Adaptive Scenario-Based Robust Optimization Device and Method for Proton Radiotherapy Planning

Abstract

The application discloses a kind of based on adaptive scene-based proton radiotherapy plan robust optimization device and method, it is related to medical science and technology field.Therein, the device is used to generate N robust scene images based on the radiotherapy plan of target object;Through dose calculation engine, the dose distribution map of each robust scene image is calculated, obtains N dose distribution map corresponding to N robust scene image;According to N dose distribution map, determine first dose distribution map and second dose distribution map;Adjusting unit is used to adjust radiotherapy plan according to first dose distribution map and second dose distribution map.This application solves the technical problem that in order to cope with the position, shape and density of tumor target area and surrounding normal tissue can change due to the change of patient's body position, physiological state or tumor itself, conventional robust optimization method is time-consuming and poor in effect.

Description

Technical Field

[0001] This application relates to the field of medical technology, and more specifically, to a device and method for robust optimization of proton radiotherapy planning based on adaptive scenario-based approaches. Background Technology

[0002] In radiotherapy, the location, shape, and density of the tumor target area and surrounding normal tissue can change due to changes in the patient's position, physiological state, or the tumor itself. These changes affect the propagation path and dose distribution of radiation, potentially leading to reduced treatment efficacy or damage to normal tissue. To address these uncertainties, radiotherapy employs robust optimization methods, such as adding extended boundaries to the tumor to incorporate potential deviations and variations into the treatment planning process. This ensures that, regardless of various possible contingencies, the treatment plan effectively covers the tumor area while maximizing the protection of normal tissue.

[0003] However, improving the robustness of radiotherapy plans by adding an outer boundary to the tumor can easily lead to excessive radiation to normal tissues. In addition, dose calculations are required for the outer boundary as well, resulting in technical problems such as time-consuming and ineffective proton radiotherapy plan optimization.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This application provides a robust optimization device and method for proton radiotherapy planning based on adaptive scenario, which at least solves the technical problem that conventional robust optimization methods are time-consuming and ineffective in order to cope with changes in the position, shape and density of the tumor target area and surrounding normal tissues due to changes in the patient's position, physiological state or changes in the tumor itself.

[0006] According to one aspect of this application, an adaptive scene-based proton radiotherapy planning robust optimization device is provided, comprising: a first processing unit, configured to generate N robust scene images based on a radiotherapy plan for a target object, wherein the N robust scene images are used to simulate the change information of M regions of interest of the target object, and N and M are both integers greater than 1; a second processing unit, configured to calculate the dose distribution map of each robust scene image through a dose calculation engine to obtain N dose distribution maps corresponding to the N robust scene images; a determination unit, configured to determine a first dose distribution map and a second dose distribution map based on the N dose distribution maps, wherein the first dose distribution map is used to record the maximum dose value of the N dose distribution maps at each grid point, and the second dose distribution map is used to record the minimum dose value of the N dose distribution maps at each grid point, and each grid point corresponds to a beam spot in radiotherapy; and an adjustment unit, configured to adjust the radiotherapy plan based on the first dose distribution map and the second dose distribution map.

[0007] Optionally, the adjustment unit includes: a first determining subunit, used to determine a first sequence matrix based on the sequence number of the robust scene image corresponding to the dose value of each grid point in the first dose distribution map; a second determining subunit, used to determine a second sequence matrix based on the sequence number of the robust scene image corresponding to the dose value of each grid point in the second dose distribution map; and an adjustment subunit, used to adjust the radiotherapy plan based on the first dose distribution map, the second dose distribution map, the first sequence matrix, and the second sequence matrix.

[0008] Optionally, the adjustment subunit includes: a first determining module, configured to determine the maximum dose distribution matrix of each region of interest based on a first dose distribution map; and to determine the minimum dose distribution matrix of each region of interest based on a second dose distribution map; a first processing module, configured to use the submatrix corresponding to each region of interest in the first sequence matrix as the first submatrix of that region of interest; a second processing module, configured to use the submatrix corresponding to each region of interest in the second sequence matrix as the second submatrix of that region of interest; and an adjustment module, configured to adjust the plan content of the radiotherapy plan for each region of interest according to the minimum dose distribution matrix, the maximum dose distribution matrix, the first submatrix, and the second submatrix of each region of interest.

[0009] Optionally, the adjustment module includes: a detection submodule for detecting the N dose distribution submaps corresponding to each region of interest in the N dose distribution maps; a determination submodule for determining the N dose volume histograms of each region of interest based on the N dose distribution submaps corresponding to each region of interest; and a first adjustment submodule for adjusting the plan content of the radiotherapy plan for each region of interest based on the N dose volume histograms, the minimum dose distribution matrix, the maximum dose distribution matrix, the first submatrix, and the second submatrix of each region of interest.

[0010] Optionally, the adjustment module includes: a second adjustment submodule, used to adjust the radiotherapy plan content related to the i-th region of interest according to the maximum dose distribution matrix and the second submatrix of the i-th region of interest when the constraint set for the i-th region of interest is detected to be a dose reduction, wherein i is an integer greater than or equal to 1; and a third adjustment submodule, used to adjust the radiotherapy plan content related to the j-th region of interest according to the minimum dose distribution matrix and the first submatrix of the j-th region of interest when the constraint set for the j-th region of interest is detected to be a dose increase, wherein j is an integer greater than or equal to 1.

[0011] Optionally, the second adjustment submodule includes: a first processing submodule, used to determine the dose constraint matrix of the i-th region of interest based on the maximum dose distribution matrix of the i-th region of interest and the constraint conditions of the i-th region of interest; a first calculation submodule, used to calculate the matrix product of the dose constraint matrix of the i-th region of interest and the second submatrix of the i-th region of interest to obtain the iterative constraint matrix of the i-th region of interest; and a second processing submodule, used to adjust the plan content of the radiotherapy plan regarding the i-th region of interest based on the iterative constraint matrix of the i-th region of interest.

[0012] Optionally, the second processing submodule includes: a third processing submodule, used to determine the gradient matrix of the i-th region of interest based on the iterative constraint matrix of the i-th region of interest; a second calculation submodule, used to calculate the matrix product of the gradient matrix of the i-th region of interest and the first submatrix of the i-th region of interest to obtain the iterative gradient matrix of the i-th region of interest; and a fourth processing submodule, used to adjust the planning content of the radiotherapy plan regarding the i-th region of interest based on the iterative gradient matrix of the i-th region of interest.

[0013] Optionally, the second processing unit includes: a first processing subunit, configured to calculate the initial dose matrix of each robust scene image based on the dose calculation engine, to obtain N initial dose matrices corresponding to N robust scene images; a second processing subunit, configured to sample the N initial dose matrices for each region of interest of the target object according to the dose constraints set for each region of interest, to obtain the dose information corresponding to each region of interest in each initial dose matrix; a third processing subunit, configured to determine N intermediate dose matrices according to the dose information corresponding to each region of interest in each initial dose matrix, wherein the N intermediate dose matrices correspond one-to-one with the N robust scene images, and each intermediate dose matrix includes the dose information corresponding to M regions of interest in an initial dose matrix; and a fourth processing subunit, configured to determine N dose distribution maps corresponding to the N robust scene images based on the N intermediate dose matrices and the weights set for each beam spot.

[0014] Optionally, the fourth processing subunit includes: a resampling module, used to resample the N intermediate dose matrices according to the parameter information of the radioactive particles to obtain N target dose matrices that correspond one-to-one with the N intermediate dose matrices; and a generation module, used to generate a dose distribution map corresponding to each target dose matrix according to each target dose matrix and the weight set for each beam spot, to obtain N dose distribution maps corresponding to the N target dose matrices.

[0015] Optionally, the adaptive scenario-based proton radiotherapy planning robust optimization device further includes: a sampling correction unit, used to determine sampling correction information through a dose calculation engine based on the radiotherapy plan of the target object and the weight set for each beam spot; and a dose adjustment unit, used to adjust the dose information of N dose distribution maps based on the sampling correction information.

[0016] According to another aspect of this application, a robust optimization method for proton radiotherapy planning based on adaptive scenario-based approaches is also provided, comprising: generating N robust scenario images based on the radiotherapy plan of the target object, wherein the N robust scenario images are used to simulate the change information of M regions of interest of the target object, and N and M are both integers greater than 1; calculating the dose distribution map of each robust scenario image through a dose calculation engine to obtain N dose distribution maps corresponding to the N robust scenario images; determining a first dose distribution map and a second dose distribution map based on the N dose distribution maps, wherein the first dose distribution map is used to record the maximum dose value of the N dose distribution maps at each grid point, and the second dose distribution map is used to record the minimum dose value of the N dose distribution maps at each grid point, and each grid point corresponds to a beam spot in radiotherapy; and adjusting the radiotherapy plan based on the first dose distribution map and the second dose distribution map.

