Radiation Therapy System, Data Processing Method, and Storage Medium

Abstract

The present application provides a radiotherapy system, a data processing method and a storage medium, which relate to the technical field of radiotherapy. The system includes a calibration database, a training module and a data processing module. The calibration database can obtain a plurality of sample data generated during the clinical treatment process. The training module can perform deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model. The data processing module can process the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data. Since the treatment algorithm model is obtained by performing deep learning on a large amount of clinical data, the reliability of the treatment algorithm model is relatively high. Correspondingly, the accuracy of the preliminary treatment data generated by the treatment algorithm model is also relatively high, which can be directly used for clinical treatment, or only requires a small amount of revision by clinical staff to be used for clinical treatment, thereby effectively improving the treatment efficiency of the radiotherapy system.

Description

[0001] This application is a divisional application of the application with the application number 201780003947.X, the application date of June 5, 2017, and the invention title of "Radiotherapy System, Data Processing Method and Storage Medium". Technical Field

[0002] This application relates to the field of radiotherapy technology, and in particular to a radiotherapy system, a data processing method and a storage medium. Background Art

[0003] Radiotherapy (abbreviation: RT) is a local treatment method for treating tumors using radiation. Modern radiotherapy technology has an increasing dependence on image information, and a new radiotherapy mode based on radiotherapy images has gradually emerged, such as the Imaging Guided Radiation Therapy (IGRT) mode and the Adaptive Radiation Therapy (ART) mode.

[0004] In the related art, the radiotherapy system based on radiotherapy images mainly includes a medical image acquisition module, a registration module, a delineation module, and a treatment plan generation module. Among them, the medical image acquisition module can include Computed Tomography (CT), cone beam CT equipment, and Magnetic Resonance Imaging (MRI) equipment, etc., which are mainly used to acquire lesion images at different treatment stages; the registration module is mainly used to register the lesion images obtained in different ways and stages, and the delineation module is mainly used to delineate the irradiation area (i.e., the target area) of the tumor and the surrounding important organs; the treatment plan generation module is mainly used to generate a preliminary treatment plan according to the acquired lesion images and the delineation results.

[0005] However, in the radiotherapy system in the related art, data processing modules such as the registration module, the delineation module, and the treatment plan generation module generally perform preliminary processing on the acquired detection data according to a preset model. The accuracy of the processing result needs to be adjusted, confirmed, and approved by radiotherapy clinical personnel at all levels for a long time before it can be used for clinical treatment. This process highly depends on the professional experience and qualities of radiotherapy clinical personnel at all levels. Therefore, there are significant differences in the treatment effects of different hospitals using the current radiotherapy system. In addition, due to the need for a large amount of manual intervention, the efficiency needs to be improved. Summary of the Invention

[0006] In order to solve the above problems, this application provides a radiotherapy system, a data processing method and a storage medium. The technical solutions are as follows:

[0007] In a first aspect, a radiotherapy system is provided, the system comprising:

[0008] a calibration database, a training module, and a data processing module;

[0009] The calibration database is configured to obtain a plurality of sample data generated during clinical treatment, wherein each sample data includes a set of detection data and a corresponding set of treatment data applied to clinical treatment;

[0010] The training module is configured to perform deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model, and send the treatment algorithm model to the data processing module;

[0011] The data processing module is configured to process the detection data received during clinical treatment according to the treatment algorithm model to generate preliminary treatment data.

[0012] Optionally, the calibration database is further configured to receive revised sample data, the revised sample data including revised data for the preliminary treatment data and detection data corresponding to the preliminary treatment data, wherein the revised data is the actual data used for clinical treatment after the clinical staff revise the preliminary treatment data;

[0013] The training module is further configured to set a weight value for the revised sample data such that the weight value of the revised sample data is greater than a preset weight value, and perform deep learning on the updated sample data in the calibration database.

[0014] Optionally, the training module is further configured to receive a feedback level for specified sample data, the feedback level being positively correlated with the quality of the treatment effect of the treatment data in the specified sample data;

[0015] The training module is further configured to adjust the weight value of the specified sample data according to the feedback level of the specified sample data, and perform deep learning on the sample data in the calibration database with the adjusted weight value, wherein the adjusted weight value of the specified sample data is positively correlated with the feedback level.

[0016] Optionally, the radiotherapy system further comprises: a communication module; the communication module is configured to send the treatment algorithm model to a cloud server so that the cloud server sends the treatment algorithm model to the local server of other hospitals;

[0017] Alternatively, the radiotherapy system is deployed in a cloud server, and the plurality of sample data stored in the calibration database is obtained from the clinical database of a specified hospital; the communication module is configured to send the treatment algorithm model to the local server of any hospital.

[0018] Optionally, the data processing module is configured to receive the detection data sent by the local server of any hospital, process the detection data according to the treatment algorithm model, generate preliminary treatment data, and send the preliminary treatment data to the local server.

[0019] Optionally, the clinical treatment process includes: an image registration stage, and the sample data generated in the image registration stage includes a plurality of medical image data and corresponding registration images; the training module is configured to: perform deep learning on the plurality of sample data generated in the image registration stage to obtain a registration model;

[0020] And / or, the clinical treatment process includes: a delineation stage, and the sample data generated in the delineation stage includes registration images and corresponding delineation result images; the training module is configured to: perform deep learning on the plurality of sample data generated in the delineation stage to obtain a delineation model;

[0021] And / or, the clinical treatment process includes: a treatment plan formulation stage, and the sample data generated in the treatment plan formulation stage includes: medical image data, delineation result images, and corresponding treatment plans; the training module is configured to: perform deep learning on the plurality of sample data generated in the treatment plan formulation stage to obtain a treatment plan formulation model.

[0022] Optionally, the clinical treatment process includes: a pre-treatment positioning stage, and the sample data generated in the pre-treatment positioning stage includes a plurality of medical image data and corresponding positioning offsets, the training module is configured to: perform deep learning on the plurality of sample data generated in the pre-treatment positioning stage to obtain a pre-treatment positioning model;

[0023] And / or, the clinical treatment process includes: a mid-treatment monitoring stage; the sample data generated in the mid-treatment monitoring stage includes monitoring data and corresponding adjustment data, wherein the monitoring data includes: patient displacement, tumor monitoring images, and remaining radiation dose, and the adjustment data includes treatment nozzle displacement, tumor displacement, and dose error; the training module is configured to: perform deep learning on the plurality of sample data generated in the mid-treatment monitoring stage to obtain a monitoring model.

[0024] Optionally, each sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information.

[0025] The data received by the data processing module during the clinical treatment process further includes additional data corresponding to the detection data, and the data processing module is further configured to process the received detection data and the corresponding additional data according to the treatment algorithm model to generate preliminary treatment data.

