Particle radiotherapy plan making method, device, medium, and particle radiotherapy system

Abstract

The application provides a particle radiotherapy plan making method, comprising: automatically delineating a critical organ and a target area through a plan CT image to create a treatment plan; generating a pseudo-CT image based on an ultrasound image, and automatically delineating a bladder and the target area on the pseudo-CT image; performing morphing registration on the pseudo-CT image and the plan CT image, and mapping the delineation result on the pseudo-CT image to a coordinate system of the plan CT image; performing first determination on target area morphing based on the result of the morphing registration and a preset target area morphing threshold; and implementing the treatment plan or adjusting the treatment plan according to the result of the first determination. The application realizes dynamic monitoring of the target area by generating a pseudo-CT in real time through ultrasound and performing morphing registration; determines the treatment path in combination with the preset morphing threshold, forms a "monitoring-evaluation-decision" closed loop, solves the problem that a traditional single plan CT cannot adapt to changes in the bladder capacity, reduces target area missed irradiation and invalid plan adjustment caused by organ displacement, and improves radiotherapy accuracy and curative effect.

Description

Technical Field

[0001] This application relates to the field of particle radiotherapy technology, and in particular to a method, equipment, medium and system for particle radiotherapy planning. Background Technology

[0002] In radiotherapy to the abdominopelvic region, precise target localization and identification of organs at risk play a crucial role in treatment efficacy. However, dynamic changes in bladder fullness become a key obstacle affecting the accuracy of treatment planning.

[0003] Traditional radiotherapy planning relies primarily on a single planning CT image. During image acquisition, the bladder's fullness is recorded at a fixed time. However, during actual radiotherapy, bladder capacity is easily affected by factors such as the patient's physiological state and drinking habits. Once the bladder fullness deviates from the state at the time of the planning CT acquisition, the location and morphology of the target area and organs at risk will shift significantly. This makes it impossible to accurately optimize the treatment plan according to the actual situation, greatly affecting the accuracy and efficacy of radiotherapy. Summary of the Invention

[0004] This application aims to solve at least one technical problem existing in the prior art mentioned above, and proposes a method, equipment, medium and particle radiotherapy system for creating a particle radiotherapy plan, which aims to improve the accuracy and efficacy of radiotherapy.

[0005] In a first aspect, embodiments of this application provide a method for creating a particle radiotherapy plan, including:

[0006] Automatic delineation of organs at risk and target areas using planned CT images, and creation of treatment plans;

[0007] Pseudo-CT images are generated based on ultrasound images, and the bladder and target area are automatically delineated on the pseudo-CT images.

[0008] The pseudo-CT image and the planned CT image are deformably registered, and the delineation results on the pseudo-CT image are mapped to the coordinate system of the planned CT image.

[0009] Based on the deformation registration results and the preset target area deformation threshold, the target area deformation is first determined;

[0010] The treatment plan will be implemented or adjusted based on the results of the first assessment.

[0011] Furthermore, the step of implementing or adjusting the treatment plan based on the result of the first determination includes:

[0012] If the target area deformation does not exceed the target area deformation threshold, then the treatment plan is implemented;

[0013] If the target area deformation exceeds the target area deformation threshold, a second determination of bladder deformation is made based on a preset bladder deformation threshold.

[0014] The treatment plan or bladder capacity may be re-established based on the results of the second assessment.

[0015] Furthermore, the step of recreating the treatment plan or adjusting bladder capacity based on the result of the second determination includes:

[0016] If the bladder deformation does not exceed the bladder deformation threshold, a new planning CT image is acquired, and a treatment plan is created based on the new planning CT image;

[0017] If the bladder deformation exceeds the bladder deformation threshold, then the following steps are performed:

[0018] Adjust bladder capacity to baseline;

[0019] Obtain the global deformation variables generated when the bladder capacity is restored to the baseline state;

[0020] The target area deformation is corrected based on the global deformation;

[0021] The corrected target area deformation was reassessed, and a treatment plan was determined based on the reassessment results.

[0022] Furthermore, the reassessment of the corrected target area deformation and the determination of a treatment plan based on the reassessment results include:

[0023] If the corrected target area deformation exceeds the target area deformation threshold, a new planning CT image is acquired, and a treatment plan is created based on the new planning CT image.

[0024] If the corrected target area deformation does not exceed the target area deformation threshold, then perform the following steps:

[0025] Adaptive planning optimization is performed based on the corrected target area deformation to obtain the optimized dose distribution;

[0026] If the dose distribution meets clinical requirements, the treatment plan is implemented; otherwise, new planning CT images are acquired, and a treatment plan is created based on the new planning CT images.

[0027] Furthermore, the step of generating a pseudo-CT image based on the ultrasound image and automatically delineating the bladder and target area on the pseudo-CT image includes:

[0028] Ultrasound images are converted into pseudo-CT images using a pre-trained neural network; the input layer of the neural network fuses ultrasound B-mode texture features with an elastic imaging stiffness distribution map; the generator of the neural network includes an anatomical attention module to preserve the features of the bladder wall and target area edges.

[0029] The patient-specific segmentation model is used to automatically delineate the contours of the bladder and target area on pseudo-CT images, and the segmentation results are mapped to the ultrasound coordinate system.

[0030] Furthermore, the step of performing deformation registration between the pseudo-CT image and the planned CT image, and mapping the delineation results on the pseudo-CT image to the coordinate system of the planned CT image, includes:

[0031] Construct a 3D spatial transformation network to fuse pseudo-CT images and planned CT images;

[0032] Based on the aforementioned three-dimensional spatial transformation network, multi-scale deformation registration is used to obtain the first deformation variable;

[0033] Based on the difference in bladder deformation between pseudo-CT images and planned CT images, organ displacement vectors are obtained;

[0034] The organ displacement vector is superimposed on the first deformation to obtain the final deformation.

[0035] The delineation results on the pseudo-CT image are mapped to the coordinate system of the planned CT image based on the final deformation.

