Radiotherapy plan generation method, apparatus, device, and medium

Abstract

The present disclosure provides a radiotherapy plan generation method, comprising: obtaining object information of a target object, the object information comprising at least medical image information and a treatment target of the target object; calculating a field intensity distribution according to the object information, the field intensity distribution comprising a plurality of dose grids and beam information of a corresponding beam of each dose grid; selecting a target dose grid from each dose grid based on a difference between an actual dose of each dose grid and a target dose; constructing an inverse optimization problem according to each target dose grid; and solving the inverse optimization problem to generate a radiotherapy plan.

Description

Technical Field

[0001] This disclosure relates to the field of medical radiation, and more specifically, to a method, apparatus, equipment, and medium for generating radiotherapy plans. Background Technology

[0002] Radiation therapy (RT) is currently one of the main methods for treating malignant tumors in clinical practice. With the development of computer technology and medical equipment, modern precision radiotherapy techniques, represented by intensity-modulated radiotherapy (IMRT) and volumetric modulated radiotherapy (VMAT), have been widely used. The core of these techniques lies in the radiotherapy planning process, which uses a computational process called "inverse optimization" to determine a set of optimal beam parameters (such as beam intensity, direction, and energy). This ensures that a sufficient lethal dose is delivered to the tumor target area while maximally protecting surrounding normal tissues and organs at risk, achieving a high-precision, highly conformal dose distribution.

[0003] In a typical inverse optimization process, the patient's anatomical structure is first discretized into tens of thousands or even millions of three-dimensional dose grids (or voxels), and a large dose influence matrix is ​​established, describing the dose contribution of each beam unit to each dose grid. Subsequently, based on clinically set dose targets (such as target area prescription dose, organ-at-risk limits, etc.), the system constructs a complex optimization mathematical problem and iteratively solves this problem to obtain the final radiotherapy plan. However, with the continuous improvement of medical image resolution and treatment precision, the number of dose grids involved in optimization has increased dramatically, resulting in an exceptionally large dimension and data scale of the dose influence matrix. This not only places extremely high demands on the memory and computing power of the computing hardware but also makes the computational workload of the inverse optimization process enormous, leading to excessively long planning times and becoming a bottleneck restricting the efficiency of clinical work. Summary of the Invention

[0004] In view of this, the present disclosure provides a method, apparatus, device and medium for generating radiotherapy plans.

[0005] One aspect of this disclosure provides a method for generating a radiotherapy plan, comprising: acquiring object information of a target object, the object information including at least medical image information of the target object and a treatment target; calculating a radiation field intensity distribution based on the object information, the radiation field intensity distribution including multiple dose grids and beam information of the corresponding beams for each dose grid; selecting a target dose grid from the dose grids based on the difference between the actual dose and the target dose of each dose grid; constructing an inverse optimization problem based on each target dose grid; and solving the inverse optimization problem to generate a radiotherapy plan.

[0006] According to embodiments of this disclosure, selecting a target dose grid from various dose grids includes: calculating a difference degree based on at least one of the absolute difference, squared difference, or weighted difference between the actual dose and the target dose of each dose grid; and selecting a target dose grid from various dose grids based on the difference degree.

[0007] According to embodiments of this disclosure, selecting a target dose grid from various dose grids includes: obtaining an objective function corresponding to each dose grid, wherein the objective function characterizes the difference between the actual dose and the target dose in the dose grid; generating a sampling probability corresponding to each dose grid based on the gradient of the objective function; and selecting a target dose grid according to the sampling probability corresponding to each dose grid.

[0008] According to embodiments of this disclosure, generating sampling probabilities corresponding to each dose grid based on the gradient of the objective function includes: calculating the norm corresponding to the gradient of the objective function; and generating sampling probabilities corresponding to each dose grid based on the norm, wherein the sampling probability is proportional to the norm.

[0009] According to embodiments of this disclosure, the sampling probability corresponding to each dose grid is generated based on the gradient of the objective function, including: calculating the norm corresponding to the gradient of the objective function; summing the norms to obtain a norm sum; and using the ratio of the norm to the norm sum as the sampling probability.

[0010] According to embodiments of this disclosure, selecting a target dose grid from various dose grids includes: calculating a compression ratio based on at least one of the dose grid resolution, the size of the data storage space, and the actual dose distribution of the dose grid; and selecting a target dose grid from various dose grids based on the compression ratio.

[0011] According to embodiments of this disclosure, the radiotherapy plan generation further includes: recalculating the sampling probability of each dose grid based on the dose distribution corresponding to the radiotherapy plan; reselecting the target dose grid based on the recalculated sampling probability; updating the inverse optimization problem based on the reselected target dose grid; solving the inverse optimization problem and updating the radiotherapy plan.

[0012] Another aspect of this disclosure provides a radiotherapy planning generation apparatus, comprising: a first acquisition module for acquiring object information of a target object, the object information including at least medical image information of the target object and a treatment target; a first calculation module for calculating a radiation field intensity distribution based on the object information, the radiation field intensity distribution including multiple dose grids and beam information of the corresponding beams for each dose grid; a first selection module for selecting a target dose grid from the dose grids based on the difference between the actual dose and the target dose of each dose grid; a first construction module for constructing an inverse optimization problem based on each target dose grid; and a first solution module for solving the inverse optimization problem to generate a radiotherapy plan.

[0013] Another aspect of this disclosure provides an electronic device including: at least one processor; and a memory connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the radiotherapy planning generation method of any of the foregoing embodiments.

[0014] Another aspect of this disclosure provides a computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to perform a radiotherapy planning method according to any of the foregoing embodiments.

[0015] Another aspect of this disclosure provides a computer program product, including a computer program / instructions, characterized in that the computer program / instructions, when executed by a processor, implement the operation of the radiotherapy planning generation method of any of the foregoing embodiments. Attached Figure Description

[0016] The above and other objects, features and advantages of this disclosure will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0017] Figure 1 A flowchart illustrating a radiotherapy plan generation method according to an embodiment of the present disclosure is shown schematically.

[0018] Figure 2 A flowchart illustrating the selection of a target dose grid in a radiotherapy planning generation method according to an embodiment of the present disclosure is shown schematically.