[0026] Optionally, each sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information;

[0027] The calibration database is further configured to classify the multiple sample data according to the content of the target patient attribute information in the additional data of each sample data, and the target patient attribute information is determined from the at least one patient attribute information;

[0028] The training module is configured to perform deep learning on each type of sample data stored in the calibration database respectively to obtain multiple types of treatment algorithm models;

[0029] The data received by the data processing module during the clinical treatment process further includes additional data corresponding to the detection data. The data processing module is further configured to determine a corresponding treatment algorithm model from the multiple types of treatment algorithm models according to the content of the target patient attribute information in the additional data, and process the received detection data according to the corresponding treatment algorithm model to generate preliminary treatment data.

[0030] In a second aspect, a data processing method is provided, which is applied to the radiotherapy system shown in the first aspect. The method includes:

[0031] Obtain a plurality of sample data generated during the clinical treatment process, where each sample data includes a set of detection data and a corresponding set of treatment data applied to the clinical treatment;

[0032] Perform deep learning on the multiple sample data stored in the calibration database to obtain a treatment algorithm model;

[0033] Process the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data.

[0034] Optionally, the method further includes:

[0035] Receive revised sample data, where the revised sample data includes revised data for the preliminary treatment data and detection data corresponding to the preliminary treatment data, and the revised data is the actual data used for clinical treatment after the clinical staff revise the preliminary treatment data;

[0036] Set a weight value for the revised sample data such that the weight value of the revised sample data is greater than a preset weight value;

[0037] Perform deep learning on the updated sample data in the calibration database.

[0038] Optionally, the method further includes:

[0039] Receive a feedback level for specified sample data, where the feedback level is positively correlated with the quality of the treatment effect of the treatment data in the specified sample data;

[0040] Adjust the weight value of the specified sample data according to the feedback level of the specified sample data, where the adjusted weight value of the specified sample data is positively correlated with the feedback level;

[0041] Perform deep learning on the sample data with adjusted weight values.

[0042] Optionally, after obtaining the treatment algorithm model, the method further includes: sending the treatment algorithm model to a cloud server so that the cloud server sends the treatment algorithm model to the local servers of other hospitals; or,

[0043] The radiotherapy system is deployed in a cloud server. The obtaining of multiple sample data generated during the clinical treatment process includes: obtaining the multiple sample data from the clinical database of a specified hospital; after obtaining the treatment algorithm model, the method further includes: sending the treatment algorithm model to the local server of any hospital.

[0044] Optionally, the processing of the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data includes: receiving the detection data sent by the local server of any hospital; processing the detection data according to the treatment algorithm model to generate preliminary treatment data; the method further includes: sending the preliminary treatment data to the local server.

[0045] Optionally, the clinical treatment process includes: an image registration stage, and the sample data generated in the image registration stage includes multiple medical image data and corresponding registration images. The performing of deep learning on the multiple sample data stored in the calibration database to obtain a treatment algorithm model includes: performing deep learning on the multiple sample data generated in the image registration stage to obtain a registration model;

[0046] And / or, the clinical treatment process includes: a contouring stage, and the sample data generated in the contouring stage includes registration images and corresponding contoured result images; the performing of deep learning on the multiple sample data stored in the calibration database to obtain a treatment algorithm model includes: performing deep learning on the multiple sample data generated in the contouring stage to obtain a contouring model;

[0047] And / or, the clinical treatment process includes: a treatment plan formulation stage, and the sample data generated in the treatment plan formulation stage includes: medical image data, a delineation result image, and a corresponding treatment plan; the deep learning of the multiple sample data stored in the calibration database to obtain a treatment algorithm model includes: the deep learning of the sample data generated in the treatment plan formulation stage to obtain a treatment plan formulation model.

[0048] Optionally, the clinical treatment process includes: a pre-treatment positioning stage, and the sample data generated in the pre-treatment positioning stage includes multiple medical image data and corresponding positioning offsets. The deep learning of the multiple sample data stored in the calibration database to obtain a treatment algorithm model includes: the deep learning of the multiple sample data generated in the pre-treatment positioning stage to obtain a pre-treatment positioning model;

[0049] And / or, the clinical treatment process includes: a treatment monitoring stage, and the sample data generated in the treatment monitoring stage includes monitoring data and corresponding adjustment data. Among them, the monitoring data includes: patient displacement, tumor monitoring images, and remaining radiation dose, and the adjustment data includes treatment nozzle displacement, tumor displacement, and dose error; the deep learning of the multiple sample data stored in the calibration database to obtain a treatment algorithm model includes: the deep learning of the multiple sample data generated in the treatment monitoring stage to obtain a monitoring model.

[0050] Optionally, each sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information;

[0051] The data received during the clinical treatment process further includes additional data corresponding to the detection data. The processing of the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data includes:

[0052] Processing the received detection data and the corresponding additional data according to the treatment algorithm model to generate preliminary treatment data.

[0053] Optionally, each sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information; the method further includes:

[0054] Classifying the multiple sample data according to the content of the target patient attribute information in the additional data of each sample data, and the target patient attribute information is determined from the at least one patient attribute information;

[0055] Performing deep learning on the multiple sample data stored in the calibration database to obtain a treatment algorithm model, including:

[0056] Performing deep learning on each type of sample data stored in the calibration database respectively to obtain multiple types of treatment algorithm models;

[0057] The data received during the clinical treatment process further includes additional data corresponding to the detection data. Processing the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data, including:

[0058] Determining a corresponding treatment algorithm model from the multiple types of treatment algorithm models according to the content of the target patient attribute information in the additional data;

[0059] Processing the received detection data according to the corresponding treatment algorithm model to generate preliminary treatment data.

[0060] In a third aspect, a computer-readable storage medium is provided. Instructions are stored in the computer-readable storage medium. When the computer-readable storage medium runs on a computer, the computer is caused to execute the data processing method provided in the second aspect.

[0061] In a fourth aspect, a computer program product containing instructions is provided. When the computer program product runs on a computer, the computer is caused to execute the data processing method provided in the second aspect.

[0062] In summary, the embodiments of the present application provide a radiotherapy system, a data processing method, and a storage medium. The radiotherapy system includes: a calibration database, a training module, and a data processing module; the calibration database can obtain multiple sample data generated during the clinical treatment process, the training module can perform deep learning on the multiple sample data stored in the calibration database to obtain a treatment algorithm model, and the data processing module can process the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data. Since the treatment algorithm model is trained by performing deep learning on a large amount of clinical data, especially clinical data with high-quality treatment effects and efficiency, the reliability of the treatment algorithm model is relatively high. Correspondingly, the accuracy of the preliminary treatment data generated by the treatment algorithm model is also relatively high, and it can be directly used for clinical treatment, or only requires a small amount of revision by clinical personnel to be used for clinical treatment, thereby effectively improving the overall effectiveness of radiotherapy and the treatment efficiency of the radiotherapy system. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0064] Figure 1 It is a schematic structural diagram of a radiotherapy system provided by an embodiment of the present application;

[0065] Figure 2 It is a schematic structural diagram of another radiotherapy system provided by an embodiment of the present application;

[0066] Figure 3 It is a schematic diagram of an application scenario of a radiotherapy system provided by an embodiment of the present application;

[0067] Figure 4 It is a schematic structural diagram of a data processing module provided by an embodiment of the present application;

[0068] Figure 5 It is a flowchart of a data processing method provided by an embodiment of the present application.