[0036] Furthermore, the multi-scale deformation registration includes:

[0037] In the coarse registration stage, the pseudo-CT image and the planned CT image are affine registered to correct large displacements;

[0038] In the fine registration stage, the pseudo-CT image and the planned CT image are registered using image pyramids at different scales with multi-resolution deformation.

[0039] Secondly, embodiments of this application provide an electronic device, including: one or more processors;

[0040] A memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors are able to perform the steps in any of the preceding manufacturing methods.

[0041] Thirdly, embodiments of this application provide a computer-readable medium storing a computer program, which, when executed by a processor, can implement the steps in any of the aforementioned manufacturing methods.

[0042] Fourthly, embodiments of this application provide a particle radiotherapy system, comprising:

[0043] The aforementioned electronic equipment is used to create particle radiotherapy plans;

[0044] A particle accelerator for generating a particle beam based on the particle radiotherapy plan.

[0045] This application provides a method for creating particle radiotherapy plans, which generates pseudo-CT scans in real time using ultrasound and performs deformation registration to achieve dynamic monitoring of the target area. Combined with a preset deformation threshold to determine the treatment path, a closed loop of "monitoring-evaluation-decision" is formed, which solves the problem that traditional single-plan CT cannot adapt to changes in bladder capacity, reduces target area omissions and ineffective plan adjustments caused by organ displacement, and improves the accuracy and efficacy of radiotherapy. Attached Figure Description

[0046] Figure 1 A core flowchart of a particle radiotherapy planning method provided in this application embodiment;

[0047] Figure 2 A detailed flowchart illustrating a method for creating a particle radiotherapy plan, provided for an embodiment of this application;

[0048] Figure 3 This is a structural block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0049] To enable those skilled in the art to better understand the technical solutions of this application, exemplary embodiments of this application are described below with reference to the accompanying drawings, including various details of the embodiments of this application to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description. Unless otherwise specified, the various embodiments of this application and the features within those embodiments can be combined with each other.

[0050] As used herein, the term “and / or” includes any and all combinations of one or more of the associated enumerated entries. The terminology used herein is for describing particular embodiments only and is not intended to limit the application. As used herein, the singular forms “a” and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated features, integrals, steps, operations, elements, and / or components is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Terms such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0051] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It should also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art and this application, and will not be interpreted as having an idealized or overly formal meaning, unless expressly so defined herein.

[0052] As described in the background section of this specification, traditional radiotherapy planning relies primarily on a single planning CT image. During image acquisition, the bladder's fullness is recorded at a fixed time. However, during actual radiotherapy, bladder capacity is highly susceptible to changes due to factors such as the patient's physiological state and drinking habits. If the bladder fullness deviates from its state at the time of the planning CT acquisition, the location and morphology of the target area and organs at risk will significantly shift. This prevents the treatment plan from being precisely optimized based on the actual situation, greatly affecting the accuracy and efficacy of radiotherapy. Therefore, this application proposes a method, equipment, medium, and system for creating particle radiotherapy plans, aiming to improve the accuracy and efficacy of radiotherapy.

[0053] refer to Figure 1 and Figure 2 One embodiment of this application proposes an adaptive particle radiotherapy planning method based on bladder capacity control, which may specifically include the following steps.

[0054] Step 1: Automatically delineate organs at risk and target areas using the patient's planned CT images, and create a treatment plan.

[0055] Planning CT images are cross-sectional images of a patient's body at a specific moment, clearly showing the anatomical structures within the abdominal and pelvic cavities. Using image processing and machine learning algorithms, the boundaries of organs at risk (such as the bladder and kidneys) and the target area (tumor site) can be automatically identified, thus completing the delineation. Based on this delineation information, combined with the dosimetry principles of radiotherapy and clinical needs, a preliminary radiotherapy plan is developed, determining parameters such as the radiotherapy dose and irradiation angle.

[0056] Step 2: Generate a pseudo-CT image from the patient's ultrasound image, and automatically delineate the bladder and target area on the pseudo-CT image.

[0057] Ultrasound images offer advantages such as real-time processing, convenience, and radiation-free operation, but their soft tissue contrast and anatomical structure information are inferior to CT images. The solution involves converting ultrasound images into pseudo-CT images with features similar to CT images. By learning the correspondence between a large number of ultrasound and CT images, pseudo-CT images are generated that are as close as possible to real CT images. On the generated pseudo-CT images, an automatic delineation algorithm is applied again to identify the bladder and target area, as the bladder's current state may differ from that at the time of the planned CT image acquisition, requiring accurate information on the current location and morphology of the bladder and target area.

[0058] In one possible implementation, generating pseudo-CT images from a patient's ultrasound images includes:

[0059] Ultrasound images are converted into pseudo-CT images using a pre-trained CycleGAN network (Cyclic Consistent Generative Adversarial Network). The input layer of the CycleGAN network fuses ultrasound B-mode texture features with elasticity distribution maps. The generator of the CycleGAN network includes an anatomical attention module to preserve bladder wall and target area edge features.

[0060] In one possible implementation, the ultrasound B-mode texture features and the elastic imaging stiffness distribution map of the CycleGAN network input layer are fused using a dynamic weighted fusion module. The dynamic weighted fusion module calculates the feature importance score of different regions through an attention mechanism and dynamically adjusts the fusion weight of the two features in different regions based on the score.

[0061] In one possible implementation, the ultrasonic B-mode texture features and the elastic imaging hardness distribution map are fused at different levels of the generator of the CycleGAN network. The lower levels perform preliminary fusion of global features, while the higher levels perform fine fusion of local features.

[0062] CycleGAN is an unsupervised image-to-image translation model consisting of a generator and a discriminator. It learns the mapping relationship from the source domain (ultrasound image) to the target domain (CT image) through adversarial training. In this embodiment, the pre-trained CycleGAN network has been trained on a large dataset of ultrasound images and corresponding CT images, enabling the generator to learn the feature transformation rules between ultrasound images and CT images, thereby converting the input ultrasound image into a pseudo-CT image with CT image features.