[0019] Figure 3 This schematically illustrates another flowchart of selecting a target dose grid in a radiotherapy planning generation method according to an embodiment of the present disclosure;

[0020] Figure 4 A flowchart illustrating the generation of sampling probabilities in a radiotherapy planning method according to an embodiment of the present disclosure is shown schematically.

[0021] Figure 5 This schematically illustrates another flowchart of the method for generating sampling probabilities in a radiotherapy planning process according to an embodiment of the present disclosure;

[0022] Figure 6 This schematically illustrates another flowchart of selecting a target dose grid in a radiotherapy planning generation method according to an embodiment of the present disclosure;

[0023] Figure 7 Another flowchart illustrating a radiotherapy plan generation method according to an embodiment of the present disclosure is shown schematically;

[0024] Figure 8A block diagram of a radiotherapy planning generation apparatus according to an embodiment of the present disclosure is illustrated schematically; and

[0025] Figure 9 A block diagram of an electronic device suitable for implementing the methods described above, according to embodiments of the present disclosure, is illustrated schematically. Detailed Implementation

[0026] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0027] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0028] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0029] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0030] In the embodiments disclosed herein, the collection, updating, analysis, processing, use, transmission, provision, disclosure, and storage of data (e.g., including but not limited to user personal information) comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures have been taken to prevent unauthorized access to user personal information data and to safeguard user personal information security, network security, and national security.

[0031] Embodiments of this disclosure provide a radiotherapy plan generation method, comprising: acquiring object information of a target object, the object information including at least medical image information and treatment target of the target object; calculating a radiation field intensity distribution based on the object information, the radiation field intensity distribution including multiple dose grids and beam information of the corresponding beams of each dose grid; selecting a target dose grid from each dose grid based on the difference between the actual dose and the target dose of each dose grid; constructing an inverse optimization problem based on each target dose grid; and solving the inverse optimization problem to generate a radiotherapy plan.

[0032] Figure 1 A flowchart illustrating a radiotherapy planning method according to an embodiment of the present disclosure is shown schematically.

[0033] like Figure 1 As shown, the radiotherapy planning generation method may include at least operations S110 to S150.

[0034] In operation S110, based on the target object's information, which includes at least the target object's medical imaging information and treatment goals, the object information defines the physical boundaries and clinical requirements of the radiotherapy plan to be executed. Medical imaging information refers to data acquired using one or more medical imaging techniques that characterize the internal anatomical structure of the target object, such as computed tomography (CT), magnetic resonance imaging (MRI), or positron emission tomography (PET) images. Treatment goals are clinical treatment parameters set for a specific lesion area, typically determined by the radiation oncologist based on clinical guidelines and the patient's specific condition. Acquiring object information serves two purposes: firstly, medical imaging information allows for precise delineation of the tumor target area and the geometry and spatial location of organs at risk that require protection; secondly, treatment goals clarify the dosimetric requirements for each region, such as the prescribed dose to be achieved in the target area and the dose limits that organs at risk can tolerate.

[0035] For example, for a patient with nasopharyngeal carcinoma, the subject information may include the patient's head CT imaging data and the treatment goals set by the physician, which stipulate that the planned target volume (PTV) should receive a prescribed dose of 66 Gy, while ensuring that the radiation dose to the optic nerve does not exceed 54 Gy and the radiation dose to the brainstem does not exceed 50 Gy.

[0036] In operation S120, based on the object information, the radiation field intensity distribution is calculated. This distribution includes multiple dose grids and the beam information corresponding to each dose grid. The radiation field intensity distribution, also known as the dose influence matrix or flux map, is a mathematical model describing the dose contribution relationship of each beam to each dose grid, typically represented as a two-dimensional matrix. A dose grid is a voxel unit formed by discretizing the anatomical structures of interest (such as tumor target areas and organs at risk) in three-dimensional space. Beam information characterizes the energy or dose deposited per unit intensity of a beam within a specific dose grid.

[0037] Calculating the beam intensity distribution establishes a linear or nonlinear mapping between beam intensity and spatial dose distribution, transforming the continuous physical irradiation process into a discrete mathematical problem, which is a prerequisite for inverse optimization. In proton or heavy ion radiotherapy, beam information can further include corrections for relative biological effect (RBE), making the calculated dose closer to the biologically equivalent dose.

[0038] For example, the patient's tumor target area is discretized into m dose grids, and n beams are designed to irradiate this target area. The calculated field intensity distribution is an m x n matrix A, where element A(i,j) represents the physical dose deposited in the i-th dose grid when the j-th beam is irradiated with a unit weight. If relative biological effects are considered, this physical dose also needs to be multiplied by a relative biological effect correction factor related to particle type, energy, and tissue characteristics.

[0039] In operation S130, a target dose grid is selected from all dose grids based on the difference between the actual dose and the target dose. The target dose grid is a subset selected from all dose grids to represent the overall dose distribution and participate in subsequent dimensionality reduction optimization calculations. The difference between the actual dose and the target dose of a dose grid characterizes the degree of deviation between the actual deposited dose of the dose grid and its expected target dose under the current irradiation parameters.

[0040] The core purpose of selecting a target dose grid is to reduce the dimensionality of the inverse optimization problem. By identifying and selecting dose grids with larger dose deviations that have a more significant impact on the optimization results as representatives, the original high-dimensional optimization problem can be approximated by a much smaller problem, thereby improving computational efficiency.

[0041] For example, a dose difference threshold can be set, and all dose grids can be iterated through. Dose grids whose actual dose differs from the target dose by more than the threshold can be selected as target dose grids. Alternatively, all dose grids can be sorted according to the magnitude of the dose difference, and the top 10% of dose grids with the largest differences can be selected as target dose grids.

[0042] In operation S140, an inverse optimization problem is constructed based on each target dose grid. In the field of radiotherapy, the inverse optimization problem refers to a computational paradigm distinct from the forward optimization problem. The forward optimization problem calculates the final dose distribution based on known radiotherapy equipment parameters (such as beam weights), while the inverse optimization problem, based on a pre-defined, desired dose distribution target, works backward to find the radiotherapy equipment parameters that optimally achieve that target.