[0017] As described above, this application effectively simulates the changes in the tumor target area and its surrounding tissues under different conditions by generating N robust scene images for the target object. Compared with the existing technology that simply uses fixed outer boundaries, this application can more accurately predict the dose distribution impact caused by changes in body position, physiological state, or tumor changes. The dose distribution is calculated for each robust scene using a dose calculation engine, resulting in N dose distribution maps that cover all possible dose distribution scenarios, providing a comprehensive data foundation for subsequent robust optimization.

[0018] Secondly, the generation of the first and second dose distribution maps records the maximum and minimum dose values ​​of the target area under various scenarios, respectively. This processing method can intuitively reflect the range of uncertainty in dose distribution. Based on these two maps, the adjustment of the radiotherapy plan can specifically reduce the overlap between the tumor's outer boundary and normal tissue, avoid excessive irradiation of normal tissue, and ensure adequate coverage of the tumor area.

[0019] Furthermore, this application achieves rapid and effective dose optimization through precise calculation and constraint adjustment during the iterative optimization process, overcoming the problems of slow optimization speed and poor performance in existing technologies. The iterative dose and constraint process ensures that adjustments are made based on the latest dose distribution information and constraints in each optimization loop, thereby achieving the optimal robust plan within a limited computation time. In addition, the algorithm's parallel processing capability further shortens the optimization time and improves efficiency.

[0020] In summary, this application not only solves the technical problems of long time consumption and poor effect of robust optimization methods in the prior art, but also can respond more accurately and flexibly to changes in the tumor target area and surrounding tissues, providing a safer and more effective dose distribution solution for radiotherapy. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0022] Figure 1 This is a schematic diagram of an adaptive scenario-based proton radiotherapy planning robust optimization device according to an embodiment of this application;

[0023] Figure 2 This is a schematic diagram of an optional set of N robust scene images according to an embodiment of this application;

[0024] Figure 3 This is an example diagram of an optional N dose distribution map according to an embodiment of this application;

[0025] Figure 4 This is an example diagram of an optional first dose distribution map according to an embodiment of this application;

[0026] Figure 5 This is an example diagram of an optional second dose distribution map according to an embodiment of this application;

[0027] Figure 6 This is a schematic diagram of an optional first sequence number matrix according to an embodiment of this application;

[0028] Figure 7 This is a flowchart of a robust optimization method for proton radiotherapy planning based on adaptive scenario-based approaches, according to an embodiment of this application.

[0029] Figure 8 This is an architecture diagram of a robust optimization method for proton radiotherapy planning based on adaptive scenario-based approaches, according to an embodiment of this application. Detailed Implementation

[0030] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0031] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application 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 of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover 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.

[0032] In one alternative embodiment, the adaptive scenario-based proton radiotherapy planning robust optimization device can be a software system or an embedded system combining software and hardware. For ease of description, the adaptive scenario-based proton radiotherapy planning robust optimization device will be simplified as a system in the following description.

[0033] According to an embodiment of this application, an embodiment of a robust optimization device for proton radiotherapy planning based on adaptive scenario-based methods is provided, wherein... Figure 1 This is a schematic diagram of an adaptive scenario-based proton radiotherapy planning robust optimization device according to an embodiment of this application, as shown below. Figure 1 As shown, the device includes: a first processing unit 101, a second processing unit 102, a determination unit 103, and an adjustment unit 104.

[0034] Optionally, the first processing unit 101 is used to generate N robust scene images based on the radiotherapy plan of the target object, wherein the N robust scene images are used to simulate the change information of M regions of interest of the target object, and N and M are both integers greater than 1.

[0035] Optionally, in the field of radiotherapy, the target subject typically refers to the patient receiving treatment. A radiotherapy plan includes all parameters and details of the radiotherapy, such as the location, intensity, irradiation angle, and duration of the radiation source, as well as patient-specific anatomical information. In this embodiment, the radiotherapy plan may be a proton radiotherapy plan.

[0036] Optionally, robust scene images can be multiple image scenarios simulated based on the patient's radiotherapy plan and possible physiological, postural, or anatomical changes (such as respiratory movements, body rotation, tumor growth, etc.). Each scenario represents the possible state of the target area and surrounding tissues under different changing conditions. These scenarios are created to evaluate the robustness of the treatment plan to these changes in practical application, i.e., whether the treatment plan can still be effectively and safely executed under various possible changes. N represents the number of robust scene images generated; N greater than 1 means that at least two scenarios were created, which ensures coverage of a sufficient number of possible changes and improves the robustness and adaptability of the radiotherapy plan.

[0037] Optionally, in radiotherapy, the region of interest (ROI) typically refers to the tumor area (target zone) and the surrounding normal organs or tissues that need to be protected (also known as organs at risk). An M greater than 1 indicates that multiple different regions are considered, which may include multiple parts of the tumor or multiple normal tissues that need to be considered, ensuring comprehensive and complex management of the treatment plan.

[0038] In one optional embodiment, the second processing unit 102 is used to calculate the dose distribution map of each robust scene image through the dose calculation engine to obtain N dose distribution maps corresponding to N robust scene images.

[0039] Optionally, the dose calculation engine is a software component in the radiotherapy planning system responsible for calculating the dose distribution of radiation as it passes through the patient's body. Based on physical and biological effect models, and taking into account factors such as radiation attenuation, scattering, and tissue heterogeneity, the dose calculation engine can accurately predict the deposited dose of radiation at various sites within the patient's body. The core task of the dose calculation engine is to predict the dose distribution of radiation irradiation on a robust scene image, given radiotherapy planning parameters.

[0040] Optionally, a dose distribution map is a visual representation of the radiation dose received by different parts of a patient's body. The dose distribution map shows the distribution of the dose in three-dimensional space, typically using different colors or grayscale levels to distinguish different dose levels. Dose distribution maps are an important component of radiotherapy planning; physicians and physicists use them to assess the safety and effectiveness of the treatment plan, ensuring that the tumor area receives an adequate dose while minimizing damage to surrounding healthy tissues.

[0041] It should be noted that the N dose distribution maps can be the original dose distribution maps (also known as the initial dose matrix) calculated by the dose calculation engine based on N robust scene images, or they can be the N dose distribution maps obtained after sampling each original dose distribution map at least once, based on the N original dose distribution maps. The conditions for each sampling can be customized.

[0042] In an optional embodiment, the determining unit 103 is used to determine a first dose distribution map and a second dose distribution map based on N dose distribution maps, wherein the first dose distribution map is used to record the maximum dose value of the N dose distribution maps at each grid point, and the second dose distribution map is used to record the minimum dose value of the N dose distribution maps at each grid point, and each grid point corresponds to a beam spot in radiotherapy.

[0043] Optionally, in robust optimization methods for radiotherapy, the "first dose distribution map" and the "second dose distribution map" are used to quantify and characterize the dose distribution uncertainty under N robust scene images. Specifically, the first dose distribution map (also known as the maximum dose distribution map) records the highest dose value that each beam spot (i.e., a specific small area of ​​the patient's body irradiated by radiation, often represented by grid points in the image) can receive under all N robust scenes. This means that for each grid point, by comparing the dose distributions in the N scenes, the highest dose value is selected as the first dose distribution value for that point. This allows for the identification of areas that may receive excessively high doses under any scene during the optimization process, especially normal tissues, thereby avoiding or mitigating potential side effects.

[0044] Conversely, the second dose distribution map (also known as the minimum dose distribution map) records the lowest dose value that each plaque can receive under N robust scenarios. For each grid point, the minimum dose value among the N scenarios is selected. This map is primarily used to ensure that the tumor region receives sufficient treatment dose in all possible scenarios, achieving the treatment goal even under the most unfavorable conditions.

[0045] Optionally, Figure 2 This is a schematic diagram of an optional set of N robust scene images according to an embodiment of this application, such as... Figure 2 As shown, when N=3, there are three robust scene images: image 1, image 2, and image 3. Each image contains two regions of interest, region A and region B. The shape and / or location of the regions of interest differ in different scene images. It should be noted that N=3 and M=2 are only examples; in practical applications, N and M are any integers greater than 1.

[0046] Optionally, Figure 3 This is an example diagram of an optional N dose distribution map according to an embodiment of this application, wherein, as shown... Figure 3 As shown, dose distribution Figure 1 This is a dose distribution map (including dose values ​​1, 2, 3, and 4) obtained based on image 1; dose distribution Figure 2 This is a dose distribution map based on image 2 (including doses of 0.5, 1.5, 5, and 3); dose distribution Figure 3 This is a dose distribution map (including doses 2, 3, 4, and 1) obtained from image 3. It should be noted that... Figure 3 This is merely an illustrative example, intended to help those skilled in the art to more easily understand the technical solutions of the embodiments of this application. In practical applications, the matrix distribution of the dose distribution map can be of various forms.

