Double-current neural network model applied to radiotherapy and dose reconstruction method
By processing two-dimensional projection images and three-dimensional CT images using a dual-stream neural network model, a three-dimensional dose distribution matrix is generated, which solves the problems of high computational resources and slow reconstruction speed in existing technologies and achieves real-time and accurate dose verification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA INST FOR RADIATION PROTECTION
- Filing Date
- 2025-12-31
- Publication Date
- 2026-05-12
AI Technical Summary
Existing methods for radiotherapy dose verification require high computational resources and have slow reconstruction speeds, failing to meet the needs for real-time and accurate verification. In particular, the computational accuracy is significantly reduced near heterogeneous tissue interfaces.
A dual-stream neural network model is adopted, including a two-dimensional processing branch, a three-dimensional processing branch, and a feature fusion and dose reconstruction branch. The projected images and CT images are processed by two-dimensional encoders and three-dimensional encoders. Combined with a geometric position embedding module, a manifold mapping dimensionality-upgrading module, and a cross-modal cross-attention fusion module, a three-dimensional dose distribution matrix is generated.
It achieves real-time 3D dose distribution matrix reconstruction with low computational complexity, meets the requirements for real-time accurate verification, and improves reconstruction speed and accuracy.
Smart Images
Figure CN122025010A_ABST
Abstract
Description
Technical Field
[0001] This specification relates to the field of radiotherapy technology, and in particular to a dual-flow neural network model and dose reconstruction method for radiotherapy. Background Technology
[0002] Radiotherapy is an important clinical treatment for cancer, and its effectiveness depends on precise dose control. During radiotherapy, accurately depositing the dose designed in the treatment plan onto the tumor target area while avoiding excessive irradiation of surrounding normal tissues or organs at risk (OAR) is key to improving efficacy and reducing side effects.
[0003] The dose distribution calculated by the treatment planning system is a real-time, idealized scenario. However, the actual dose delivered to the patient may deviate due to factors such as machine performance fluctuations, patient positioning errors, and changes in anatomical structure. Therefore, validating the actual delivered dose is crucial for ensuring the safety and effectiveness of treatment. An X-ray flat panel detector can measure the emitted X-ray beam information passing through the phantom / patient before or during treatment, providing a means for dose validation.
[0004] However, existing treatment validation methods have high computational resource requirements, slow reconstruction speed, and rely on some simplification assumptions, which leads to a significant reduction in computational accuracy near heterogeneous tissue interfaces, making it impossible to meet the needs of real-time accurate validation. Summary of the Invention
[0005] This specification provides a dual-flow neural network model and dose reconstruction method for radiotherapy, which at least partially solves the aforementioned problems existing in the prior art.
[0006] The following technical solution is adopted in this specification: This specification provides a two-stream neural network model for radiotherapy, including a two-dimensional processing branch, a three-dimensional processing branch, and a feature fusion and dose reconstruction branch; The two-dimensional processing branch includes a two-dimensional encoder, which is used to process the input two-dimensional projection image to determine the two-dimensional beam feature map. The three-dimensional processing branch includes a three-dimensional encoder, which is used to process the input three-dimensional CT images to determine three-dimensional anatomical feature maps. The dose reconstruction branch includes a geometric location embedding module, a manifold mapping dimensionality enhancement module, and a cross-modal cross-attention fusion module. The geometric location embedding module is used to determine the coordinate index mapping from the three-dimensional voxel space to the two-dimensional feature plane based on the field geometry parameters. The manifold mapping dimensionality enhancement module is used to sample and differ the two-dimensional beam feature map along the ray direction under the constraint of the coordinate index mapping to generate a three-dimensional beam feature volume. The cross-modal cross-attention fusion module is used to perform weighted fusion of the beam features based on the three-dimensional anatomical feature map and the three-dimensional beam feature volume through a voxel-level cross-modal attention mechanism to generate a fused feature tensor. Based on the fused feature tensor, a three-dimensional dose regression network is used to output the corresponding three-dimensional dose distribution matrix.
