A tumor radiotherapy control system and storage medium

Abstract

The present invention discloses a tumor radiotherapy control system and a storage medium. The system comprises: a dose prediction module, which is used to input a patient's image and a structure set containing a boost target area, a tumor target area and an area where an organ at risk is located in the image into a dose prediction model, so as to use the dose prediction model to predict and process the dose indexes of the tumor and the organ at risk; a judgment module, which is used to judge whether the predicted dose indexes of the tumor and the organ at risk meet the plan requirements; and an optimization and adjustment module, which is used to automatically adjust the boost target area within the range of the tumor target area when the plan requirements are not met. After the optimization and adjustment module obtains the adjusted boost target area, the prediction process and the judgment process are repeated until the predicted dose indexes of the tumor and the organ at risk meet the plan requirements, thereby shortening the planning process time, improving the consistency of the planning quality and reducing the planning cost.

Description

Technical Field

[0001] The present invention relates to the field of medical devices, and in particular to a tumor radiotherapy control system and a storage medium. Background Art

[0002] Tumor burden refers to the number of cancer cells in the human body, the size of the tumor, and the total number of cancer lesions. Generally, cancer patients have a large tumor burden and are closely related to the surrounding normal tissues. The internal blood supply is insufficient, and the effects of surgery, chemotherapy, and conventional fractionated radiotherapy (abbreviated as radiotherapy) are not satisfactory. In recent decades, radiation oncologists have been committed to developing new tumor radiotherapy programs to solve this problem.

[0003] In tumor radiotherapy, the biological effective dose (BED) refers to the total dose required for a given regimen to produce the same cell killing when the dose rate is infinitely low or the dose is infinitely long; the physical dose refers to the radiation dose absorbed by the human body; the tumor target area refers to the radiation irradiation area formed by expanding the tumor area visible in the image by a certain distance, which is used for radiotherapy plan design; the tumor boost target area (abbreviated as boost target area, GTVb) refers to the area within the tumor target area where the radiation dose is increased; and the tissue or organ at risk (OAR) refers to the human tissue or organ that needs to be protected during radiotherapy.

[0004] According to the theory of radiobiology, the higher the BED of the tumor area, the higher the local control rate and the better the tumor radiotherapy effect. Stereotactic Ablation Radiotherapy (SABR) is a new radiotherapy method that can accurately deliver high-dose radiation to the tumor target area, significantly improving the BED of the tumor area. However, due to the dose limitations of adjacent tissues and organs, it is generally believed that SABR is only suitable for small tumors. The reason is that as the tumor size increases, the tumor is often closely related to the surrounding normal tissues, and the toxic reaction of SABR to the tumor as a whole is large, and the risk of radiation side effects such as radiation pneumonitis and radiation enteritis in patients is greatly increased.

[0005] To solve the problems of SABR, partial stereotactic ablation boost radiotherapy (Partial SABR, P-SABR) has emerged. P-SABR can maximize the BED of the tumor without increasing the dose to OAR, and achieve the purpose of improving the local control rate of the tumor with equal toxicity (that is, when the patient has the same radiotherapy side effects). Specifically, P-SABR is a modified treatment plan of conventional fractionated radiotherapy (CFRT) combined with SABR. The implementation plan is: the first course is planned to use SABR to give the maximum local boost in the tumor, that is, to specify one or more areas in the tumor area to give a higher radiation dose than the surrounding area, such as Figure 1a As shown in the figure, the larger enclosed area in the figure is the whole tumor area, and a smaller enclosed area in the tumor area is the designated area, that is, the boost target area; the second course uses CFRT to supplement the prescribed dose of the whole tumor area, such as Figure 1b As shown in the figure, the larger enclosed area is the whole tumor area. In clinical practice, P-SABR has been applied to various pathological types of cancers, such as squamous cell carcinoma, adenocarcinoma, fibromyxoid sarcoma, transitional cell carcinoma, malignant mesothelioma, etc. In the treatment of tumors in multiple parts of the body, such as ureteral tumors, lung cancer, thymic cancer, abdominal wall tumors, prostate cancer, etc., significant therapeutic effects have been achieved, the local control rate has been significantly improved, and the incidence of adverse reactions is low.

[0006] However, in actual application, the P-SABR plan contains iterative steps based on dose (including physical dose and / or biological effective dose) judgment, and the workflow is quite complicated, such as Figure 2 As shown in the figure, it mainly includes: import and registration of patient positioning CT and other imaging data, delineation of GTVb and other structures, plan design, dose calculation, plan evaluation, GTVb adjustment based on evaluation results and plan output. In the P-SABR plan, the range of GTVb is crucial to improving the BED of the tumor area. In clinical practice, it mainly relies on senior radiotherapists to manually delineate the tumor, especially the giant tumor and its adjacent OAR after empirical evaluation. Then, the radiotherapist needs to hand over the structure set containing GTVb and other organs to the radiotherapy physicist for calculation to obtain the dose index of the tumor and OAR. According to the dose index of the tumor and OAR, the clinician evaluates and modifies (for example, expands or reduces) the range of GTVb until the dose index of the tumor and OAR meets the clinical requirements, that is, the dose index of the tumor reaches the prescribed dose, and the dose index of the OAR is lower than the clinical dose index limit. In order to maximize the BED in the tumor without increasing the radiotherapy toxicity of normal tissues, this process often needs to be repeated several times or even more than a dozen times, which takes 8-20 working days.

[0007] Clearly, the P-SABR planning process has the following problems:

[0008] 1. The range of the push target area cannot be quantified.

[0009] In actual operation, the target area is manually delineated by clinicians based on their experience. If the target area is too large, the dose to the surrounding normal organs will increase, resulting in adverse reactions caused by radiotherapy; if the target area is too small, the BED of the tumor area will not be significantly improved, and the local control of the tumor will be poor. In clinical practice, it is often necessary to repeatedly iteratively calculate and modify the target area range to find the limit range of the increased dose target area.

[0010] 2. Inconsistent quality.