[0063] Ultrasound B-mode texture features can reflect the internal structure and texture information of tissues, while elastography stiffness distribution maps can show the elastic properties of tissues. The two contain different but complementary information; the introduction of the dynamic weighted fusion module aims to combine these two features more effectively.

[0064] The attention mechanism can automatically learn the importance of different regions in an image. In this module, the attention mechanism analyzes ultrasound B-mode texture features and elastography stiffness maps to calculate the feature importance score for each region. For example, in some regions, ultrasound B-mode texture features may more accurately reflect the anatomical structure of the tissue, so the importance score of the ultrasound B-mode texture features in that region will be higher; while in other regions, elastography stiffness maps may be more helpful in distinguishing different tissue types, so the importance score of the elastography stiffness maps in that region will be higher.

[0065] Based on the calculated feature importance scores, the dynamic weighted fusion module assigns different fusion weights to ultrasound B-mode texture features and elastography stiffness distribution maps in different regions. In regions with high feature importance scores, the corresponding features will occupy a larger proportion in the fusion process, so that the fused features can more accurately reflect the tissue characteristics of the region. This dynamic adjustment method can avoid using fixed weights for fusion in all regions, thereby improving the flexibility and accuracy of feature fusion.

[0066] In the lower layers of the CycleGAN network's generator, image features possess high spatial resolution and a wider receptive field, enabling the capture of global image information. At this level, the fusion of ultrasound B-mode texture features and elastography stiffness distribution maps primarily involves preliminary integration of the global information from these two types of features. For example, by fusing the low-frequency components of the two features, a global feature representation encompassing the overall tissue structure and approximate elasticity distribution can be obtained. This preliminary fusion provides a more comprehensive foundation for subsequent feature processing, helping the network better understand the overall features of the image.

[0067] As the network depth increases, the spatial resolution of features gradually decreases, but the level of abstraction and semantic information of the features gradually increases, enabling the capture of fine local information in the image. At higher levels of fusion, the focus is primarily on local details and features of specific regions in ultrasound B-mode texture features and elastography stiffness distribution maps. For example, in key areas such as the bladder wall and target area edges, fusing high-level feature representations of these two types of features can more accurately preserve the edge features and detailed information of these areas. The resulting pseudo-CT images can more clearly display the boundaries of the bladder wall and target area, providing more accurate anatomical information for subsequent radiotherapy planning.

[0068] The anatomical attention module plays a crucial role in the generator network (the generator of the CycleGAN network) by focusing on key anatomical structures. During the conversion of ultrasound images into pseudo-CT images, this module automatically identifies key anatomical structures such as the bladder wall and target area, enhancing and preserving the features of these regions. Through the attention mechanism, the module learns the characteristic patterns of these key regions, giving them more attention and weight during feature processing and image generation. This ensures that the generated pseudo-CT images accurately preserve the edge features of the bladder wall and target area, improving image quality and diagnostic value.

[0069] In one possible implementation, the bladder and target area are automatically delineated on a pseudo-CT image; including:

[0070] The patient-specific segmentation model was used to delineate the contours of the bladder and target area on pseudo-CT images, and the segmentation results were mapped to the ultrasound coordinate system.

[0071] Patient-specific segmentation models are built using deep learning technology, typically based on convolutional neural networks (CNNs), such as U-Net and Mask R-CNN. When building the model, a large amount of medical image data containing the bladder and target area is used for training. The data comes from CT images, ultrasound images, etc. of multiple patients, and the bladder and target area in the images have been manually labeled by professional doctors.

[0072] To improve the accuracy and specificity of segmentation, the model learns the characteristics of a specific patient by collecting the patient's multimodal image data (such as CT images and ultrasound images from different periods) and combining the patient's clinical information (such as age, gender, and medical history). The model analyzes and processes this data to extract unique feature patterns related to the patient's bladder and target area, thereby enabling more accurate identification and segmentation of the bladder and target area in the patient's pseudo-CT images.

[0073] When a pseudo-CT image is input into a patient-specific segmentation model, the model extracts features from the image through operations such as convolutional layers and pooling layers. Convolutional layers extract local features of the image, such as edges and textures; pooling layers reduce the dimensionality of the feature map, thus reducing computational cost. After multiple layers of feature extraction, the model learns the feature representations of the bladder and target area in the pseudo-CT image. Finally, the model restores the feature map to the same size as the input image through deconvolutional layers or upsampling operations, and outputs the segmentation results, namely the contours of the bladder and target area.

[0074] The pseudo-CT image and the ultrasound image have a one-to-one correspondence. Therefore, their coordinate system and geometric information are exactly the same. The bladder and target area contour information obtained from the pseudo-CT image can be directly correlated with the ultrasound image for more intuitive observation and analysis on the ultrasound image. Combined with the real-time dynamic information of the ultrasound image, the radiotherapy plan can be formulated and adjusted more accurately.

[0075] One possible implementation employs a multi-scale, multi-resolution mapping strategy. First, a coarse mapping of the segmentation contours on the pseudo-CT image is performed at low resolution to quickly determine the approximate location. Then, the resolution is gradually increased, using the mapping results from the previous scale as initialization for finer mapping adjustments. At each scale, the mapping transformation parameters are optimized by combining local and global image features. For example, at high resolution, the focus is on the detailed features of the bladder wall and target area edges, making the mapping of the segmentation contours in the ultrasound coordinate system more closely match the actual anatomical structure and reducing mapping errors caused by resolution differences.

[0076] Step 3: Perform deformation registration between the pseudo-CT image and the planned CT image, and map the delineation results on the pseudo-CT image to the coordinate system of the planned CT image.

[0077] Because patients' body positions and bladder fullness vary at different times, the location and morphology of organs in pseudo-CT images and planned CT images will differ. The purpose of deformation registration is to find a transformation relationship that aligns the pseudo-CT image and the planned CT image as closely as possible in terms of anatomical structure. Commonly used methods include feature-based registration and grayscale-based registration. By calculating the deformation variable, the contours of the bladder and target area automatically drawn on the pseudo-CT image are transformed according to this deformation variable and mapped to the coordinate system of the planned CT image, thereby achieving information unification between different images.