[0007] Preferably, the dual-stream neural network model applied to radiotherapy includes a preprocessing module; The preprocessing module is used to perform filtering and / or image / dose calibration processing on the input two-dimensional projection image.
[0008] Preferably, the two-dimensional encoder is used to extract a two-dimensional beam feature map containing beam intensity information and the shape of the multi-leaf collimator from the two-dimensional projected image layer by layer.
[0009] Preferably, the dual-stream neural network model is configured with a hybrid loss function, which includes two influencing factors: voxel-level error and dose distribution structure error.
[0010] On the other hand, this specification provides a dose reconstruction method for radiotherapy, utilizing the dual-stream neural network model for radiotherapy provided above, the method comprising: Acquire 3D CT images and capture 2D projection images of the patient in real time during the treatment process; The three-dimensional CT image and the two-dimensional projection image are input into the two-stream neural network model to determine the three-dimensional dose distribution matrix output by the two-stream neural network model; The three-dimensional dose distribution matrix is registered and compared with the planned dose in the treatment plan, and the comparison result is determined in real time.
[0011] Preferably, the step of registering and comparing the three-dimensional dose distribution matrix with the planned dose in the treatment plan, and determining the comparison result in real time, includes: Based on the three-dimensional dose distribution matrix, the real-time three-dimensional throughput and the real-time dose-volume histogram of organs at risk are determined. Based on the planned dose, the planned three-dimensional pass rate and the dose-volume histogram of the planned organs at risk are determined; The real-time 3D pass rate is compared with the planned 3D pass rate, and the real-time dose-volume histogram of organs at risk is compared with the planned dose-volume histogram of organs at risk, and the comparison results are determined in real time.
[0012] Preferably, the method further includes: Based on the comparison results, a treatment accident was determined; Issue an alert to remind doctors to pause treatment or adjust subsequent treatment plans.
[0013] On the other hand, the computer-readable storage medium provided in this specification stores a computer program that, when executed by a processor, implements the dose reconstruction method for radiotherapy provided in one aspect above.
[0014] On the other hand, this specification provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the dose reconstruction method for radiotherapy provided in one aspect above.
[0015] On the other hand, this specification provides a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to implement the dose reconstruction method for radiotherapy provided in the above-mentioned aspect.
[0016] The above-mentioned technical solutions adopted in this specification can achieve the following beneficial effects: Based on the above, the dual-stream neural network model includes a two-dimensional processing branch, a three-dimensional processing branch, and a feature fusion and dose reconstruction branch. The two-dimensional processing branch includes a two-dimensional encoder, which processes the input two-dimensional projection image to determine a two-dimensional beam feature map. The three-dimensional processing branch includes a three-dimensional encoder, which processes the input three-dimensional CT image to determine a three-dimensional anatomical feature map. The dose reconstruction branch includes a geometric location embedding module, a manifold mapping dimensionality enhancement module, and a cross-modal cross-attention fusion module. The geometric location embedding module is used to determine the coordinate index mapping from the three-dimensional voxel space to the two-dimensional feature plane based on the field geometry parameters. The manifold mapping dimensionality enhancement module is used to sample and differ the two-dimensional beam feature map along the ray direction under the geometric location embedding constraint to generate a three-dimensional beam feature volume. The cross-modal cross-attention fusion module is used to perform weighted fusion of the beam features based on the three-dimensional anatomical feature map and the three-dimensional beam feature volume through a voxel-level cross-modal attention mechanism to generate a fused feature tensor. Based on the fused feature tensor, the corresponding three-dimensional dose distribution matrix is output through a three-dimensional dose regression network.
[0017] As can be seen, the model has low computational complexity, can determine the three-dimensional dose distribution matrix in real time based on real-time acquired data, and has a fast reconstruction speed, which can meet the needs of real-time accurate verification. Attached Figure Description
[0018] The accompanying drawings, which are included to provide a further understanding of this specification and form part of this specification, illustrate exemplary embodiments and are used to explain this specification, but do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the architecture of a two-stream neural network model applied to radiotherapy in this specification; Figure 2 This is a schematic diagram of the operation flow of a dual-flow neural network model applied to radiotherapy in this specification; Figure 3 This is a flowchart illustrating a dose reconstruction method for radiotherapy provided in this specification. Figure 4 The embodiment provided in this specification corresponds to Figure 3 A schematic diagram of the structure of an electronic device. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this specification will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments in this specification without creative effort are within the scope of protection of this application.