[0047] Optionally, the second processing unit is further configured to detect the dose value recorded at the x-th grid point of each dose distribution map, where x is an integer greater than or equal to 1; and to determine the maximum dose value recorded at the x-th grid point of the N dose distribution maps as the value of the first dose distribution map at the x-th grid point based on the dose value recorded at the x-th grid point of each dose distribution map.

[0048] Optionally, Figure 4 This is an example diagram of an optional first dose distribution map according to an embodiment of this application, such as... Figure 4 As shown, based on dose distribution Figure 1 Dose distribution Figure 2 and dose distribution Figure 3Given that the grid points in the upper left corner are 1, 0.5, and 2, the maximum dose value at the grid point in the upper left corner of the three dose distribution maps can be determined to be 2 (from the dose distribution). Figure 3 Similarly, the maximum dose value at the grid point in the upper right corner of the three dose distribution maps is 5 (from the dose distribution). Figure 2 The maximum dose value at the grid point in the lower left corner of the three dose distribution maps is 3 (from the dose distribution). Figure 3 The maximum dose value at the grid point in the lower right corner of the three dose distribution maps is 4 (from the dose distribution). Figure 1 ).

[0049] Optionally, the second processing unit is further configured to detect the dose value recorded at the y-th grid point of each dose distribution map, where y is an integer greater than or equal to 1; and based on the dose value recorded at the y-th grid point of each dose distribution map, determine the minimum dose value recorded at the y-th grid point of the N dose distribution maps as the value of the second dose distribution map at the y-th grid point.

[0050] Optionally, Figure 5 This is an example diagram of an optional second dose distribution map according to an embodiment of this application, such as... Figure 5 As shown, based on dose distribution Figure 1 Dose distribution Figure 2 and dose distribution Figure 3 In this regard, the minimum dose value at the top left grid point of the three dose distribution maps is 0.5 (from the dose distribution). Figure 2 The minimum dose value at the grid point in the upper right corner of the three dose distribution maps is 3 (from the dose distribution). Figure 1 The minimum dose value at the bottom left grid point of the three dose distribution maps is 1.5 (from the dose distribution). Figure 2 The minimum dose value at the grid point in the lower right corner of the three dose distribution maps is 1 (from the dose distribution). Figure 3 ).

[0051] As can be seen from the above, the first and second dose distribution maps, by recording the upper and lower limits of dose distribution under N robust scenarios, provide key information for the robust design of radiotherapy, helping to achieve safer and more effective radiotherapy plan dose optimization.

[0052] In an optional embodiment, the adjustment unit 104 is used to adjust the radiotherapy plan based on the first dose distribution map and the second dose distribution map.

[0053] Optionally, in the robust optimization process of radiotherapy, adjustments to the radiotherapy plan based on the "first dose distribution map" and the "second dose distribution map" ensure that the treatment plan remains effective and safe in the face of uncertainty. This adjustment process aims to reduce the range of fluctuations in dose distribution, particularly in the tumor target area and surrounding normal tissues, to ensure treatment accuracy and reduce side effects.

[0054] Optionally, regarding how to adjust the radiotherapy plan based on the first dose distribution map and the second dose distribution map, this application provides at least the following two implementation methods:

[0055] The first implementation involves identifying the parts of normal tissue most likely to receive excessive radiation, corresponding to areas with high dose values ​​in the first dose distribution map. For these areas, the radiotherapy plan needs to be adjusted to reduce radiation exposure. The second dose distribution map identifies locations where the tumor target area's dose value is below the treatment threshold, meaning that in the worst-case scenario, these parts may not receive sufficient treatment dose.

[0056] Based on the above analysis, the direction and intensity of the beam can be adjusted. For example, for areas with excessive exposure to normal tissue, direct irradiation can be avoided by changing the beam direction or reducing the beam intensity. For areas of tumor with insufficient dose, the beam intensity can be increased or a more effective multi-beam irradiation angle can be found to ensure that the necessary therapeutic dose is achieved.

[0057] A second implementation method provides an adaptive dose compensation strategy based on the dose non-uniformity regions identified in the first and second dose distribution maps. For example, additional shielding materials or adjustable dose compensators are used to dynamically adjust the dose distribution during radiotherapy to achieve a uniform dose distribution.

[0058] Based on information from the first and second dose distribution maps, the shape and density distribution of the dose compensator are designed or adjusted. This requires the compensator to precisely correspond to the high-dose and low-dose regions shown in the dose distribution map to reduce excessive irradiation of normal tissues and increase the dose to tumor areas.

[0059] During radiotherapy, real-time monitoring technology can be used to monitor the actual dose distribution within the patient's body. If a significant deviation is found between the dose distribution and the first or second dose distribution map, an adaptive dose compensation strategy is immediately activated for adjustment, ensuring robustness and precise dose control during treatment.

[0060] Both of the above embodiments demonstrate the guiding role of the first and second dose distribution maps. By adjusting the specific parameters of radiotherapy or introducing auxiliary tools, the robustness of the treatment plan is enhanced, ensuring that the best treatment effect can still be achieved under various possible uncertain scenarios, while protecting the patient's health to the greatest extent.

[0061] It should be noted that this application also allows for further analysis based on the first dose distribution map and the second dose distribution map. For example, it can analyze the sub-map corresponding to each region of interest in the first dose distribution map and the sub-map corresponding to each region of interest in the second dose distribution map, and combine the robust scene image corresponding to the sub-map corresponding to each region of interest with the N dose distribution sub-maps corresponding to each region of interest in the N dose distribution maps, so as to adjust the planning content of each region of interest in the radiotherapy plan.

[0062] As described above, this application effectively simulates the changes in the tumor target area and its surrounding tissues under different conditions by generating N robust scene images for the target object. Compared with the existing technology that simply uses fixed outer boundaries, this application can more accurately predict the dose distribution impact caused by changes in body position, physiological state, or tumor changes. The dose distribution is calculated for each robust scene using a dose calculation engine, resulting in N dose distribution maps that cover all possible dose distribution scenarios, providing a comprehensive data foundation for subsequent robust optimization.

[0063] Secondly, the generation of the first and second dose distribution maps records the maximum and minimum dose values ​​of the target area under various scenarios, respectively. This processing method can intuitively reflect the range of uncertainty in dose distribution. Based on these two maps, the adjustment of the radiotherapy plan can specifically reduce the overlap between the tumor's outer boundary and normal tissue, avoid excessive irradiation of normal tissue, and ensure adequate coverage of the tumor area.

[0064] Furthermore, this application achieves rapid and effective dose optimization through precise calculation and constraint adjustment during the iterative optimization process, overcoming the problems of slow optimization speed and poor performance in existing technologies. The iterative dose and constraint process ensures that adjustments are made based on the latest dose distribution information and constraints in each optimization loop, thereby achieving the optimal robust plan within a limited computation time. In addition, the algorithm's parallel processing capability further shortens the optimization time and improves efficiency.

[0065] In one optional embodiment, the adjustment unit 104 includes: a first determining subunit, configured to determine a first sequence matrix based on the sequence number of the robust scene image corresponding to the dose value of each grid point in the first dose distribution map; a second determining subunit, configured to determine a second sequence matrix based on the sequence number of the robust scene image corresponding to the dose value of each grid point in the second dose distribution map; and an adjustment subunit, configured to adjust the radiotherapy plan based on the first dose distribution map, the second dose distribution map, the first sequence matrix, and the second sequence matrix.

[0066] Optionally, in the embodiments of this application, the "first sequence matrix" and the "second sequence matrix" further provide "source" information of extreme values ​​in the dose distribution, that is, which robust scenario generated the maximum dose at each grid point in the first dose distribution map and the minimum dose at each grid point in the second dose distribution map. This enables the adjustment of radiotherapy plans to be more than just a simple adjustment of the dose, but to perform targeted optimization of the dose distribution in specific scenarios, thereby improving the personalization and robustness of treatment.

[0067] Optionally, the radiotherapy plan adjustment device further includes: a sequence number determination unit, used to take the sequence number of the robust scene image corresponding to the dose value of the xth grid point in the first dose distribution map as the value at the xth matrix position in the first sequence number matrix.

