Dose distribution generation method, apparatus, and electronic device

Abstract

The application provides a dose distribution generation method and device and electronic equipment, which can be applied to the technical field of medical image processing. The method comprises the following steps: determining an initial intensity distribution map according to a predicted dose distribution matched with target tissue contour data of a target object; optimizing the initial intensity distribution map by using a target optimization function and a target constraint condition to obtain a target intensity distribution map, the target optimization function is used to optimize the boundary beam intensity of the initial intensity distribution map in a boundary region, the target constraint condition is used to constrain the proportion of target voxels in the boundary region, which exceeds a preset boundary dose threshold, to be less than a preset proportion threshold, the boundary region represents the region of the boundary between the organ tissue region and the tumor tissue region, and the boundary conversion dose is obtained by converting the boundary beam intensity based on an intensity dose conversion factor; and determining a target dose distribution representing the deposition of simulated radiation irradiating the tumor tissue region and the organ tissue region according to the target intensity distribution map.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical image processing, in particular to a dose distribution generation method and device and electronic equipment. BACKGROUND

[0002] Intensity-Modulated Radiation Therapy (IMRT) technology can effectively protect the surrounding normal tissue while giving high dose to the tumor target area by modulating the photon flux. However, the treatment planning process of IMRT is very complex, usually relying on the manual "trial and error" method, not only time-consuming and laborious, but also the process of dose planning is highly dependent on the personal experience of the physicist, resulting in large error and low precision of the dose distribution. SUMMARY

[0003] In view of the above problems, the present application provides a dose distribution generation method and device and electronic equipment.

[0004] According to a first aspect of the present application, a dose distribution generation method is provided, comprising: determining an initial intensity distribution map according to a predicted dose distribution adapted to target tissue contour data of a target object, the target tissue contour data characterizing a tumor tissue region and an organ tissue region adjacent to the tumor tissue region, and the initial intensity distribution map characterizing beam intensity corresponding to the predicted dose distribution; optimizing the initial intensity distribution map by using a target optimization function and a target constraint condition to obtain a target intensity distribution map, wherein the target optimization function is used to optimize the boundary beam intensity of the initial intensity distribution map in a boundary region, and the target constraint condition is used to constrain the proportion of target voxels in the boundary region whose boundary conversion dose exceeds a preset boundary dose threshold to be less than a preset proportion threshold, the boundary region characterizing the region of the boundary between the organ tissue region and the tumor tissue region, and the boundary conversion dose being converted from the boundary beam intensity based on an intensity dose conversion factor; and determining a target dose distribution characterizing the deposition of a simulated radiation on the tumor tissue region and the organ tissue region according to the target intensity distribution map.

[0005] Optionally, the optimization of the initial intensity distribution map by using the target optimization function and the target constraint condition to obtain the target intensity distribution map comprises: fusing a plurality of initial intensity distribution maps to obtain an intermediate intensity distribution map based on the number of initial intensity distribution maps and the beam intensity of the projected beam region in the initial intensity distribution map, the dimension of the intermediate intensity distribution map being lower than that of the initial intensity distribution map; and optimizing the intermediate intensity distribution map by using the target optimization function and the target constraint condition to obtain the target intensity distribution map.

[0006] Optionally, the optimizing the intermediate intensity distribution map comprises performing the target operation: obtaining a dose deviation of the i-1th target region according to a deviation between a preset target region prescribed dose and a dose of the i-1th target region, wherein the dose of the i-1th target region is obtained by converting a target region beam intensity of the i-1th intermediate intensity distribution map in the tumor tissue region based on an intensity dose conversion factor; obtaining a dose deviation of the i-1th reference region according to the i-1th spatial weighting matrix, a dose of the i-1th reference region and a preset upper limit of the reference region dose, wherein the dose of the i-1th reference region is obtained by converting a reference region beam intensity of the i-1th intermediate intensity distribution map in the organ tissue region based on the intensity dose conversion factor, and the i-1th spatial weighting matrix represents a distance between a voxel in the organ tissue region and the boundary region in the i-1th intermediate intensity distribution map; adjusting a first weight of the i-1th intermediate intensity distribution map in the voxel of the tumor tissue region, an adaptive penalty weight of the voxel in the boundary region or a distance-aware weight of the voxel in the organ tissue region to obtain the i-th intermediate intensity distribution map, when the dose deviation of the i-1th target region is greater than a preset target region dose deviation threshold, the dose deviation of the i-1th reference region is greater than a preset reference region dose deviation threshold or a proportion of the target voxel in the boundary region of the i-1th intermediate intensity distribution map is greater than a preset proportion threshold; and determining the target intensity distribution map according to the i-th intermediate intensity distribution map when the dose deviation of the i-th target region is less than or equal to the preset target region dose deviation threshold, the dose deviation of the i-th reference region is less than or equal to the preset reference region dose deviation threshold and the proportion of the target voxel in the boundary region is less than or equal to the preset proportion threshold, wherein I≥i>1.

[0007] Optionally, the adjusting the first weight of the i-1th intermediate intensity distribution map in the voxel of the tumor tissue region, the adaptive penalty weight of the voxel in the boundary region or the distance-aware weight of the voxel in the organ tissue region to obtain the i-th intermediate intensity distribution map comprises: adjusting the adaptive penalty weight of the voxel in the boundary region in the i-1th intermediate intensity distribution map according to a preset boundary intensity adjustment rule, the proportion of the target voxel and the preset proportion threshold to obtain an adjusted adaptive penalty weight; adjusting the distance-aware weight of the voxel in the organ tissue region in the i-1th intermediate intensity distribution map according to the dose deviation of the i-1th reference region and the distance between the voxel in the organ tissue region and the boundary region in the i-1th spatial weighting matrix to obtain an adjusted distance-aware weight; and updating the i-1th intermediate intensity distribution map according to the adjusted first weight, the adjusted adaptive penalty weight or the adjusted distance-aware weight to obtain the i-th intermediate intensity distribution map.

[0008] Optionally, the target constraint condition is further used to constrain the smoothness of the i-1th intermediate intensity distribution map to be greater than a preset smoothness threshold and a dose volume constraint value in the organ tissue region to be within a preset dose volume constraint range.

[0009] Optionally, the initial intensity distribution map is determined according to the estimated dose distribution adapted to the target tissue contour data of the target object, including: performing intensity prediction on the target tissue contour data in at least one preset beam direction according to the estimated dose distribution by using an intensity prediction network to obtain an initial intensity distribution map corresponding to each of the at least one preset beam direction, wherein the preset beam direction represents a preset spatial angle of the beam irradiating the target object.

[0010] Optionally, the target dose distribution deposited by the simulated rays irradiating the tumor tissue region and the organ tissue region is obtained according to the target intensity distribution map, including: performing decomposition processing on the target intensity distribution map to obtain a plurality of rectangular apertures, wherein the rectangular aperture represents spatial shape information and intensity information of the tissue region irradiated by the beam; performing sequence optimization on the plurality of rectangular apertures by using a heuristic optimization function to obtain a multileaf collimator leaf sequence, the multileaf collimator leaf sequence being used to determine the spatiotemporal distribution and intensity distribution of the beam exit, and the optimization objective of the heuristic optimization function being to minimize the leaf total travel index of the multileaf collimator leaf sequence; and obtaining the target dose distribution based on the multileaf collimator leaf sequence.

[0011] Optionally, the target dose distribution is obtained based on the multileaf collimator leaf sequence, including: generating a simulated ray beam based on the multileaf collimator leaf sequence by using an accelerator model; and obtaining the target dose distribution by using a physical action model to process the beam energy change information obtained by transporting the simulated ray beam to a digital volume film representing the target tissue contour data of the target object.

[0012] The second aspect of the present application provides a dose distribution generation device, including: a prediction module configured to determine an initial intensity distribution map according to an estimated dose distribution adapted to target tissue contour data of a target object, the target tissue contour data representing a tumor tissue region and an organ tissue region adjacent to the tumor tissue region, and the initial intensity distribution map representing a beam intensity corresponding to the estimated dose distribution; an optimization module configured to optimize the initial intensity distribution map by using a target optimization function and a target constraint condition to obtain a target intensity distribution map, wherein the target optimization function is used to optimize a boundary beam intensity of the initial intensity distribution map in a boundary region, the target constraint condition is used to constrain a proportion of target voxels in the boundary region whose boundary conversion dose exceeds a preset boundary dose threshold to be less than a preset proportion threshold, and the boundary region represents a region of a boundary between the organ tissue region and the tumor tissue region, and the boundary conversion dose is obtained by converting the boundary beam intensity based on an intensity dose conversion factor; and a generation module configured to determine a target dose distribution deposited by simulated rays irradiating the tumor tissue region and the organ tissue region according to the target intensity distribution map.