[0020] In the description of this invention, it should be noted that the term "or" is generally used to include the meaning of "and / or" unless otherwise expressly stated in the content.
[0021] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. Furthermore, in the description of this application, the terms "first," "second," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.
[0022] Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0023] The technical solutions provided in the various embodiments of this specification are described in detail below with reference to the accompanying drawings.
[0024] Figure 1 This is a schematic diagram of the architecture of a two-stream neural network model applied to radiotherapy, as described in this specification. Figure 1 As shown, the dual-flow neural network model applied to radiotherapy includes a two-dimensional processing branch, a three-dimensional processing branch, and a feature fusion and dose reconstruction branch.
[0025] Preferably, the two-dimensional processing branch includes a two-dimensional encoder.
[0026] Preferably, the two-dimensional encoder is used to process the input two-dimensional projected image to determine the two-dimensional beam feature map.
[0027] Preferably, the two-dimensional beam feature map is obtained from Monte Carlo simulation.
[0028] Preferably, the two-dimensional beam feature map is acquired via an X-ray imaging plate during radiotherapy for the patient.
[0029] Preferably, the dual-stream neural network model applied to radiotherapy includes a preprocessing module.
[0030] Preferably, the preprocessing module is used to perform filtering and / or image / dose calibration processing on the input two-dimensional projection image.
[0031] Preferably, the three-dimensional processing branch includes a three-dimensional encoder.
[0032] Preferably, the three-dimensional encoder is used to process the input three-dimensional CT image to determine the three-dimensional anatomical feature map.
[0033] Preferably, the dose reconstruction branch includes a geometric location embedding module, a manifold mapping dimensionality enhancement module, and a cross-modal cross-attention fusion module.
[0034] Figure 2 This is a schematic diagram illustrating the operation flow of a dual-flow neural network model applied to radiotherapy, as described in this specification. Figure 2 As shown.
[0035] Preferably, the geometric position embedding module is used to determine the coordinate index mapping from the three-dimensional voxel space to the two-dimensional feature plane based on the field geometry parameters.
[0036] Preferably, the geometric position embedding module is used to construct a differentiable mesh based on the field geometry parameters and establish a coordinate index mapping from three-dimensional voxel space to two-dimensional feature plane.
[0037] By employing the above method, differentiable dimensionality upgrades of 2D beam features to 3D space are achieved, supporting end-to-end gradient backpropagation and enabling the network to learn the optimal feature upscaling representation. Furthermore, ray geometric constraint information is preserved, with the feature mapping position of each 3D voxel strictly corresponding to the ray trajectory, ensuring the spatial correspondence between the beam and voxels. This allows the subsequent cross-attention fusion module to precisely modulate the beam features in space. Compared to traditional direct 3D convolution operations, this reduces computational complexity and improves efficiency.
[0038] Preferably, the field geometry parameters, two-dimensional projection image, and three-dimensional projection image are all data input into the model.
[0039] Preferably, the manifold mapping dimension-upgrading module is used to sample and differ the two-dimensional beam feature map along the ray direction under the constraint of the coordinate index mapping to generate a three-dimensional beam feature body.
[0040] Preferably, the manifold mapping dimensionality upscaling module is used to sample and interpolate the two-dimensional beam feature map along the ray direction based on the coordinate index mapping generated by the geometric position embedding module, and implicitly map the two-dimensional beam feature map to the three-dimensional feature space along the ray trajectory to generate an initial "three-dimensional beam feature volume". This process supports end-to-end backpropagation of gradients, allowing the network to adaptively learn the optimal feature dimensionality upscaling representation.