[0069] Through the above accompanying drawings, the clear embodiments of the present application have been shown, and there will be more detailed descriptions hereinafter. These accompanying drawings and text descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. Specific Embodiments

[0070] To make the objectives, technical solutions, and advantages of the present application clearer, the following will further describe the embodiments of the present application in detail with reference to the accompanying drawings.

[0071] Figure 1 It is a schematic structural diagram of a radiotherapy system provided by an embodiment of the present application. Referring to Figure 1 , the system may specifically include: a calibration database 10, a training module 20, and a data processing module 30.

[0072] Among them, the calibration database 10 is used to obtain a plurality of sample data generated during the clinical treatment process, where each sample data includes a set of detection data and a corresponding set of treatment data applied to the clinical treatment.

[0073] The calibration database 10 can be connected to the clinical database of the hospital through a preset database interface. The clinical database block can store the detection data and treatment data of multiple patients. Each patient corresponds to a set of detection data and a set of treatment data, and the treatment data of each patient is the data actually used for clinical treatment verified by clinical staff during the treatment process. Therefore, the reliability of the treatment data is relatively high. Moreover, since the amount of calibrated sample data required for deep learning is large, by directly obtaining the sample data generated during the clinical treatment process, it is no longer necessary to generate sample data through manual calibration, thus effectively reducing the difficulty of obtaining sample data.

[0074] In the embodiment of the present application, the calibration database 10 can obtain the detection data and the corresponding treatment data from the clinical database in real time or periodically, and store a set of detection data and the corresponding set of treatment data as a sample data.

[0075] Figure 2 It is a schematic diagram of another radiotherapy system provided by the embodiment of the present application. Refer to Figure 2 As can be seen, the clinical treatment process of radiotherapy generally includes multiple treatment stages such as treatment plan generation, pre-treatment positioning, and in-treatment monitoring. The detection data and treatment data in each stage are different. For example, for the treatment plan formulation stage, the detection data may include medical image data, and the corresponding treatment data may be the treatment plan; for the pre-treatment positioning stage, the detection data may include the medical image data obtained before and during treatment, and the corresponding treatment data may include the positioning offset of the patient; for the in-treatment monitoring stage, the detection data may include the displacement of the patient during treatment, tumor monitoring images, and remaining radiation dose, etc., and the treatment data may include the displacement of the treatment muzzle, tumor displacement, and dose error, etc. In order to be able to generate corresponding treatment algorithm models for each treatment stage respectively, the calibration data 10 can classify and store the sample data generated in different treatment stages. The radiotherapy system provided by the present application can be applied to Figure 2 the entire clinical treatment process shown in

[0076] The training module 20 is used to perform deep learning on the multiple sample data stored in the calibration database to obtain a treatment algorithm model, and send the treatment algorithm model to the data processing module 30.

[0077] The training module 20 can be connected to the calibration database 10 through a preset interface. When the training module 20 detects that the number of sample data stored in the calibration database 10 reaches the preset data volume requirement, it can start deep learning on the sample data stored in the calibration database 10. For example, deep learning can be performed through a supervised learning method to obtain a treatment algorithm model.

[0078] In the embodiment of the present application, as Figure 2 shown, the training module 20 can perform deep learning on the sample data corresponding to each treatment stage stored in the calibration database 10 respectively, and then obtain a treatment algorithm model for each treatment stage.

[0079] It should be noted that as the data in the clinical database is continuously updated, the sample data stored in the calibration database 10 will also be continuously updated. Correspondingly, when the training module 20 detects that the sample data in the calibration database 10 has been updated, it can perform deep learning on the data stored in the calibration database 10 again, or it can also perform deep learning on the sample data stored in the calibration database 10 periodically, so as to continuously optimize and improve the treatment algorithm model.

[0080] The data processing module 30 is used to process the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data.

[0081] In the embodiment of the present application, after clinical staff obtains the detection data of a patient during the clinical treatment process, it can be directly input into the data processing module 30. The data processing module 30 can process the detection data according to the treatment algorithm model obtained through deep learning to generate preliminary treatment data. Since the treatment algorithm model is obtained by performing deep learning on a large amount of clinical data, the reliability of the treatment algorithm model is relatively high. Correspondingly, the accuracy of the preliminary treatment data generated by the treatment algorithm model is also relatively high, and it can be directly used for clinical treatment, or only a small amount of revision by clinical staff is required to be used for clinical treatment, thereby effectively improving the treatment efficiency of the radiotherapy system.

[0082] Furthermore, if clinical staff believes that the preliminary treatment data generated by the data processing module 30 has defects and uses it for clinical treatment only after revising it, the operator can directly upload the revised data and the corresponding treatment data as revised sample data to the calibration database 10, or alternatively, the revised data and the corresponding treatment data can be stored in the clinical database, and then the calibration database 10 can obtain the revised sample data from the clinical database.

[0083] After the calibration database 10 in the radiotherapy system receives the revised sample data, the training module 20 can set the weight value of the revised sample data so that the weight value of the revised sample data is greater than the preset weight value, and perform deep learning on the updated data in the calibration database 10 again.

[0084] In the embodiment of the present application, when the training module 20 initially trains the data stored in the calibration database 10, an initial weight value can be assigned to each sample data, and the initial weight values of the sample data in this initial stage can be the same, and the preset weight value can be this initial weight value. Since the revised sample data is the data revised by clinical staff, it indicates that there may still be defects in the original treatment algorithm model based on which the data processing module 30 generates the preliminary treatment data. Therefore, the weight value of the revised sample data can be set relatively high, so that when the training module 20 performs deep learning on the updated data again, the influence of the revised sample data on the treatment algorithm model can be improved, thereby further improving the treatment algorithm model.

[0085] It should be noted that, in order to facilitate each module in the radiotherapy system to identify the revised sample data, the operator can add a revision identifier to the revised sample data when storing or uploading the revised sample data, and this revision identifier is used to indicate that the sample data is the sample data revised by clinical staff.

[0086] In addition, if the preliminary treatment data generated by the data processing module 30 is directly used for clinical treatment without revision after being verified by clinical staff, it indicates that the treatment algorithm model is already relatively perfect. Therefore, this treatment data and the corresponding detection data can be no longer uploaded to the calibration database 10. Of course, the operator can also upload this treatment data and the corresponding detection data to the calibration database 10 to further increase the sample size in the calibration database.