[0043] The core function of constructing an inverse optimization problem is to transform clinical treatment goals (e.g., providing a high dose to the tumor target while protecting adjacent organs at risk) into a mathematical optimization model that can be solved by a computer. This model typically includes at least one objective function and / or one or more sets of constraints. The objective function quantifies the difference between the actual dose distribution and the target dose distribution, while the constraints set insurmountable dose limits. Solving this optimization problem involves finding a set of beam weights that minimizes the objective function while satisfying all constraints. In this embodiment, the mathematical model is built based on a selected target dose grid, resulting in an inverse optimization problem with significantly reduced dimensionality and easier solution.

[0044] For example, if the original field intensity distribution matrix A has a dimension of m×n, and after filtering, C target dose grids are obtained, then a new reduced-dimensional field intensity distribution matrix Acomp with a dimension of C×n is constructed. Simultaneously, the optimization objective function is modified from calculating the sum of dose deviations for all m dose grids to calculating only the sum of dose deviations for these C target dose grids. This constructs an inverse optimization problem where the number of variables to be solved remains unchanged (still n beam weights), but the computational cost of constraints or the objective function is significantly reduced.

[0045] It should be noted that the above example is only a simplified illustration. In specific implementations, the objective function used to select the target dose grid (i.e., downsampling) may differ from the objective function used in the final inverse optimization process. For example, the downsampling process can use a simple squared difference function to quickly assess the optimization importance of each grid, while the final inverse optimization can use a more complex objective function that better reflects clinical needs. Furthermore, the objective function used in inverse optimization does not necessarily have to be the sum of the objective function components of all target dose grids; it can also be other functional forms that can evaluate the quality of dose distribution, such as the p-norm of the dose deviation vector based on these C target dose grids. The core of this invention lies in reducing the dimensionality and computational complexity of the problem to be optimized by selecting representative target dose grids, rather than limiting the specific mathematical form of the objective function.

[0046] In operation S150, an inverse optimization problem is solved to generate a radiotherapy plan. The radiotherapy plan is the final scheme guiding the radiotherapy equipment to perform irradiation; its core content lies in the calculated weights of each beam and other relevant irradiation parameters. By solving this dimensionality-reduced optimization problem, a set of approximately optimal beam weights can be quickly obtained. This set of weights ensures that the dose distribution of the selected target dose grid optimally approximates its target dose. The final generated radiotherapy plan includes this set of beam weights, as well as information such as the position and energy of each beam.

[0047] For example, optimization algorithms such as gradient descent, conjugate gradient, or Newton's method can be used to iteratively solve the dimensionality reduction inverse optimization problem constructed in S140. Once the objective function converges or reaches the preset number of iterations, the obtained beam weight vector x is output. This vector x, combined with the preset beam geometry information, constitutes a complete and executable radiotherapy plan.

[0048] According to embodiments of this disclosure, a targeted dimensionality reduction strategy is achieved by selecting target dose grids based on the difference between the actual dose and the target dose of each dose grid. Unlike random sampling or downsampling methods based on static anatomical information, this approach directly utilizes dynamic information—dose deviation—during the optimization process. This ensures that the selected target dose grids precisely represent the regions in the current dose distribution that have the largest discrepancy with the clinical target and are most in need of optimization. Therefore, subsequent optimization calculations can concentrate limited computational resources on solving the most critical dose deviation problem, avoiding redundant calculations on dose grids that already meet the requirements. This significantly improves optimization speed while ensuring the accuracy of the optimization direction, enabling faster convergence to a high-quality radiotherapy plan.

[0049] Figure 2 The flowchart illustrating the selection of a target dose grid in a radiotherapy planning method according to an embodiment of the present disclosure is shown schematically.

[0050] like Figure 2 As shown, based on the aforementioned embodiments, operation S130 may include operations S210 to S220.

[0051] In operation S210, the difference degree is calculated based on at least one of the absolute difference, squared difference, or weighted difference between the actual dose and the target dose of each dose grid. The difference degree is a quantitative indicator used to characterize the degree to which the current dose value of a single dose grid deviates from its preset target value. In inverse optimization problems, the overall objective function is usually constructed as the sum of the objective function components of all dose grids to measure the quality of the overall dose distribution. Therefore, the difference degree in this embodiment can also be understood as the contribution value of a single dose grid to the overall objective function.

[0052] The purpose of calculating the difference is to assign an evaluation value to each dose grid that reflects its optimization urgency. A higher difference indicates a stronger "boost" effect on the overall objective function value, making it the part most in need of correction in the current optimization iteration. This calculation can take several mathematical forms, which typically correspond to the specific construction method of the objective function: one method is to calculate the absolute difference, directly reflecting the magnitude of the deviation, a simple and straightforward calculation; another method is to calculate the squared difference, which amplifies the impact of larger deviations, giving higher attention to the grids with the most severe dose deviations; yet another method is to calculate the weighted difference, allowing different weighting factors to be assigned to different anatomical regions or different dose grids, thus incorporating clinical importance (e.g., overdoses to organs at risk are more concerning than underdoses within the target area) into the difference calculation.

[0053] For example, for the i-th dose grid, its actual dose is dose. i The target dose is t i Its contribution component to the objective function (i.e., the degree of difference) can be calculated as follows:

[0054] If the absolute difference is used, the degree of difference is |dose i -t i Assuming an actual dose of 55 Gy and a target dose of 50 Gy in a grid, the difference is 5.

[0055] If the difference of squares is used, the degree of difference is (dose) i -t i The difference is (55-50)^2. For the above grid, the difference is (55-50)^2=25.

[0056] If a weighted difference is used, its degree of difference is w. i *(dose i -t i )^2. If the grid is located within a critical organ at risk, assign it a weight w. i If the value is 10, then its difference is 10*(55-50)^2=250, which significantly increases the likelihood of it being selected.

[0057] In operation S220, the target dose grid is selected from each dose grid based on the degree of difference. Selection based on degree of difference means that, according to the degree of difference evaluation value calculated in operation S210, which characterizes the contribution of each dose grid to the objective function, a subset of dose grids are selected as target dose grids through a set set of rules.