[0078] Traditional radiotherapy planning is often based on a single-moment planning CT scan, failing to adequately consider the dynamic changes in bladder fullness during radiotherapy. This embodiment, however, generates pseudo-CT images from acquired ultrasound images, reflecting the bladder's current state in real time. By comparing and registering these pseudo-CT images with the planning CT images, it can accurately capture shifts in the target area and the position and shape of organs at risk caused by changes in bladder capacity. For example, in actual radiotherapy, bladder fullness may displace surrounding tissues, altering the target area's position. This method can precisely measure this change, ensuring accurate radiation dose coverage of the target area and avoiding residual tumor cells or unnecessary damage to surrounding healthy tissues due to positional deviations.

[0079] Automatically delineating the bladder and target area on pseudo-CT images and mapping the delineation results to the planning CT images using deformation registration effectively integrates the advantages of different image modalities. Ultrasound images provide real-time soft tissue information, helping to more clearly define the boundaries of the bladder and target area. The generated pseudo-CT images also possess anatomical structural features similar to CT images, making target area delineation more accurate, reducing radiotherapy deviations caused by delineation errors, and improving the spatial accuracy of radiotherapy. The automatic delineation of organs at risk and target areas, as well as the automatic generation of pseudo-CT images, greatly reduces the workload and time cost of manual operation, significantly improving the efficiency of radiotherapy planning.

[0080] In one possible implementation, deformation registration is performed between the pseudo-CT image and the planned CT image, and the delineation results on the pseudo-CT image are mapped to the coordinate system of the planned CT image, including:

[0081] A three-dimensional spatial transformation network is constructed, whose input layer fuses pseudo-CT images and planned CT images;

[0082] Based on the aforementioned three-dimensional spatial transformation network, multi-scale deformation registration is used to obtain the first deformation variable;

[0083] Based on the difference in bladder deformation between pseudo-CT images and planned CT images, organ displacement vectors are calculated using a hyperelastic constitutive model; the organ displacement vectors are then superimposed onto the first deformation variable to obtain the final deformation variable.

[0084] The delineation results are mapped to the coordinate system of the planned CT image based on the final deformation.

[0085] One possible implementation employs multi-scale deformation registration, including:

[0086] In the coarse registration stage, the pseudo-CT image and the planned CT image are affine registered to correct large displacements;

[0087] In the fine registration stage, the pyramid strategy is used to perform deformation registration of pseudo-CT images and planned CT images at different scales.

[0088] In one possible implementation, the loss function used during training of the 3D spatial transformation network includes: a distance loss term for anatomical landmarks, a deformation differential smoothing constraint term, and a regularization term for the bladder wall elastic modulus parameter.

[0089] A three-dimensional spatial transformation network was constructed to learn the spatial transformation relationship between pseudo-CT images and planned CT images. Through this network, a suitable transformation method can be found to make the pseudo-CT images as spatially aligned as possible with the planned CT images, laying the foundation for subsequent accurate deformation registration and result mapping.

[0090] The input layer of the 3D spatial transformation network fuses pseudo-CT images and planning CT images, aiming to allow the network to simultaneously acquire information features from both types of images. The pseudo-CT images contain relevant tissue information converted from ultrasound images, while the planning CT images are the original standard images used to formulate radiotherapy plans. After fusing the two, the network can analyze and compare the differences and similarities of corresponding anatomical structures in the two images during subsequent processing, and then learn how to transform the structures in the pseudo-CT images to a spatial location consistent with the planning CT images.

[0091] Coarse Registration Stage (Affine Registration Corrects Large Displacements): In the coarse registration stage, affine registration is performed on the pseudo-CT image and the planned CT image. Affine transformation is a linear transformation that can perform operations such as translation, rotation, scaling, and shearing on images. During radiotherapy, factors such as changes in patient position and overall organ displacement can cause significant positional deviations between the pseudo-CT image and the planned CT image. Affine registration can quickly correct these large displacements. By calculating the affine transformation matrix, the pseudo-CT image is initially adjusted to a state similar to the planned CT image in overall spatial position, providing more favorable initial conditions for subsequent fine registration, reducing the search space in the subsequent registration process, and improving registration efficiency.

[0092] In the fine registration stage, a pyramid strategy is used for deformation registration at different scales. First, an image pyramid is constructed by downsampling the original pseudo-CT image and the planned CT image to generate a series of image layers with different resolutions, forming a pyramid structure. Registration begins at the top of the pyramid (low-resolution image) because low-resolution images contain less detail but require less computation, allowing for a quick capture of the approximate correspondence between images and obtaining a preliminary deformation variable. Then, this preliminary deformation variable is used as the initial value and passed to the next higher-resolution image layer for registration. Based on the results of the previous layer, further optimization is performed, considering more image details, and gradually obtaining a more accurate deformation variable. This coarse-to-fine, multi-scale registration method fully utilizes the advantages of images with different resolutions, enabling both rapid alignment of the overall structure and precise adjustment of local details, thereby improving registration accuracy.

[0093] If the difference in bladder deformation exceeds a preset threshold during treatment, it indicates that the deformation of the target area may be caused by changes in bladder capacity. Therefore, bladder capacity control is necessary, which may include catheter drainage or injection of saline solution. To avoid unnecessary ultrasound acquisition, simulation can be performed by calculating the organ displacement vector after the bladder capacity returns to the baseline level.

[0094] Based on the differences in bladder deformation between pseudo-CT images and planned CT images, organ displacement vectors are calculated using a hyperelastic constitutive model. As a deformable organ, the deformation of the bladder in different images affects the position of surrounding organs. The hyperelastic constitutive model can describe the deformation behavior of bladder tissue under stress. By analyzing the differences in shape and size of the bladder between pseudo-CT and planned CT images, this model can be used to calculate the displacement vectors of surrounding organs (including the target area) caused by bladder deformation. This vector reflects the direction and extent of the influence of bladder deformation on the spatial position of surrounding organs.