[0041] The above method achieves precise dimensionality upscaling of two-dimensional beam features to three-dimensional space. Under coordinate index mapping constraints, the two-dimensional beam features are sampled and interpolated along the ray direction to generate a three-dimensional beam feature volume. This ensures that the features of each three-dimensional voxel correctly correspond to the two-dimensional information on its ray path. Furthermore, it guarantees beam set constraints, providing accurate spatial positioning for subsequent cross-modal feature fusion. This improves the accuracy and efficiency of three-dimensional dose reconstruction.
[0042] Preferably, the cross-modal cross-attention fusion module is used to perform weighted fusion of the beam features based on the three-dimensional anatomical feature map and the three-dimensional beam feature volume through a voxel-level cross-modal attention mechanism to generate a fused feature tensor.
[0043] Preferably, the cross-modal cross-attention fusion module is used to output the corresponding three-dimensional dose distribution matrix based on the fused feature tensor through a three-dimensional dose regression network.
[0044] Preferably, the cross-modal attention fusion module is used to automatically calculate the attention weight map based on the voxel-level cross-modal attention mechanism, using "three-dimensional anatomical feature map" as the query and "three-dimensional beam feature volume" as the key and value, based on the density features of anatomical structures (such as bones or lungs), and to perform spatial modulation and weighted fusion of the beam features to generate a fusion feature tensor containing anatomical context information.
[0045] By employing the above method, adaptive fusion of beam characteristics and anatomical structures can be achieved, thereby enhancing the spatial accuracy of dose reconstruction.
[0046] Based on the above, the dual-stream neural network model includes a two-dimensional processing branch, a three-dimensional processing branch, and a feature fusion and dose reconstruction branch. The two-dimensional processing branch includes a two-dimensional encoder, which processes the input two-dimensional projection image to determine a two-dimensional beam feature map. The three-dimensional processing branch includes a three-dimensional encoder, which processes the input three-dimensional CT image to determine a three-dimensional anatomical feature map. The dose reconstruction branch includes a geometric location embedding module, a manifold mapping dimensionality enhancement module, and a cross-modal cross-attention fusion module. The geometric location embedding module is used to determine the coordinate index mapping from the three-dimensional voxel space to the two-dimensional feature plane based on the field geometry parameters. The manifold mapping dimensionality enhancement module is used to sample and differ the two-dimensional beam feature map along the ray direction under the geometric location embedding constraint to generate a three-dimensional beam feature volume. The cross-modal cross-attention fusion module is used to perform weighted fusion of the beam features based on the three-dimensional anatomical feature map and the three-dimensional beam feature volume through a voxel-level cross-modal attention mechanism to generate a fused feature tensor. Based on the fused feature tensor, the corresponding three-dimensional dose distribution matrix is output through a three-dimensional dose regression network.
[0047] As can be seen, the model has low computational complexity, can determine the three-dimensional dose distribution matrix in real time based on real-time acquired data, and has a fast reconstruction speed, which can meet the needs of real-time accurate verification.
[0048] Preferably, the two-dimensional encoder is used to extract a two-dimensional beam feature map containing beam intensity information and the shape of the multi-leaf collimator from the two-dimensional projected image layer by layer.
[0049] Preferably, the dual-stream neural network model is configured with a hybrid loss function, which incorporates two influencing factors: voxel-level error and dose distribution structure error.
[0050] The hybrid loss function can be determined by the following formula: Where R is the loss value, , Here are the parameters: A is the voxel-level error, and B is the dose distribution structure error.
[0051] Preferred, , The value of is related to gradient descent.
[0052] Preferably, during the training of the model, a high-precision three-dimensional dose distribution in the patient body calculated by Monte Carlo (MC) simulation can be used as the gold standard (Ground Truth).
[0053] It should be noted that all actions involving the acquisition of signals, information, or data in this application are carried out in compliance with the relevant data protection laws and policies of the country where the application is located, and with the authorization granted by the owner of the relevant device.
[0054] The above are one or more embodiments of the dual-flow neural network model for radiotherapy provided in this specification. Based on the same idea, this specification also provides a corresponding dose reconstruction method for radiotherapy.