[0013] A third aspect of the present invention provides an electronic device comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors perform the method described above.

[0014] According to the dose distribution generation method, apparatus, and electronic device provided by the present invention, by first predicting a suitable estimated dose distribution based on target tissue contour data, and then using the estimated dose distribution as prior knowledge to predict the intensity of the target tissue contour data, a high-quality initial intensity distribution map that closely approximates the clinical global radiotherapy plan is obtained. This initial intensity distribution map can serve as the optimal initial solution for the optimization process. Using the initial intensity distribution map as a starting point, the refined solution for local intensity is initiated, thereby transforming the complex global intensity optimization search into an efficient local intensity refinement. This significantly shortens the iteration process and convergence time during optimization while improving dose prediction accuracy, greatly enhancing clinical work efficiency. The constraint condition limits the proportion of target voxels whose boundary conversion dose exceeds a preset boundary dose threshold to less than a preset proportion threshold. This allows for moderate dose tolerance in boundary regions adjacent to tumor tissue and strict dose constraints in boundary regions far from tumor tissue. This not only corrects the predicted intensity error of the target intensity distribution map but also improves the uniformity of the target dose distribution. It ensures that the target dose distribution in the tumor tissue region is highly matched with the clinically approved plan. At the same time, it provides fine-grained dose protection for organs in the boundary region and organs in the non-boundary region. This enables the rapid and stable generation of high-quality treatment plans with clinically deployable target dose distributions. Attached Figure Description

[0015] The above and other objects, features and advantages of the present invention will become clearer from the following description of embodiments of the invention with reference to the accompanying drawings.

[0016] Figure 1 A flowchart of a dose distribution generation method according to an embodiment of the present invention is shown.

[0017] Figure 2 A schematic diagram of a dose prediction network and an intensity prediction network according to an embodiment of the present invention is shown.

[0018] Figure 3 A schematic diagram of generating a target dose distribution according to an embodiment of the present invention is shown.

[0019] Figure 4A The dose-product histogram curves in the target area based on different methods according to embodiments of the present invention are shown.

[0020] Figure 4B The dose-product histogram curves in the bladder region based on different methods according to embodiments of the present invention are shown.

[0021] Figure 4C A dose-volume histogram curve in a rectum region based on different methods according to an embodiment of the present application is shown.

[0022] Figure 4D A dose-volume histogram curve in a small intestine region based on different methods according to an embodiment of the present application is shown.

[0023] Figure 5 A structure block diagram of a dose distribution generation apparatus according to an embodiment of the present application is shown.

[0024] Figure 6 A block diagram of an electronic device adapted to implement a dose distribution generation method according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0025] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. It is to be understood, however, that these descriptions are merely exemplary and are intended to interpret the scope of the present application. In the following detailed description of the embodiments of the present application, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to one skilled in the art that the embodiments of the present application can be practiced without these specific details. In other instances, well-known structures and functions have not been described in detail in order to avoid obscuring aspects of the present application.

[0026] The terms used herein are merely used to describe specific embodiments, and are not intended to limit the present application. The terms "include" and "have" and the like used herein indicate the presence of the features, steps, operations and / or components, but do not exclude the presence or addition of one or more other features, steps, operations or components.

[0027] All terms used herein, including technical and scientific terms, have the same meanings as those generally understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having meanings consistent with the context of the present specification, and should not be interpreted in an idealized or overly formal manner.

[0028] In the case of using expressions similar to "at least one of A, B, and C, etc.", it is generally construed that the meaning of the expression is understood by one of ordinary skill in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, a system having B alone, a system having C alone, a system having both A and B, a system having both A and C, a system having both B and C, and / or a system having A, B, and C together, etc.).

[0029] In the process of implementing the present application, it is found that existing automatic dose prediction methods are mainly divided into two categories: one is a traditional optimization algorithm: an iterative algorithm based on gradient, which takes accurate satisfaction of clinical dose constraints as the convergence condition, and generates high-precision dose, but the iterative calculation speed is slow and a large amount of manual interaction is required; the other is a data-driven or deep learning algorithm, which directly predicts the fluence map from the patient's anatomical structure, but usually as a single heuristic method, lacks clear dosimetric guidance, resulting in limited generalization ability on complex anatomical structures, and sometimes cannot strictly satisfy all clinical constraints.

[0030] Therefore, embodiments of the present application provide a dose distribution generation method and device and electronic equipment. The method comprises: determining an initial intensity distribution map according to a predicted dose distribution matched with target tissue contour data of a target object; optimizing the initial intensity distribution map by using a target optimization function and a target constraint condition to obtain a target intensity distribution map, the target optimization function is used to optimize the boundary beam intensity of the initial intensity distribution map in a boundary region, and the target constraint condition is used to constrain the proportion of target voxels in the boundary region whose boundary conversion dose exceeds a preset boundary dose threshold to be less than a preset proportion threshold, the boundary region represents a region of a boundary between an organ tissue region and a tumor tissue region, and the boundary conversion dose is obtained by converting the boundary beam intensity based on an intensity dose conversion factor; and determining a target dose distribution deposited by simulating a radiation beam irradiating the tumor tissue region and the organ tissue region according to the target intensity distribution map.

[0031] In the technical solution of the present application, the user information (including but not limited to user personal information, user image information, user equipment information such as location information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards, necessary security measures are taken, do not violate public order and good customs, and corresponding operation entrances are provided for users to choose authorization or refusal.

[0032] It should be noted that the serial numbers of the various operations in the following method are only used to represent the operation and to describe, and should not be regarded as representing the execution order of the various operations. Unless explicitly stated, the method does not need to be executed in the order shown.

[0033] Figure 1 A flowchart of a dose distribution generation method according to an embodiment of the present application is shown.

[0034] As shown in Figure 1 , the method comprises operation S110 to operation S130.

[0035] At operation S110, an initial intensity map is determined according to a predicted dose distribution adapted to the target tissue contour data of the target object.

[0036] At operation S120, the initial intensity map is optimized by using a target optimization function and a target constraint condition to obtain a target intensity map.

[0037] At operation S130, a target dose distribution deposited in the tumor tissue region and the organ tissue region by the simulated radiation is determined according to the target intensity map.

[0038] The target object is a tumor patient, and the target tissue contour data characterizes the tumor tissue region and the organ tissue region adjacent to the tumor tissue region.

[0039] The target tissue contour data can be a tomographic image in which a tumor tissue region contour and an organ tissue region contour are delineated. The tomographic image is contrast data obtained by scanning the tumor tissue region and the organ tissue region of the target object using a computed tomography (CT) technique.

[0040] The target tissue contour data of the target object can be processed by using a dose prediction network to obtain a predicted dose distribution. The predicted dose distribution is a preliminary prediction of the three-dimensional dose distribution of the radiation after the radiation passes through the tumor tissue region and the organ tissue region.

[0041] The dose prediction network can be constructed based on an improved residual neural network module (Residual Network-101, ResNet-101) and an atrous spatial pyramid pooling module (Atrous Spatial Pyramid Pooling, ASPP).

[0042] The intensity prediction network can be used to perform intensity prediction on the target tissue contour data according to the projection of the predicted dose distribution in the field direction as guide information to obtain an initial intensity map. The intensity prediction network can be constructed by a double residual neural network module (Residual Network-50, ResNet-50) and a feature co-learning module (Feature Co-learning).

[0043] The initial intensity map represents the beam intensity corresponding to the predicted dose distribution. The higher the pixel value of a pixel point in the initial intensity map, the higher the intensity of the beam passing through the corresponding pixel point on the tomographic image.

[0044] The target optimization function is used to optimize the boundary beam intensity of the initial intensity map in the boundary region. The boundary region represents the region of the boundary between the organ tissue region and the tumor tissue region.

[0045] For example, the objective optimization function can be constructed based on a projected gradient descent algorithm.