[0011] The planning stage involves a lot of subjective judgment by physicists and manual operations by clinicians, which leads to large differences in plan quality. Therefore, P-SABR is not easy to promote and use.

[0012] 3. The planning cost is high.

[0013] The delineation of the target volume and the adjustment of the range require doctors and physicists to spend a lot of time, constantly balancing between improving BED and protecting normal organs. At the same time, the high-intensity and high-demand planning and design work will cause fatigue and anxiety in clinical workers in actual work, affecting the progress of clinical radiotherapy. In other words, the time cost and economic cost are high. Summary of the invention

[0014] The embodiments of the present invention provide a tumor radiotherapy control system and a storage medium to solve the above problems.

[0015] The embodiment of the present invention provides a tumor radiotherapy control system, the system comprising: a dose prediction module, used to input the patient's image and a structure set containing a boost target area, a tumor target area and an area where an organ at risk is located in the image into a dose prediction model, so as to use the dose prediction model to predict the dose index of the tumor and the organ at risk; a judgment module, used to judge whether the predicted dose index of the tumor and the organ at risk meets the plan requirements; an optimization and adjustment module, used to automatically adjust the boost target area within the range of the tumor target area when it is judged that the predicted dose index of the tumor and the organ at risk does not meet the plan requirements, so as to obtain an adjusted boost target area; wherein, after the optimization and adjustment module obtains the adjusted boost target area, the prediction processing of the dose prediction module and the judgment processing of the judgment module are repeated until the dose index of the tumor and the organ at risk predicted by the dose prediction module meets the plan requirements.

[0016] Preferably, the dose prediction module is also used to train the dose prediction model using historical data to obtain a trained dose prediction model, so as to perform prediction processing using the trained dose prediction model; wherein the historical data includes: imaging images of historical patients, a structure set containing the boost target area, tumor target area and the area where organs at risk are located in the imaging images, and dose indicators of tumors and organs at risk that meet the planning requirements.

[0017] Preferably, the boost target area during the first prediction processing is the tumor target area, and the optimization and adjustment module is specifically used to divide the boost target area with the geometric center of the boost target area as the centroid and at intervals of a first preset angle to obtain multiple first dividing lines; taking the position point where each first dividing line intersects with the edge of the boost target area as the starting point, moving a first preset distance toward the centroid along the first dividing line where the starting point is located to obtain a new position point; and determining the closed area obtained by connecting adjacent new position points in pairs as the adjusted boost target area, so as to repeat the prediction processing of the dose prediction module and the judgment processing of the judgment module, until the dose indicators of the tumor and the endangered organs predicted by the dose prediction module meet the plan requirements, and the corresponding boost target area is determined as the boost target area for radiotherapy of the patient.

[0018] Preferably, the optimization and adjustment module is specifically used to divide the boost target area with the geometric center of the boost target area as the centroid and the second preset angle as the interval to obtain multiple second dividing lines when the boost target area is within the range of the tumor target area and the dose indicators of the tumor and the organs at risk predicted based on the boost target area meet the plan requirements; take the position point where each second dividing line intersects with the edge of the boost target area as the starting point, move along the second dividing line where the starting point is located to the edge of the tumor target area by a second preset distance to obtain a new position point; determine the closed area obtained by connecting adjacent new position points in pairs as the adjusted boost target area, so as to repeat the prediction processing of the dose prediction module and the judgment processing of the judgment module, until the judgment module determines that the dose indicators of the tumor and the organs at risk predicted by the dose prediction module do not meet the plan requirements, and then determine the boost target area that can meet the plan requirements before adjustment as the boost target area for radiotherapy of the patient.

[0019] Preferably, the system further comprises: a transmission module for transmitting a structure set containing corresponding boost target areas, tumor target areas and areas where organs at risk are located in the image to a radiotherapy planning system when the dose indicators of the tumor and organs at risk meet the planning requirements; or, when the dose indicators of the tumor and organs at risk predicted based on the boost target area before adjustment do not meet the planning requirements and the boost target area after adjustment is not completely within the range of the tumor target area, transmitting a structure set containing the boost target area before adjustment, tumor target area and areas where organs at risk are located in the image to a radiotherapy planning system.

[0020] Preferably, the volume of the tumor is greater than or equal to 5 cubic centimeters.

[0021] An embodiment of the present invention provides a storage medium, on which a program is stored. When the program is executed by a processor, the following steps are implemented: inputting an image of a patient and a structure set containing a boost target area, a tumor target area, and an area where an organ at risk is located in the image into a dose prediction model, so as to use the dose prediction model to predict the dose indicators of the tumor and the organ at risk; judging whether the predicted dose indicators of the tumor and the organ at risk meet the plan requirements; when judging that the predicted dose indicators of the tumor and the organ at risk do not meet the plan requirements, automatically adjusting the boost target area within the range of the tumor target area to obtain an adjusted boost target area, and after obtaining the adjusted boost target area, repeating the prediction process and the judgment process until the predicted dose indicators of the tumor and the organ at risk meet the plan requirements.

[0022] Preferably, the boost target area during the first prediction processing is the tumor target area, and the program stored on the storage medium specifically implements the following steps when executed by the processor: taking the geometric center of the boost target area as the centroid and taking the first preset angle as the interval, the boost target area is divided to obtain multiple first dividing lines; taking the position point where each first dividing line intersects with the edge of the boost target area as the starting point, moving a first preset distance along the first dividing line where the starting point is located toward the centroid to obtain a new position point; determining the closed area obtained by connecting adjacent new position points in pairs as the adjusted boost target area; repeating the prediction processing and the judgment processing until the corresponding boost target area is determined as the boost target area for radiotherapy of the patient when the predicted dose indicators of the tumor and the organs at risk meet the plan requirements.

