability weight model, initial energy sampling is performed based on the particle energy probability density distribution corresponding to the sampled particle. The medium physical parameters corresponding to the current space are called according to the material composition index, the particle trajectory is tracked and energy deposition is accumulated to generate a three-dimensional dose rate distribution. Based on preset pharmacokinetic parameters, the three-dimensional dose rate distribution is processed to obtain the three-dimensional targeted isotope absorbed dose distribution of the target object.

2. The method for determining the targeted isotope dose according to claim 1, characterized in that, The step of mapping the CT three-dimensional image data into a non-uniform medium density distribution and material composition index includes: Based on the mapping relationship between Huntsfield units and chemometrics school quasi-functionalities, the CT three-dimensional image data is converted voxel by voxel; The Huntsfield unit values ​​of each voxel in the CT three-dimensional image data are converted into electron density tensors through the first mapping relationship. Through the second mapping relationship, the Huntsfield unit values ​​of each voxel in the CT three-dimensional image data are converted into material composition index tensors. The material composition index tensors are used to distinguish between bone, soft tissue, lung and cavity regions. The non-uniform medium density distribution includes the electron density tensor. The first mapping relationship is as follows: ； In the formula, In Huntsfield units, The conversion function, determined based on a chemometrics school lookup table, is used to map HU values ​​to the electron density of the corresponding tissue at a preset reference energy. , Preset reference energy, preset reference energy Determined based on the decay characteristics of the targeted isotope; The second mapping relationship is: ； In the formula, Spatial location The index value at that location, Operators for finding the minimum parameter, For spatial location The Huntsfield unit value of the CT image corresponding to the voxel at that location. Let be the preset HU threshold for the i-th index, where i is the index independent variable.

3. The method for determining the targeted isotope dose according to claim 1, characterized in that, Determining the activity value of each voxel in the PET 3D image data, and constructing a 3D non-uniform source location probability weighting model for Monte Carlo sampling based on the activity value of each voxel, includes: Extract the standard uptake values ​​of each voxel from the PET 3D image data; Obtain the total injection activity, weight, and preset cross-calibration coefficient of the target subject; Based on the total injected activity, the target subject's weight, the preset cross-calibration coefficient, and the standard intake value of each voxel, the activity value of each voxel is determined in conjunction with the first formula. The activity values ​​of all voxels are normalized in three-dimensional space to obtain the probability density function of the non-uniform source location in three-dimensional space. The probability density function of the three-dimensional non-uniform source location is used as the three-dimensional non-uniform source location probability weighting model for the Monte Carlo sampling. The first formula is: ； In the formula, Spatial location Location, Time The initial activity value at that time, This represents the standard intake value for voxels. Total injectable activity, For the target subject's weight, Preset cross-calibration coefficients; The probability density function of the non-uniform source location in three-dimensional space is: ； In the formula, Let be the probability density function of the non-uniform source location in three-dimensional space. The denominator is the total activity value of the three-dimensional space of the target object.

4. The method for determining the targeted isotope dose according to claim 1, characterized in that, The decay profile parameters in the nuclear structure database are called to construct a structure containing alpha particles and beta particles. - Multi-particle composite source terms for particles and gamma rays include: Access a pre-defined nuclear structure database to extract decay pattern parameters of the target isotope and its daughter nuclei. These decay pattern parameters include at least: α decay energy, β decay energy, and so on. - Maximum decay energy, gamma-ray energy, branching ratio of each decay mode, alpha particle yield, and beta particle yield. - Particle production, gamma-ray production; Based on the decay framework parameters, α particles and β particles are constructed respectively. - Single-particle energy probability density distribution of particles and gamma rays , and ,in, Constructed based on discrete energy level transitions Based on Fermi theory Constructed based on characteristic spectral lines; Based on the aforementioned branch ratio, Output, β - Output The output is obtained by weighting and combining the single-particle energy probability density distribution to generate the full-particle energy probability density distribution of the multi-particle composite source term. The multi-particle composite source term is constructed by fusing the full-particle energy probability density distribution with the voxel activity weights. The multi-particle composite source term is represented as follows: ； In the formula, For multi-particle composite source terms, it represents the spatial location. The energy is The particle emission probability density; Spatial location Voxel activity weights at locations; For the first Various radiation modes, including particle, Particles and The yield or branching ratio of rays; For the first The single-particle energy probability density distribution of a certain type of particle.