[0046] The boundary converted dose is converted from the boundary beam intensity based on an intensity-to-dose conversion factor. The intensity-to-dose conversion factor is a dose deposition matrix that describes the conversion relationship between intensity and dose.

[0047] The preset boundary dose threshold can be a prescribed dose distribution preset by medical staff in the boundary region based on experience rules. The boundary converted dose exceeding the preset boundary dose threshold represents that the beam irradiation of the boundary region based on the boundary beam intensity is harmful to the tissue of the boundary region.

[0048] The boundary converted dose of the boundary region close to the tumor tissue is given a moderate dose tolerance, and the boundary converted dose of the boundary region far from the tumor tissue is given a strict dose constraint, allowing a certain proportion of voxels in the boundary region to have a dose exceeding the preset boundary dose threshold, thereby maximizing the radiotherapy of the tumor tissue and the protection of the organ tissue.

[0049] The target constraint condition is used to constrain the proportion of target voxels in the boundary region whose boundary converted dose exceeds the preset boundary dose threshold to be less than a preset proportion threshold.

[0050] The target intensity distribution map is used to constrain the proportion of target voxels in the boundary region whose boundary converted dose corresponding to the boundary beam intensity in the boundary region exceeds the preset boundary dose threshold to be less than a preset proportion threshold.

[0051] The target intensity distribution map is converted into a physically executable multi-leaf collimator leaf sequence, and the multi-leaf collimator leaf sequence is input into a graphics processor accelerated Monte Carlo dose calculation engine that simulates the transport process of a photon beam matching the target intensity distribution map in a target object, and then determines a target dose distribution representing the deposited dose distribution of the simulated radiation beam irradiating the tumor tissue region and the organ tissue region. The target dose distribution is converted based on the deposited energy distribution of the radiation beam.

[0052] Optionally, since the target tissue contour data is first predicted based on the adapted estimated dose distribution, and then the intensity prediction of the target tissue contour data is performed based on the estimated dose distribution as prior knowledge, an initial intensity distribution map of high quality close to the clinical global radiotherapy plan is obtained; the initial intensity distribution map can be used as the optimal initial solution of the optimization process, and the local intensity refinement solution is started from the initial intensity distribution map as the hot start point, so that the complex global intensity optimization search is converted into an efficient local intensity refinement, which not only improves the dose prediction accuracy, but also greatly shortens the iteration process and convergence time in the optimization process, and greatly improves the clinical work efficiency; the target constraint condition restricts the proportion of the target voxels whose boundary conversion dose exceeds the preset boundary dose threshold to be less than the preset proportion threshold, so that the boundary conversion dose of the boundary region close to the tumor tissue is given a moderate dose tolerance, and the boundary conversion dose of the boundary region away from the tumor tissue is given a strict dose constraint, and the target dose distribution is determined based on the obtained target intensity distribution map, which not only corrects the prediction intensity error of the target intensity distribution map, but also improves the uniformity of the target dose distribution, ensures that the target dose distribution in the tumor tissue region has a high matching degree with the clinical approved plan, and at the same time, the fine differentiated dose protection of the organ tissue in the boundary region and the organ tissue in the non-boundary region.

[0053] Optionally, the initial intensity distribution map is optimized by using the target optimization function and the target constraint condition to obtain a target intensity distribution map, including: fusing a plurality of initial intensity distribution maps based on the number of initial intensity distribution maps and the beam intensity of the projection beam region in the initial intensity distribution map to obtain an intermediate intensity distribution map, the dimension of the intermediate intensity distribution map being lower than that of the initial intensity distribution map; and optimizing the intermediate intensity distribution map by using the target optimization function and the target constraint condition to obtain the target intensity distribution map.

[0054] The intensity prediction network can generate a plurality of initial intensity distribution maps corresponding to different field directions respectively. The initial intensity distribution map is a two-dimensional high-resolution map with a resolution of 512x512, and a plurality of initial intensity distribution maps are fused and reduced in dimension to obtain a one-dimensional intermediate intensity distribution map.

[0055] In an embodiment, the intermediate intensity distribution map is calculated as shown in formula (1):

[0056] (1).

[0057] wherein, represents the initial intensity distribution map generated under the jth field, represents a downsampling function, which is used to reduce the high-resolution to a low-resolution grid (such as 101x101) required by the optimizer, denotes an effective beam mask under the jth field, denotes a beam intensity of a projected beam region in the initial intensity profile under the jth field, denotes a vectorization dimension reduction function.

[0058] The effective beam mask in intensity-modulated radiation therapy (IMRT) refers to an aperture defined by the projection of the tumor tissue region in the field direction. All grid values located inside the projection contour are set to 1, and the grid values located outside the projection contour are set to 0.

[0059] The intermediate intensity profile is optimized by using the target optimization function and the target constraint condition to obtain the target intensity profile.

[0060] Optionally, the optimization of the intermediate intensity profile comprises performing a target operation: obtaining a dose deviation of the i-1th target region according to a deviation between the dose of the i-1th target region and a preset target region prescribed dose, wherein the dose of the i-1th target region is obtained by converting the beam intensity of the target region in the tumor tissue region of the i-1th intermediate intensity profile based on the intensity dose conversion factor; obtaining a dose deviation of the i-1th reference region according to the i-1th spatial weighting matrix, the dose of the i-1th reference region and a preset upper limit of the reference region dose, wherein the dose of the i-1th reference region is obtained by converting the beam intensity of the reference region in the organ tissue region of the i-1th intermediate intensity profile based on the intensity dose conversion factor, and the i-1th spatial weighting matrix represents the distance between the voxels located in the organ tissue region in the i-1th intermediate intensity profile and the boundary region; in the case that the dose deviation of the i-1th target region is greater than a preset target region dose deviation threshold, the dose deviation of the i-1th reference region is greater than a preset reference region dose deviation threshold, or the proportion of the target voxels in the boundary region of the i-1th intermediate intensity profile is greater than a preset proportion threshold, adjusting the first weight of the voxels in the tumor tissue region, the adaptive penalty weight of the voxels in the boundary region or the distance-aware weight of the voxels in the organ tissue region of the i-1th intermediate intensity profile to obtain the i-th intermediate intensity profile; in the case that the dose deviation of the i-th target region of the i-th intermediate intensity profile is less than or equal to the preset target region dose deviation threshold, the dose deviation of the i-th reference region is less than or equal to the preset reference region dose deviation threshold, and the proportion of the target voxels in the boundary region is less than or equal to the preset proportion threshold, the target intensity profile is determined according to the i-th intermediate intensity profile, and i≥i>1.

[0061] The preset target region prescribed dose can be an upper limit value of the prescribed dose preset by medical personnel in the tumor tissue region based on experience rules.

[0062] The preset reference region dose upper limit value can be a prescribed dose upper limit value preset by medical staff based on experience rules in the organ tissue region.

[0063] The first target region dose is obtained by converting the beam intensity of the first intermediate intensity distribution map in the target region of the tumor tissue region based on the intensity dose conversion factor.

[0064] The first target region dose deviation is obtained according to the deviation between the first target region dose and the preset target region prescribed dose.

[0065] The first reference region dose is obtained by converting the beam intensity of the first intermediate intensity distribution map in the reference region of the tumor tissue region based on the intensity dose conversion factor.

[0066] The first reference region dose deviation is obtained by adjusting the deviation between the first reference region dose and the preset reference region dose upper limit value by using the first spatial weighting matrix.

[0067] In a case where the first target region dose deviation is greater than the preset target region dose deviation threshold value, or the first reference region dose deviation is greater than the preset reference region dose deviation threshold value, or the proportion of target voxels in the boundary region in the first intermediate intensity distribution map is greater than the preset proportion threshold value, the second intermediate intensity distribution map is obtained by adjusting the first weight of the voxels in the tumor tissue region, the adaptive penalty weight of the voxels in the boundary region or the distance perception weight of the voxels in the organ tissue region in the first intermediate intensity distribution map.

[0068] In the process of optimizing the intermediate intensity distribution map, the i-1 target region dose deviation is obtained according to the deviation between the i-1 target region dose and the preset target region prescribed dose. The i-1 reference region dose deviation is obtained by adjusting the deviation between the i-1 reference region dose and the preset reference region dose upper limit value by using the first spatial weighting matrix. In a case where the i-1 target region dose deviation is greater than the preset target region dose deviation threshold value, or the i-1 reference region dose deviation is greater than the preset reference region dose deviation threshold value, or the proportion of target voxels in the boundary region in the i-1 intermediate intensity distribution map is greater than the preset proportion threshold value, the i intermediate intensity distribution map is obtained by adjusting the first weight of the voxels in the tumor tissue region, the adaptive penalty weight of the voxels in the boundary region or the distance perception weight of the voxels in the organ tissue region in the i-1 intermediate intensity distribution map.