[0068] Optionally, Figure 6 This is a schematic diagram of an optional first sequence number matrix according to an embodiment of this application, such as... Figure 6 As shown, since the dose distribution Figure 1 Dose distribution Figure 2 and dose distribution Figure 3 The maximum dose value at the grid point in the top left corner of the three dose distribution maps is 2 (from the dose distribution). Figure 3 The maximum dose value at the grid point in the upper right corner of the three dose distribution maps is 5 (from the dose distribution). Figure 2 The maximum dose value at the grid point in the lower left corner of the three dose distribution maps is 3 (from the dose distribution). Figure 3 The maximum dose value at the grid point in the lower right corner of the three dose distribution maps is 4 (from the dose distribution). Figure 1 The top left corner of the first sequence matrix is ​​"3" (representing dose distribution). Figure 3 The top right corner is 2 (representing dose distribution). Figure 2 The bottom left corner is 3 (representing dose distribution). Figure 3 The bottom right corner is 1 (representing dose distribution). Figure 1 ).

[0069] Optionally, the radiotherapy planning adjustment device further includes: a sequence number determination unit, used to take the sequence number of the robust scene image corresponding to the dose value of the y-th grid point in the second dose distribution map as the value at the y-th matrix position in the second sequence number matrix.

[0070] The process of determining the second sequence matrix can be referenced from the process of determining the first sequence matrix, as the principles are similar and will not be repeated here.

[0071] Optionally, regarding how to adjust the radiotherapy plan based on the first dose distribution map, the second dose distribution map, the first sequence matrix, and the second sequence matrix, this application provides at least two implementation methods.

[0072] Method 1: Adjust the beam current to adapt to the most unfavorable scenarios. For example, use the "first sequence matrix" and "second sequence matrix" to identify which robust scenarios correspond to the maximum and minimum doses for each grid point. Identify the scenarios most likely to cause over-irradiation of normal tissue and under-irradiation of the tumor region. For grid points in the first dose distribution map where normal tissue may be over-irradiated, find their corresponding most unfavorable scenario number in the first sequence matrix, and adjust the beam direction or intensity to avoid direct irradiation of these areas. For grid points in the second dose distribution map where the tumor region has a low dose, find their corresponding most unfavorable scenario number in the second sequence matrix, and selectively enhance the beam current to ensure that the tumor region receives the required treatment dose in all scenarios.

[0073] Method 2: Introduce a dose-adaptive control mechanism. For example, based on a "first sequence matrix" and a "second sequence matrix," an adaptive dose control system can be designed. This system can dynamically adjust beam parameters according to the patient's actual condition during radiotherapy to cope with the most unfavorable dose distribution scenario. During treatment, real-time image-guided technology (such as CBCT or EPID) is used to monitor changes in the patient's tumor location and surrounding tissues. Based on the monitoring results, the system quickly searches the first and second sequence matrices to identify the robust scenario closest to the current change, and then immediately adjusts the beam intensity and direction to match the optimal dose distribution under this scenario. During treatment, the adaptive control system continuously calculates the dose distribution based on the current changing scenario and compares it with the first and second dose distribution maps. If the actual dose distribution is found to be close to the first dose distribution map (risk of over-irradiation) or the second dose distribution map (risk of under-irradiation), the system will automatically adjust the beam parameters to ensure that the dose distribution can cover the tumor area without causing excessive damage to normal tissues.

[0074] Both embodiments focus on using a first dose distribution map, a second dose distribution map, and first and second sequence matrices to provide a strategy for in-depth adjustment of radiotherapy plans. This approach not only considers numerical dose variations but also incorporates contextual information that generates these variations, making treatment plan adjustments more precise and personalized. It effectively addresses various uncertainties during treatment, improving treatment outcomes and patient safety.

[0075] In one optional embodiment, the adjustment subunit includes: a first determining module, configured to determine the maximum dose distribution matrix of each region of interest based on a first dose distribution map; and to determine the minimum dose distribution matrix of each region of interest based on a second dose distribution map; a first processing module, configured to use the sub-matrix corresponding to each region of interest in the first index matrix as the first sub-matrix of that region of interest; a second processing module, configured to use the sub-matrix corresponding to each region of interest in the second index matrix as the second sub-matrix of that region of interest; and an adjustment module, configured to adjust the plan content of the radiotherapy plan for each region of interest according to the minimum dose distribution matrix, the maximum dose distribution matrix, the first sub-matrix, and the second sub-matrix of each region of interest.

[0076] Optionally, the first dose distribution map records the maximum dose value that all beam spots may receive under N robust scenarios. For each region of interest (ROI), the maximum dose value of all beam spots in that region can be extracted from the first dose distribution map to construct the maximum dose distribution matrix for that ROI. This matrix visually displays the upper limit of the dose that the ROI can receive under the most unfavorable scenario. For example, in Figure 4 In the first dose distribution diagram shown, the matrix formed by the two dose values ​​(2 and 3) on the left is the maximum dose distribution matrix corresponding to region of interest A; the matrix formed by the two dose values ​​(5 and 4) on the right is the maximum dose distribution matrix corresponding to region of interest B.

[0077] Similarly, the second dose distribution map records the minimum dose value that all beam spots can receive under N robust scenarios. For each region of interest (ROI), the minimum dose values ​​of all beam spots in that region are extracted from the second dose distribution map, constructing a minimum dose distribution matrix for that ROI. This matrix helps identify the lower limit of the dose that the ROI can receive under any scenario, especially ensuring that the tumor region receives a sufficient therapeutic dose even under the most unfavorable conditions. For example, in Figure 5 In the second dose distribution diagram shown, the matrix formed by the two dose values ​​(0.5 and 1.5) on the left is the minimum dose distribution matrix corresponding to region of interest A; the matrix formed by the two dose values ​​(3 and 1) on the right is the minimum dose distribution matrix corresponding to region of interest B.

[0078] Additionally, the first and second index matrices indicate the robust scene from which the maximum and minimum doses of each beam spot originate. For each region of interest (ROI), a submatrix corresponding to that ROI is extracted from the first index matrix; this submatrix reveals the scene information that led to the maximum dose in the ROI beam spot. For example, in... Figure 6 In the first indexed matrix shown, the submatrix formed by the two indices on the left (3 and 3) is the first submatrix corresponding to region of interest A; the submatrix formed by the two indices on the right (2 and 1) is the first submatrix corresponding to region of interest B. Similarly, the submatrix corresponding to this ROI is extracted from the second indexed matrix, i.e., the second submatrix, which reveals the scene information that leads to the minimum dose of the ROI beam spot.

[0079] Finally, based on the maximum dose distribution matrix, minimum dose distribution matrix, first submatrix, and second submatrix for each ROI, the dose distribution within the radiotherapy plan for each region of interest can be optimized and adjusted. For example, by analyzing the first submatrix and the maximum dose distribution matrix, the potential over-irradiation of normal tissues under the most unfavorable scenario can be identified, and the beam direction and intensity can be adjusted accordingly to reduce the dose to normal tissues and protect them from damage. Simultaneously, based on the second submatrix and the minimum dose distribution matrix, ensuring that the tumor region receives sufficient treatment dose under the most favorable scenario may require enhancing the beam or optimizing the beam path to overcome dose inadequacy under unfavorable scenarios. This adjustment process is an iterative optimization process that may require multiple iterations until an optimal radiotherapy plan is found that satisfies dose limitations and treatment objectives under all robust scenarios.

[0080] Through the above steps, this application provides a robust optimization method based on quantitative analysis, which not only considers the numerical changes in dose distribution but also incorporates the sequence information of the changing scenarios, enabling precise adjustment of the radiotherapy plan and improving the efficiency and safety of treatment, especially in dealing with complex cases with significant uncertainties during treatment.

[0081] In one optional embodiment, the adjustment module includes: a detection submodule, configured to detect N dose distribution sub-maps corresponding to each region of interest in the N dose distribution maps; a determination submodule, configured to determine N dose volume histograms of the region of interest based on the N dose distribution sub-maps corresponding to each region of interest; and a first adjustment submodule, configured to adjust the radiotherapy plan content related to each region of interest based on the N dose volume histograms of each region of interest, the minimum dose distribution matrix, the maximum dose distribution matrix, the first sub-matrix, and the second sub-matrix.

[0082] Optionally, firstly, for the N dose distribution maps obtained from N robust scenario simulations, the proton radiotherapy planning robust optimization device can automatically detect and extract the dose distribution information of each region of interest (including the tumor target area and key normal tissues) under each scenario, forming N dose distribution sub-maps. These sub-maps provide the dose change trend of the ROI under different scenarios, which is the basis for subsequent analysis and adjustment.

[0083] Optionally, a dose-volume histogram (DVH) is a statistical tool used to describe the volume of tissue irradiated within a certain dose range. Based on N dose distribution subplots for each ROI, N corresponding DVH plots can be generated. This step essentially quantifies the dose range and frequency received by different parts of the ROI under various scenarios.