[0058] This selection process is a deterministic screening mechanism. One approach is to set a difference threshold and select all dose grids with a difference higher than the threshold as target dose grids; another approach is to sort all dose grids according to their difference and select the top N or top N% dose grids with the highest difference as target dose grids, where N is a preset value.

[0059] For example, a difference threshold can be set, and all dose grids can be iterated through to identify the dose grids with a difference greater than the threshold as target dose grids. Alternatively, all dose grids can be sorted from highest to lowest based on their difference values, and the top 10% of dose grids after sorting can be selected as target dose grids for subsequent optimization calculations.

[0060] According to embodiments of this disclosure, by correlating the degree of dissimilarity with the contribution value of the objective function, the selection process of the target dose grid has a clear optimization orientation. Instead of blindly reducing data points, it precisely identifies the dose grids that have the greatest impact on the current optimization objective (i.e., minimizing the objective function). By prioritizing these grids with high dissimilarity, the optimization algorithm can solve the most critical contradictions in each iteration, thereby achieving a rapid decrease in the objective function value with fewer computational resources, improving the convergence efficiency of the optimization and the quality of the final plan.

[0061] Figure 3 Another flowchart illustrating the selection of a target dose grid in a radiotherapy planning method according to an embodiment of the present disclosure is shown.

[0062] like Figure 3 As shown, based on the aforementioned embodiments, operation S130 may include operations S310 to S330.

[0063] In operation S310, the objective function corresponding to each dose grid is obtained. The objective function characterizes the difference between the actual dose and the target dose in the dose grid. The objective function corresponding to each dose grid can refer to the mathematical components related to a single dose grid in the overall inverse optimization objective function. The overall objective function is usually constructed as the sum of all these components to evaluate the overall quality of the current radiotherapy plan.

[0064] For example, for the i-th dose grid, its corresponding objective function component f i It can be defined as a weighted difference of squares:

[0065]

[0066] in, This is the actual dose for that grid. For the target dose, Weighted by clinical importance. The number of dose grids contained in tissue S.

[0067] It should be understood that the above description and examples are merely one specific implementation method for ease of understanding. In other implementations, the objective function used to guide sampling in this step may differ from the overall objective function used in the subsequent inverse optimization solution stage. Furthermore, the overall inverse optimization objective function does not necessarily have to be the sum of all dose grid objective function components. For example, this step may use a squared difference function to calculate the gradient to guide sampling, while the subsequent inverse optimization may use a complex objective function based on the dose volume histogram (DVH), or directly optimize the p-norm of the dose deviation vector.

[0068] In operation S320, sampling probabilities are generated for each dose grid based on the gradient of the objective function. The gradient of the objective function, obtained in operation S310, refers to the partial derivative of the objective function component for each dose grid with respect to the beam weight vector to be optimized. This gradient is a vector representing the rate and direction of change of the objective function value for that dose grid relative to the components in the beam weight vector. The sampling probability is a value between 0 and 1 assigned to each dose grid to guide the subsequent random selection process.

[0069] Gradient-based sampling probability generation aims to transform the multidimensional vector information of the gradient into a scalar probability value to guide selection. This transformation follows a fundamental principle: the higher the sensitivity of the dose grid, as represented by the gradient, to beam weight adjustments, the higher the sampling probability assigned to that dose grid. Here, "sensitivity" is a comprehensive consideration, reflecting the potential impact of adjusting the beam weights on the dose deviation of that grid.

[0070] For example, calculating the objective function component f of the i-th dose grid. i The gradient g relative to the beam weight vector x i The gradient vector g i It is then processed into a single numerical value, namely the sampling probability prob, through a preset mapping relationship or transformation function. i This mapping ensures the gradient g i The stronger the overall sensitivity reflected, the higher the probability of generating probabilities. i The larger the value, the better.

[0071] In operation S330, target dose grids are selected based on the sampling probabilities corresponding to each dose grid. Selection based on sampling probabilities is a random sampling process, where the probability of each dose grid being selected is determined by the sampling probability assigned to it in operation S320. By presetting the total number of target dose grids to be selected, algorithms such as weighted random sampling are used to extract a corresponding number of grids from all dose grids as target dose grids. Grids with higher probabilities have a greater chance of being selected, but grids with lower probabilities still have a possibility of being selected.

[0072] For example, suppose we need to select C target dose grids. We can use a roulette wheel selection algorithm to accumulate and normalize the sampling probabilities of all dose grids, generating a probability distribution. Then, we perform C independent random samplings, each based on this probability distribution, to select the C target dose grids.

[0073] According to embodiments of this disclosure, the dimensionality reduction process is made more in-depth and forward-looking by employing sampling probabilistic selection based on the gradient of the objective function. The gradient not only contains information about dose differences, but more importantly, it also contains information about the correlation between dose differences and all beam weights—that is, "how to change" to most effectively "eliminate differences." Therefore, gradient-based sampling can prioritize the selection of dose grids that are most sensitive to optimization adjustments and can most effectively guide the optimization direction. Furthermore, the introduction of a probabilistic selection mechanism avoids the risk of the optimization process getting trapped in local optima by focusing on only a few grids with the largest differences, increasing the globality and diversity of the selection and contributing to a better overall radiotherapy plan.

[0074] Figure 4 A flowchart illustrating the generation of sampling probabilities in a radiotherapy planning method according to an embodiment of the present disclosure is shown.

[0075] like Figure 4 As shown, based on the aforementioned embodiments, operation S320 may include operations S410~S420.

[0076] In operation S410, the norm corresponding to the gradient of the objective function is calculated. The norm is a mathematical operation that maps the gradient vector to a non-negative scalar value. Since the gradient is a vector existing in a high-dimensional beam weight space, the role of the norm is to synthesize its multi-dimensional information (i.e., the sensitivity of the objective function to changes in each beam weight) into a single numerical value to measure the overall "magnitude" or "strength" of the gradient.

[0077] The purpose of norm calculation is to reduce the dimensionality of complex vector information to obtain a scalar index that can intuitively reflect the sensitivity of the dose grid optimization. There are various ways to calculate the norm, such as L2 norm (Euclidean norm) or L1 norm (Manhattan norm).