[0095] The organ displacement vectors calculated based on bladder deformation are superimposed onto the first deformation variable obtained through a 3D spatial transformation network and multi-scale deformation registration. The first deformation variable primarily considers the overall spatial transformation of the image, but it does not fully capture the subtle effects of local bladder deformation on surrounding organs. By superimposing the organ displacement vectors, the first deformation variable can be supplemented and corrected, resulting in a more accurate and comprehensive final deformation variable. This final deformation variable comprehensively considers both the overall spatial transformation of the image and the impact of local bladder deformation on surrounding organs, thus more realistically reflecting the actual deformation from the pseudo-CT image to the planned CT image.

[0096] The final deformation variable maps the delineation results to the coordinate system of the planning CT image. This final deformation variable contains all the deformation information needed to accurately transform the pseudo-CT image to its spatial position in the planning CT image. The delineation result is initially obtained in the coordinate system of the pseudo-CT image; by applying this final deformation variable, the coordinates in the delineation result are transformed accordingly. For example, for each point on the contour of the bladder or target area delineated by ultrasound, the new position of that point in the coordinate system of the planning CT image is calculated based on the displacement vector and transformation relationship in the final deformation variable, thus accurately mapping the entire ultrasound delineation result to the coordinate system of the planning CT image. In this way, based on the planning CT image, the information provided by the ultrasound image can be comprehensively utilized to provide more accurate data support for the formulation and adjustment of radiotherapy plans.

[0097] Anatomical landmarks are points in medical images that have clear anatomical significance and are easily identifiable, such as joints of bones and specific boundary points of organs. During the training of a 3D spatial transformation network, the distance (e.g., Euclidean distance) between the predicted and actual locations of anatomical landmarks is calculated as a loss term. This loss term aims to ensure that the spatial transformation learned by the network accurately transforms anatomical landmarks in pseudo-CT images to locations close to their corresponding locations in the planned CT image. By continuously minimizing this distance loss, the network can optimize its transformation parameters, achieving greater consistency in the registered images at the anatomical landmark level, thereby improving the overall registration accuracy.

[0098] Deformation variables describe the displacement of each point in an image. To ensure the rationality and continuity of these deformation variables, a differential smoothing constraint term is introduced. In actual organ deformation processes, the deformation of organ tissues is continuous and smooth, without sudden jumps or discontinuous deformations. By constraining the first or second derivatives of the deformation variables (e.g., limiting the magnitude or range of the derivatives) and incorporating them into the loss function, the generated deformation variables are encouraged to meet smoothness requirements during network training. This avoids unreasonable local deformations, making the registration results more consistent with actual physiological conditions and improving the reliability of the registration results.

[0099] The elastic modulus of the bladder wall is a crucial parameter reflecting the mechanical properties of bladder tissue and is closely related to bladder deformation. When training the network, a regularization term for the bladder wall elastic modulus parameter is introduced to ensure that the bladder deformation learned by the network matches the actual elastic mechanical properties of the bladder. By constraining the difference between the predicted bladder wall elastic modulus parameter and a known or reasonably estimated elastic modulus parameter, and incorporating this constraint into the loss function, the network is guided to consider the elastic properties of the bladder wall when learning the deformation relationship between pseudo-CT images and planned CT images. This allows for a more accurate simulation of the bladder deformation process, improving the accuracy of calculations based on bladder deformation differences and further enhancing the precision of the entire registration process.

[0100] Step 4: Based on the deformation registration results and the target area deformation threshold, make a first determination on the target area deformation; implement the treatment plan or adjust the treatment plan directly based on the result of the first determination.

[0101] In one possible implementation, the treatment plan is directly implemented or adjusted based on the result of the first determination, including:

[0102] If the target area deformation does not exceed the target area deformation threshold, then the treatment plan is implemented directly.

[0103] If the target area deformation exceeds the target area deformation threshold, a second determination of bladder deformation is made based on the bladder deformation threshold, and the treatment plan is recreated or the bladder capacity is adjusted according to the result of the second determination.

[0104] After registration and result conversion, the original target area contour on the planned CT image and the target area contour mapped from the pseudo-CT image are compared to calculate the target area deformation (such as displacement and volume change). The deformation is compared with a preset target area deformation threshold. If the deformation does not exceed the threshold, it means that the current position and morphological changes of the target area are within an acceptable range, and the treatment plan can be implemented directly. If the deformation exceeds the threshold, it indicates that the target area has changed significantly, which may affect the accuracy of radiotherapy. Further adjustments to the treatment plan are needed to ensure that radiotherapy can accurately target the target area while reducing damage to organs at risk. Subsequent adjustments will be made based on bladder deformation, such as recreating the treatment plan or adjusting bladder capacity, to achieve adaptive optimization of the radiotherapy plan.

[0105] After completing the deformation registration of the pseudo-CT image and the planning CT image, and mapping the delineation results on the pseudo-CT image to the planning CT, the deformation of the target area can be obtained. This deformation includes the displacement of the target area, changes in shape, and changes in volume. The deformation of the target area is compared with a pre-set target area deformation threshold. This threshold is determined based on a large amount of clinical data and medical research, and represents the maximum allowable deformation range of the target area without affecting the radiotherapy effect.

[0106] When the target area deformation does not exceed the target area deformation threshold, it means that the current change in the target area is within an acceptable range, and the initial treatment plan can still accurately target the target area for radiotherapy, while the impact on organs at risk is also within a controllable range. Therefore, directly implementing the treatment plan can save medical resources and avoid the waste of time and manpower costs caused by unnecessary adjustments to the treatment plan.

[0107] If the target area deformation exceeds the target area deformation threshold, it indicates that the current treatment plan may not be effective and further investigation of the cause and measures are needed. Since changes in bladder fullness are an important factor leading to target area deformation, a second judgment on bladder deformation is made based on the bladder deformation threshold. Similarly, the bladder deformation threshold is also set based on clinical experience and research to measure whether bladder deformation has reached the level that requires intervention.