[0023] Preferably, when the program stored on the storage medium is executed by the processor, the following steps are specifically implemented: when the boost target area is within the range of the tumor target area and the dose indicators of the tumor and organs at risk predicted based on the boost target area meet the plan requirements, the boost target area is divided with the geometric center of the boost target area as the centroid and with a second preset angle as the interval to obtain a plurality of second dividing lines; with the position point where each second dividing line intersects with the edge of the boost target area as the starting point, the second dividing line where the starting point is located is moved toward the edge of the tumor target area by a second preset distance to obtain a new position point; the closed area obtained by connecting adjacent new position points in pairs is determined as the adjusted boost target area; and the prediction process and the judgment process are repeated until the predicted dose indicators of the tumor and organs at risk do not meet the plan requirements, and the boost target area that can meet the plan requirements before adjustment is determined as the boost target area for radiotherapy of the patient.

[0024] Preferably, when the program stored on the storage medium is executed by the processor, the following steps are also implemented: when the dose indicators of the tumor and the organs at risk meet the planning requirements, the structure set containing the corresponding boost target area, the tumor target area and the area where the organs at risk are located in the image is transmitted to the radiotherapy planning system; or, when the dose indicators of the tumor and the organs at risk predicted based on the boost target area before adjustment do not meet the planning requirements and the boost target area after adjustment is not completely within the range of the tumor target area, the structure set containing the boost target area before adjustment, the tumor target area and the area where the organs at risk are located in the image is transmitted to the radiotherapy planning system.

[0025] A tumor radiotherapy control system and storage medium provided by an embodiment of the present invention use a dose prediction model to predict the dose indicators of tumors and organs at risk, and automatically adjust the boost target area when it is judged that the predicted dose indicators of tumors and organs at risk do not meet the plan requirements. Through continuous prediction, judgment, and automatic adjustment processes, a boost target area with a suitable size and position can be automatically determined quickly and accurately, so that the dose indicators of tumors and organs at risk meet the plan requirements, shortening the planning process time, improving the consistency of planning quality, and reducing planning costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1a This is a real case diagram of applying the SABR plan;

[0027] Figure 1b This is a real case diagram of applying the CFRT plan;

[0028] Figure 2 It is a flowchart of clinical P-SABR program design;

[0029] Figure 3is a schematic structural block diagram of a tumor radiotherapy control system provided by an embodiment of the present invention;

[0030] Figure 4 is a dose prediction flow chart provided by an embodiment of the present invention;

[0031] Figure 5 is an optimization adjustment flow chart provided by an embodiment of the present invention;

[0032] Figure 6 It is a moving pattern diagram of the coordinate points of the target area provided by an embodiment of the present invention;

[0033] Figure 7 This is a flow chart of tumor radiotherapy control provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0035] In the subsequent description, the suffixes such as "module", "component" or "unit" used to represent elements are only used to facilitate the description of the present invention, and have no special meanings. Therefore, "module", "component" or "unit" can be used in a mixed manner.

[0036] Embodiment 1

[0037] Figure 3 is a schematic structural block diagram of a tumor radiotherapy control system provided by an embodiment of the present invention, such as Figure 3 As shown, the system includes a dose prediction module, a judgment module and an optimization and adjustment module.

[0038] 1. The dose prediction module

[0039] The dose prediction module is used to input the patient's image and the structure set containing the boost target area, tumor target area and the area where the organs at risk are located in the image into the dose prediction model, so as to use the dose prediction model to predict the dose indicators of the tumor and the organs at risk.

[0040] In addition, the dose prediction module is also used to use historical data to train the dose prediction model to obtain a trained dose prediction model, so as to use the trained dose prediction model to perform prediction processing; wherein the historical data includes: imaging images of historical patients, a structure set containing the boost target area, the tumor target area and the area where the organs at risk are located in the imaging images, and dose indicators of tumors and organs at risk that meet the plan requirements.

[0041] The dose prediction model may adopt a deep learning model, such as a convolutional neural network, a recurrent neural network, a long short-term memory network, U-Net, etc.

[0042] The imaging images include but are not limited to CT imaging images and magnetic resonance imaging images.

[0043] The structure set containing the region where the boost target region, the tumor target region and the organ at risk are located in the image is obtained after outlining the region where the boost target region, the tumor target region and the organ at risk are located in the image. The image can be outlined manually or automatically, or a combination of manual and automatic outlines can be used when necessary.

[0044] 2. The Judgment Module

[0045] The judgment module is used to judge whether the predicted dose index of the tumor and the organ at risk meets the plan requirements. Specifically, the judgment module judges whether the predicted dose index of the tumor reaches the prescribed dose and whether the predicted dose index of the organ at risk is lower than the clinical dose limit.

[0046] 3. The Optimization and Adjustment Module

[0047] The optimization and adjustment module is used to automatically adjust the boost target area within the range of the tumor target area to obtain an adjusted boost target area when it is determined that the predicted dose index of the tumor and the organ at risk does not meet the plan requirements. After the optimization and adjustment module obtains the adjusted boost target area each time, the prediction processing of the dose prediction module and the judgment processing of the judgment module are repeated until the dose index of the tumor and the organ at risk predicted by the dose prediction module meets the plan requirements.

[0048] In a first embodiment, the boost target area during the first prediction processing is the tumor target area, and the optimization and adjustment module finds a suitable boost target area by continuously narrowing the range of the boost target area. Specifically, the optimization and adjustment module divides the boost target area with the geometric center of the boost target area as the centroid and the first preset angle as the interval to obtain multiple first dividing lines; takes the position point where each first dividing line intersects with the edge of the boost target area as the starting point, moves along the first dividing line where the starting point is located toward the centroid by a first preset distance to obtain a new position point; and determines the closed area obtained by connecting adjacent new position points in pairs as the adjusted boost target area, so as to repeat the prediction processing of the dose prediction module and the judgment processing of the judgment module, until the dose indicators of the tumor and the endangered organ predicted by the dose prediction module meet the plan requirements, and the corresponding boost target area is determined as the boost target area for radiotherapy of the patient.