[0078] For example, for the i-th dose grid, its gradient g i It is an n-dimensional vector, where n is the total number of beams. If the L2 norm is used, its norm value is ||g|. i The value of ||2 is calculated by taking the square root of the sum of the squares of the n components in the gradient vector. The larger this value, the steeper the trend of the objective function value of the dose grid at the current point.

[0079] In operation S420, based on the norm, the sampling probability corresponding to each dose grid is generated, where the sampling probability is proportional to the norm. The sampling probability being proportional to the norm refers to the sampling probability prob of the i-th dose grid. i Its gradient norm ||g is calculated in operation S410 i There is a linear proportional relationship between ||, i.e., prob i =k*||g i ||, where k is a proportionality constant applicable to all dose grids.

[0080] The sampling probability generation based on the norm aims to directly convert the norm value, which characterizes the optimization sensitivity obtained in the previous operation, into the probability that the dose grid will be selected during the random selection process. The proportionality constant k serves to normalize the probability distribution, ensuring that the sum of the sampling probabilities of all dose grids is 1, thus forming an effective probability distribution.

[0081] For example, suppose there are two dose grids, A and B, with gradient norms of 10 and 5 respectively. Then, the sampling probability of dose grid A being selected will be twice that of dose grid B. By introducing a uniform scaling constant k, their specific sampling probabilities can be calculated, for example, prob... A =k*10, prob B =k*5.

[0082] According to embodiments of this disclosure, by calculating the norm of the gradient, a clear and efficient approach is provided to transform high-dimensional gradient vector information into a scalar value that can directly quantify the importance of optimization. Using the norm as a benchmark for sampling probability tightly couples the selection rules with the geometric properties of the optimization problem itself (i.e., the steepness of the objective function at the current point), ensuring that the dose grid most sensitive to beam weight adjustment receives a proportionally higher chance of being selected. This makes the dimensionality reduction sampling process more reasonable and evidence-based.

[0083] Figure 5 Another flowchart illustrating the generation of sampling probabilities in a radiotherapy planning method according to an embodiment of the present disclosure is shown.

[0084] like Figure 5As shown, based on the aforementioned embodiments, operation S320 may include operations S510 to S530.

[0085] In operation S510, the norm corresponding to the gradient of the objective function is calculated. This is similar to operation S410 in the previous embodiment and will not be described again here.

[0086] In operation S520, the individual norms are summed to obtain the norm sum. The norm sum refers to the total value obtained by arithmetically summing the gradient norm values ​​calculated in operation S510 for all dose grids involved in the optimization. Calculating the norm sum aims to obtain a global normalization factor. This sum value represents the overall sensitivity of the entire dose field to beam weight adjustments under the current optimization state.

[0087] For example, suppose the entire optimization problem contains only three dose grids, and their gradient norms are calculated to be 5, 10, and 15 respectively using operation S510. Then, the sum of the norms is 5 + 10 + 15 = 30.

[0088] In operation S530, the ratio of the norm to the sum of norms is used as the sampling probability. Using the ratio of the norm to the sum of norms as the sampling probability means that for any dose grid, its sampling probability is determined by dividing its own gradient norm value by the sum of norms calculated in operation S520.

[0089] Then the sampling probability prob of the i-th dose grid i :

[0090]

[0091] Where N is the total number of dose grids.

[0092] Continuing with the previous example, the sum of the norms is 30.

[0093] The sampling probability of the first dose grid (norm 5) is 5 / 30≈0.167.

[0094] The sampling probability of the second dose grid (norm 10) is 10 / 30≈0.333.

[0095] The sampling probability of the third dose grid (norm 15) is 15 / 30 = 0.5.

[0096] These probability values ​​can be directly used in the subsequent weighted random sampling process.

[0097] According to embodiments of this disclosure, a specific, explicit, and self-consistent probability calculation method is provided by using the ratio of a single norm to the sum of norms as the sampling probability. This method not only ensures that the sampling probability is proportional to the gradient norm, but also, through global normalization, ensures that the sampling probabilities of all dose grids constitute a strict probability distribution. This allows the probability of each dose grid being selected to accurately reflect its relative contribution to the current overall optimization objective, providing a directly usable, standardized input for subsequent random selection operations without additional processing, thus enhancing the method's standardization and feasibility.

[0098] Figure 6 Another flowchart illustrating the selection of a target dose grid in a radiotherapy planning method according to an embodiment of the present disclosure is shown.

[0099] like Figure 6 As shown, based on the aforementioned embodiments, operation S130 may include operations S610 to S620.

[0100] In operation S610, the compression ratio is calculated based on at least one of the following: the resolution of the dose grid, the size of the data storage space, and the actual dose distribution of the dose grid. The compression ratio is a parameter used to determine the proportional relationship between the number of target dose grids and the original total number of dose grids, or it can directly refer to the target number of target dose grids.

[0101] The compression ratio is calculated to dynamically or pre-determine the degree of dimensionality reduction based on actual hardware constraints or the complexity of the dose distribution. This calculation can be based on one or more of the following factors:

[0102] One approach is based on the resolution of the dose grid. The higher the resolution, the more grids there are in total. To ensure computational efficiency, a higher compression ratio may need to be set.

[0103] Another approach is to determine a minimum compression ratio that meets hardware requirements by calculating the core data such as the reduced field intensity distribution matrix, based on the size of the data storage space, such as computer memory or video memory capacity.

[0104] Another approach is based on the actual dose distribution of the dose grid. In regions where the dose gradient changes drastically or has a complex distribution, the compression ratio can be dynamically reduced (i.e., more target grids are selected) to ensure optimization accuracy, while in regions with flat dose, the compression ratio can be increased accordingly.

[0105] For example, if the available video memory is 8 gigabytes (GB), while the complete field intensity distribution matrix requires 20 gigabytes of storage, and the problem can be computed in video memory, it is calculated that the data volume needs to be compressed to less than 40% of its original size. Therefore, a compression ratio is set so that the number of selected target dose grids does not exceed 40% of the total. If, based on dose distribution, the system analyzes the dose-volume histogram after one iteration and finds that the dose curve for organs at risk is too flat and fails to meet the clinical requirement of a steep descent, the system can adaptively calculate and adjust the compression ratio, increasing the selection ratio of target dose grids from 10% to 15% to increase the granularity of dose optimization for that region.