[0084] Optionally, adjustments to the radiotherapy plan are based on a comprehensive assessment of the dose distribution of the ROI across N scenarios. By analyzing N DVHs, the dose received by tumors and normal tissues under all potential physical variations can be determined, identifying dose distribution limits and high-risk areas. Furthermore, the minimum and maximum dose distribution matrices provide the lower and upper limits of the ROI dose distribution, while the first and second sub-matrices identify the scenarios leading to these extreme dose distributions. This matrix information, combined with data from N DVHs, provides a multi-dimensional perspective for adjusting beam parameters, optimizing dose distribution, reducing normal tissue dose, and ensuring tumor dose coverage.

[0085] For example, for normal tissues, by analyzing the maximum dose distribution matrix and the first sub-matrix, the risk of excessive radiation can be identified, and the beam direction, dose rate, or shielding methods can be adjusted to reduce the radiation received by these tissues. For tumor regions, the minimum dose distribution matrix and the second sub-matrix are referenced to ensure that the tumor receives the dose level required for treatment, even in the most unfavorable scenarios. This may mean increasing the intensity of certain beams or optimizing the beam path to overcome dose inadequacy issues in certain scenarios.

[0086] Based on the analysis of N dose-dependent radiation (DVH) values, the radiotherapy plan can be iteratively adjusted as needed until the dose distribution of all regions of interest (ROIs) meets predetermined clinical criteria and robustness is improved. In each iteration, the system recalculates the DVH and evaluates the effect of the adjustment until an optimal solution is found.

[0087] In one optional embodiment, the adjustment module includes: a second adjustment submodule, configured to adjust the radiotherapy plan content regarding the i-th region of interest based on the maximum dose distribution matrix and a second sub-matrix of the i-th region of interest when the constraint condition set for the i-th region of interest is detected to be a dose reduction, wherein i is an integer greater than or equal to 1; and a third adjustment submodule, configured to adjust the radiotherapy plan content regarding the j-th region of interest based on the minimum dose distribution matrix and a first sub-matrix of the j-th region of interest when the constraint condition set for the j-th region of interest is detected to be a dose increase, wherein j is an integer greater than or equal to 1.

[0088] In one optional embodiment, the adjustment module includes: a second adjustment submodule, configured to adjust the radiotherapy plan content regarding the i-th region of interest based on the maximum dose distribution matrix and a first sub-matrix of the i-th region of interest when the constraint condition set for the i-th region of interest is detected to be a dose reduction, wherein i is an integer greater than or equal to 1; and a third adjustment submodule, configured to adjust the radiotherapy plan content regarding the j-th region of interest based on the minimum dose distribution matrix and a second sub-matrix of the j-th region of interest when the constraint condition set for the j-th region of interest is detected to be a dose increase, wherein j is an integer greater than or equal to 1.

[0089] Optionally, when the constraint set for the i-th region of interest (ROI) is dose reduction: for the i-th ROI, its maximum dose distribution matrix is ​​first located. This is because the dose reduction constraint aims to prevent the region from receiving excessive radiation dose, and the maximum dose distribution matrix contains the highest dose values ​​that can be reached under all robust scenarios. Furthermore, the first sub-matrix provides robust scenario information where the dose contribution is greatest at each beam spot (i.e., the beam incident point) in the maximum dose distribution matrix. In other words, the first sub-matrix helps the system identify which scenarios cause the highest dose to the i-th ROI, thereby locating the beam or scenario that needs adjustment.

[0090] Based on the analysis results of the maximum dose distribution matrix and the first sub-matrix, the system adjusts the portion of the radiotherapy plan involving the i-th ROI. Adjustment strategies include: reducing the intensity of certain beams, adjusting the beam direction, introducing beam blockage, or adjusting the irradiation time, to ensure that the i-th ROI does not exceed the set dose limit in all possible scenarios.

[0091] Optionally, when the constraint set for the j-th region of interest (ROI) is an increased dose, the system can lock the minimum dose distribution matrix of the j-th ROI. This is to ensure that the region receives sufficient dose under all robust scenarios to ensure the effectiveness of the treatment. Secondly, the second submatrix of the j-th ROI indicates which scenarios in the minimum dose distribution matrix contribute the least to the dose of the j-th ROI. This helps the algorithm identify beams that need to be strengthened or optimized to increase the radiation dose to the j-th ROI.

[0092] Optionally, based on the analysis of the minimum dose distribution matrix and the second sub-matrix, the system may increase the intensity of certain beams, optimize the beam direction, or extend the irradiation time if necessary, to ensure that the j-th ROI can reach or approach the lower limit of the dose required for treatment even under the most unfavorable robust scenario.

[0093] It's important to note that, regardless of whether the dose is reduced or increased, the key to the above process lies in precisely locating the beam and scene requiring adjustment using the maximum or minimum dose distribution matrix and its associated first or second sub-matrices. Based on this information, the system automatically adjusts the specific parameters of the radiotherapy plan to adapt to the dose constraints of different ROIs, ensuring that the radiotherapy plan still achieves safe and effective tumor coverage and normal tissue protection even when facing individual patient differences and uncertainties in the treatment process. This dynamic adjustment mechanism based on the dose distribution matrix and sub-matrices provides an efficient and intelligent solution for robust optimization of radiotherapy.

[0094] It should also be noted that the above process can improve the optimization speed of radiotherapy techniques. For example, if N dose values ​​are obtained for the target area in N scenarios, according to existing technologies, optimization calculations need to be performed for each dose value separately under any constraint, which means N calculations are required. However, under specific constraints, it is not necessary to focus on each dose value. Taking the target area as an example, the target area is usually subject to a constraint of increasing the dose. In this case, according to the technical solution of this application, only the minimum dose value among the N dose values ​​is considered, and then the dose is optimized by combining the minimum dose value with the constraint conditions. For the larger dose values ​​among the N dose values, no further optimization calculation is required. This is equivalent to reducing N-1 calculations, thereby improving computational efficiency and saving computational resources. Moreover, calculating N dose values ​​simultaneously can actually introduce errors. For example, it can easily cause the neural network model to "illusion" because it is still optimizing even when a larger dose value already meets the constraint conditions, which can mislead the neural network model's judgment direction. However, the technical solution of this application can avoid this problem.

[0095] In one optional embodiment, the second adjustment submodule includes: a first processing submodule, configured to determine the dose constraint matrix of the i-th region of interest based on the maximum dose distribution matrix of the i-th region of interest and the constraint conditions of the i-th region of interest; a first calculation submodule, configured to calculate the matrix product of the dose constraint matrix of the i-th region of interest and the first submatrix of the i-th region of interest to obtain the iterative constraint matrix of the i-th region of interest; and a second processing submodule, configured to adjust the plan content of the radiotherapy plan regarding the i-th region of interest based on the iterative constraint matrix of the i-th region of interest.

[0096] Optionally, radiotherapy plans typically set a dose cap for each region of interest (ROI) to prevent excessive irradiation of normal tissues or critical organs, while also setting a dose floor to ensure the tumor region receives a sufficient dose. The primary focus for dose reduction constraints is the dose cap. Based on the dose cap constraint for the ROI, the system generates a dose constraint matrix for that ROI. This matrix is ​​a threshold matrix, where each element represents the dose constraint value at the corresponding beam spot location, i.e., the maximum allowable dose value at that location.

[0097] Optionally, by performing a matrix multiplication operation on the dose constraint matrix and the first submatrix, the beam spot positions that exceed the dose constraints under the most unfavorable scenario can be identified. Through matrix multiplication, an iterative constraint matrix can be obtained, which highlights the beam spot positions that need adjustment to satisfy the dose constraints.

[0098] Optionally, the system adjusts the radiotherapy plan based on an iterative constraint matrix. This matrix indicates which beam spots or beam parameters need to be modified to reduce over-irradiation of the i-th ROI. Possible adjustments include: reducing beam intensity, adjusting the beam angle or incident point, increasing beam shielding, using shielding to reduce the dose to specific areas, or adjusting the irradiation sequence or timing to minimize the impact on sensitive areas.

[0099] Optionally, the above adjustment process can be iterative until a preset optimization criterion is reached. After each adjustment, the maximum dose distribution matrix is ​​recalculated, and the iterative constraint matrix is ​​updated until the dose distribution of the i-th ROI stabilizes within the dose constraint range, thus achieving the goal of robust optimization.

[0100] Through the above process, this application can accurately identify the beam or beam spot that needs adjustment for the dose constraints of a specific ROI in the radiotherapy plan. Through iterative optimization, it ensures that the dose distribution effectively covers the tumor area without causing unnecessary damage to normal tissues or critical organs. This robust optimization method based on matrix operations significantly improves the safety and personalization of radiotherapy.