[0069] In addition, the adaptive penalty weight of the voxels in the non-boundary region in the organ tissue region can also be adjusted to minimize the proportion of target voxels exceeding the preset reference region dose upper limit value in the non-boundary region, so as to strictly control the dose in the organ tissue region away from the boundary region to be less than the preset reference region dose upper limit value, and protect the organ tissue.

[0070] The target constraint is also used to constrain the i-1th target region dose deviation to be less than or equal to a preset target region dose deviation threshold and the i-1th reference region dose deviation to be less than or equal to a preset reference region dose deviation threshold.

[0071] The target intensity distribution map is obtained until the target region dose deviation of the optimized intermediate intensity distribution map is less than or equal to the preset target region dose deviation threshold, the reference region dose deviation is less than or equal to the preset reference region dose deviation threshold, and the proportion of the target voxels in the boundary region is less than or equal to the preset proportion threshold.

[0072] The target region dose deviation less than or equal to the preset target region dose deviation threshold represents that the dose in the tumor tissue region is constrained not to deviate from the acceptable range of the tumor tissue in the clinical radiotherapy.

[0073] The reference region dose deviation less than or equal to the preset reference region dose deviation threshold represents that the dose in the organ tissue region is constrained not to deviate from the acceptable range of the organ tissue in the clinical radiotherapy.

[0074] In an embodiment, the target operation process is performed As shown in formula (2):

[0075] (2).

[0076] Wherein, N represents the number of voxels in the tumor tissue region, represents the first weight of the n-th voxel, represents the intensity dose conversion factor of the n-th voxel, represents the preset target region prescribed dose of the n-th voxel, represents the i-1th intermediate intensity distribution map, Y represents the number of voxels in the organ tissue region, represents the second weight of the y-th voxel in the organ tissue region, represents the adaptive penalty weight of the y-th voxel, represents the i-1th spatial weighting matrix of the y-th voxel, represents the distance-aware weight of the y-th voxel, represents the intensity dose conversion factor of the n-th voxel, represents the preset reference region dose upper limit value of the y-th voxel, is a smoothness parameter.

[0077] Optionally, the target constraint condition restricts the target region dose deviation to be less than or equal to a preset target region dose deviation threshold, so that the dose converted from the intensity distribution map optimized by the target optimization function in the tumor tissue region does not deviate from the acceptable range of the tumor tissue in the clinical radiotherapy; the target constraint condition restricts the reference region dose deviation to be less than or equal to a preset reference region dose deviation threshold, so that the dose converted from the intensity distribution map optimized by the target optimization function in the organ tissue region does not deviate from the acceptable range of the organ tissue in the clinical radiotherapy; the target constraint condition minimizes the proportion of the target voxels in the non-boundary region of the organ tissue region that exceed the preset reference region dose upper limit, so as to strictly control the dose of the organ tissue far from the boundary region to be less than the preset reference region dose upper limit, and maximize the protection of the organ tissue; and the target constraint condition restricts the proportion of the target voxels in the boundary region to be less than a preset proportion threshold, so as to give a moderate dose tolerance to the organ tissue in the boundary region whose dose is greater than the preset boundary dose threshold, that is, to allow the over-standard voxels of the preset proportion threshold, so as to maximize the radiotherapy of the tumor tissue.

[0078] Optionally, adjusting the first weight of the voxel in the tumor tissue region, the adaptive penalty weight of the voxel in the boundary region, or the distance-aware weight of the voxel in the organ tissue region in the i-1 intermediate intensity distribution map to obtain the i intermediate intensity distribution map comprises: adjusting the adaptive penalty weight of the voxel located in the boundary region in the i-1 intermediate intensity distribution map according to a preset boundary intensity adjustment rule, the target voxel proportion, and the preset proportion threshold to obtain an adjusted adaptive penalty weight; adjusting the distance-aware weight of the voxel located in the organ tissue region in the i-1 intermediate intensity distribution map according to the i-1 reference region dose deviation and the distance between the voxel in the organ tissue region and the boundary region in the i-1 spatial weighting matrix to obtain an adjusted distance-aware weight; and updating the i-1 intermediate intensity distribution map according to the adjusted first weight, the adjusted adaptive penalty weight, or the adjusted distance-aware weight to obtain the i intermediate intensity distribution map.

[0079] In the optimization iteration process, the system monitors the number of voxels in the boundary region that exceed the constraint in real time. If it is found that the over-standard target voxel proportion is greater than the preset proportion threshold in the i-1 iteration, the system will multiply the adaptive penalty weight of the target voxel by a specified multiple in the next iteration, so that the weight becomes larger, and the optimizer will pay more attention to the intensity optimization of the boundary region in the next iteration, overcoming the contradiction between the cold and hot spots and the insufficient protection, until convergence is achieved.

[0080] At the same time, the system also monitors the number of voxels in the non-boundary region of the organ tissue region that exceed the constraint in real time, and when it is found that the over-standard voxel proportion is greater than a lower preset proportion threshold, the adaptive penalty weight of the related organ tissue voxel is multiplied by a specified multiple in the next iteration, so as to pay attention to the intensity optimization of the related organ tissue voxel.

[0081] In an embodiment, the distance-aware weight of the y-th voxel As shown in equation (3):

[0082] (3).

[0083] wherein, characterizes the distance value of the y-th voxel in the i-1-th spatial weighting matrix table, characterizes the distance decay parameter, characterizes the basic penalty strength.

[0084] In the optimization process, the distance between the voxels in the organ tissue region and the boundary region in the i-1-th spatial weighting matrix determines whether the voxels are close to the boundary region. The distance-aware weight applies a higher penalty weight to the over-standard voxels in the organ tissue region away from the boundary region, and gives a moderate physical tolerance to the over-standard voxels close to the boundary region.

[0085] The i-1-th reference region dose deviation and the distance between the voxels in the organ tissue region and the boundary region in the i-1-th spatial weighting matrix can be processed by the projection gradient descent method to update the distance-aware weight.

[0086] The distance decay parameter is adaptively set according to the geometric scale of the tumor tissue region. For a tumor tissue region with a small volume, the distance decay parameter is set to a small value (such as 3 mm) to form a steep penalty gradient; for a large volume distance decay parameter, the distance decay parameter is appropriately increased (such as 5-8 mm) to allow natural drop of physical scattering.

[0087] The basic penalty strength is graded according to the functional importance of the organ. For serial organs, it is set to a high value; for parallel organs, it is set to a low value to allow partial low-dose overflow.

[0088] After obtaining the adjusted first weight, the second weight, the adjusted adaptive penalty weight, or the adjusted distance-aware weight, the fixed weight is updated by minimizing the dose prediction error to obtain the i-th intermediate intensity distribution map.

[0089] Optionally, by introducing the distance decay parameter, a tolerance buffer zone that conforms to the physical scattering law is constructed. In the vicinity of the boundary region, the algorithm gives a moderate physical exemption, ensuring the edge coverage of the tumor tissue region (avoiding cold spots); and in the region away from the boundary region, the algorithm applies a higher penalty to eliminate unnecessary low-dose diffusion (avoiding hot spots). This differentiated processing in accordance with the physical topology significantly improves the conformal index of the plan.

[0090] Optionally, the target constraint condition further constrains the smoothness of the i-1th intermediate intensity distribution map to be greater than a preset smoothness threshold and the dose volume constraint value of the intermediate intensity distribution map in the organ tissue region to be within a preset dose volume constraint range.

[0091] The constraint term of the target constraint condition further includes a smoothness constraint term and a dose volume constraint term.

[0092] The smoothness constraint term is a regularization term, which is used to constrain the smoothness of the intermediate intensity distribution map in the optimization process to be greater than a preset smoothness threshold.

[0093] The dose volume constraint term is used to constrain the dose volume constraint value of the intermediate intensity distribution map in the optimization process in the organ tissue region to be within a preset dose volume constraint range.