[0106] In operation S620, target dose grids are selected from each dose grid based on the compression ratio. Selection based on the compression ratio means using the compression ratio calculated in operation S610 as the basis for determining the final number of target dose grids, combined with a selection strategy to complete the screening. This operation is a specific step in the dimensionality reduction process. The compression ratio is first used to calculate the specific number C of target dose grids. Subsequently, the system uses a screening mechanism (e.g., random selection, or selection based on the difference or sampling probability in the aforementioned embodiments) to select C grids as target dose grids from all dose grids.

[0107] For example, assuming the total number of dose grids is 1,000,000, and the compression ratio calculated in operation S610 requires the selection of 20% of the grids, then the number of target dose grids C is determined to be 200,000. Subsequently, weighted random sampling can be performed according to a preset sampling probability distribution until 200,000 non-repeating dose grids are selected as the final target dose grid set.

[0108] According to embodiments of this disclosure, by introducing the calculation of the compression ratio, the degree of dimensionality reduction is no longer fixed or empirically determined, but rather has a clear basis and adaptability. Linking the compression ratio to hardware resources (such as storage space) ensures the feasibility of the optimization problem in a specific computing environment, avoiding computational failures or inefficiencies due to excessive data volume. Correlating the compression ratio to the characteristics of the actual dose distribution endows the dimensionality reduction process with "intelligence," enabling it to dynamically adjust the granularity of information extraction according to the difficulty of the optimization task, retaining more information in areas requiring fine-tuning, and performing greater compression in simpler areas, thereby achieving a better balance between computational efficiency and plan quality.

[0109] Figure 7 Another flowchart illustrating a radiotherapy planning method according to an embodiment of the present disclosure is shown schematically.

[0110] like Figure 7As shown, based on the aforementioned embodiments, the radiotherapy plan generation method may include operations S710~S740.

[0111] In operation S710, the sampling probability of each dose grid is recalculated based on the dose distribution corresponding to the radiotherapy plan. "Based on the dose distribution corresponding to the radiotherapy plan" means calculating the latest three-dimensional dose distribution using the beam weights obtained from the previous inverse optimization solution. "Recalculating the sampling probability" means repeating the process of calculating the sampling probability in the aforementioned embodiments based on this latest dose distribution. As optimization progresses, areas with previously large dose deviations may improve, and their optimization priority will decrease; simultaneously, other areas may exhibit new dose deviations that require attention, and their optimization priority will increase accordingly.

[0112] For example, after the Nth iteration, a new radiotherapy plan is obtained. A new dose distribution, dose(N), is calculated based on this plan. Based on the difference between dose(N) and the target dose t, the objective function component f for each dose grid is recalculated. i And then calculate its gradient g. i and the new sampling probability prob i(N+1) .

[0113] In operation S720, the target dose grid is reselected based on the recalculated sampling probabilities. Reselection based on recalculated sampling probabilities means using the updated sampling probability distribution obtained in operation S710 to perform weighted random sampling again, resulting in a completely new set of target dose grids. This ensures that the representative grid set used for dimensionality reduction is no longer static but dynamically evolves with the optimization process. The newly selected target dose grid set more accurately reflects the region that most needs optimization at the current stage.

[0114] For example, using a new sampling probability distribution prob i(N+1) Then, randomly select C dose grids again. This newly selected set may include some grids that were not selected before, or may remove some grids that were selected before but whose dose requirements are now met.

[0115] In operation S730, the inverse optimization problem is updated based on the newly selected target dose grid. Updating the inverse optimization problem refers to constructing a new, dimension-reduced inverse optimization problem based on the target dose grid set reselected in operation S720. Specifically, row vectors corresponding to the newly selected target dose grid are extracted from the original, complete field intensity distribution matrix to construct a new dimension-reduced matrix A. comp(N+1) .

[0116] For example, if the newly selected target dose grid index is c new Then, c is extracted from the complete field intensity distribution matrix A. new The corresponding rows form a new dimension-reduced matrix A. comp(N+1) At the same time, the objective function is updated to sum only over this new target dose grid set.

[0117] Similar to the preceding description of the objective function, the above description and examples are merely one specific implementation method for ease of understanding. In other implementations, the objective function used to guide sampling in this step may differ from the overall objective function used in the subsequent inverse optimization solution stage. Furthermore, the overall inverse optimization objective function does not necessarily have to be the sum of all dose grid objective function components.

[0118] In operation S740, the inverse optimization problem is solved to update the radiotherapy plan. Solving the updated inverse optimization problem refers to solving the newly constructed dimensionality reduction problem in operation S730 to obtain a set of updated beam weights. Updating the radiotherapy plan means using this new set of beam weights as the current optimal radiotherapy plan. The entire process (S710 to S740) can be repeated until a certain termination condition is met, such as reaching a preset maximum number of iterations, or the change in the overall objective function value of the radiotherapy plan is less than a threshold.

[0119] For example, for A-based comp(N+1) The constructed optimization problem is solved to obtain a new beam weight vector x(N+1), which is the updated radiotherapy plan. Then, the process can proceed to the (N+2)th iteration, or determine whether the termination condition is met.

[0120] According to embodiments of this disclosure, by introducing an iterative update mechanism, the selection process of the target dose grid can dynamically adapt to the optimization process. This method avoids the bias that may arise from a single static sampling, where the initially selected representative grid may not maintain its representativeness throughout the optimization process. By periodically re-evaluating and selecting the dose grid that has the greatest impact on the optimization direction based on the current dose distribution, computational resources can be continuously focused on the region most in need of improvement, thereby more effectively guiding the optimization process, preventing it from getting trapped in local optima, and ultimately converging to a higher-quality radiotherapy plan with greater efficiency.

[0121] Figure 8 A block diagram of a radiotherapy planning device according to an embodiment of the present disclosure is shown schematically.

[0122] like Figure 8 As shown, the radiotherapy planning generation device 800 may include a first acquisition module 810, a first calculation module 820, a first selection module 830, a first construction module 840, and a first solution module 850.

[0123] The first acquisition module 810 is used to acquire object information of the target object, which includes at least the medical image information and treatment target of the target object. In some embodiments, the first acquisition module 810 can be used to perform operation S110 in the radiotherapy plan generation method described above, which will not be elaborated here.