[0101] Optionally, the third adjustment submodule is also used to determine the dose constraint matrix of the j-th region of interest based on the minimum dose distribution matrix of the j-th region of interest and the constraint conditions of the j-th region of interest; calculate the matrix product of the dose constraint matrix of the j-th region of interest and the second submatrix of the j-th region of interest to obtain the iterative constraint matrix of the j-th region of interest; and adjust the plan content of the radiotherapy plan regarding the j-th region of interest according to the iterative constraint matrix of the j-th region of interest.

[0102] In one optional embodiment, the second processing submodule includes: a third processing submodule, configured to determine the gradient matrix of the i-th region of interest based on the iterative constraint matrix of the i-th region of interest; a second calculation submodule, configured to calculate the matrix product of the gradient matrix of the i-th region of interest and the first submatrix of the i-th region of interest to obtain the iterative gradient matrix of the i-th region of interest; and a fourth processing submodule, configured to adjust the planning content of the radiotherapy plan regarding the i-th region of interest based on the iterative gradient matrix of the i-th region of interest.

[0103] Optionally, the iterative constraint matrix highlights beam spot locations where dose adjustments are needed to ensure that dose constraints for the ROI are not violated. Based on this matrix, the system can calculate the gradient of the dose constraint change for each beam spot location, i.e., how much the dose should be reduced to satisfy the constraints for each beam spot location. The gradient matrix reflects the rate of change of dose in space and is crucial for optimizing beam path and intensity. In the context of robust optimization, the gradient matrix helps identify which small adjustments to parameters in the treatment plan can effectively reduce or control the dose received by a specific ROI.

[0104] Optionally, the system performs a matrix multiplication operation between the gradient matrix of the i-th ROI and the first submatrix. The result of the matrix multiplication is an iterative gradient matrix, which contains the proposed gradient values ​​for dose adjustment for each beam spot under the most unfavorable robust scenario. These gradient values ​​guide the adjustment direction of beam parameters (such as intensity and direction) to effectively reduce the dose of the i-th ROI and achieve the goal of reducing dose constraints.

[0105] Based on the iterative gradient matrix, the system can determine which beam parameters need to be adjusted and how to adjust them to reduce the dose to the i-th ROI. Possible adjustments include: reducing beam intensity, especially at beam spot locations that contribute the most to the ROI dose; adjusting the beam direction or incident point to avoid or reduce irradiation of sensitive ROIs; introducing or adjusting beam shielding, using shielding materials to reduce the dose to the ROI from specific beam spots.

[0106] It should be noted that the above adjustment process is iterative. That is, after each adjustment, the dose distribution is recalculated, generating a new iterative constraint matrix and a first submatrix. The iterative gradient matrix is ​​then recalculated until the dose distribution of the i-th ROI stabilizes and satisfies all dose constraints. This process continuously approaches the treatment target, ensuring the robustness and effectiveness of the dose distribution.

[0107] Through the above technical process, this application provides a method for precisely adjusting beam parameters to optimize the dose distribution of a specific ROI, which can not only effectively control the dose received in sensitive areas, but also improve the adaptability and safety of treatment plans in the face of individual patient differences and treatment uncertainties.

[0108] Optionally, the proton radiotherapy planning robust optimization device is also used to determine the gradient matrix of the j-th region of interest based on the iterative constraint matrix of the j-th region of interest; calculate the matrix product of the gradient matrix of the j-th region of interest and the second submatrix of the j-th region of interest to obtain the iterative gradient matrix of the j-th region of interest; and adjust the planning content of the radiotherapy plan regarding the j-th region of interest based on the iterative gradient matrix of the j-th region of interest.

[0109] In an optional embodiment, the second processing unit includes: a first processing subunit, configured to calculate an initial dose matrix for each robust scene image based on a dose calculation engine, to obtain N initial dose matrices corresponding to N robust scene images; a second processing subunit, configured to sample the N initial dose matrices for each region of interest of the target object according to dose constraints set for each region of interest, to obtain dose information corresponding to each region of interest in each initial dose matrix; a third processing subunit, configured to determine N intermediate dose matrices based on the dose information corresponding to each region of interest in each initial dose matrix, wherein the N intermediate dose matrices correspond one-to-one with the N robust scene images, and each intermediate dose matrix includes dose information corresponding to M regions of interest in an initial dose matrix; and a fourth processing subunit, configured to determine N dose distribution maps corresponding to the N robust scene images based on the N intermediate dose matrices and weights set for each beam spot.

[0110] Optionally, firstly, a dose calculation engine (e.g., AI-based dose prediction model, Monte Carlo simulation, convolutional neural network, etc.) is used to predict the dose distribution of images under N robust scenarios, resulting in N initial dose matrices. Each dose matrix contains dose information for all beam spot locations within the patient's body under a specific scenario. Then, dose constraints are set for each region of interest (ROI) of the target object. Each ROI (e.g., tumor target area, critical organ, etc.) has its specific dose constraints, designed to ensure that the tumor region receives a sufficient dose while the dose to surrounding normal tissues remains within a safe range. Constraints include lower dose limits, upper dose limits, or mean doses.

[0111] Optionally, the system applies dose constraints for each Region of Interest (ROI) to N initial dose matrices and extracts the dose distribution of each ROI under each robust scenario through sampling. For example, sampling can be performed by statistically analyzing or averaging the dose values ​​of a specific region (ROI) in the matrix to obtain the dose information of that ROI under a specific scenario. Optionally, based on the dose information of each ROI in each initial dose matrix, the system constructs N intermediate dose matrices. Each intermediate dose matrix contains dose distribution data for M ROIs under a certain robust scenario, where M represents the number of ROIs. The role of the intermediate dose matrices is to extend the dose of a single beam spot to the dose description of the entire ROI, providing more comprehensive information for the subsequent generation of dose distribution maps.

[0112] Optionally, the system can generate N final dose distribution maps using an intermediate dose matrix and weights (W). The weights reflect the contribution of each beam spot in the treatment plan to the overall treatment and can be set based on physical parameters (such as beam intensity and direction) or clinical parameters (such as dose optimization goals). The calculation involves multiplying the dose value at each beam spot location in the intermediate dose matrix by its corresponding weight, and then summing the dose values ​​of all beam spots to obtain the overall dose distribution for each robust scenario.

[0113] In an optional embodiment, the fourth processing subunit includes: a resampling module, configured to resample the N intermediate dose matrices according to the parameter information of the radioactive particles to obtain N target dose matrices that correspond one-to-one with the N intermediate dose matrices; and a generation module, configured to generate a dose distribution map corresponding to each target dose matrix according to each target dose matrix and the weight set for each beam spot, thereby obtaining N dose distribution maps corresponding to the N target dose matrices.

[0114] Optionally, in radiotherapy, different types of radioactive particles (such as photons, protons, and electrons) possess different physical properties, such as penetrating power, scattering characteristics, and range. These properties directly affect the dose distribution pattern within the patient's body. The system resamples the previously generated N intermediate dose matrices, a process that takes into account the specific parameters of the radioactive particles. Resampling includes adjusting the resolution of the dose matrix, recalculating the dose distribution based on the particle's penetration depth, or correcting the dose distribution based on the particle's range and scattering effects to ensure that the dose calculation more accurately reflects the actual behavior of the particles within the patient's body.

[0115] Optionally, after resampling, the intermediate dose matrix for each robust scenario is converted into a target dose matrix that more accurately reflects the radiation particle dose distribution. These matrices contain the dose values ​​for each beam spot in a specific scenario, taking into account particle characteristics, providing a basis for the generation of subsequent dose distribution maps.

[0116] Optionally, the weight (W) of each beam spot reflects its contribution to the overall dose distribution. The weights are typically set based on physical parameters (such as beam intensity and beam direction) and clinical needs (such as tumor coverage and normal tissue protection). The system can multiply the beam spot dose value in each target dose matrix by its corresponding weight, and then sum the weighted dose values ​​of all beam spots to generate a dose distribution map for this robust scenario. The dose distribution map visually demonstrates which regions of the patient receive higher or lower doses, considering the radiation particle parameters and beam spot weights, as well as the uniformity and coverage of the dose distribution.

[0117] Repeat the above steps to generate corresponding dose distribution maps for each of the N robust scenarios, ultimately resulting in a set of N dose distribution maps. These maps not only reflect the dose distribution of each beam spot in each scenario, but also take into account the influence of radiation particle characteristics and beam spot weights, providing a comprehensive perspective on the robustness of treatment planning.

[0118] It should be noted that, in the embodiments of this application, the sampling process is not a necessary step; the purpose of sampling is to improve the optimization speed of radiotherapy planning. Furthermore, even if sampling is performed, the number of sampling times can be customized, for example, one sampling or two samplings.