[0094] The dose volume constraint value represents the volume of voxels whose dose in the reference region of the intermediate intensity distribution map exceeds a preset upper limit of the reference region dose.

[0095] The preset dose volume constraint range includes preset dose volume constraint ranges corresponding to different organ tissue types.

[0096] Optionally, the target constraint condition constrains the smoothness of the i-1th intermediate intensity distribution map to be greater than a preset smoothness threshold, so as to ensure that the target intensity distribution map obtained through final optimization has certain smoothness and collimation, thereby being more suitable for generating a multi-leaf collimator leaf sequence that is physically and graphically executable; in addition, the dose volume constraint allows most of the organ tissues to receive a lower dose, and accurately quantifies the protection of the organ tissues.

[0097] Optionally, the initial intensity distribution map is determined according to an estimated dose distribution adapted to the target tissue contour data of the target object, including: using an intensity prediction network to perform intensity prediction on the target tissue contour data in at least one preset field direction respectively according to the estimated dose distribution, to obtain an initial intensity distribution map corresponding to each of the at least one preset field direction, wherein the preset field direction represents a preset spatial angle of the beam irradiating the target object.

[0098] The estimated dose distribution and the target tissue contour data are projected in each preset field direction respectively to obtain a two-dimensional dose projection map and a two-dimensional contour projection map corresponding to each preset field direction.

[0099] The two-dimensional dose projection map and the two-dimensional contour projection map of each preset field direction are subjected to feature extraction and intensity prediction by using the intensity prediction network to obtain the initial intensity distribution map of the preset field direction.

[0100] Figure 2 A schematic diagram of a dose prediction network and an intensity prediction network according to an embodiment of the present application is shown.

[0101] As Figure 2 shown, input the target tissue contour data into the dose prediction network to obtain a plurality of estimated dose slices, stack the plurality of estimated dose slices to obtain an estimated dose distribution; project the estimated dose distribution and the target tissue contour data in each preset field direction to obtain a two-dimensional dose projection map and a two-dimensional contour projection map corresponding to each set field direction respectively; input the two-dimensional dose projection map and the two-dimensional contour projection map into the intensity prediction network to obtain an initial intensity distribution map. Wherein, the dose prediction network comprises a residual neural network module (ResNet-101), a cavity space pyramid pooling module, a transposed convolution module, a convolution module and an activation module; the intensity prediction network comprises two residual neural network modules (ResNet-50), a double attention module, two convolution modules, two activation modules and a transposed convolution module.

[0102] Optionally, according to the target intensity distribution map, a target dose distribution representing the deposition of simulated radiation on the tumor tissue region and the organ tissue region is obtained, comprising: decomposing the target intensity distribution map to obtain a plurality of rectangular apertures, wherein the rectangular aperture represents the spatial shape information and intensity information of the tissue region of the target object irradiated by the beam; sorting and optimizing the plurality of rectangular apertures by using a heuristic optimization function to obtain a multileaf collimator leaf sequence, the multileaf collimator leaf sequence is used to determine the spatiotemporal distribution and intensity distribution of the beam exit, and the optimization target of the heuristic optimization function is to minimize the total travel index of the multileaf collimator leaf sequence; based on the multileaf collimator leaf sequence, the target dose distribution is obtained.

[0103] In an embodiment, the decomposition process of the target intensity distribution map is shown in formulas (4), (5) and (6):

[0104] (4).

[0105] (5).

[0106] (6).

[0107] Wherein, the target intensity distribution map, K represents the total number of segmentation indexes, the coefficient of the kth rectangular aperture, the kth rectangular aperture, total machine jumps, field segment number.

[0108] The decomposition process of the target intensity distribution map aims to minimize the total machine jumps and the field segment number.

[0109] The rectangular aperture is a binary rectangular grid, and the bright grid 1 represents a beam, and the dark grid 0 represents no beam. After the combination of multiple rectangular apertures, the spatial shape information and intensity information of the tissue region of the target object irradiated by the beam can be reflected.

[0110] The Lin-Kernighan-Helsgaun algorithm is introduced into the heuristic optimization function for optimization.

[0111] In an embodiment, the heuristic optimization function is shown in formulas (7) and (8).

[0112] (7).

[0113] (8).

[0114] Q represents the total number of leaf pairs in the multileaf collimator leaf sequence, 、 respectively, the left and right positions of the qth pair of leaves in the kth segment, respectively, the left and right positions of the qth pair of leaves in the k+1th segment, represents the maximum single-leaf movement distance required for switching from the rectangular aperture in the kth segment to the rectangular aperture in the k+1th segment, represents the maximum single-leaf movement distance required for switching from the rectangular aperture in the kth segment to the rectangular aperture in the k+1th segment, represents the minimized leaf total travel index.

[0115] The leaf total travel represents the motion trajectory of the leaves.

[0116] Based on the image processor accelerated Monte Carlo dose calculation method, the multileaf collimator leaf sequence is processed to obtain the target dose distribution.

[0117] Optionally, the target intensity distribution map is decomposed into multiple rectangular aperture systems to minimize the total machine jumps and field segments while maintaining high accuracy of the flow; the heuristic optimization function is used to sort and optimize the multiple rectangular apertures, which improves the transmission efficiency and provides a clinically deliverable multileaf collimator leaf sequence generation flow with integrity, higher accuracy, and shorter machine motion time after intensity prediction.

[0118] Optionally, based on the multileaf collimator leaf sequence, the target dose distribution is obtained, including: based on the multileaf collimator leaf sequence, using an accelerator model to generate a simulated ray beam; using a physical action model to process the simulated ray beam to transport the beam energy change information obtained by transporting the simulated ray beam to the digital volume film representing the target tissue contour data of the target object, to obtain the target dose distribution.

[0119] The multileaf collimator leaf sequence is a modelable sequence of an accelerator, the multileaf collimator leaf sequence is input into the accelerator model in the image processor, and a simulated ray beam is generated.

[0120] The intensity distribution of the simulated ray beam matches the target intensity distribution map.

[0121] The physical action model can be a Monte Carlo dose calculation model.

[0122] The transport thread of the image processor is used to simulate the transport of the ray beam to the digital volume film of the target tissue contour data representing the target object, the physical action occurs in the transport process, the beam energy change information changes, the physical action model is used to process the beam energy change information of the ray beam, the deposited energy is obtained, and the target dose distribution is obtained by converting the deposited energy.

[0123] For example, the beam energy change information can include reaction released energy information and exit energy information.

[0124] Figure 3 A schematic diagram of generating a target dose distribution according to an embodiment of the present application is shown.

[0125] As shown in Figure 3 In the intensity prediction stage, the target tissue contour data is input into the dose prediction network to obtain an estimated dose distribution; the estimated dose distribution and the target tissue contour data are input into the intensity prediction network to obtain an initial intensity distribution map. In the optimization process, the initial intensity distribution map is used as a hot start point, the target optimization function is used to optimize the initial intensity distribution map, when the intermediate intensity distribution map in the iterative optimization does not satisfy the target constraint condition, the first weight, the adaptive penalty weight and the distance perception weight are updated, so that the intermediate intensity distribution map is updated, until convergence is reached, and the target intensity distribution map is obtained. In the generation of the multileaf collimator leaf sequence stage, the target intensity distribution map is decomposed to obtain a plurality of rectangular apertures; the heuristic optimization function is used to sort and optimize the plurality of rectangular apertures to obtain the multileaf collimator leaf sequence. Finally, the multileaf collimator leaf sequence is processed based on the Monte Carlo dose calculation method to obtain the target dose distribution.

[0126] The method proposed in the present application is verified on a group of clinical data containing 20 cervical cancer patients, and the comparison table of the clinical dose plan according to the embodiment of the present application and the target dose distribution is shown in Table 1.

[0127] Table 1

[0128]

[0129] As shown in Table 1, in the plan evaluation, in order to evaluate the coverage of the target area (tumor tissue region), the , Dose metrics representing the dose received by 98% and 2% of the volume of the target region, respectively; to evaluate the sparing of organ tissue regions, V 50 and Dose metrics, V 50 representing the percentage of the volume of the organ receiving at least 50 Gy dose (unit: %, the lower the value, the better the protection); representing the maximum dose within the organ (unit: Gy, the lower the value, the better the protection). The target dose distribution generated by the method of the present application is comparable to the clinically approved clinical dose plan in terms of target coverage, while showing better performance in the sparing of organ tissue regions, for example, significantly reduced irradiation dose to the bladder region and small intestine region (significance p < 0.05), with a steeper dose drop and fewer hot spots. In addition, the average absolute error of the target intensity distribution predicted by the method of the present application is only 0.055, and the structural similarity is 0.94, which is significantly better than existing methods, with high precision and stability; compared with intensity optimization from zero, the "prediction-reoptimization" strategy adopted by the present application reduces the optimization time by 89%, and the entire automated process can be completed within 2 minutes, greatly improving the clinical work efficiency.