[0124] The first calculation module 820 is used to calculate the radiation field intensity distribution based on the object information. The radiation field intensity distribution includes multiple dose grids and the beam information of the corresponding beam for each dose grid. In some embodiments, the first calculation module 820 can be used to perform operation S120 in the radiotherapy planning method described above, which will not be elaborated here.

[0125] The first selection module 830 is used to select a target dose grid from the various dose grids based on the difference between the actual dose and the target dose of each dose grid. In some embodiments, the first selection module 830 may be used to perform operation S130 in the radiotherapy planning method described above, which will not be elaborated here.

[0126] The first construction module 840 is used to construct an inverse optimization problem based on each target dose grid. In some embodiments, the first construction module 840 can be used to perform operation S140 in the radiotherapy planning method described above, which will not be elaborated here.

[0127] The first solver module 850 is used to solve the inverse optimization problem and generate a radiotherapy plan. In some embodiments, the first solver module 850 can be used to perform operation S150 in the radiotherapy plan generation method described above, which will not be elaborated here.

[0128] According to embodiments of this disclosure, the first selection module may include a second calculation module and a second selection module.

[0129] The second calculation module is used to calculate the degree of difference based on at least one of the absolute difference, squared difference, or weighted difference between the actual dose and the target dose for each dose grid. In some embodiments, the second calculation module can be used to perform operation S210 in the radiotherapy planning method described above, which will not be elaborated here.

[0130] The second selection module is used to select a target dose grid from each dose grid based on the degree of difference. In some embodiments, the second selection module can be used to perform operation S220 in the radiotherapy planning method described above, which will not be elaborated here.

[0131] According to embodiments of this disclosure, the first selection module may include a second acquisition module, a second generation module, and a third selection module.

[0132] The second acquisition module is used to acquire the objective function corresponding to each dose grid, whereby the objective function characterizes the difference between the actual dose and the target dose in the dose grid. In some embodiments, the second acquisition module can be used to perform operation S310 in the radiotherapy planning method described above, which will not be elaborated here.

[0133] The second generation module is used to generate the sampling probability corresponding to each dose grid based on the gradient of the objective function. In some embodiments, the second generation module can be used to perform operation S320 in the radiotherapy planning method described above, which will not be elaborated here.

[0134] The third selection module is used to select the target dose grid based on the sampling probability corresponding to each dose grid. In some embodiments, the third selection module can be used to perform operation S330 in the radiotherapy planning method described above, which will not be elaborated here.

[0135] According to embodiments of this disclosure, the second generation module may include a third calculation module and a third generation module.

[0136] The third calculation module is used to calculate the norm corresponding to the gradient of the objective function. In some embodiments, the third calculation module can be used to perform operation S410 in the radiotherapy planning method described above, which will not be elaborated here.

[0137] The third generation module is used to generate the sampling probability corresponding to each dose grid based on the norm, wherein the sampling probability is proportional to the norm. In some embodiments, the third generation module can be used to perform operation S420 in the radiotherapy planning method described above, which will not be elaborated here.

[0138] According to embodiments of this disclosure, the second generation module may include a fourth calculation module, a summation module, and a fourth generation module.

[0139] The fourth calculation module is used to calculate the norm corresponding to the gradient of the objective function. In some embodiments, the fourth calculation module can be used to perform operation S510 in the radiotherapy planning method described above, which will not be elaborated here.

[0140] The summation module is used to sum the various norms to obtain a norm sum. In some embodiments, the summation module can be used to perform operation S520 in the radiotherapy planning method described above, which will not be elaborated here.

[0141] The fourth generation module is used to take the ratio of the norm and the sum of the norms as the sampling probability. In some embodiments, the fourth generation module can be used to perform operation S530 in the radiotherapy planning method described above, which will not be elaborated here.

[0142] According to embodiments of this disclosure, the first selection module may include a fifth calculation module and a fourth selection module.

[0143] The fifth calculation module is used to calculate the compression ratio based on at least one of the dose grid resolution, the data storage space size, and the actual dose distribution of the dose grid. In some embodiments, the fifth calculation module can be used to perform operation S610 in the radiotherapy planning method described above, which will not be elaborated here.

[0144] The fourth selection module is used to select a target dose grid from the various dose grids based on the compression ratio. In some embodiments, the fourth selection module can be used to perform operation S620 in the radiotherapy planning method described above, which will not be elaborated here.

[0145] According to embodiments of this disclosure, the first selection module may include a first calculation module, a first selection module, a first update module, and a second update module.

[0146] The first calculation module is used to recalculate the sampling probability of each dose grid according to the dose distribution corresponding to the radiotherapy plan. In some embodiments, the first calculation module can be used to perform operation S710 in the radiotherapy plan generation method described above, which will not be elaborated here.

[0147] The first selection module is used to reselect the target dose grid based on the recalculated sampling probability. In some embodiments, the first selection module can be used to perform operation S720 in the radiotherapy planning method described above, which will not be elaborated here.

[0148] The first update module is used to update the inverse optimization problem based on the reselected target dose grid. In some embodiments, the first update module can be used to perform operation S730 in the radiotherapy planning method described above, which will not be elaborated here.

[0149] The second update module is used to solve the inverse optimization problem and update the radiotherapy plan. In some embodiments, the second update module can be used to perform operation S740 in the radiotherapy plan generation method described above, which will not be elaborated here.

[0150] Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure, or at least part of the functions of any one or more of them, can be implemented in one module. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be implemented by dividing them into multiple modules. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be at least partially implemented as hardware circuitry, such as a Field-Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a System-on-Chip, a System-on-a-Substrate, a System-on-Package, an Application-Specific Integrated Circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be at least partially implemented as computer program modules, which, when run, can perform corresponding functions.