[0119] In an optional embodiment, the radiotherapy plan adjustment device further includes: a sampling correction unit, used to determine sampling correction information through a dose calculation engine based on the radiotherapy plan of the target object and the weight set for each beam spot; and a dose adjustment unit, used to adjust the dose information of N dose distribution maps based on the sampling correction information.

[0120] Optionally, sampling offers the advantage of accelerating radiotherapy planning optimization. However, sampling also introduces errors. To correct these errors, the computer or dose calculation engine compares the initially generated N dose distribution maps with the treatment target, identifying differences between the dose distribution and the expected target, thus generating sampling correction information. This information may include specific descriptions and quantifications of issues such as dose level deviations, dose distribution inhomogeneity, insufficient ROI coverage, or over-irradiation. The system then adjusts the dose information in the N initial dose distribution maps based on the sampling correction information. This may involve recalculating the beam spot dose, adjusting the beam angle or intensity, or changing the irradiation time to correct deviations in the initial dose distribution maps.

[0121] For example, the dose calculation engine can extract all necessary parameters from the radiotherapy plan, including but not limited to specific patient information, delineation data of tumor and normal tissue, beam angle and intensity, irradiation time, and dose constraints for each region of interest (ROI). Secondly, a beam spot refers to the beam incident point designated for a specific dose distribution in the treatment plan, usually predefined in the plan file. Weights reflect the importance of each beam spot in the overall dose distribution and are used to adjust the dose contribution of each beam spot. Weights are typically set based on the beam spot's physical parameters (such as beam intensity and incident angle) and clinical needs (such as tumor dose coverage and normal tissue protection).

[0122] Optionally, based on the radiotherapy planning parameters and beam spot weights obtained during the data preparation phase, the dose calculation engine simulates the initial distribution of radiation dose within the patient's body. This calculation may be coarse or rapid, used to initially estimate the dose distribution. Subsequently, the dose calculation engine compares the initial dose distribution map with the treatment target, which includes dose constraints for each ROI, such as upper dose limit, lower dose limit, or average dose. Through analysis, beam spots or regions that have not reached the expected dose target, as well as non-uniformity or deviation in the dose distribution, are identified. Finally, mathematical methods are used to quantify the difference between the initial dose distribution and the treatment target, generating an error matrix or error distribution map. This may include calculating metrics such as the average error, root mean square error, or maximum deviation of the dose distribution.

[0123] Based on the error analysis results, the beam spots or dose distribution areas requiring correction are identified, along with specific correction methods. For example, it might be necessary to adjust the beam intensity to reduce high-dose areas or increase beam shielding to protect normal tissue. Finally, combined with the correction strategy, sampling correction information for each beam spot or region is generated. This could be a list containing parameter changes before and after correction for each beam spot, or guidance for fine-tuning the input parameters of the dose calculation engine.

[0124] In one alternative embodiment, Figure 7This is a flowchart of a robust optimization method for proton radiotherapy planning based on adaptive scenario-based approaches, according to an embodiment of this application, wherein, as... Figure 7 As shown, the method includes the following steps:

[0125] Step S701: Generate N robust scene images based on the radiotherapy plan of the target object.

[0126] Among them, N robust scene images are used to simulate the change information of M regions of interest of the target object, where N and M are both integers greater than 1;

[0127] Step S702: Calculate the dose distribution map of each robust scene image using the dose calculation engine to obtain N dose distribution maps corresponding to N robust scene images.

[0128] Step S703: Determine the first dose distribution map and the second dose distribution map based on the N dose distribution maps.

[0129] The first dose distribution map is used to record the maximum dose value of N dose distribution maps at each grid point, and the second dose distribution map is used to record the minimum dose value of N dose distribution maps at each grid point. Each grid point corresponds to a beam spot in radiotherapy.

[0130] Step S704: Adjust the radiotherapy plan based on the first dose distribution map and the second dose distribution map.

[0131] Optionally, Figure 8 This is an architecture diagram of a robust optimization method for proton radiotherapy planning based on adaptive scenario-based approaches, according to an embodiment of this application. Figure 8 As shown, the entire architecture consists of four stages: "before optimization begins", "iterative dose", "iterative constraints", and "gradient calculation".

[0132] The "Before Optimization Begins" phase includes the following steps:

[0133] Obtain CT images of the target object and the corresponding delineation files;

[0134] Generate an initial radiotherapy plan for the target patient;

[0135] Robust scenes are generated based on the initial radiotherapy plan of the target object. This includes: images from a non-robust scene and N robust scene images, where the HU value and / or location of each pixel in the N robust scene images are uncertain;

[0136] The dose influence matrix corresponding to N robust images is calculated based on the dose calculation engine (optional AI engine, Monte Carlo engine, convolution engine, etc.). (This corresponds to the initial dose matrix mentioned above.) Figure 8 In Each ROI is based on N robust scene images to obtain N initial dose matrices (ROI1, ROI2...ROIN);

[0137] Based on M Regions of Interest (ROIs) with overdose constraints, N dose influence matrices are sampled. Sampling of each ROI is performed in parallel on the GPU (e.g., sampling N millimeters below the Bragg peak, whole-image sampling as a hammer pattern below the central axis of the light spot, etc.). (During sampling, only the region containing the ROI is sampled).

[0138] The dose effect matrices from parallel sampling are merged;

[0139] Based on the characteristics of radioactive particles, further sampling is performed on the N dose influence matrices after sampling M ROIs. Parameters that can be set include, but are not limited to, cutoff distance, cutoff position, sampling shape, etc.

[0140] Set the weight W for each beam spot in the dose influence matrix.

[0141] After the above steps, a dose distribution map in a non-robust scenario can be obtained after two samplings (i.e., Figure 8 Non-robust scenarios ) and N dose distribution maps under N robust scenarios (i.e. Figure 8 Robust scenarios in ...robust scenarios ).

[0142] Optionally, the iterative dosing phase includes the following steps:

[0143] Obtain the weight w corresponding to each point in the current dose influence matrix;

[0144] Calculate the dose distribution for N scenarios based on the weight w and the sampled dose influence matrix;

[0145] The dose calculation engine is invoked to perform a complete dose calculation, correcting the dose distribution calculated after sampling (i.e. Figure 8 Intermediate dose correction;

[0146] Obtain the maximum value at each grid point in the N dose distributions to obtain the maximum dose distribution (i.e., obtain the first dose distribution map mentioned above).

[0147] Take the minimum value at each grid point in the N dose distributions to obtain the minimum dose distribution (i.e., obtain the second dose distribution map mentioned above).

[0148] The dose at each grid point in the maximum dose distribution is calculated from the nth robust scene to obtain the maximum sequence image (i.e., the first sequence matrix mentioned above), as shown below. Figure 8As shown, we can obtain the mask of the maximum existing dose in non-robust scenarios and the masks of the maximum existing dose in N robust scenarios;

[0149] The dose at each grid point in the minimum dose distribution is calculated to originate from the nth robust scene, yielding the minimum sequence image (i.e., the second sequence matrix mentioned above), as shown below. Figure 8 As shown, the mask for the minimum existing dose in non-robust scenarios and the mask for the minimum existing dose in N robust scenarios can be obtained;

[0150] By trimming N dose distributions, N dose distributions corresponding to each ROI are obtained, including obtaining the maximum dose distribution and the minimum dose distribution corresponding to each ROI;

[0151] Calculate the DVH for each ROI in N dose distributions;

[0152] Crop the image with the largest index corresponding to M ROIs (i.e. Figure 8 (ROI maximum dose sequence number Mask).

[0153] Crop the smallest sequence number image corresponding to M ROIs (i.e. Figure 8 The minimum dose sequence number of the ROI (Mask) in the data.

[0154] Optionally, the iterative constraint phase includes the following steps:

[0155] Based on non-robust scenarios and N robust scenarios, obtain the ROI matrix for the current constraint: For constraints of decreasing dose type, obtain the maximum dose matrix (e.g., the maximum dose distribution matrix of the i-th region of interest); for constraints of increasing dose type, obtain the minimum dose matrix (e.g., the minimum dose distribution matrix of the j-th region of interest). For mean-type constraints, obtain the dose matrices under N robust scenarios.

[0156] Calculate the FGAP matrix constrained under the corresponding dose matrix, including: calculating the FGAP matrix constrained under the minimum dose matrix and calculating the FGAP matrix constrained under the maximum dose matrix.

[0157] Determine the ROI sequence image, including: for dose reduction constraints, obtain the ROI maximum sequence mask (e.g., the first sub-matrix of the i-th region of interest); for dose increase constraints, obtain the ROI minimum sequence mask (e.g., the second sub-matrix of the j-th region of interest); for mean constraints, it is not necessary to obtain the sequence image.

[0158] Multiply the FGAP matrix by the corresponding ROI index Mask to obtain the FGAP matrix for the corresponding constraint (including the minimum dose FGAP matrix and the maximum dose FGAP matrix for the constraint). For mean-based constraints, skip this step and directly obtain the FGAP for each scene dose.