[0087] Furthermore, in the embodiment of the present application, after the patient completes the treatment, the operator can also regularly visit the patient to track the treatment effect, and can determine the feedback level of the sample data corresponding to the patient according to the quality of the treatment effect, and then upload this feedback level to the training module 20 in the radiotherapy system. Exemplarily, this feedback level and the quality of the treatment effect can be positively correlated, that is, the better the treatment effect, the higher the feedback level.

[0088] After the training module 20 receives the feedback level for the specified sample data, it can adjust the weight value of the specified sample data according to the feedback level of the specified sample data, and the adjusted weight value of the specified sample data is positively correlated with the feedback level, that is, the adjusted weight value of the specified sample data is positively correlated with the quality of its treatment effect.

[0089] If the feedback level of the specified sample data is relatively high, the training module 20 can determine that the treatment effect of the treatment data in the specified sample data is relatively good. Therefore, the corresponding weight value can also be adjusted to be relatively high to enhance the influence of the specified sample data on the treatment algorithm model. Correspondingly, if the feedback level of the specified sample data is relatively low, the training module 20 can determine that the treatment effect of the treatment data in the specified sample data is relatively poor. Therefore, the corresponding weight value can also be adjusted to be relatively low to reduce the influence of the specified sample data on the treatment algorithm model. If the feedback level of the specified sample data is lower than the preset lower threshold value, the training module 20 can set the weight value of the specified sample data to 0, or can directly delete the specified sample data from the calibration database 10.

[0090] After completing the adjustment of the weight value of the specified sample data, the training module 20 can perform deep learning on the sample data with the adjusted weight value in the calibration database 10 again, so as to continuously optimize the treatment algorithm model.

[0091] Figure 3 It is a schematic diagram of an application scenario of a radiotherapy system provided by an embodiment of the present application. In an implementable manner, the radiotherapy system can be deployed in the local server of a specified hospital. For example, it can be deployed in server 01 of a certain Class III Grade A hospital. The radiotherapy system can also include a communication module. The radiotherapy system deployed in the local server 01 can upload the trained treatment algorithm model to the cloud server through the communication module, so that the cloud server can send the treatment algorithm model to the local servers of other hospitals, such as server 02 to server 04, so that other hospitals can also obtain excellent medical resources.

[0092] In another implementable manner, the radiotherapy system can also be deployed in the cloud server. When the radiotherapy system is deployed in the cloud server, the multiple sample data stored in the calibration database 10 can be obtained from the clinical database of a specified hospital. Since the current medical levels in different regions are unevenly distributed and the medical levels of hospitals in different regions vary greatly. To integrate excellent treatment resources, Class A hospitals in provinces and cities with relatively high medical levels can be selected as the specified hospitals, and sample data can be obtained from the clinical databases of the specified hospitals to ensure the reliability of the trained treatment algorithm model.

[0093] When the radiotherapy system is deployed in a cloud server, on the one hand, the radiotherapy system may further include: a communication module, which is used to send the treatment algorithm model trained by the training module 20 to the local server of any hospital, so that the local server of the hospital can process the detection data in the clinical treatment process according to the treatment algorithm model to obtain preliminary treatment data. Wherein, any of the hospitals may be a hospital that has established a communication connection with the cloud server, and any of the hospitals may include the designated hospital or other hospitals with relatively low medical levels.

[0094] Exemplarily, assuming that in Figure 3 the system shown, the hospitals to which Server 01 and Server 02 belong are designated hospitals, then the radiotherapy system deployed in the cloud server can obtain sample data from the clinical databases of Server 01 and Server 02 for training to obtain a treatment algorithm model; thereafter, the radiotherapy system can send the treatment algorithm model to any one of Server 01 to Server 04 through the communication module.

[0095] When the radiotherapy system is deployed in a cloud server, on the other hand, the data processing module 30 in the radiotherapy system can also receive the detection data sent by the local server of any hospital, for example, it can receive the detection data sent by any one of Server 01 to Server 04, and then process the received detection data according to the treatment algorithm model to generate preliminary treatment data, and finally feedback the preliminary treatment data to the local server of the corresponding hospital.

[0096] Through the above method, township hospitals with relatively low medical levels can obtain efficient remote diagnosis and treatment. The process of remote diagnosis and treatment does not require the physical participation of excellent medical personnel in other centers, eliminating the drawback of sacrificing the time of these professionals for online remote guidance; at the same time, since there is no need for medical personnel in other centers to access patient data, the security of patient data is further ensured.

[0097] Certainly, the calibration database 10 and the training module 20 in the radiotherapy system can also be deployed in the cloud server, and the data processing module can be deployed in the local servers of hospitals everywhere. The local servers of these hospitals can establish a communication connection with the cloud server through a cloud interface to obtain the treatment algorithm model trained by the training module 20. If the training module 20 optimizes the treatment algorithm model, the data processing module deployed in the local server can also upgrade the stored treatment algorithm model in real time.

[0098] In the embodiment of the present application, referring to Figure 2, the clinical treatment process of radiotherapy can specifically include: a treatment plan generation stage, a pre-treatment positioning stage, and a treatment monitoring stage. Among them, the treatment plan generation stage can be further divided into an image registration stage, a delineation stage, and a treatment plan formulation stage. In the radiotherapy system provided by the embodiments of the present application, the calibration database 10 can obtain sample data of at least one of these stages for the training module to perform deep learning to obtain a corresponding treatment algorithm model. Refer to Figure 4 , the data processing module 30 can specifically include at least one of the following modules: an image registration sub-module 301, a delineation sub-module 302, a treatment plan formulation sub-module 303, a positioning sub-module 304, and a monitoring sub-module 305.

[0099] Correspondingly, the training module 20 can specifically be used to execute at least one of the methods shown in the following (1) to (5) to obtain at least one treatment algorithm model:

[0100] (1) Perform deep learning on multiple sample data generated in the image registration stage to obtain a registration model.

[0101] Among the sample data generated in the image registration stage, the detection data can be multiple medical image data, and the corresponding treatment data can be a registered image. Among them, the multiple medical image data can be image data of different treatment stages obtained by means of CT or magnetic resonance imaging, etc. In the embodiments of the present application, after the training module 20 trains and obtains a registration model, it can send the registration model to the image registration sub-module 301. The image registration sub-module 301 can process the medical image data obtained during the clinical treatment process according to the registration model and output a preliminary registered image.

[0102] (2) Perform deep learning on multiple sample data generated in the delineation stage to obtain a delineation model.