[0049] The first preset angle may be any angle between 0 and 10, such as 5 degrees. Each time the thrust target area is adjusted, the first preset angle may be the same, for example, the first preset angle is 5 degrees, but the position of the first dividing line for any two adjustments is random. Of course, the first preset angle may also be different, for example, the first preset angle may be 5 degrees for the first adjustment, and the first preset angle may be 4 degrees for the second adjustment. The embodiment of the present invention does not limit this.

[0050] The first preset distance may be any distance between 0 and 10 mm, for example, 5 mm. Each time the thrust target area is adjusted, the first preset distance may be the same, for example, the first preset distance is 5 mm, or the first preset distance may be different, for example, the first preset distance may be 5 mm for the first adjustment and 4 mm for the second adjustment. The embodiment of the present invention does not limit this.

[0051] Generally, the smaller the first preset angle and the smaller the first preset distance, the better the thrust target area found. In specific implementation, the specific values ​​of the first preset angle and the first preset distance can be set according to the situation.

[0052] In addition, all first dividing lines can be located in the same plane, for example, dividing a horizontal plane at a first preset angle, and all first dividing lines obtained are located in the horizontal plane. The first dividing lines can also be located in a three-dimensional space, for example, starting from the centroid and dividing the three-dimensional space at intervals of a first preset angle.

[0053] In a second embodiment, the boost target area during the first prediction processing is within the range of the tumor target area, and when the boost target area is within the range of the tumor target area and the dose indicators of the tumor and organs at risk predicted based on the boost target area meet the plan requirements, the optimization and adjustment module divides the boost target area with the geometric center of the boost target area as the centroid and the second preset angle as the interval to obtain a plurality of second dividing lines; takes the position point where each second dividing line intersects with the edge of the boost target area as the starting point, moves along the second dividing line where the starting point is located to the edge of the tumor target area by a second preset distance to obtain a new position point; and determines the closed area obtained by connecting adjacent new position points in pairs as the adjusted boost target area, so as to repeat the prediction processing of the dose prediction module and the judgment processing of the judgment module, until the judgment module determines that the dose indicators of the tumor and organs at risk predicted by the dose prediction module do not meet the plan requirements, and determines the boost target area that can meet the plan requirements before adjustment as the boost target area for radiotherapy of the patient.

[0054] The second preset angle may be any angle between 0 and 10, such as 5 degrees. Each time the thrust target area is adjusted, the second preset angle may be the same, for example, the second preset angle is 5 degrees, but the position of the first dividing line for any two adjustments is random. Of course, the second preset angle may also be different, for example, the second preset angle may be 5 degrees for the first adjustment, and the second preset angle may be 4 degrees for the second adjustment. The embodiment of the present invention does not limit this.

[0055] The second preset distance may be any distance between 0 and 10 mm, for example, 5 mm. Each time the thrust target area is adjusted, the second preset distance may be the same, for example, the second preset distance is 5 mm. The second preset distance may also be different, for example, the second preset distance may be 5 mm for the first adjustment and 4 mm for the second adjustment. The embodiment of the present invention does not limit this.

[0056] Generally, the smaller the second preset angle and the smaller the second preset distance, the better the thrust target area found. In specific implementation, the specific values ​​of the second preset angle and the second preset distance can be set according to the situation.

[0057] In addition, all second dividing lines can be located in the same plane, for example, dividing the horizontal plane at the second preset angle, and all the second dividing lines obtained are located in the horizontal plane. The second dividing lines can also be located in the three-dimensional space, for example, starting from the centroid and dividing the three-dimensional space at intervals of the second preset angle.

[0058] In a third embodiment, the above two embodiments can be combined. Specifically, the first embodiment is first executed, for example, the tumor target area is used as the boost target area, and the range of the boost target area is continuously reduced based on the first preset angle and the first preset distance, until the dose index of the tumor and organs at risk predicted by the dose prediction module based on the reduced boost target area meets the plan requirements, and the boost target area used in this prediction is recorded. For the convenience of explanation, it is recorded as the first boost target area. Then, the second embodiment is executed, for example, based on the second preset angle and the second preset distance, the range is continuously expanded on the basis of the first boost target area, until the dose index of the tumor and organs at risk predicted by the dose prediction module based on the expanded boost target area does not meet the plan requirements, which means that the boost target area used in this prediction is not suitable, and the boost target area used in the previous prediction should be used.

[0059] In the third embodiment, the first preset angle is greater than the second preset angle, and the first preset distance is greater than the second preset distance.

[0060] The system further includes a transmission module (not shown in the figure), which is used to communicate with the radiotherapy planning system. Specifically, when the dose indicators of the tumor and the organs at risk meet the planning requirements, the structure set containing the corresponding boost target area, the tumor target area and the area where the organs at risk are located in the image is transmitted to the radiotherapy planning system; or, when the dose indicators of the tumor and the organs at risk predicted based on the boost target area before adjustment do not meet the planning requirements and the boost target area after adjustment is not completely within the range of the tumor target area, the structure set containing the boost target area before adjustment, the tumor target area and the area where the organs at risk are located in the image is transmitted to the radiotherapy planning system.

[0061] The system provided by the embodiment of the present invention performs prediction and judgment processing based on the new structure set of the updated boost target area, and can quickly find the location of the appropriate boost target area. It is applicable to tumors of all volumes, especially tumors with a volume greater than or equal to 5 cubic centimeters.

[0062] A tumor radiotherapy control system provided by an embodiment of the present invention uses a dose prediction model to predict the dose indicators of tumors and organs at risk, and automatically adjusts the boost target area when it is judged that the predicted dose indicators of tumors and organs at risk do not meet the plan requirements. Through continuous prediction, judgment, and automatic adjustment processes, it is possible to quickly and accurately automatically determine a boost target area with a suitable size and position, so that the dose indicators of tumors and organs at risk meet the plan requirements, shorten the planning process time, improve the consistency of planning quality, and reduce planning costs.