[0055] Figure 3 This specification provides a flowchart illustrating a dose reconstruction method applied in radiotherapy, as shown below. Figure 3 As shown, the method specifically includes the following steps.
[0056] S900: Acquires 3D CT images and captures 2D projection images of patients during treatment in real time.
[0057] Preferably, this method can be executed by an electronic device such as a server or computer capable of running the dual-stream neural network model for radiotherapy provided in one or more of the foregoing embodiments; this specification does not limit this. In this specification, the execution of the method by an electronic device is used as an example for illustration.
[0058] Preferably, the electronic device stores and is capable of running the dual-stream neural network model for radiotherapy provided in one or more of the foregoing embodiments.
[0059] As mentioned in the background section, real-time dose monitoring is necessary during radiotherapy to ensure the treatment process aligns with the treatment plan. By constructing this dual-flow neural network model for radiotherapy, the three-dimensional dose distribution matrix can be determined in real-time during treatment.
[0060] S902: Input the three-dimensional CT image and the two-dimensional projection image into the dual-stream neural network model to determine the three-dimensional dose distribution matrix output by the dual-stream neural network model.
[0061] S904: The three-dimensional dose distribution matrix is registered and compared with the planned dose in the treatment plan, and the comparison result is determined in real time.
[0062] More preferably, in step S904, the electronic device can determine the real-time three-dimensional throughput and the real-time dose-volume histogram of organs at risk based on the three-dimensional dose distribution matrix. It can also determine the planned three-dimensional throughput and the planned dose-volume histogram of organs at risk based on the planned dose. Then, the electronic device can compare the real-time three-dimensional throughput with the planned three-dimensional throughput, and compare the real-time dose-volume histogram of organs at risk with the planned dose-volume histogram of organs at risk, determining the comparison result in real time.
[0063] Further preferably, when the electronic device determines that a treatment accident has occurred when the real-time 3D pass rate is less than a preset threshold or a preset percentage compared to the planned 3D pass rate, it issues an alarm to remind the doctor to suspend treatment or adjust the subsequent treatment plan. The preset percentage can be 80%, 85%, 90%, 95%, etc. The first planned percentage is based on the planned 3D pass rate.
[0064] More preferably, the electronic device determines a treatment accident when it determines that the dose-volume histogram of any one or more organs at risk exceeds the dose-volume histogram of the organs at risk in the plan, and issues an alarm to remind the doctor to suspend treatment or adjust the subsequent treatment plan.
[0065] The above are one or more embodiments of the dose reconstruction method for radiotherapy provided in this specification. Based on the same idea, this specification also provides a corresponding dose reconstruction device for radiotherapy, which stores the two-stream neural network model for radiotherapy provided in the above one or more embodiments.
[0066] The dose reconstruction device used in radiotherapy includes: The acquisition module is used to acquire three-dimensional CT images and to acquire two-dimensional projection images of the patient in real time during the treatment process; The generation module is used to input the three-dimensional CT image and the two-dimensional projection image into the dual-stream neural network model to determine the three-dimensional dose distribution matrix output by the dual-stream neural network model; The comparison module is used to register and compare the three-dimensional dose distribution matrix with the planned dose in the treatment plan, and determine the comparison result in real time.
[0067] This specification also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 4 A dose reconstruction method is provided for use in radiotherapy.
[0068] This specification also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 4 A dose reconstruction method is provided for use in radiotherapy.
[0069] This specification also provides a computer program product in which instructions, when executed by the processor of an electronic device, cause the electronic device to perform the above-described functions. Figure 3 A dose reconstruction method is provided for use in radiotherapy.
[0070] This instruction manual also provides Figure 4 The diagram shows a schematic structural representation of the electronic device. Figure 4 At the hardware level, the electronic device includes a processor, internal bus, network interface, memory, and non-volatile memory, and may also include other hardware required for the business operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then runs it to achieve the above-mentioned functions. Figure 3 The described dose reconstruction method is applied to radiotherapy. Of course, besides software implementation, this specification does not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution entity of the following processing flow is not limited to individual logic units, but can also be hardware or logic devices.