[0103] Among the sample data generated in the delineation stage, the detection data can be the registered image generated in the image registration stage, and the corresponding treatment data can be a delineation result image, in which the irradiation area (i.e., the target area) of the tumor and the surrounding important organs are delineated. In the embodiments of the present application, after the training module 20 trains and obtains a delineation model, it can send the delineation model to the delineation sub-module 302. The delineation sub-module 302 can process the registered image obtained during the clinical treatment process according to the delineation model and output a preliminary delineation result image.

[0104] (3) Perform deep learning on multiple sample data generated in the treatment plan formulation stage to obtain a treatment plan formulation model.

[0105] In the sample data generated in the treatment plan formulation stage, the detection data can be medical image data and the contoured result image, and the corresponding treatment data can be the treatment plan, which can specifically include data such as the radiotherapy cycle, the duration of each radiotherapy, the radiation dose, and the conformal shape of the irradiated target area. After the training module 20 trains the treatment plan formulation model, it can send the treatment plan formulation model to the treatment plan formulation sub-module 303. The treatment plan formulation sub-module 303 can process the medical image data and the contoured result image obtained during the clinical treatment process according to the treatment plan formulation model, and output a preliminary treatment plan.

[0106] In addition, it should be noted that in order to ensure the reliability of the generated treatment plan, the detection data obtained in this treatment plan formulation stage can also include the patient's age, gender, and other data related to the physical condition.

[0107] (4) Perform deep learning on multiple sample data generated in the pre-treatment setup stage to obtain a pre-treatment setup model.

[0108] In the sample data generated in the pre-treatment setup stage, the detection data can include the medical image data obtained in the image registration stage and the medical image data obtained in this setup stage. The corresponding treatment data can include the setup offset, which can refer to the offset between the patient's current position and the target position. The operator can adjust the patient's body position through this setup offset to ensure that the radiotherapy rays can accurately irradiate the target area. After the training module 20 trains the setup model, it can send the setup model to the setup sub-module 304. The setup sub-module 304 can process the medical image data obtained in the image registration stage and the medical image data obtained in the setup stage during the clinical treatment process according to the setup model, and output a preliminary setup offset.

[0109] (5) Perform deep learning on multiple sample data generated in the in-treatment monitoring stage to obtain a monitoring model.

[0110] In the sample data generated during the monitoring phase of this treatment, the detection data can be the monitoring data during the treatment, and the corresponding treatment data can be the adjustment data. Among them, the monitoring data can include: patient displacement, tumor monitoring images, remaining radiation dose, etc., and the adjustment data can include treatment muzzle displacement, tumor displacement, dose error, etc. Among them, the patient displacement in the monitoring data can refer to the offset of the patient's position relative to the target position detected by sensors or monitoring images during the treatment process, and the remaining radiation dose can be the dose remaining after the radiotherapy rays detected by the detector pass through the patient's body during the treatment process. The dose error in the adjustment data can refer to the error of the radiation dose in the treatment plan. After the training module 20 trains and obtains the monitoring model, it can send the monitoring model to the monitoring sub-module 305. The monitoring sub-module 305 can process the monitoring data obtained during the monitoring phase in the clinical treatment process according to the monitoring model and output preliminary adjustment data.

[0111] Further, referring to Figure 2 , the image registration phase in the treatment plan generation phase can specifically include deformation registration and image fusion, the delineation phase can specifically include image segmentation and contour delineation, the treatment plan generation phase can specifically include treatment template selection and treatment plan generation, as well as other treatment preparation operations. The pre-treatment positioning phase can specifically include 2D low-dose projection, 3D low-dose projection, image denoising, rigid registration, and other pre-treatment positioning operations; the in-treatment monitoring phase can specifically include motion monitoring, tumor tracking, dose verification, and other in-treatment monitoring operations. Correspondingly, the calibration data 10 can store the sample data generated by each processing operation in each phase, and the training module 20 can generate a treatment algorithm model for each processing operation according to the sample data corresponding to each processing operation.

[0112] In an alternative implementation manner of the embodiment of the present application, each sample data stored in the calibration data 10 can further include a set of additional data, and the set of additional data can include at least one patient attribute information. Therefore, the influence of the above additional data is also comprehensively considered in the treatment algorithm model obtained after the training module 20 trains the sample data, making the treatment algorithm model further improved. Correspondingly, when performing clinical treatment according to the treatment algorithm model, the additional data of the patient can be used as part of the detection data and input into the treatment algorithm model.

[0113] Exemplarily, referring to Figure 2, this set of additional data may include patient identity data, lesion data, treatment provider data, and other classification data. Among them, the identity data may include multiple patient attribute information such as gender, region, and age. The lesion data may include multiple patient attribute information such as the location and stage of the tumor. The treatment provider data may include multiple patient attribute information such as the attending physician, prescribed dose, and follow-up survival rate.

[0114] In another alternative implementation manner of the embodiment of the present application, after the calibration database obtains the additional data in each sample data, it may also classify the multiple sample data according to the content of the target patient attribute information in the additional data of each sample data. Among them, the target patient attribute information may be predetermined from the at least one patient attribute information, and the division rule based on which the multiple sample data are classified according to the content of the target patient attribute information may also be pre-configured.

[0115] Exemplarily, assuming that the target attribute information is: gender, the calibration database may classify the sample data with gender "female" in the additional data into one category and the sample data with gender "male" into another category among the multiple sample data. Or, if the target attribute information is: age, the calibration database may classify the sample data with age under 20 years old in the additional data into one category, the sample data with age between 20 and 50 years old into one category, and the sample data with age over 50 years old into one category among the multiple sample data.

[0116] Furthermore, the training module 20 may also perform deep learning on each category of sample data stored in the calibration database 10 respectively, so as to obtain various types of treatment algorithm models.

[0117] Exemplarily, if the sample data stored in the calibration database 10 are classified into two categories according to the gender of the patient, the training module 20 may perform deep learning on the two categories of sample data respectively, and obtain a treatment algorithm model for patient attribute information "male" and a treatment algorithm model for patient attribute information "female" respectively.

[0118] Correspondingly, the data received by the data processing module 30 during the clinical treatment process may also include additional data corresponding to the detection data. The data processing module 30 may also determine the corresponding treatment algorithm model from the various types of treatment algorithm models according to the content of the target patient attribute information in the additional data, and process the received detection data according to the corresponding treatment algorithm model to generate preliminary treatment data.

[0119] Exemplarily, assume that during the clinical treatment process, the target patient attribute information is: gender. Among the additional data corresponding to the detection data received by the data processing module 30, the gender of the patient is "female". Then, the data processing module 30 can select, from these two types of treatment algorithm models, the treatment algorithm model for patients with the attribute information of "female", and process the detection data according to the selected treatment algorithm model.