[0063] Example 2

[0064] In order to overcome the defects of the target area range that cannot be quantified, the quality is inconsistent, and the time is consuming when designing the P-SABR plan, the embodiment of the present invention applies deep learning to the design of clinical radiotherapy plans, and uses artificial intelligence to assist in the design of the push target area. In the previously designed P-SABR plan, the outline range of the P-SABR push target area and the dose data established with this target area and the set dose are saved. The target area and dose of the previous P-SABR plan are deeply learned to obtain a dose prediction model that can be used for radiotherapy dose prediction. Then, with the help of the powerful computing and self-learning capabilities of the computer, the optimal target area range that meets the dose limit is found, so as to achieve the purpose of automatically generating the optimized target area, greatly increase the equivalent biological dose of the tumor, and at the same time do not increase the risk of normal organs and tissues around the tumor, so as to provide a solution for the clinical safety and efficient formulation of radiotherapy plans. In addition, it also solves the limitations of equipment and experience between hospitals and doctors, and provides the possibility for further promotion of this radiotherapy method.

[0065] The following Figures 4 to 6 , the present invention is described in detail.

[0066] Figure 4 is a dose prediction flow chart provided by an embodiment of the present invention, such as Figure 4 As shown, the details are as follows:

[0067] Step S401: Input patient imaging images and dose limit requirements.

[0068] The user inputs patient imaging data, including but not limited to CT images and MRI images, and manually defines the radiation exposure area and dose limits for organs at risk based on the patient's clinical information and treatment guidelines.

[0069] Step S402: manually and / or automatically outlining the patient's initial boost target volume, tumor target volume, and organs at risk based on the image.

[0070] Automatic delineation method: Automatic segmentation and delineation of planned tumor target volume and organs at risk based on machine learning, generating the initial boost target volume based on the planned tumor target volume range;

[0071] Manual outlining method: Based on the experience of imaging and radiotherapy doctors, the initial boost target area (a designated area within the tumor area is given a higher radiation dose than the surrounding area), the planned tumor target area (referred to as the tumor target area, that is, the area used for radiotherapy planning after the solid tumor visible in the image is expanded to a certain range) and the structure of the organ at risk are manually outlined.

[0072] Step S403: Obtain relevant data of previous P-SABR plans, including image and structure set data, and dose data, to construct a dose prediction model based on deep learning (i.e., a physical dose or biological dose model).

[0073] Step S404: The patient data set (i.e., image and structure data set) is input into the dose prediction model to automatically predict and generate the dose distribution (i.e., physical dose or biological dose distribution) and dose volume histogram (DVH) of the tumor and organs at risk.

[0074] The dose prediction module imports historical P-SABR plans containing images, structure sets (files containing the target area and the area where the organs at risk are located in the image), plans and other data. The historical P-SABR plans are used to train deep learning models such as convolutional neural networks, recurrent neural networks, long short-term memory networks, U-Net, etc., to learn and establish the correlation between the anatomical position and dose (physical dose or biological dose) between the target area and the organs at risk, and the trained deep learning model is used as a dose prediction model to predict the dose indicators of tumors and organs at risk. Specifically, the organ dose indicator (i.e., the volume percentage of different doses on the target area or OAR), the global dose distribution (i.e., the dose values ​​corresponding to different spatial positions on the image), and the DVH map (i.e., a display of the percentage of different doses on the target area or OAR) can be predicted.

[0075] For a newly input patient data set, the dose prediction module automatically extracts and organizes the data and then performs dose prediction (i.e., physical dose or biological dose prediction) to obtain the patient's organ dose index, global dose distribution, and DVH map.

[0076] Figure 5 is an optimization adjustment flow chart provided by an embodiment of the present invention, such as Figure 5 As shown, the details are as follows:

[0077] Step S501: Acquire the patient's boost target volume, tumor target volume and surrounding organ structures.

[0078] If this is the first dose prediction for this patient, the initial boost target volume can be the tumor target volume.

[0079] Step S502: The dose prediction module inputs the patient's imaging image and the structure set containing the boost target area, the tumor target area and the surrounding organ structures in the imaging image into the dose prediction model, so as to use the dose prediction model to predict the dose index of the tumor and the organs at risk.

[0080] Step S503: Based on the dose limit in the aforementioned step S401 and the prediction result obtained in step S502, determine whether there is a dose index of an organ at risk that exceeds the dose limit. If so, execute steps S504 to S508 to reversely optimize and adjust the tumor push target area. Otherwise, execute step S509.

[0081] According to the prediction result of step S502, the target dose and the dose limit of the organ at risk are automatically evaluated for the P-SABR plan. If, in the prediction result, there is a dose index of the organ at risk that exceeds the dose limit in the aforementioned step S401, for example, one or more or all dose indexes of the organ at risk are greater than the dose limit, the plan data set is transferred to the optimization and adjustment module, otherwise, step S509 is executed.

[0082] Step S504: The optimization and adjustment module performs three-dimensional reconstruction of the tissue structure on the image, taking the geometric center of the initial thrust target area as the center of mass or the center of the circle, and divides the horizontal plane into 360° clockwise intervals of n°, where n is a natural number divisible by 360.

[0083] Step S505: Taking the centroid as the starting point, reverse projection is performed along each dividing line toward the target area, and the intersection point with the boundary of the target area is recorded as m i , where i=1, 2,…, (360° / n°).

[0084] Step S506: Calculate each intersection point m i The straight-line distance from the centroid l mi , take the distance change interval as kmm, k can be any value within 0-10mm; m i Move along the dividing line at equal intervals to obtain u i .

[0085] Step S507: Determine u i Is it within the planned tumor target area? If yes, then set u i Update the target volume, and transmit the image and the updated structure set to the dose prediction module to execute the dose prediction step of step S503, and repeat the dose prediction and target volume update. Otherwise, execute step S509.