[0071] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must also be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also understand that by simply performing some logic programming on the method flow using one of these hardware description languages and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.
[0072] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.
[0073] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.
[0074] For ease of description, the above devices are described in terms of function, divided into various units. Of course, in implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware components.
[0075] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0076] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0077] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0078] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0079] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0080] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0081] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0082] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0083] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this specification may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0084] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.
[0085] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.
[0086] The above description is merely an embodiment of this specification and is not intended to limit this specification. Various modifications and variations can be made to this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this specification should be included within the scope of the claims of this application.
Claims
1. A two-stream neural network model for radiotherapy, characterized in that, This includes two-dimensional processing branches, three-dimensional processing branches, and feature fusion and dose reconstruction branches; The two-dimensional processing branch includes a two-dimensional encoder, which is used to process the input two-dimensional projection image to determine the two-dimensional beam feature map. The three-dimensional processing branch includes a three-dimensional encoder, which is used to process the input three-dimensional CT images to determine three-dimensional anatomical feature maps. The dose reconstruction branch includes a geometric location embedding module, a manifold mapping dimensionality enhancement module, and a cross-modal cross-attention fusion module. The geometric location embedding module is used to determine the coordinate index mapping from the three-dimensional voxel space to the two-dimensional feature plane based on the field geometry parameters. The manifold mapping dimensionality enhancement module is used to sample and differ the two-dimensional beam feature map along the ray direction under the constraint of the coordinate index mapping to generate a three-dimensional beam feature volume. The cross-modal cross-attention fusion module is used to perform weighted fusion of the beam features based on the three-dimensional anatomical feature map and the three-dimensional beam feature volume through a voxel-level cross-modal attention mechanism to generate a fused feature tensor. Based on the fused feature tensor, a three-dimensional dose distribution matrix is output through a three-dimensional dose regression network.
2. The dual-stream neural network model for radiotherapy according to claim 1, characterized in that, The dual-stream neural network model applied to radiotherapy includes a preprocessing module; The preprocessing module is used to perform filtering and / or image / dose calibration processing on the input two-dimensional projection image.
3. The dual-stream neural network model for radiotherapy according to claim 1, characterized in that, The two-dimensional encoder is used to extract two-dimensional beam feature maps containing beam intensity information and multi-leaf collimator shape from the two-dimensional projection image layer by layer.
4. The dual-stream neural network model for radiotherapy according to any one of claims 1-3, characterized in that, The dual-stream neural network model is configured with a hybrid loss function, which includes two influencing factors: voxel-level error and dose distribution structure error.
5. A dose reconstruction method applied to radiotherapy, characterized in that, The method, utilizing the dual-flow neural network model for radiotherapy proposed in any one of claims 1-4, comprises: Acquire 3D CT images and capture 2D projection images of the patient in real time during the treatment process; The three-dimensional CT image and the two-dimensional projection image are input into the two-stream neural network model to determine the three-dimensional dose distribution matrix output by the two-stream neural network model; The three-dimensional dose distribution matrix is registered and compared with the planned dose in the treatment plan, and the comparison result is determined in real time.
6. The dose reconstruction method for radiotherapy according to claim 5, characterized in that, The step of registering and comparing the three-dimensional dose distribution matrix with the planned dose in the treatment plan, and determining the comparison result in real time, includes: Based on the three-dimensional dose distribution matrix, the real-time three-dimensional throughput and the real-time dose-volume histogram of organs at risk are determined. Based on the planned dose, the planned three-dimensional pass rate and the dose-volume histogram of the planned organs at risk are determined; The real-time 3D pass rate is compared with the planned 3D pass rate, and the real-time dose-volume histogram of organs at risk is compared with the planned dose-volume histogram of organs at risk, and the comparison results are determined in real time.
7. The method according to claim 6, characterized in that, The method further includes: Based on the comparison results, a treatment accident was determined; Issue an alert to remind doctors to pause treatment or adjust subsequent treatment plans.
8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method described in any one of claims 5 to 7.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method described in any one of claims 5 to 7.
10. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device causes the electronic device to perform the method as described in any one of claims 5 to 7.