[0130] Figure 4A The dose-volume histogram curves in the target region based on different methods according to an embodiment of the present application are shown.

[0131] As shown in Figure 4A , the dose-volume histogram curve in the target region is determined based on the target dose distribution, the abscissa represents the dose in the target region, the higher to the right represents the higher dose, and the ordinate represents the relative volume, that is, the percentage of the volume of the tissue in the target region receiving irradiation of this dose to the total volume. Any point (x, y) on the graph represents that at least y% of the volume in the region receives a dose greater than or equal to x Gy. The dose-volume histogram curves in the target region are obtained by three methods, respectively, according to the deep learning guided optimization method (DL-Guided optimization), the zero initialized optimization method (Zero-Initialized optimization) and the optimization process based on the initial intensity distribution (warm-start strategy) proposed by the present application, and compared. In the target region, it is hoped that 100% of the tumor volume must be sufficiently irradiated, for example, at a dose of 50 Gy, the method of the present application realizes clinically acceptable target coverage.

[0132] Figure 4B The dose-volume histogram curves in the bladder region based on different methods according to an embodiment of the present application are shown.

[0133] As shown in Figure 4BAs shown, the horizontal axis represents the dose in the target area, with higher doses further to the right. The vertical axis represents the relative volume, i.e., the percentage of the bladder region's tissue volume irradiated by this dose. The dose-product histogram curves in the bladder region obtained using three methods—DL-Guided optimization, Zero-Initialized optimization, and the warm-start strategy proposed in this invention—are compared. The method of this invention shows the best performance in terms of the percentage of irradiated volume in the bladder region.

[0134] Figure 4C The dose-product histogram curves in the rectal region based on different methods according to embodiments of the present invention are shown.

[0135] like Figure 4C As shown, the horizontal axis represents the dose in the rectal region, with higher doses further to the right. The vertical axis represents the relative volume, i.e., the percentage of the total rectal volume that received this dose. The dose-product histogram curves in the rectal region were compared using three methods: deep learning-guided optimization, zero-initialized optimization, and the warm-start strategy proposed in this invention. The method of this invention showed the best performance in terms of the percentage of irradiated volume in the rectal region.

[0136] Figure 4D The dose-product histogram curves in the small intestine region based on different methods according to embodiments of the present invention are shown.

[0137] like Figure 4D As shown, the horizontal axis represents the dose in the target area, with higher doses further to the right. The vertical axis represents the relative volume, i.e., the percentage of the small intestine region's tissue volume irradiated by this dose. The dose-product histogram curves in the small intestine region were compared using three methods: deep learning-guided optimization, zero-initialized optimization, and the warm-start strategy proposed in this invention. The method of this invention shows the best performance in terms of the percentage of irradiated volume in the small intestine region. For example, the percentage of irradiated volume (V) in the small intestine region with a 20 Gy dose is... 20In the method of the present application, it is reduced to 38%, while the values of this number for the optimization method based on zero initialization and the optimization method based on deep learning guidance are 40% and 43% respectively.

[0138] In terms of optimization efficiency, the deep learning guided optimization method (DL-Guided optimization) converges on average after 27 ± 8 iterations, corresponding to an average optimization time of 14.7 ± 4.3 seconds. In contrast, the zero-initialized optimization method (Zero-Initialized optimization) requires an average of 213 ± 27 iterations, corresponding to a time of 84.8 ± 10.9 seconds. The results show that the optimization process based on the initial intensity distribution map (warm-start strategy) proposed in the present application significantly improves the core optimization efficiency, reducing the average optimization time by 82.6%.

[0139] Based on the above dose distribution generation method, the present application also provides a dose distribution generation device. The following will be combined with Figure 5 The device will be described in detail.

[0140] Figure 5 The structure block diagram of the dose distribution generation device according to the embodiment of the present application is shown.

[0141] As Figure 5 shown, the dose distribution generation device 500 of this embodiment includes a prediction module 510, an optimization module 520 and a generation module 530.

[0142] The prediction module 510 is configured to determine an initial intensity distribution map according to an estimated dose distribution adapted to target tissue contour data of a target object, the target tissue contour data characterizing a tumor tissue region and an organ tissue region adjacent to the tumor tissue region, and the initial intensity distribution map characterizing beam intensity corresponding to the estimated dose distribution. In an embodiment, the prediction module 510 can be configured to perform the operation S110 described above, which will not be repeated here.

[0143] The optimization module 520 is configured to optimize the initial intensity distribution map using a target optimization function and a target constraint condition to obtain a target intensity distribution map, wherein the target optimization function is used to optimize the boundary beam intensity of the initial intensity distribution map in a boundary region, and the target constraint condition is used to constrain the proportion of target voxels in the boundary region whose boundary conversion dose exceeds a preset boundary dose threshold to be less than a preset proportion threshold, the boundary region characterizing the region of the boundary between the organ tissue region and the tumor tissue region, and the boundary conversion dose being converted based on the intensity dose conversion factor. In an embodiment, the optimization module 520 can be configured to perform the operation S120 described above, which will not be repeated here.

[0144] The generating module 530 is configured to determine a target dose distribution representing a simulation of a radiation beam irradiating the tumor tissue region and the organ tissue region according to the target intensity distribution map. In an embodiment, the generating module 530 can be configured to perform the operation S130 described above, and details are not repeated here.

[0145] Optionally, the optimization module 520 includes a first optimization sub-module and a second optimization sub-module.

[0146] The first optimization sub-module is configured to fuse the plurality of initial intensity distribution maps to obtain an intermediate intensity distribution map based on the number of initial intensity distribution maps and the beam intensity of the projection beam region in the initial intensity distribution maps, wherein a dimension of the intermediate intensity distribution map is lower than a dimension of the initial intensity distribution map.

[0147] The second optimization sub-module is configured to optimize the intermediate intensity distribution map to obtain the target intensity distribution map using the target optimization function and the target constraint condition.

[0148] Optionally, the second optimization sub-module includes a first optimization unit, a second optimization unit, a third optimization unit, and a fourth optimization unit.

[0149] The first optimization unit is configured to obtain an i-1th target region dose deviation based on a deviation between the i-1th target region dose and a preset target region prescription dose, wherein the i-1th target region dose is obtained by converting a target region beam intensity of the i-1th intermediate intensity distribution map in the tumor tissue region based on an intensity dose conversion factor.

[0150] The second optimization unit is configured to obtain an i-1th reference region dose deviation based on an i-1th spatial weighting matrix, an i-1th reference region dose, and a preset reference region dose upper limit value, wherein the i-1th reference region dose is obtained by converting a reference region beam intensity of the i-1th intermediate intensity distribution map in the organ tissue region based on an intensity dose conversion factor, and the i-1th spatial weighting matrix represents a distance between a voxel in the organ tissue region and a boundary region in the i-1th intermediate intensity distribution map.

[0151] The third optimization unit is configured to adjust a first weight of a voxel in the tumor tissue region, an adaptive penalty weight of a voxel in the boundary region, or a distance-aware weight of a voxel in the organ tissue region of the i-1th intermediate intensity distribution map to obtain an i-th intermediate intensity distribution map in a case that the i-1th target region dose deviation is greater than a preset target region dose deviation threshold, the i-1th reference region dose deviation is greater than a preset reference region dose deviation threshold, or a proportion of a target voxel in the boundary region in the i-1th intermediate intensity distribution map is greater than a preset proportion threshold.

[0152] The fourth optimization unit is configured to determine the target intensity distribution map according to the i-th intermediate intensity distribution map when the i-th target region dose deviation of the i-th intermediate intensity distribution map is less than or equal to a preset target region dose deviation threshold, the i-th reference region dose deviation is less than or equal to a preset reference region dose deviation threshold, and the proportion of the target voxels in the boundary region is less than or equal to a preset proportion threshold, where i≥i>1.