5. The method for determining the targeted isotope dose according to claim 1, characterized in that, Within the geometric space defined by the non-uniform medium density distribution and material composition index, for each sampled particle obtained from the non-uniform source location probability weight model, initial energy sampling is performed based on the particle energy probability density distribution corresponding to the sampled particle. Then, according to the material composition index, the corresponding medium physical parameters at the current space are called, particle trajectories are tracked, and energy deposition is accumulated to generate a three-dimensional dose rate distribution, including: Based on the particle energy probability density distribution corresponding to the sampled particle, the initial kinetic energy of the particle is determined by combining the inverse transformation sampling method. Based on the material composition index at the current space location, the corresponding medium physical parameters are retrieved and called from a pre-set physical parameter database. These medium physical parameters include at least mass density, electron density, atomic number, photoelectric effect cross section / Compton scattering cross section / pair effect cross section for gamma rays, and cross section for alpha particles and beta particles. - Data on the particle's stopping power; Solve the linear Boltzmann transport equation for gamma rays; Targeting alpha particles and beta particles - For charged particles, solve the Fokker-Planck collision evolution equations based on the continuous deceleration approximation; When a particle crosses a medium interface defined by different material composition indices, it is determined whether the remaining range of the current particle is less than the current voxel boundary distance; if so, the step size is reduced to the medium interface, and the material composition index and the corresponding medium physical parameters are updated after crossing the interface. The tracked particle energy deposition values ​​are accumulated into the corresponding voxel score array, and the particle flux and deposition energy in each voxel are statistically analyzed to generate the three-dimensional dose rate distribution. The linear Boltzmann transport equation is as follows: ； In the formula, Spatial location Particle emission energy and direction of movement Particle angular flux density at that location, Let E be the spatial position and E be the particle's emission energy. The unit vector in the direction of motion; The directional derivative characterizes the spatial migration and leakage terms of particles. This represents the overall macroscopic cross-section of the corresponding medium; It is a multi-particle composite source term distribution; The macroscopic differential scattering cross section represents the probability of a particle being scattered. The incident energy of the particle; The direction of motion of the incident particle. Spatial location Particle incident energy and direction of movement angular flux density of particles at that location; The Fock-Planck collision evolution equation is: ； In the formula, The stopping power of charged particles in the current medium characterizes the energy loss rate under the continuous deceleration approximation. It is the energy differential operator; The momentum partial differential term, used to characterize multiple Coulomb scattering, is used to describe the angular deflection of a charged particle's trajectory.

6. The method for determining the targeted isotope dose according to claim 1, characterized in that, Based on preset pharmacokinetic parameters, the three-dimensional dose rate distribution is processed to obtain the three-dimensional targeted isotope absorbed dose distribution of the target object, including: Based on preset pharmacokinetic parameters, the biological metabolic constant and physical decay constant of the target isotope in the target voxel are obtained. Based on the aforementioned biological metabolic constants and physical decay constants, the effective decay constant of the target isotope within the target voxel is determined. Based on the effective decay constant, a time-activity retention functional for the target voxel is constructed. The physical dose rate tensor obtained from Monte Carlo simulation is integrated with the time activity retention functional of the target voxel in the time domain to obtain the three-dimensional target isotope absorbed dose distribution of the target object. The integral formula is as follows: ； In the formula, For the target voxel in space The time-activity retention functional (time integral activity) at a given location. For time The activity concentration of the target isotope within the target voxel; For integration time; ； In the formula, For the target object in space Three-dimensional targeted isotope cumulative absorbed dose distribution at the location; This is the physical dose rate tensor of the unit activity of the corresponding voxel generated by Monte Carlo simulation.

7. The method for determining the targeted isotope dose according to claim 1, characterized in that, The process of tracking particle trajectories and accumulating energy deposition includes: During Monte Carlo particle transport, real-time monitoring of abrupt changes in energy deposition of charged particles at voxel boundaries is conducted. When a discontinuity in energy deposition is detected due to abrupt changes in the material composition index, a boundary energy deposition smoothing algorithm is activated to subdivide the particle step size across different medium interfaces.

8. A targeted isotope dose determination device, characterized in that, The device includes: The 3D image data acquisition module is used to acquire PET 3D image data and CT 3D image data of the target object; The mapping module is used to map the CT three-dimensional image data into a non-uniform medium density distribution and material composition index; A three-dimensional non-uniform source location probability weight model construction module is used to determine the activity value of each voxel in the PET three-dimensional image data, and construct a three-dimensional non-uniform source location probability weight model for Monte Carlo sampling based on the activity value of each voxel. The multi-particle composite source term calling module is used to call decay pattern parameters in the nuclear structure database to construct a multi-particle composite source term containing α particles, β-particles and γ rays. The multi-particle composite source term integrates voxel activity weights and particle energy probability density distributions. The voxel activity weights are obtained based on the normalization of the voxel activity distribution. The three-dimensional dose rate distribution generation module is used to generate a three-dimensional dose rate distribution for each sampled particle sampled from the non-uniform source location probability weight model within the geometric space defined by the non-uniform medium density distribution and material composition index. It performs initial energy sampling based on the particle energy probability density distribution corresponding to the sampled particle, calls the medium physical parameters corresponding to the current space according to the material composition index, tracks the particle trajectory and accumulates energy deposition. The absorbed dose distribution determination module is used to process the three-dimensional dose rate distribution based on preset pharmacokinetic parameters to obtain the three-dimensional targeted isotope absorbed dose distribution of the target object.

9. An electronic device, characterized in that, The electronic device includes a memory and at least one processor, the memory storing instructions; the at least one processor invokes the instructions in the memory to cause the electronic device to perform the steps of the targeted isotope dose determination method as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the targeted isotope dose determination method as described in any one of claims 1-7.

Images