[0108] In one possible implementation, the treatment plan is recreated or the bladder capacity is adjusted based on the result of the second determination, including:

[0109] If the bladder deformation does not exceed the bladder deformation threshold, a new planned CT image is acquired, and a treatment plan is generated based on the new planned CT image; the pseudo CT image is deformed and registered with the new planned CT image, and the delineation result on the pseudo CT image is deformed onto the new planned CT image, and the aforementioned determination steps are repeated.

[0110] When bladder deformation exceeds the bladder deformation threshold, the following steps are performed:

[0111] Adjust bladder capacity to a baseline state; the baseline state is the bladder capacity state of the patient at the time of the planned CT scan.

[0112] Calculate the global deformation variables generated when the bladder capacity is restored to the baseline state, and correct the target area deformation variables based on the global deformation variables;

[0113] The corrected target area deformation was reassessed, and a treatment plan was determined based on the reassessment results.

[0114] If the bladder deformation does not exceed the bladder deformation threshold, it indicates that the bladder change is relatively small. However, if the target area deformation exceeds the allowable range, it may be due to target area displacement or deformation caused by other complex factors. In this case, acquiring new planning CT images can more comprehensively and accurately reflect the actual anatomical location and morphology of the target area and organs at risk within the patient's body. Based on the new planning CT images, an updated treatment plan can be generated, allowing for the re-planning of key parameters such as radiation dose distribution, irradiation angle, and timing to adapt to changes in the target area and ensure the precision of radiotherapy.

[0115] When bladder deformation exceeds the bladder deformation threshold, it indicates that changes in bladder capacity have a significant impact on target area deformation. Adjusting bladder capacity to a baseline state (usually the bladder state at the time of planned CT image acquisition) is crucial because the initial treatment plan developed under this state provides the best prediction of radiotherapy efficacy. By calculating the global deformation resulting from restoring bladder capacity to the baseline state, the influence of bladder state changes on surrounding tissues, including the target area, can be understood. This calculation avoids re-enhancing ultrasound imaging, directly calculating the actual changes in the target area on ultrasound images after bladder capacity control by superimposing the global bladder deformation with the target area deformation. Correcting the target area deformation based on the calculated global deformation allows for a more accurate prediction of the target area's position and morphology after the bladder returns to its baseline state. Re-evaluating the corrected target area deformation and determining the final treatment plan based on the evaluation results allows for the development of a radiotherapy scheme that better suits the patient's actual condition, taking bladder factors into account, thereby improving the success rate and treatment efficacy of radiotherapy.

[0116] By rigorously comparing target area deformation with target area deformation thresholds, treatment plans are implemented directly when target area changes are within acceptable ranges, and subsequent adjustments are made when changes exceed these ranges. This ensures that radiotherapy consistently targets the target area precisely, improving its effectiveness. Different strategies are employed for different target areas and bladder deformation conditions. For example, new CT images are acquired to generate updated treatment plans when bladder deformation does not exceed the threshold, or bladder capacity is adjusted and target area deformation is corrected when bladder deformation exceeds the threshold. This allows radiotherapy plans to better adapt to the dynamic changes in the patient's organs, improving the targeting and adaptability of the radiotherapy protocol, thereby enhancing treatment outcomes. Full consideration of target area and bladder deformation during the assessment process allows for the timely detection of changes in the position of organs at risk due to bladder distension. By adjusting the treatment plan, organs at risk that have changed position due to bladder displacement can be precisely avoided. Implementation of the treatment plan directly when target area deformation does not exceed the threshold avoids unnecessary examinations and treatment plan adjustments, improving the efficiency of medical resource utilization. When bladder deformation exceeds the threshold, the target area deformation is corrected by calculating the global deformation generated when the bladder capacity is restored to the baseline state, thus avoiding the need for repeated ultrasound examinations.

[0117] In one possible implementation, the modified target area deformation is reassessed, and a treatment plan is determined based on the reassessment results, including:

[0118] If the corrected target area deformation exceeds the target area deformation threshold, a new planned CT image is acquired, and a treatment plan is generated based on the new planned CT image; the pseudo CT image is deformed and registered with the new planned CT image, and the delineation result on the pseudo CT image is deformed onto the new planned CT image, and the aforementioned determination steps are repeated.

[0119] If the corrected target area deformation does not exceed the target area deformation threshold, then perform the following steps:

[0120] Adaptive planning optimization based on deformation variables is performed to calculate the optimized dose distribution.

[0121] The treatment plan is implemented when the dose distribution meets clinical requirements; otherwise, new planning CT images are acquired and a treatment plan is generated based on the new planning CT images.

[0122] The pseudo-CT image is deformably registered with the new planned CT image, and the delineation result on the pseudo-CT image is deformed onto the new planned CT image. The aforementioned determination steps are repeated.

[0123] After adjusting the bladder capacity, the global deformation is calculated and the target area deformation is corrected. At this point, the target area deformation needs to be calculated again. The target area deformation calculated here is similar to that before, covering multiple parameters such as displacement of the target area, changes in shape, and changes in volume. The corrected target area deformation is compared again with the pre-set target area deformation threshold. This threshold is still determined based on a large amount of clinical data and medical research, and represents the maximum allowable deformation range of the target area without affecting the radiotherapy effect.

[0124] By re-evaluating the corrected target area deformation and making decisions based on the results, we avoid blindly conducting new planned CT scans and updating treatment plans when the target area deformation has not exceeded the threshold, thus reducing the waste of medical resources.

[0125] When the corrected target volume deformation exceeds the target volume deformation threshold, it indicates that even after previous bladder capacity adjustments and target volume deformation corrections, the position and morphology of the target volume have still changed significantly. The current radiotherapy plan may not be able to precisely treat the target volume, necessitating the acquisition of new planning CT images. These new planning CT images reflect the patient's current anatomical structure, including the location and morphology of the target volume and organs at risk. An updated treatment plan is generated based on these new planning CT images, and key radiotherapy parameters are redesigned to adapt to changes in the target volume, ensuring the precision of radiotherapy.