[0153] Optionally, the third optimization unit comprises a first optimization subunit, a second optimization subunit and a third optimization subunit.

[0154] The first optimization subunit is configured to adjust the adaptive penalty weight of the voxels located in the boundary region in the i-1-th intermediate intensity distribution map according to the preset boundary intensity adjustment rule, the proportion of the target voxels and the preset proportion threshold, to obtain an adjusted adaptive penalty weight.

[0155] The second optimization subunit is configured to adjust the distance-aware weight of the voxels located in the organ tissue region in the i-1-th intermediate intensity distribution map according to the i-1-th reference region dose deviation and the distance between the voxels in the organ tissue region and the boundary region in the i-1-th spatial weighting matrix, to obtain an adjusted distance-aware weight.

[0156] The third optimization subunit is configured to update the i-1-th intermediate intensity distribution map according to the adjusted first weight, the adjusted adaptive penalty weight or the adjusted distance-aware weight, to obtain the i-th intermediate intensity distribution map.

[0157] Optionally, the prediction module 510 comprises a prediction sub-module.

[0158] The prediction sub-module is configured to perform intensity prediction on the target tissue contour data in at least one preset field direction respectively according to the estimated dose distribution by using an intensity prediction network, to obtain an initial intensity distribution map corresponding to each of the at least one preset field direction, where the preset field direction represents a preset spatial angle of beam irradiation on a target object.

[0159] Optionally, the generation module 530 comprises a first generation sub-module, a second generation sub-module and a third generation sub-module.

[0160] The first generation sub-module is configured to perform decomposition processing on the target intensity distribution map, to obtain a plurality of rectangular apertures, where the rectangular aperture represents spatial shape information and intensity information of the beam irradiation on the tissue region of the target object.

[0161] The second generation sub-module is configured to perform sorting optimization on the plurality of rectangular apertures by using a heuristic optimization function, to obtain a multileaf collimator leaf sequence, the multileaf collimator leaf sequence being used to determine a spatiotemporal distribution and an intensity distribution of the beam emission, and an optimization objective of the heuristic optimization function being to minimize a leaf total travel index of the multileaf collimator leaf sequence.

[0162] a third generation sub-module, configured to obtain the target dose distribution based on the multileaf collimator leaf sequence.

[0163] Optionally, the third generation sub-module comprises a first generation unit and a second generation unit.

[0164] The first generation unit is configured to generate a simulated ray beam based on the multileaf collimator leaf sequence and the accelerator model.

[0165] The second generation unit is configured to obtain the target dose distribution by processing the beam energy variation information obtained by transporting the simulated ray beam onto the digital volume film representing the target tissue contour data of the target object using the physical action model.

[0166] Optionally, any of the prediction module 510, the optimization module 520 and the generation module 530 can be combined in one module, or any of the modules can be split into multiple modules. Alternatively, at least part of the function of one or more of the modules can be combined with at least part of the function of the other modules, and implemented in one module. Optionally, at least one of the prediction module 510, the optimization module 520 and the generation module 530 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on board, a system in package, an application specific integrated circuit (ASIC), or any other reasonable way of integrating or packaging a circuit, etc. hardware or firmware, or in any one of software, hardware and firmware or in any appropriate combination of any of them. Alternatively, at least one of the prediction module 510, the optimization module 520 and the generation module 530 can be at least partially implemented as a computer program module which, when executed, can perform the corresponding function.

[0167] Figure 6 A block diagram of an electronic device suitable for implementing the dose distribution generation method according to an embodiment of the present application is shown.

[0168] Figure 6 The electronic device shown is only an example and should not impose any limitation on the function and use range of the embodiments of the present application.

[0169] As Figure 6As shown, the computer electronic device 600 according to embodiments of the present application includes a processor 601 which can perform various appropriate actions and processes according to programs stored in a ROM 602 (Read Only Memory) or loaded into a RAM 603 (Random Access Memory) from a storage section 608. The processor 601 can include, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor, and / or a related chipset, and / or a special purpose microprocessor (e.g., an application specific integrated circuit (ASIC)), etc. The processor 601 can also include on-board memory for cache purposes. The processor 601 can include a single processing unit or multiple processing units to perform the various actions of the method processes according to embodiments of the present application.

[0170] In the RAM 603, various programs and data required for the operation of the electronic device 600 are stored. The processor 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. The processor 601 performs various operations of the method processes according to embodiments of the present application by executing the programs in the ROM 602 and / or the RAM 603. Note that the programs can also be stored in one or more memories other than the ROM 602 and the RAM 603. The processor 601 can also perform various operations of the method processes according to embodiments of the present application by executing the programs stored in the one or more memories.

[0171] Optionally, the electronic device 600 can also include an input / output (I / O) interface 605 which is also connected to the bus 604. The electronic device 600 can also include one or more of the following components connected to the input / output (I / O) interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a display such as a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as necessary. A removable medium 611 such as a magnetic disk, a magneto-optical disk, a semiconductor memory, etc. is attached to the drive 610 as necessary, so that a computer program read out from it is installed into the storage section 608 as necessary.

[0172] Optionally, the method flow according to the embodiments of the present application can be implemented as a computer software program. For example, the embodiments of the present application include a computer program product comprising a computer program carrying computer program code for executing the method shown in the flow chart. In such embodiments, the computer program can be downloaded and installed from a network through the communication part 609, and / or installed from the detachable medium 611. When the computer program is executed by the processor 601, the above-mentioned functions defined in the system / apparatus according to the embodiments of the present application are executed. Optionally, the system, apparatus, device, module, unit, etc. described above can be implemented by computer program modules.

[0173] The present application also provides a computer readable storage medium, which can be included in the device / apparatus / system described in the above embodiments, or exist separately without being assembled into the device / apparatus / system. The computer readable storage medium carries one or more programs, which, when executed, implement the dose distribution generation method according to the embodiments of the present application.

[0174] Optionally, the computer readable storage medium can be a non-volatile computer readable storage medium. For example, it can include but is not limited to portable computer diskette, hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any appropriate combination thereof. In the present application, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus or device.

[0175] For example, optionally, the computer readable storage medium can include the ROM 602 and / or the RAM 603 described above and / or one or more memories other than the ROM 602 and the RAM 603.

[0176] The embodiments of the present application also include a computer program product comprising a computer program containing program code for executing the method provided by the embodiments of the present application, which program code is used to make the electronic device implement the dose distribution generation method provided by the embodiments of the present application when the computer program product is run on the electronic device.

[0177] When the computer program is executed by the processor 601, the above-mentioned functions defined in the system / apparatus according to the embodiments of the present application are executed. Optionally, the system, apparatus, device, module, unit, etc. described above can be implemented by computer program modules.

[0178] In one embodiment, the computer program can be tangibly embodied, e.g., in a non-transitory computer-readable storage medium such as one or more of a magnetic, optical, or semiconductor storage medium. In another embodiment, the computer program can be tangibly embodied, e.g., in a transmission medium, e.g., in a carrier wave transmitted over an electronic, optical, or electromagnetic network. In yet another embodiment, the computer program can be tangibly embodied, e.g., in a combination of one or more of a magnetic, optical, or semiconductor storage medium, and a carrier wave transmitted over an electronic, optical, or electromagnetic network. As an example, the computer program can be tangibly embodied in one or more of a magnetic, optical, or semiconductor storage medium, and / or in a carrier wave transmitted over an electronic, optical, or electromagnetic network. The computer program can also be distributed, e.g., over network(s) and / or for download over the network(s): as an example, the

[0179] Optionally, program code implementing a computer program for executing embodiments of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++, or the like, and / or procedural programming languages, such as the "C" programming language, python, or the like, and / or assembly or machine language. Program code can execute entirely on the user's computing device, partly on the user's computing device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network ("LAN") or a wide area network ("WAN"), or the connection can be made to an external computing device, e.g., through the Internet using an Internet Service Provider ("ISP"). It will be appreciated that program code can be tangibly embodied, e.g., in a non-transitory computer-readable storage medium.