[0120] It should be noted that in the embodiments of the present application, the calibration database 10 can also use each patient attribute information as the target attribute information in turn to classify the multiple sample data multiple times. For example, the sample data can be classified multiple times according to patient attribute information such as age, region, gender, tumor location, and attending physician. Correspondingly, the training module 20 can train multiple treatment algorithm models respectively according to multiple types of sample data corresponding to each patient attribute information. For example Figure 2 In [example], for the sample data in the deformation registration stage, the training module 20 can train 12 corresponding treatment algorithm models A1 to A12 according to 12 patient attribute information such as age and region. Each type of treatment algorithm model can also include multiple sub-models according to the different content of the patient attribute information. Exemplarily, assume that the A1 model is a treatment algorithm model for the patient attribute information: age. Then, the A1 model can specifically include sub-models for patient attribute information of "under 20 years old", sub-models for patient attribute information of "20 to 50 years old", and sub-models for patient attribute information of "over 50 years old". Assume that the A2 model is a treatment algorithm model for the patient attribute information: gender. Then, the A2 model can specifically include sub-models for patient attribute information of "male", and sub-models for patient attribute information of "female".

[0121] In addition, a post-intelligent module can also be set in the radiotherapy system provided in the embodiments of the present application. The post-intelligent module can streamline and statistically analyze the sample data in the calibration database according to changes in various parameters of other modules in the radiotherapy system, such as changes in the sample quantity in the calibration database, adjustments of the weight values of the sample data by the training module, and changes in the quantity of the revised sample data, and inversely deduce the implicit expression model of the treatment algorithm model trained by the training module.

[0122] In summary, the embodiment of the present application provides a radiotherapy system, which includes a calibration database, a training module, and a data processing module. The calibration database can obtain a plurality of sample data generated during the clinical treatment process. The training module can perform deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model. The data processing module can process the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data. Since the treatment algorithm model is trained by performing deep learning on a large amount of clinical data, the reliability of the treatment algorithm model is relatively high. Correspondingly, the accuracy of the preliminary treatment data generated by the treatment algorithm model is also relatively high, and it can be directly used for clinical treatment, or only requires a small amount of revision by clinical staff to be used for clinical treatment, thereby effectively improving the treatment efficiency of the radiotherapy system.

[0123] Figure 5 is a flowchart of a data processing method provided by an embodiment of the present application, and this method can be applied to Figure 1 or Figure 2 the radiotherapy system shown in the reference Figure 5 , and this method may include:

[0124] Step 101: Obtain a plurality of sample data generated during the clinical treatment process, where each sample data includes a set of detection data and a corresponding set of treatment data applied to clinical treatment.

[0125] Step 102: Perform deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model.

[0126] Step 103: Process the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data.

[0127] A data processing method provided by the present application can be applied to Figure 1 or Figure 2 the radiotherapy system shown in the reference Figure 5 , and as shown in the reference

[0128] Step 104: Receive revised sample data, where the revised sample data includes revised data for the preliminary treatment data and the detection data corresponding to the preliminary treatment data.

[0129] Among them, the revised data is the data actually used for clinical treatment after the clinical staff revise the preliminary treatment data.

[0130] Step 105: Set the weight value of the revised sample data so that the weight value of the revised sample data is greater than the preset weight value.

[0131] After that, the radiotherapy system can execute step 102 again, that is, continue to perform deep learning on the updated sample data in the calibration database.

[0132] A data processing method provided by this application, as Figure 5 shown, the method may further include:

[0133] Step 106: Receive the feedback level for the specified sample data, where the feedback level is positively correlated with the quality of the treatment effect of the treatment data in the specified sample data.

[0134] Step 107: Adjust the weight value of the specified sample data according to the feedback level of the specified sample data, and the adjusted weight value of the specified sample data is positively correlated with the feedback level.

[0135] After that, the radiotherapy system can execute step 102 again, that is, continue to perform deep learning on the sample data after adjusting the weight value.

[0136] In an alternative implementation, the radiotherapy system can be deployed in the local server of a specified hospital. After obtaining the treatment algorithm model in the above step 102, the method may further include:

[0137] Send the treatment algorithm model to the cloud server so that the cloud server can send the treatment algorithm model to the local servers of other hospitals.

[0138] In another alternative implementation, the radiotherapy system can be deployed in the cloud server. The above step 101 may specifically include: obtaining the multiple groups of detection data and the corresponding multiple groups of treatment data from the clinical database of a specified hospital. After obtaining the treatment algorithm model in the above step 102, the method may further include: sending the treatment algorithm model to the local server of any hospital.

[0139] In yet another alternative implementation, the radiotherapy system can be deployed in the cloud server. The above step 103 may specifically include:

[0140] Step 1031a: Receive the detection data sent by the local server of any hospital.

[0141] Step 1032a: Process the detection data according to the treatment algorithm model to generate preliminary treatment data. Further, the method may further include:

[0142] Step 1033a: Send the preliminary treatment data to the local server.

[0143] Optionally, in the embodiments of the present application, the clinical treatment process may include at least one of the following stages: an image registration stage, a delineation stage, and a treatment plan formulation stage; wherein, the sample data generated in the image registration stage may include multiple medical image data and corresponding registration images, the sample data generated in the delineation stage includes registration images and corresponding delineation result images, and the sample data generated in the treatment plan formulation stage includes: medical image data, delineation result images, and corresponding treatment plans.

[0144] Correspondingly, step 102 may specifically include at least one of the following steps 1021 to 1023:

[0145] Step 1021: Perform deep learning on the multiple sample data generated in the image registration stage to obtain a registration model.

[0146] Step 1022: Perform deep learning on the multiple sample data generated in the delineation stage to obtain a delineation model.

[0147] Step 1023: Perform deep learning on the sample data generated in the treatment plan formulation stage to obtain a treatment plan formulation model.

[0148] Furthermore, the clinical treatment process may further include: a pre-treatment positioning stage and / or a mid-treatment monitoring stage. The sample data generated in the pre-treatment positioning stage includes multiple medical image data and corresponding positioning offsets, and the sample data generated in the mid-treatment monitoring stage includes monitoring data and corresponding adjustment data, wherein the monitoring data may include: patient displacement, tumor monitoring images, and remaining radiation dose, and the adjustment data includes treatment nozzle displacement, tumor displacement, and dose error.

[0149] Step 102 may further include at least one of the following steps 1024 and 1025:

[0150] Step 1024: Perform deep learning on the multiple sample data generated in the pre-treatment positioning stage to obtain a pre-treatment positioning model.

[0151] Step 1025: Perform deep learning on the multiple sample data generated in the mid-treatment monitoring stage to obtain a monitoring model.

[0152] In an optional implementation manner of the embodiments of the present application, each sample data may further include a set of additional data, and the set of additional data includes at least one patient attribute information. Therefore, in step 102 above, the training model obtained after performing deep learning on the sample data also comprehensively considers the influence of the additional data, making the treatment algorithm model further improved. Correspondingly, when performing clinical treatment according to the treatment algorithm model, the additional data of the patient may be used as part of the detection data and input into the treatment algorithm model.