[0126] If the corrected target area deformation does not exceed the target area deformation threshold, it indicates that the location and morphology of the target area are within an acceptable range. In this case, adaptive planning optimization is performed based on the corrected deformation, utilizing the information about changes in the target area and surrounding tissues contained in the deformation to further optimize the radiotherapy plan.

[0127] The core of adaptive treatment planning optimization is calculating the optimized dose distribution. Using the dose calculation algorithm within the radiotherapy planning system, and considering the shape and location of the target area, as well as the condition of surrounding organs at risk, and taking into account the impact of corrected deformation on the dose distribution, the optimal dose distribution for the target area under the current conditions is calculated. This process comprehensively considers various factors, such as the uniformity of dose distribution at different irradiation angles, the minimum and maximum doses received by the target area, and the dose limitations imposed on organs at risk.

[0128] The optimized dose distribution is then compared with clinical requirements. Clinical requirements are typically formulated based on factors such as tumor type, size, location, and the patient's overall condition. These requirements include the minimum prescribed dose to be achieved at the target volume, dose homogeneity indices, and the maximum tolerated dose for organs at risk. When the dose distribution meets clinical requirements, it indicates that the current radiotherapy plan can control damage to normal tissues within an acceptable range while ensuring therapeutic efficacy, and the treatment plan can be implemented. If the dose distribution does not meet clinical requirements, it means that the current radiotherapy plan may not be effective in treating the tumor or may cause excessive damage to normal tissues. Therefore, it is necessary to acquire new CT images of the plan and generate an updated treatment plan, and to re-optimize and evaluate the dose distribution until it meets clinical requirements.

[0129] In one possible implementation, adaptive planning optimization is performed based on deformation variables to calculate the optimized dose distribution; including:

[0130] Obtain the corrected target area deformation and deformation information of organs at risk;

[0131] Based on the deformation information, the radiation field parameters in the radiotherapy plan are adjusted, including the size, shape, and angle of the radiation field;

[0132] Based on the adjusted irradiation field parameters, the dose calculation algorithm in the radiotherapy planning system is used to calculate the optimized dose distribution, taking into account factors such as tissue density and radiation energy attenuation. The dose calculation algorithm adopts the Monte Carlo algorithm or the convolution superposition algorithm, and the dose grid is finely divided during the calculation process to improve the calculation accuracy.

[0133] During radiotherapy, a patient's physiological state may change, leading to alterations in the position and shape of the target area and organs at risk. Preliminary image registration and deformation correction procedures allow for the acquisition of corrected target area deformation and organ deformation information. This information accurately reflects the dynamic changes in the spatial location and morphology of the organs, forming the basis for subsequent adaptive treatment planning optimization.

[0134] Based on the obtained deformation information of the target area and organs at risk, the new position and shape of the organs are analyzed. If the target area has shifted or changed shape, the original irradiation field may not be able to completely cover the target area, or may cause unnecessary irradiation to surrounding organs at risk. Therefore, the irradiation field parameters need to be adjusted according to the specific deformation.

[0135] After adjusting the irradiation field parameters, the dose calculation algorithm in the radiotherapy planning system is used to calculate the optimized dose distribution. In actual radiotherapy, the propagation of rays in human tissue is affected by various physical factors such as tissue density and ray energy attenuation. Different tissues (such as bone, muscle, and fat) have different absorption and scattering characteristics for rays; the higher the tissue density, the faster the ray energy attenuates. Therefore, these factors must be fully considered in the dose calculation process to accurately simulate the propagation and energy deposition of rays within the human body. Commonly used dose calculation algorithms include the Monte Carlo algorithm and the convolutional superposition algorithm.

[0136] The embodiments of this application automatically delineate organs at risk and target areas using the patient's planned CT images, and utilize a pre-trained CycleGAN network to convert ultrasound images into pseudo-CT images, automatically delineating the bladder and target areas on the pseudo-CT images. This allows for more accurate determination of the location and morphology of the target areas and organs at risk. The input layer of the CycleGAN network integrates ultrasound B-mode texture features and elastography stiffness distribution maps, combined with an anatomical attention module, to preserve bladder wall and target area edge features, resulting in more accurate delineation and providing a precise basis for subsequent radiotherapy planning. Deformation registration is performed between the pseudo-CT images and planned CT images using a three-dimensional spatial transformation network and multi-scale deformation registration method. This considers the bladder deformation differences between the pseudo-CT images and planned CT images, and obtains the final deformation by superimposing organ displacement vectors. This allows for more accurate mapping of the ultrasound delineation results to the coordinate system of the planned CT image, ensuring that radiation accurately irradiates the target area during radiotherapy and reducing damage to surrounding normal tissues. Based on deformation registration results and target area deformation thresholds, target area deformation is assessed. Depending on the assessment results, different decisions are made: direct implementation of the treatment plan, adjustment of the treatment plan, re-creation of the treatment plan, or adjustment of bladder capacity. This hierarchical decision-making mechanism dynamically adjusts the radiotherapy plan according to the patient's actual condition, allowing the treatment plan to better adapt to the dynamic changes in the patient's organs and improving the effectiveness of radiotherapy. By calculating the global deformation resulting from restoring bladder capacity to its baseline state, the impact of bladder changes on surrounding tissues, including the target area, can be understood. This calculation avoids re-enhancing ultrasound imaging; by superimposing the global deformation of the bladder with the target area deformation, the true changes in the target area on ultrasound images after bladder capacity control are directly calculated. This allows for the development of radiotherapy plans that better suit the patient's actual condition, taking bladder factors into account, thereby improving the success rate and treatment efficacy of radiotherapy.

[0137] Based on the same inventive concept, embodiments of this application also provide an electronic device. Figure 3 This is a structural block diagram of an electronic device provided in an embodiment of this application. Figure 3As shown in the embodiments of this application, an electronic device includes: one or more processors 101, a memory 102, and one or more I / O interfaces 103. The memory 102 stores one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement any of the particle radiotherapy planning methods described in the above embodiments; the one or more I / O interfaces 103 are connected between the processor and the memory, configured to enable information interaction between the processor and the memory.