[0180] The computer program can also be tangibly embodied, e.g., in one or more computer program storage media - e.g., in one or more physical computer-readable storage media. The physical computer-readable storage media can include non-transitory computer-readable storage media, and / or communication media, and / or transmission media. The physical computer-readable storage media can include, but is not limited to, magnetic, optical, semiconductor, and / or other storage media. The computer program storage media can be tangible and / or non-transitory. The computer program can be downloaded from a network, e.g., the Internet, or can be downloaded from another computer program storage medium, e.g., a magnetic, optical, semiconductor, and / or other computer program storage medium. The computer program can be downloaded from a network, e.g., the Internet, or can be downloaded from another computer program storage medium, e.g., a magnetic, optical, semiconductor, and / or other computer program storage medium. The computer program can be downloaded from a network, e.g., the Internet, or can be downloaded from another computer program storage medium, e.g., a magnetic, optical, semiconductor, and / or other computer program storage medium.

[0181] The embodiments of the application have been described. However, these embodiments are merely for illustration and are not intended to limit the scope of the application. Although each embodiment is described above separately, this does not mean that the measures in each embodiment cannot be used advantageously in combination. Various alternatives and modifications to the embodiments described herein will be apparent to those skilled in the art in view of the foregoing without departing from the scope of the application.

Claims

1. A method of dose distribution generation, characterized by, The method comprises: determining an initial intensity distribution map according to a predicted dose distribution adapted to target tissue profile data of a target object, the target tissue profile data characterizing a tumor tissue region and an organ tissue region adjacent to the tumor tissue region, the initial intensity distribution map characterizing beam intensities corresponding to the predicted dose distribution; optimizing the initial intensity distribution map using a target optimization function and a target constraint condition to obtain a target intensity distribution map, wherein the target optimization function is used to optimize boundary beam intensities of the initial intensity distribution map in a boundary region, the target constraint condition is used to constrain a proportion of target voxels in the boundary region whose boundary conversion dose exceeds a preset boundary dose threshold to be less than a preset proportion threshold, the boundary region characterizes a region of a boundary between the organ tissue region and the tumor tissue region, and the boundary conversion dose is converted from the boundary beam intensities based on an intensity dose conversion factor; determining a target dose distribution deposited by simulated radiation irradiating the tumor tissue region and the organ tissue region according to the target intensity distribution map.

2. The method of claim 1, wherein, The optimization of the initial intensity distribution map using the target optimization function and the target constraint condition to obtain the target intensity distribution map comprises: fusing a plurality of initial intensity distribution maps to obtain an intermediate intensity distribution map based on a number of the initial intensity distribution maps and beam intensities of projected beam regions in the initial intensity distribution maps, the intermediate intensity distribution map having a dimension lower than that of the initial intensity distribution map; optimizing the intermediate intensity distribution map using the target optimization function and the target constraint condition to obtain the target intensity distribution map.

3. The method of claim 2, wherein, The optimization of the intermediate intensity distribution map comprises performing a target operation: obtaining an i-1th target region dose deviation according to a deviation between an i-1th target region dose and a preset target region prescription dose, wherein the i-1th target region dose is converted from target region beam intensities of an i-1th intermediate intensity distribution map in the tumor tissue region based on the intensity dose conversion factor; obtaining an i-1th reference region dose deviation according to an i-1th spatial weighting matrix, an i-1th reference region dose and a preset reference region dose upper limit value, wherein the i-1th reference region dose is converted from reference region beam intensities of the i-1th intermediate intensity distribution map in the organ tissue region based on the intensity dose conversion factor, and the i-1th spatial weighting matrix characterizes distances between voxels in the organ tissue region in the i-1th intermediate intensity distribution map and the boundary region; in a case where the i-1th target region dose deviation is greater than a preset target region dose deviation threshold, the i-1th reference region dose deviation is greater than a preset reference region dose deviation threshold, or a proportion of target voxels in the i-1th intermediate intensity distribution map in the boundary region is greater than a preset proportion threshold, adjusting a first weight of voxels in the tumor tissue region, an adaptive penalty weight of voxels in the boundary region or a distance-aware weight of voxels in the organ tissue region of the i-1th intermediate intensity distribution map to obtain an i-th intermediate intensity distribution map. In a case that the first target region dose deviation of the first intermediate intensity distribution map is less than or equal to the preset target region dose deviation threshold, the first reference region dose deviation of the first intermediate intensity distribution map is less than or equal to the preset reference region dose deviation threshold, and the proportion of the target voxels in the boundary region is less than or equal to the preset proportion threshold, the target intensity distribution map is determined according to the first intermediate intensity distribution map, I≥i>1.

4. The method of claim 3, wherein, The adjusting the first weight of the (i-1)th intermediate intensity distribution map, the adaptive penalty weight of the voxels in the boundary region, or the distance-aware weight of the voxels in the organ tissue region, to obtain the ith intermediate intensity distribution map, includes: According to the target voxels proportion, the preset proportion threshold, and the preset boundary intensity adjustment rule, the adaptive penalty weight of the voxels in the boundary region in the (i-1)th intermediate intensity distribution map is adjusted to obtain an adjusted adaptive penalty weight; According to the (i-1)th reference region dose deviation and the distance between the voxels in the organ tissue region and the boundary region in the (i-1)th spatial weighting matrix, the distance-aware weight of the voxels in the organ tissue region in the (i-1)th intermediate intensity distribution map is adjusted to obtain an adjusted distance-aware weight; According to the adjusted first weight, the adjusted adaptive penalty weight, or the adjusted distance-aware weight, the (i-1)th intermediate intensity distribution map is updated to obtain the ith intermediate intensity distribution map.

5. The method of claim 3, wherein, The target constraint condition is further used to constrain the smoothness of the (i-1)th intermediate intensity distribution map to be greater than a preset smoothness threshold, and the dose volume constraint value in the organ tissue region to be within a preset dose volume constraint range.

6. The method of claim 1, wherein, The initial intensity distribution map is determined according to a predicted dose distribution matched with target tissue contour data of a target object, including: Intensity prediction is performed on the target tissue contour data in at least one preset field direction respectively by using an intensity prediction network according to the predicted dose distribution, to obtain an initial intensity distribution map corresponding to each of the at least one preset field direction, wherein the preset field direction represents a preset spatial angle of beam irradiation on the target object.

7. The method of claim 1, wherein, The target dose distribution deposited by simulated rays irradiating the tumor tissue region and the organ tissue region is obtained according to the target intensity distribution map, including: The target intensity distribution map is decomposed to obtain a plurality of rectangular apertures, wherein the rectangular aperture represents spatial shape information and intensity information of the tissue region irradiated by the beam to the target object; The plurality of rectangular apertures are sorted and optimized by using a heuristic optimization function to obtain a multileaf collimator leaf sequence, the multileaf collimator leaf sequence is used to determine the spatiotemporal distribution and intensity distribution of the beam exit, and the optimization objective of the heuristic optimization function is to minimize the total travel index of the multileaf collimator leaf sequence; The target dose distribution is obtained based on the multileaf collimator leaf sequence.

8. The method of claim 7, wherein, The target dose distribution is obtained based on the multileaf collimator leaf sequence, including: A simulated ray beam is generated by using an accelerator model based on the multileaf collimator leaf sequence; The beam energy variation information obtained by transporting the simulated ray beams onto a digital volume film of target tissue profile data representing a target object using a physical action model processing is used to obtain a target dose distribution.

9. A dose distribution generation apparatus comprising: a prediction module configured to determine an initial intensity distribution map according to a predicted dose distribution adapted to target tissue profile data of a target object, the target tissue profile data representing a tumor tissue region and an organ tissue region adjacent to the tumor tissue region, the initial intensity distribution map representing beam intensities corresponding to the predicted dose distribution; an optimization module configured to optimize the initial intensity distribution map using a target optimization function and a target constraint condition to obtain a target intensity distribution map, wherein the target optimization function is configured to optimize boundary beam intensities of the initial intensity distribution map at a boundary region, the target constraint condition is configured to constrain a proportion of target voxels in the boundary region whose boundary converted dose exceeds a preset boundary dose threshold to be less than a preset proportion threshold, the boundary region representing a region of a boundary between the organ tissue region and the tumor tissue region, and the boundary converted dose is converted from the boundary beam intensities based on an intensity-to-dose conversion factor; a generation module configured to determine a target dose distribution representing a dose deposited by simulated ray irradiation of the tumor tissue region and the organ tissue region according to the target intensity distribution map.

10. An electronic device comprising: one or more processors; a memory configured to store one or more computer programs, wherein the one or more processors are configured to invoke the one or more computer programs to implement steps of the method according to any one of claims 1-8.

Images