[0086] In this embodiment, when the optimization module calculates the range of the boosted tumor target area, it defaults to using the original tumor target area as the boost area for the dose assessment of the organ at risk. If it meets the requirements, step S509 is directly executed. If it does not meet the requirements, the boosted target area range is adjusted. Figure 6 As shown in the figure, when adjusting, take the geometric center 0 of the existing thrust target area as the center, and evenly set connecting lines in a 360-degree clockwise direction with an angle interval of 5 degrees. The connecting line is a ray with the geometric center of the target area as the endpoint. When a connecting line intersects with the boundary of the existing target area, the intersection point is set as m i , i = 1, 2, ..., 72. Then the intersection point m i Move 5mm toward the geometric center 0 in the direction of the connecting line, and set the point at the moved position to u i , at this time u i The distance from the geometric center 0 is denoted as lui . Then l mi and m i The set of ui and u i , and update the push target area range based on this. Import the updated push target area structure set into the dose prediction model to perform dose prediction, that is, execute step S503. If the prediction result does not meet the dose limit input in the above step S401, repeat steps S504 to S508, otherwise execute step S509.

[0087] Step S509: Based on the updated boost target data, an executable radiotherapy plan is automatically generated. In specific implementation, the structure set file containing the final boost target is transmitted back to the radiotherapy planning system, and the radiotherapy planning system calculates and generates an executable radiotherapy plan. Then, the radiotherapy planning system outputs the radiotherapy plan and related reports, including the intensity modulated radiotherapy plan (the dose distribution map of the irradiation field is determined by inverse design) and the stereotactic (simultaneous use of coplanar and non-coplanar irradiation fields) radiotherapy plan.

[0088] In actual operation, first import the patient's Dicom format image data into the system, input the number of irradiation times and total dose, and select the generation mode of the radiotherapy target area and the corresponding organs at risk. For example, if automatic generation is selected, the system automatically determines the range of the target area and organs at risk on the image based on the input image. If manual delineation is selected, the structure set data file corresponding to the target area and organs at risk that match the image data needs to be manually imported. Secondly, manually input the dose and volume requirements that the organs at risk need to meet, select the required dose distribution, and the dose definition in the DVH diagram (i.e., biological dose or physical dose). If physical dose is selected, check the physical dose unit to be displayed. If biological dose is selected, enter the biological model parameter values ​​corresponding to different organs at risk. Then, the system determines the maximum range of the push target area based on the input parameters-organ at risk dose and volume limit, image data, total irradiation dose and other parameters, and completes the generation and writing of the corresponding structure, and outputs it as a structure set file. The system outputs the DVH diagram corresponding to the structure set data, showing the predicted global dose distribution diagram.

[0089] Embodiment 3

[0090] An embodiment of the present invention provides a storage medium storing one or more programs for tumor radiotherapy control, wherein the one or more programs can be executed by one or more processors to implement Figure 7 The following specific steps are shown:

[0091] Step S701: Inputting the patient's imaging image and the structure set containing the boost target area, the tumor target area and the area where the organs at risk are located in the imaging image into a dose prediction model, so as to use the dose prediction model to predict the dose indicators of the tumor and the organs at risk.

[0092] The imaging images include but are not limited to CT imaging images and magnetic resonance imaging images.

[0093] The structure set containing the region where the boost target region, the tumor target region and the organ at risk are located in the image is obtained after outlining the region where the boost target region, the tumor target region and the organ at risk are located in the image. The image can be outlined manually or automatically, or a combination of manual and automatic outlines can be used when necessary.

[0094] Step S702: Determine whether the predicted dose indicators of the tumor and organs at risk meet the planning requirements.

[0095] Specifically, it is determined whether the predicted dose index of the tumor reaches the prescribed dose and whether the predicted dose index of the organ at risk is lower than the clinical dose limit.

[0096] Step S703: When it is determined that the predicted dose indicators of the tumor and organs at risk do not meet the planned requirements, the boost target volume is automatically adjusted within the range of the tumor target volume to obtain an adjusted boost target volume.

[0097] Step S704: After the adjusted boost target volume is obtained, the prediction process and the judgment process are repeated until the predicted dose indicators of the tumor and organs at risk meet the planning requirements.

[0098] In a first embodiment, the boost target area during the first prediction processing is the tumor target area, and steps S703 and S704 implemented when the program stored on the storage medium is executed by the processor can be specifically as follows: taking the geometric center of the boost target area as the centroid and taking the first preset angle as the interval, the boost target area is divided to obtain multiple first dividing lines; taking the position point where each first dividing line intersects with the edge of the boost target area as the starting point, moving along the first dividing line where the starting point is located toward the centroid by a first preset distance to obtain a new position point; determining the closed area obtained by connecting adjacent new position points in pairs as the adjusted boost target area; repeating the prediction processing and the judgment processing until the corresponding boost target area is determined as the boost target area for radiotherapy of the patient when the predicted dose indicators of the tumor and the organs at risk meet the plan requirements.

[0099] The first preset angle may be any angle between 0 and 10, such as 5 degrees. Each time the thrust target area is adjusted, the first preset angle may be the same, for example, the first preset angle is 5 degrees, but the position of the first dividing line for any two adjustments is random. Of course, the first preset angle may also be different, for example, the first preset angle may be 5 degrees for the first adjustment, and the first preset angle may be 4 degrees for the second adjustment. The embodiment of the present invention does not limit this.

[0100] The first preset distance may be any distance between 0 and 10 mm, for example, 5 mm. Each time the thrust target area is adjusted, the first preset distance may be the same, for example, the first preset distance is 5 mm, or the first preset distance may be different, for example, the first preset distance may be 5 mm for the first adjustment and 4 mm for the second adjustment. The embodiment of the present invention does not limit this.

[0101] Generally, the smaller the first preset angle and the smaller the first preset distance, the better the thrust target area found. In specific implementation, the specific values ​​of the first preset angle and the first preset distance can be set according to the situation.

[0102] In addition, all first dividing lines can be located in the same plane, for example, dividing a horizontal plane at a first preset angle, and all first dividing lines obtained are located in the horizontal plane. The first dividing lines can also be located in a three-dimensional space, for example, starting from the centroid and dividing the three-dimensional space at intervals of a first preset angle.