[0151] For example, any and more of the first acquisition module 810, first calculation module 820, first selection module 830, first construction module 840, and first solution module 850 can be combined into one module / unit / subunit, or any one of these modules / units / subunits can be split into multiple modules / units / subunits. Alternatively, at least some of the functions of one or more of these modules / units / subunits can be combined with at least some of the functions of other modules / units / subunits and implemented in one module / unit / subunit. According to embodiments of this disclosure, at least one of the first acquisition module 810, the first calculation module 820, the first selection module 830, the first construction module 840, and the first solution module 850 can be at least partially implemented as hardware circuits, such as field-programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems-on-a-chip, systems-on-a-substrate, systems-on-package, application-specific integrated circuits (ASICs), or any other reasonable means of integrating or packaging circuits, or implemented in software, hardware, or firmware, or in any appropriate combination of any of these three implementation methods. Alternatively, at least one of the first acquisition module 810, the first calculation module 820, the first selection module 830, the first construction module 840, and the first solution module 850 can be at least partially implemented as computer program modules, which can perform corresponding functions when the computer program module is run.

[0152] It should be noted that the data processing system part in the embodiments of this disclosure corresponds to the data processing method part in the embodiments of this disclosure. The specific description of the data processing system part is referred to in the data processing method part, and will not be repeated here.

[0153] Figure 9 A block diagram of an electronic device suitable for implementing the methods described above, according to embodiments of the present disclosure, is illustrated schematically. Figure 9 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0154] like Figure 9 As shown, an electronic device 900 according to an embodiment of the present disclosure includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage portion 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0155] RAM 903 stores various programs and data required for the operation of electronic device 900. Processor 901, ROM 902, and RAM 903 are interconnected via bus 904. Processor 901 performs various operations of the method flow according to embodiments of the present disclosure by executing programs in ROM 902 and / or RAM 903. It should be noted that the programs may also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in said one or more memories.

[0156] According to embodiments of this disclosure, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to a bus 904. The electronic device 900 may also include one or more of the following components connected to the input / output (I / O) interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 910 as needed so that computer programs read from it can be installed into the storage section 908 as needed.

[0157] According to embodiments of this disclosure, the method flow according to embodiments of this disclosure can be implemented as a computer software program. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network via communication section 909, and / or installed from removable medium 911. When the computer program is executed by processor 901, it performs the functions defined in the system of embodiments of this disclosure. According to embodiments of this disclosure, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0158] This disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of this disclosure.

[0159] According to embodiments of this disclosure, the computer-readable storage medium can be a non-volatile computer-readable storage medium. Examples include, but are not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0160] For example, according to embodiments of this disclosure, a computer-readable storage medium may include one or more memories other than ROM 902 and / or RAM 903 described above.

[0161] Embodiments of this disclosure also include a computer program product comprising a computer program containing program code for performing the methods provided in the embodiments of this disclosure. When the computer program product is run on an electronic device, the program code is used to enable the electronic device to implement the radiotherapy plan generation method provided in the embodiments of this disclosure.

[0162] When the computer program is executed by the processor 901, it performs the functions defined in the system / apparatus of this disclosure embodiments. According to embodiments of this disclosure, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0163] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices or magnetic storage devices. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via communication section 909, and / or installed from removable medium 911. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof. According to embodiments of this disclosure, program code for executing the computer programs provided in embodiments of this disclosure can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code may be executed entirely on a user computing device, partially on a user device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing devices can be connected to user computing devices via any type of network, including local area networks (LANs) or wide area networks (WANs), or they can be connected to external computing devices (e.g., via the Internet using an Internet service provider).

[0164] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions. Those skilled in the art will understand that the features described in the various embodiments of the present disclosure can be combined and / or combined in various ways, even if such combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments of this disclosure may be combined and / or combined in various ways without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0165] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A method for generating a radiotherapy plan, characterized in that, include: Obtain object information of the target object, wherein the object information includes at least the medical image information and treatment target of the target object; Based on the object information, the field intensity distribution is calculated, which includes multiple dose grids and the beam information of the beam corresponding to each dose grid. Based on the difference between the actual dose and the target dose in each of the dose grids, a target dose grid is selected from each of the dose grids; Based on each of the target dose grids, construct an inverse optimization problem; Solve the aforementioned inverse optimization problem to generate a radiotherapy plan; The step of selecting a target dose grid from each of the dose grids includes: Obtain the objective function corresponding to each of the dose grids, wherein the objective function characterizes the difference between the actual dose and the target dose in the dose grid; Based on the gradient of the objective function, the sampling probability corresponding to each of the dose grids is generated; The target dose grid is selected based on the sampling probability corresponding to each dose grid.

2. The method according to claim 1, characterized in that, The step of generating the sampling probability corresponding to each dose grid based on the gradient of the objective function includes: Calculate the norm corresponding to the gradient of the objective function; Based on the norm, a sampling probability corresponding to each of the dose grids is generated, wherein the sampling probability is proportional to the norm.

3. The method according to claim 1, characterized in that, The step of generating the sampling probability corresponding to each dose grid based on the gradient of the objective function includes: Calculate the norm corresponding to the gradient of the objective function; Summing each of the aforementioned norms yields the total norm; The ratio of the norm to the sum of the norms is used as the sampling probability.

4. The method according to claim 1, characterized in that, Also includes: Based on the dose distribution corresponding to the radiotherapy plan, the sampling probability of each dose grid is recalculated; The target dose grid is reselected based on the recalculated sampling probability; The inverse optimization problem is updated based on the newly selected target dose grid. Solve the inverse optimization problem to update the radiotherapy plan.

5. A radiotherapy planning device, characterized in that, include: The first acquisition module acquires object information of the target object, wherein the object information includes at least the medical image information and treatment target of the target object; The first calculation module calculates the field intensity distribution based on the object information. The field intensity distribution includes multiple dose grids and the beam information of the beam corresponding to each dose grid. The first selection module selects a target dose grid from each dose grid based on the difference between the actual dose and the target dose of each dose grid. The first construction module constructs an inverse optimization problem based on each of the target dose grids; The first solution module solves the inverse optimization problem and generates a radiotherapy plan; The step of selecting a target dose grid from each of the dose grids includes: Obtain the objective function corresponding to each of the dose grids, wherein the objective function characterizes the difference between the actual dose and the target dose in the dose grid; Based on the gradient of the objective function, the sampling probability corresponding to each of the dose grids is generated; The target dose grid is selected based on the sampling probability corresponding to each dose grid.

6. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 4.

Images