[0153] In another alternative implementation manner of the embodiment of the present application, each sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information. After the above step 101, the method may further include:

[0154] Classify the multiple sample data according to the content of the target patient attribute information in the additional data of each sample data, and the target patient attribute information is determined from the at least one patient attribute information.

[0155] The above step 102 may specifically include: performing deep learning on each type of sample data stored in the calibration database respectively to obtain multiple types of treatment algorithm models.

[0156] Correspondingly, the data received during the clinical treatment process may further include additional data corresponding to the detection data, and the above step 103 may include:

[0157] Step 1031b: Determine the corresponding treatment algorithm model from the multiple types of treatment algorithm models according to the content of the target patient attribute information in the additional data.

[0158] Step 1032b: Process the received detection data according to the corresponding treatment algorithm model to generate preliminary treatment data.

[0159] It should be noted that the order of the steps of the data processing method provided in the embodiment of the present application can be appropriately adjusted, and the steps can also be increased or decreased accordingly according to the situation. For example, the above step 106 and step 107 can also be executed before step 104. Any method of change that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application, so it will not be elaborated here.

[0160] It should also be noted that for the convenience and brevity of description, in the above-described data processing method, step 101 and step 104 may be executed by the calibration database 10 in the radiotherapy system, step 102, step 105 to step 107 may be executed by the training module 20, and step 103 may be executed by the data processing module. The specific working processes of these steps can refer to the function descriptions of each module in the foregoing system embodiment, and will not be elaborated here.

[0161] In summary, the embodiments of the present application provide a data processing method. This method can obtain multiple sample data generated during the clinical treatment process, perform deep learning on the multiple sample data to obtain a treatment algorithm model, and then process the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data. Since the treatment algorithm model is trained by performing deep learning on a large amount of clinical data, the reliability of the treatment algorithm model is relatively high. Correspondingly, the accuracy of the preliminary treatment data generated by the treatment algorithm model is also relatively high, and it can be directly used for clinical treatment, or only requires a small amount of revision by clinical staff to be used for clinical treatment, effectively improving the treatment efficiency.

[0162] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above embodiments can be completed by hardware, or can be completed by instructing relevant hardware through instructions. The instructions can be stored in a computer-readable storage medium. The storage medium mentioned above can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that integrates one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium, or a semiconductor medium (for example, a solid-state drive), etc.

[0163] In addition, the term "and / or" in the present application is merely a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0164] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A radiotherapy system, characterized in that, The radiotherapy system includes: a calibration database, a training module, and a data processing module; The calibration database is used to obtain a plurality of sample data generated during clinical treatment from a clinical database, where each sample data includes a set of detection data and a corresponding set of treatment data applied to clinical treatment, and the treatment data is the data actually used for clinical treatment verified by clinical staff during the treatment process; The training module is used to, when the number of sample data in the calibration database reaches a preset quantity requirement, perform deep learning on the plurality of sample data stored in the calibration database by means of supervised learning to obtain a treatment algorithm model, and send the treatment algorithm model to the data processing module; The data processing module is used to process the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data; The training module is further used to receive a feedback level for specified sample data, and the feedback level is positively correlated with the quality of the treatment effect of the treatment data in the specified sample data; The training module is further used to adjust the weight value of the specified sample data according to the feedback level of the specified sample data, and perform deep learning on the sample data in the calibration database after adjusting the weight value, and the adjusted weight value of the specified sample data is positively correlated with the feedback level.

2. The radiotherapy system according to claim 1, wherein the calibration database is further used to receive revised sample data, and the revised sample data includes revised data for the preliminary treatment data and the detection data corresponding to the preliminary treatment data, where the revised data is the data actually used for clinical treatment obtained after clinical staff revise the preliminary treatment data; The training module is further used to set the weight value of the revised sample data such that the weight value of the revised sample data is greater than a preset weight value, and perform deep learning on the updated sample data in the calibration database.

3. The radiotherapy system according to claim 1, wherein The radiotherapy system further includes: a communication module; The communication module is used to send the treatment algorithm model to a cloud server so that the cloud server sends the treatment algorithm model to the local server of other hospitals; Or, the radiotherapy system is deployed in a cloud server, and the plurality of sample data stored in the calibration database is obtained from the clinical database of a specified hospital, and the communication module is used to send the treatment algorithm model to the local server of any hospital.

4. The radiotherapy system according to claim 1, wherein The radiotherapy system is deployed in a cloud server, and the plurality of sample data stored in the calibration database is obtained from the clinical database of a specified hospital; The data processing module is used to receive the detection data sent by the local server of any hospital, process the detection data according to the treatment algorithm model to generate preliminary treatment data, and send the preliminary treatment data to the local server.

5. The radiotherapy system according to any one of claims 1 to 4, wherein The clinical treatment process includes: an image registration stage, and the sample data generated in the image registration stage includes a plurality of medical image data and corresponding registered images; the training module is configured to: perform deep learning on the plurality of sample data generated in the image registration stage to obtain a registration model; And / or, the clinical treatment process includes: a delineation stage, and the sample data generated in the delineation stage includes registered images and corresponding delineation result images; the training module is configured to: perform deep learning on the plurality of sample data generated in the delineation stage to obtain a delineation model; And / or, the clinical treatment process includes: a treatment plan formulation stage, and the sample data generated in the treatment plan formulation stage includes: medical image data, delineation result images, and corresponding treatment plans; the training module is configured to: perform deep learning on the plurality of sample data generated in the treatment plan formulation stage to obtain a treatment plan formulation model.

6. The radiotherapy system according to any one of claims 1 to 4, characterized in that, The clinical treatment process includes: a pre-treatment positioning stage, and the sample data generated in the pre-treatment positioning stage includes a plurality of medical image data and corresponding positioning offsets; the training module is configured to: perform deep learning on the plurality of sample data generated in the pre-treatment positioning stage to obtain a pre-treatment positioning model; And / or, the clinical treatment process includes: an in-treatment monitoring stage; the sample data generated in the in-treatment monitoring stage includes monitoring data and corresponding adjustment data, wherein the monitoring data includes: patient displacement, tumor monitoring images, and remaining radiation dose, and the adjustment data includes treatment nozzle displacement, tumor displacement, and dose error; the training module is configured to: perform deep learning on the plurality of sample data generated in the in-treatment monitoring stage to obtain a monitoring model.

7. The radiotherapy system according to any one of claims 1 to 4, wherein Each of the sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information; The additional data corresponding to the detection data is further included in the data received by the data processing module during the clinical treatment process, and the data processing module is configured to process the received detection data and the corresponding additional data according to the treatment algorithm model to generate the preliminary treatment data.