[0103] In a second embodiment, the boost target area during the first prediction processing is within the range of the tumor target area, and steps S703 and S704 implemented when the program stored on the storage medium is executed by the processor can be specifically as follows: when the boost target area is within the range of the tumor target area and the dose indicators of the tumor and organs at risk predicted based on the boost target area meet the plan requirements, the boost target area is divided with the geometric center of the boost target area as the centroid and with the second preset angle as the interval to obtain a plurality of second dividing lines; with the position point where each second dividing line intersects with the edge of the boost target area as the starting point, the second dividing line where the starting point is located is moved toward the edge of the tumor target area by a second preset distance to obtain a new position point; a closed area obtained by connecting adjacent new position points in pairs is determined as the adjusted boost target area; and the prediction processing and the judgment processing are repeated until the predicted dose indicators of the tumor and organs at risk do not meet the plan requirements, and the boost target area that can meet the plan requirements before adjustment is determined as the boost target area for radiotherapy of the patient.

[0104] The second preset angle may be any angle between 0 and 10, such as 5 degrees. Each time the thrust target area is adjusted, the second preset angle may be the same, for example, the second preset angle is 5 degrees, but the position of the first dividing line for any two adjustments is random. Of course, the second preset angle may also be different, for example, the second preset angle may be 5 degrees for the first adjustment, and the second preset angle may be 4 degrees for the second adjustment. The embodiment of the present invention does not limit this.

[0105] The second preset distance may be any distance between 0 and 10 mm, for example, 5 mm. Each time the thrust target area is adjusted, the second preset distance may be the same, for example, the second preset distance is 5 mm. The second preset distance may also be different, for example, the second preset distance may be 5 mm for the first adjustment and 4 mm for the second adjustment. The embodiment of the present invention does not limit this.

[0106] Generally, the smaller the second preset angle and the smaller the second preset distance, the better the thrust target area found. In specific implementation, the specific values ​​of the second preset angle and the second preset distance can be set according to the situation.

[0107] In addition, all second dividing lines can be located in the same plane, for example, dividing the horizontal plane at the second preset angle, and all the second dividing lines obtained are located in the horizontal plane. The second dividing lines can also be located in the three-dimensional space, for example, starting from the centroid and dividing the three-dimensional space at intervals of the second preset angle.

[0108] In a third embodiment, the two aforementioned embodiments may be combined. Specifically, when the program stored on the storage medium is executed by the processor, step S703 and step S704 implemented may be specifically as follows: first, the first embodiment is performed, for example, the tumor target area is used as the boost target area, and the range of the boost target area is continuously reduced based on the first preset angle and the first preset distance, until the dose index of the tumor and organs at risk predicted based on the reduced boost target area meets the plan requirements, and the boost target area used in this prediction is recorded, and for ease of description, it is recorded as the first boost target area. Then, the second embodiment is performed, for example, based on the second preset angle and the second preset distance, the range is continuously expanded on the basis of the first boost target area, until the dose index of the tumor and organs at risk predicted based on the expanded boost target area does not meet the plan requirements, and at this time, it is indicated that the boost target area used in this prediction is inappropriate, and the boost target area used in the previous prediction should be used.

[0109] In the third embodiment, the first preset angle is greater than the second preset angle, and the first preset distance is greater than the second preset distance.

[0110] Furthermore, when the program stored on the storage medium is executed by the processor, the following steps can also be implemented: using historical data to train the dose prediction model to obtain a trained dose prediction model, and using the trained dose prediction model to perform prediction processing. Wherein, the historical data includes: images of historical patients, a structure set containing the boost target area, the tumor target area and the area where the organs at risk are located in the image, and the dose indicators of the tumor and organs at risk that meet the plan requirements. The dose prediction model can use a deep learning model, such as a convolutional neural network, a recurrent neural network, a long short-term memory network, U-Net, etc.

[0111] Furthermore, when the program stored on the storage medium is executed by the processor, the following steps are also implemented: when the dose indicators of the tumor and the organs at risk meet the planning requirements, the structure set containing the corresponding boost target area, the tumor target area and the area where the organs at risk are located in the image is transmitted to the radiotherapy planning system; or, when the dose indicators of the tumor and the organs at risk predicted based on the boost target area before adjustment do not meet the planning requirements and the boost target area after adjustment is not completely within the range of the tumor target area, the structure set containing the boost target area before adjustment, the tumor target area and the area where the organs at risk are located in the image is transmitted to the radiotherapy planning system.

[0112] The storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by the computer.

[0113] The present invention can achieve the following technical effects:

[0114] 1. Assisting radiotherapists and physicists in determining the range of the target area for boosting, solving the limitation of manual delineation (estimated dose drop distance based on experience) that cannot maximize the tumor boosting target area. It is applicable to tumors of various sizes, especially, it can significantly improve the target area for tumors with a volume greater than 5 cm. 3 The radiotherapy dose for tumors can be reduced, thereby improving the tumor control rate, benefiting patients' survival, saving medical costs, and having obvious social benefits.

[0115] 2. It can improve the quality consistency of P-SABR plans. By taking advantage of the data volume and technology, it can fill the gap in the field of complex tumor target area delineation, further promote the intelligence of clinical work, and promote the promotion of boost radiotherapy.

[0116] 3. It can significantly shorten the time of P-SABR planning process.

[0117] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the present invention is not limited thereby. Any modification, equivalent substitution and improvement made by those skilled in the art without departing from the scope and essence of the present invention shall be within the scope of the present invention.