[0138] The processor 101 is a device with data processing capabilities, including but not limited to a central processing unit (CPU); the memory 102 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read / write interface) 103 is connected between the processor 101 and the memory 102, and can realize information interaction between the processor 101 and the memory 102, including but not limited to a data bus (Bus).

[0139] In some embodiments, the processor 101, memory 102, and I / O interface 103 are interconnected via bus 104, and thus connected to other components of the computing device.

[0140] In some embodiments, the one or more processors 101 include a field-programmable gate array.

[0141] This application also provides a particle radiotherapy system, including the aforementioned electronic device and particle accelerator.

[0142] The aforementioned electronic device is used to create particle radiotherapy plans.

[0143] A particle accelerator for generating high-energy particle beams based on the particle radiotherapy plan and for use in radiotherapy.

[0144] This application also provides a computer-readable medium. The computer-readable medium stores a computer program, which, when executed by a processor, implements the steps of any of the particle radiotherapy planning methods described in the above embodiments. The computer-readable storage medium may be volatile or non-volatile.

[0145] This application also provides a computer program product, including computer-readable code, or a non-volatile computer-readable storage medium carrying computer-readable code. When the computer-readable code is run in the processor of an electronic device, the processor in the electronic device executes the above-described particle radiotherapy planning method.

[0146] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer storage media (or non-transitory media) and communication media (or transient media).

[0147] As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable program instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technologies, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable program instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0148] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0149] The computer program instructions used to perform the operations of this application may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuits, such as programmable logic circuits, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), are personalized by utilizing the status information of the computer-readable program instructions. These electronic circuits can execute the computer-readable program instructions to implement various aspects of this application.

[0150] The computer program product described herein can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0151] Various aspects of this application are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should 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-readable program instructions.

[0152] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0153] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0154] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0155] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for general illustrative purposes only and should not be construed as limiting. In some embodiments, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in connection with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in connection with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of this application as set forth by the appended claims.

Claims

1. A method for creating a particle radiotherapy plan, characterized in that, include: Automatic delineation of organs at risk and target areas using planned CT images, and creation of initial treatment plans; Pseudo-CT images are generated based on ultrasound images, and the bladder and target area are automatically delineated on the pseudo-CT images. The pseudo-CT image and the planned CT image are deformably registered, and the delineation results on the pseudo-CT image are mapped to the coordinate system of the planned CT image. Based on the deformation registration results and the preset target area deformation threshold, the target area deformation is first determined; If the target area deformation does not exceed the target area deformation threshold, then the initial treatment plan is determined to be the execution plan for the current segment. If the target area deformation exceeds the target area deformation threshold, a second determination of bladder deformation is made based on a preset bladder deformation threshold. If the bladder deformation does not exceed the bladder deformation threshold, a new planning CT image is acquired, and a new treatment plan is created based on the new planning CT image. If the bladder deformation exceeds the bladder deformation threshold, then the following steps are performed: Adjust the bladder capacity to a baseline state, where the baseline state is the bladder capacity state at the time of taking the planned CT image; Obtain the global deformation resulting from restoring the bladder capacity to the baseline state; The target area deformation is modified based on the global deformation; The corrected target area deformation is re-evaluated, and the execution plan for the current segment is determined based on the re-evaluation results; The step of re-evaluating the corrected target area deformation and determining the execution plan for the current segment based on the re-evaluation results includes: If the corrected target area deformation exceeds the target area deformation threshold, a new planning CT image is acquired, and a new treatment plan is created based on the new planning CT image. If the corrected target area deformation does not exceed the target area deformation threshold, then perform the following steps: Adaptive planning optimization is performed based on the corrected target area deformation to obtain the optimized dose distribution; If the dose distribution meets clinical requirements, the optimized plan is determined as the execution plan for the current fraction; otherwise, a new plan CT image is acquired, and a new treatment plan is created based on the new plan CT image. The step of generating a pseudo-CT image based on an ultrasound image and automatically delineating the bladder and target area on the pseudo-CT image includes: Ultrasound images are converted into pseudo-CT images using a pre-trained neural network; the input layer of the neural network fuses ultrasound B-mode texture features with an elastic imaging stiffness distribution map; the generator of the neural network includes an anatomical attention module to preserve the features of the bladder wall and target area edges. The patient-specific segmentation model is used to automatically delineate the contours of the bladder and target area on pseudo-CT images, and the segmentation results are mapped to the ultrasound coordinate system.

2. The manufacturing method according to claim 1, characterized in that, The step of performing deformation registration between the pseudo-CT image and the planned CT image, and mapping the delineation results on the pseudo-CT image to the coordinate system of the planned CT image, includes: Construct a 3D spatial transformation network to fuse pseudo-CT images and planned CT images; Based on the aforementioned three-dimensional spatial transformation network, multi-scale deformation registration is used to obtain the first deformation variable; Based on the difference in bladder deformation between pseudo-CT images and planned CT images, organ displacement vectors are obtained; The organ displacement vector is superimposed on the first deformation to obtain the final deformation. The delineation results on the pseudo-CT image are mapped to the coordinate system of the planned CT image based on the final deformation.

3. The manufacturing method according to claim 2, characterized in that, The multi-scale deformation registration method includes: In the coarse registration stage, the pseudo-CT image and the planned CT image are affine registered to correct large displacements; In the fine registration stage, the pseudo-CT image and the planned CT image are registered using image pyramids at different scales with multi-resolution deformation.

4. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to perform the steps in the manufacturing method as described in any one of claims 1 to 3.

5. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it can perform the steps in the manufacturing method as described in any one of claims 1 to 3.

6. A particle radiotherapy system, characterized in that, include: The electronic device of claim 4 is used for creating particle radiotherapy plans; A particle accelerator for generating a particle beam based on the particle radiotherapy plan.

Images