8. The radiotherapy system according to any one of claims 1 to 4, wherein Each of the sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information; The calibration database is further configured to classify the plurality of sample data according to the content of the target patient attribute information in the additional data of each sample data, and the target patient attribute information is determined from the at least one patient attribute information; The training module is configured to perform deep learning on each type of sample data stored in the calibration database respectively to obtain multiple types of treatment algorithm models; The data received by the data processing module during the clinical treatment process further includes additional data corresponding to the detection data. The data processing module is configured to determine a corresponding treatment algorithm model from the multiple types of treatment algorithm models according to the content of the target patient attribute information in the additional data, and process the received detection data according to the corresponding treatment algorithm model to generate the preliminary treatment data.

9. The radiotherapy system according to any one of claims 1 to 4, characterized in that, The radiotherapy system includes a data interface, a model training interface, or a cloud interface to generate data for a preliminary treatment plan using an algorithm model respectively.

10. The radiotherapy system according to claim 9, wherein The radiotherapy system further includes a deformation registration interface, an image fusion interface, an image segmentation interface, a contour drawing interface, a template selection interface, a treatment plan formulation interface, a low-dose projection interface, a low-dose reconstruction interface, an image denoising interface, a rigid registration interface, a motion monitoring interface, a tumor tracking interface, a dose verification interface, or a treatment operation interface.

11. A data processing method, characterized in that, Applied to a radiotherapy system, the method includes: Obtaining a plurality of sample data generated during the clinical treatment process, where each sample data includes a set of detection data and a corresponding set of treatment data applied to the clinical treatment, and the treatment data is the data actually used for clinical treatment verified by clinical staff during the treatment process; When the number of sample data in the calibration database reaches a preset quantity requirement, performing deep learning on the plurality of sample data stored in the calibration database by means of supervised learning to obtain a treatment algorithm model; Processing the detection data received during the clinical treatment process according to the treatment algorithm model to generate preliminary treatment data; Receiving a feedback level for a specified sample data, where the feedback level is positively correlated with the quality of the treatment effect of the treatment data in the specified sample data; Adjusting the weight value of the specified sample data according to the feedback level of the specified sample data, and the adjusted weight value of the specified sample data is positively correlated with the feedback level; Performing deep learning on the sample data after adjusting the weight value.

12. The method according to claim 11, wherein The method further includes: Receiving revised sample data, where the revised sample data includes revised data for the preliminary treatment data and the detection data corresponding to the preliminary treatment data, and the revised data is the data actually used for clinical treatment obtained after clinical staff revise the preliminary treatment data; Setting the weight value of the revised sample data such that the weight value of the revised sample data is greater than a preset weight value; Performing deep learning on the updated sample data in the calibration database.

13. The method according to claim 11, wherein After obtaining the treatment algorithm model, the method further includes: sending the treatment algorithm model to a cloud server so that the cloud server sends the treatment algorithm model to the local server of other hospitals; or The radiotherapy system is deployed in a cloud server, and obtaining a plurality of sample data generated during the clinical treatment process includes: obtaining the plurality of sample data from the clinical database of a specified hospital. After obtaining the treatment algorithm model, the method further includes: sending the treatment algorithm model to the local server of any hospital.

14. The method according to claim 11, wherein The radiotherapy system is deployed in a cloud server, and obtaining a plurality of sample data generated during a clinical treatment process includes: obtaining the plurality of sample data from the clinical database of a designated hospital; According to the treatment algorithm model, processing the detection data received during the clinical treatment process to generate preliminary treatment data, includes: receiving the detection data sent by the local server of any hospital; processing the detection data according to the treatment algorithm model to generate preliminary treatment data; The method further includes: sending the preliminary treatment data to the local server.

15. The method according to any one of claims 11 to 14, wherein the clinical treatment process includes: an image registration stage, and the sample data generated in the image registration stage includes a plurality of medical image data and corresponding registration images. Performing deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model includes: performing deep learning on the plurality of sample data generated in the image registration stage to obtain a registration model; and / or, the clinical treatment process includes: a delineation stage, and the sample data generated in the delineation stage includes registration images and corresponding delineation result images; Performing deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model includes: performing deep learning on the plurality of sample data generated in the delineation stage to obtain a delineation model; and / or, the clinical treatment process includes: a treatment plan formulation stage, and the sample data generated in the treatment plan formulation stage includes: medical image data, delineation result images, and corresponding treatment plans; Performing deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model includes: performing deep learning on the plurality of sample data generated in the treatment plan formulation stage to obtain a treatment plan formulation model.

16. The method according to any one of claims 11 to 14, wherein the clinical treatment process includes: a pre-treatment positioning stage, and the sample data generated in the pre-treatment positioning stage includes a plurality of medical image data and corresponding positioning offsets. Performing deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model includes: performing deep learning on the plurality of sample data generated in the pre-treatment positioning stage to obtain a pre-treatment positioning model; and / or, the clinical treatment process includes: a treatment monitoring stage, and the sample data generated in the treatment monitoring stage includes monitoring data and corresponding adjustment data, wherein the monitoring data includes: patient displacement, tumor monitoring images, and remaining radiation dose, and the adjustment data includes treatment nozzle displacement, tumor displacement, and dose error; Performing deep learning on the plurality of sample data stored in the calibration database to obtain a treatment algorithm model includes: performing deep learning on the plurality of sample data generated in the treatment monitoring stage to obtain a monitoring model.

17. The method according to any one of claims 11 to 14, wherein each of the sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information; the data received during the clinical treatment process further includes additional data corresponding to the detection data, and according to the treatment algorithm model, the detection data received during the clinical treatment process is processed to generate preliminary treatment data, including: processing the received detection data and the corresponding additional data according to the treatment algorithm model to generate the preliminary treatment data.

18. The method according to any one of claims 11 to 14, wherein each of the sample data further includes a set of additional data, and the set of additional data includes at least one patient attribute information; the method further includes: classifying the multiple sample data according to the content of the target patient attribute information in the additional data of each sample data, and the target patient attribute information is determined from the at least one patient attribute information; the deep learning of the multiple sample data stored in the calibration database to obtain a treatment algorithm model includes: performing deep learning on each category of sample data stored in the calibration database respectively to obtain multiple types of treatment algorithm models; the data received during the clinical treatment process further includes additional data corresponding to the detection data, and according to the treatment algorithm model, the detection data received during the clinical treatment process is processed to generate preliminary treatment data, including: determining a corresponding treatment algorithm model from the multiple types of treatment algorithm models according to the content of the target patient attribute information in the additional data; processing the received detection data according to the corresponding treatment algorithm model to generate the preliminary treatment data.

19. A computer-readable storage medium, characterized in that, Instructions are stored in the computer-readable storage medium, and when the computer-readable storage medium runs on a computer, the computer is caused to execute the data processing method according to any one of claims 11 to 18.

Images