[0159] Optionally, the gradient calculation stage includes the following steps:

[0160] Calculate the gradient matrix for each constraint.

[0161] For dose-increasing constraints, multiply the gradient matrix of the ROI by the smallest index Mask of that ROI. For dose-decreasing constraints, multiply the gradient matrix of the ROI by the largest index Mask of that ROI. Mean constraints are skipped.

[0162] All gradients are superimposed into a single complete matrix.

[0163] The above four stages can be repeated in a loop until the optimizer converges and stops.

[0164] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0165] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0166] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0167] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0168] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0169] 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 application, 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 described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0170] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A robust optimization device for proton radiotherapy planning based on adaptive scenario-based methods, characterized in that, The method comprises the steps of: a first processing unit is configured to generate N robust scenario images based on a radiotherapy plan of a target object, wherein the N robust scenario images are used to simulate variation information of M regions of interest of the target object, and N and M are both integers greater than 1; a second processing unit is configured to calculate a dose distribution map of each robust scenario image by a dose calculation engine, thereby obtaining N dose distribution maps corresponding to the N robust scenario images; a determination unit is configured to determine a first dose distribution map and a second dose distribution map according to the N dose distribution maps, wherein the first dose distribution map is used to record maximum dose values of the N dose distribution maps in each grid point, and the second dose distribution map is used to record minimum dose values of the N dose distribution maps in each grid point, and each grid point corresponds to a beam spot in radiotherapy; an adjustment unit is configured to adjust the radiotherapy plan according to the first dose distribution map and the second dose distribution map; the adjustment unit comprises: a first determination sub-unit configured to determine a first serial number matrix according to a serial number of a robust scenario image corresponding to a dose value of each grid point in the first dose distribution map; a second determination sub-unit configured to determine a second serial number matrix according to a serial number of a robust scenario image corresponding to a dose value of each grid point in the second dose distribution map; a first determination module configured to determine a maximum dose distribution matrix of each region of interest based on the first dose distribution map, and determine a minimum dose distribution matrix of the each region of interest based on the second dose distribution map; a first processing module configured to take a sub-matrix corresponding to each region of interest in the first serial number matrix as a first sub-matrix of the region of interest; a second processing module configured to take a sub-matrix corresponding to the each region of interest in the second serial number matrix as a second sub-matrix of the region of interest; and an adjustment module configured to adjust a plan content about the each region of interest in the radiotherapy plan according to the minimum dose distribution matrix, the maximum dose distribution matrix, the first sub-matrix and the second sub-matrix of the each region of interest.

2. The adaptive scene-based proton radiotherapy plan robust optimization apparatus of claim 1, wherein, The adjustment module comprises: a detection sub-module configured to detect N dose distribution sub-maps corresponding to the each region of interest in the N dose distribution maps; a determination sub-module configured to determine N dose volume histograms of the each region of interest according to the N dose distribution sub-maps corresponding to the each region of interest; a first adjustment sub-module configured to adjust the plan content about the each region of interest in the radiotherapy plan according to the N dose volume histograms, the minimum dose distribution matrix, the maximum dose distribution matrix, the first sub-matrix and the second sub-matrix of the each region of interest.

3. The adaptive scene-based proton radiotherapy plan robust optimization apparatus of claim 1, wherein, The adjustment module comprises: a second adjustment sub-module configured to adjust the plan content about an i-th region of interest in the radiotherapy plan according to a maximum dose distribution matrix and a first sub-matrix of the i-th region of interest when a constraint condition set for the i-th region of interest is detected to be dose reduction, wherein i is an integer greater than or equal to 1. The third adjusting sub-module is configured to, when it is detected that the constraint condition set for the jth region of interest is to increase the dose, adjust the planning content in the radiotherapy plan related to the jth region of interest according to the minimum dose distribution matrix of the jth region of interest and the second sub-matrix, where j is an integer greater than or equal to 1.

4. The adaptive scene-based proton radiotherapy plan robust optimization apparatus of claim 3, wherein, The second adjusting sub-module comprises: The first processing sub-module is configured to determine the dose constraint matrix of the ith region of interest according to the maximum dose distribution matrix of the ith region of interest and the constraint condition of the ith region of interest; The first calculating sub-module is configured to calculate the matrix product of the dose constraint matrix of the ith region of interest and the first sub-matrix of the ith region of interest, to obtain the iterative constraint matrix of the ith region of interest; The second processing sub-module is configured to adjust the planning content in the radiotherapy plan related to the ith region of interest according to the iterative constraint matrix of the ith region of interest.

5. The adaptive scene-based proton radiotherapy plan robust optimization apparatus of claim 4, wherein, The second processing sub-module comprises: The third processing sub-module is configured to determine the gradient matrix of the ith region of interest according to the iterative constraint matrix of the ith region of interest; The second calculating sub-module is configured to calculate the matrix product of the gradient matrix of the ith region of interest and the first sub-matrix of the ith region of interest, to obtain the iterative gradient matrix of the ith region of interest; The fourth processing sub-module is configured to adjust the planning content in the radiotherapy plan related to the ith region of interest according to the iterative gradient matrix of the ith region of interest.

6. The adaptive scene-based proton radiotherapy plan robust optimization apparatus of claim 1, wherein, The second processing unit comprises: The first processing sub-unit is configured to calculate the initial dose matrix of each robust scenario image based on the dose calculation engine, to obtain N initial dose matrices corresponding to the N robust scenario images; The second processing sub-unit is configured to, for each region of interest of the target object, sample the N initial dose matrices according to the dose constraint condition set for each region of interest, to obtain the dose information corresponding to each region of interest in each initial dose matrix; The third processing sub-unit is configured to determine N intermediate dose matrices according to the dose information corresponding to each region of interest in each initial dose matrix, where the N intermediate dose matrices correspond to the N robust scenario images one by one, and each intermediate dose matrix includes the dose information corresponding to the M regions of interest in one initial dose matrix; The fourth processing sub-unit is configured to determine N dose distribution maps corresponding to the N robust scenario images according to the N intermediate dose matrices and the weight set for each beam spot.

7. The adaptive scene-based proton radiotherapy plan robust optimization apparatus of claim 6, wherein, The fourth processing sub-unit comprises: The resampling module is configured to resample the N intermediate dose matrices according to the parameter information of the radioactive particles, to obtain N target dose matrices corresponding to the N intermediate dose matrices one by one; and generating a dose distribution map corresponding to each target dose matrix according to the target dose matrix and the weight set for each beam spot, to obtain N dose distribution maps corresponding to the N target dose matrices.

8. The adaptive scene-based proton radiotherapy plan robust optimization apparatus of claim 6 or 7, wherein, The adaptive scene-based proton radiotherapy plan robust optimization device further includes: a sampling correction unit configured to determine sampling correction information by the dose calculation engine according to the radiotherapy plan of the target object and the weight set for each beam spot; a dose adjustment unit configured to adjust dose information of the N dose distribution maps according to the sampling correction information.

9. A method for adaptive scenario-based robust optimization of proton therapy plans, characterized in that, comprises: generating N robust scene images based on a radiotherapy plan of a target object, wherein the N robust scene images are used to simulate change information of M regions of interest of the target object, and N and M are integers greater than 1; calculating a dose distribution map of each robust scene image by a dose calculation engine to obtain N dose distribution maps corresponding to the N robust scene images; determining a first dose distribution map and a second dose distribution map according to the N dose distribution maps, wherein the first dose distribution map is used to record maximum dose values of the N dose distribution maps in each grid point, and the second dose distribution map is used to record minimum dose values of the N dose distribution maps in each grid point, each grid point corresponding to a beam spot in radiotherapy; adjusting the radiotherapy plan according to the first dose distribution map and the second dose distribution map; wherein adjusting the radiotherapy plan according to the first dose distribution map and the second dose distribution map comprises: determining a first serial number matrix according to serial numbers of robust scene images corresponding to dose values of each grid point in the first dose distribution map; determining a second serial number matrix according to serial numbers of robust scene images corresponding to dose values of each grid point in the second dose distribution map; determining a maximum dose distribution matrix of each region of interest based on the first dose distribution map; determining a minimum dose distribution matrix of each region of interest based on the second dose distribution map; taking a sub-matrix corresponding to each region of interest in the first serial number matrix as a first sub-matrix of the region of interest; taking a sub-matrix corresponding to each region of interest in the second serial number matrix as a second sub-matrix of the region of interest; and adjusting plan content about each region of interest in the radiotherapy plan according to the minimum dose distribution matrix, the maximum dose distribution matrix, the first sub-matrix and the second sub-matrix of each region of interest.

Images