Claims

1. A target area optimization control system for tumor radiotherapy, characterized in that: The boost target area refers to an area within the tumor target area where the radiation dose is increased. The system comprises: A dose prediction module is used to predict the dose index of the tumor and the organ at risk under the current boost target area by inputting the patient's image and the structure set containing the boost target area, the tumor target area and the area where the organ at risk is located in the image into the dose prediction model; wherein the initial boost target area is generated according to the range of the tumor target area; A judgment module is used to judge whether the dose index of the tumor and the organ at risk under the predicted current boost target volume meets the plan requirements, wherein the plan requirements refer to that the dose index of the tumor reaches the prescribed dose and the dose index of the organ at risk is lower than the clinical dose limit; An optimization and adjustment module, for automatically adjusting the range of the boost target area within the range of the tumor target area to obtain a boost target area with an adjusted range when it is determined that the dose indicators of the tumor and organs at risk under the predicted current boost target area do not meet the plan requirements; The maximum range of the boost target area that can make the dose index of the tumor and the organ at risk meet the plan requirements is determined by repeating the prediction processing of the dose prediction module, the judgment processing of the judgment module and the automatic adjustment processing of the optimization adjustment module.

2. The system according to claim 1, characterized in that The dose prediction module is also used to train the dose prediction model using historical data to obtain a trained dose prediction model, so as to perform prediction processing using the trained dose prediction model; wherein the historical data includes: imaging images of historical patients, a structure set containing boost target areas, tumor target areas and areas where organs at risk are located in the imaging images, and dose indicators of tumors and organs at risk that meet planning requirements.

3. The system according to claim 1, characterized in that The push target area during the first prediction process is the tumor target area, and the optimization and adjustment module is specifically used to divide the push target area with the geometric center of the push target area as the centroid and the first preset angle as the interval to obtain a plurality of first dividing lines; Taking the position point where each first dividing line intersects with the edge of the boost target area as the starting point, move along the first dividing line where the starting point is located toward the centroid by a first preset distance to obtain a new position point; the closed area obtained by connecting adjacent new position points in pairs is determined as the adjusted boost target area, so as to repeat the prediction processing of the dose prediction module and the judgment processing of the judgment module, until the dose indicators of the tumor and the endangered organ predicted by the dose prediction module meet the plan requirements, the corresponding boost target area is determined as the boost target area for radiotherapy of the patient.

4. The system according to claim 1, characterized in that The optimization and adjustment module is specifically used to divide the boost target area with the geometric center of the boost target area as the centroid and the second preset angle as the interval to obtain a plurality of second dividing lines when the boost target area is within the range of the tumor target area and the dose indicators of the tumor and the organs at risk predicted based on the boost target area meet the planning requirements; Taking the position point where each second dividing line intersects with the edge of the boost target area as the starting point, move along the second dividing line where the starting point is located to the edge of the tumor target area by a second preset distance to obtain a new position point; the closed area obtained by connecting adjacent new position points in pairs is determined as the adjusted boost target area, so as to repeat the prediction processing of the dose prediction module and the judgment processing of the judgment module, until the judgment module determines that the dose indicators of the tumor and the endangered organ predicted by the dose prediction module do not meet the plan requirements, and the boost target area that can meet the plan requirements before adjustment is determined as the boost target area for radiotherapy of the patient.

5. The system according to any one of claims 1 to 4, characterized in that: The system further comprises: A transmission module is used to transmit a structure set containing the corresponding boost target area, tumor target area and the area where the organs at risk are located in the image to a radiotherapy planning system when the dose indicators of the tumor and the organs at risk meet the planning requirements; or, when the dose indicators of the tumor and the organs at risk predicted based on the boost target area before adjustment do not meet the planning requirements and the boost target area after adjustment is not completely within the range of the tumor target area, transmit the structure set containing the boost target area before adjustment, the tumor target area and the area where the organs at risk are located in the image to the radiotherapy planning system.

6. The system according to claim 1, characterized in that The volume of the tumor is greater than or equal to 5 cubic centimeters.

7. A storage medium, characterized in that: The storage medium stores a program, and when the program is executed by the processor, the following steps are performed to achieve the optimization control of the target area during tumor radiotherapy: The dose indexes of the tumor and the organs at risk under the current boost target area are predicted by inputting the patient's image and the structure set containing the boost target area, the tumor target area and the area where the organs at risk are located in the image into the dose prediction model; wherein the boost target area refers to the area within the tumor target area where the radiation dose is increased, and the initial boost target area is generated according to the range of the tumor target area; Determine whether the dose index of the tumor and organs at risk under the predicted current boost target volume meets the planned requirements, which means that the dose index of the tumor reaches the prescribed dose and the dose index of the organs at risk is lower than the clinical dose limit; When it is determined that the dose indicators of the tumor and organs at risk under the predicted current boost target volume do not meet the plan requirements, the range of the boost target volume is automatically adjusted within the range of the tumor target volume to obtain a boost target volume with an adjusted range; By repeating prediction processing, judgment processing and automatic adjustment processing, the maximum range of the boost target area that can make the dose indicators of tumors and organs at risk meet the plan requirements is determined.

8. The storage medium according to claim 7, characterized in that The target volume for the first prediction process is the tumor target volume. When the program stored in the storage medium is executed by the processor, the following steps are specifically implemented: Taking the geometric center of the thrust target area as the centroid and taking the first preset angle as the interval, the thrust target area is divided to obtain a plurality of first dividing lines; Taking the position point where each first dividing line intersects with the edge of the thrust target area as the starting point, moving along the first dividing line where the starting point is located toward the centroid by a first preset distance to obtain a new position point; The closed area obtained by connecting two adjacent new position points is determined as the adjusted push target area; The prediction process and the judgment process are repeated until the predicted dose indicators of the tumor and the organs at risk meet the plan requirements, and the corresponding boost target area is determined as the boost target area for radiotherapy of the patient.

9. The storage medium according to claim 7, characterized in that: When the program stored in the storage medium is executed by the processor, the following steps are specifically implemented: When the boost target area is within the range of the tumor target area and the dose indicators of the tumor and organs at risk predicted based on the boost target area meet the planning requirements, the boost target area is divided with the geometric center of the boost target area as the centroid and the second preset angle as the interval to obtain a plurality of second dividing lines; Taking the position point where each second dividing line intersects with the edge of the target area as the starting point, moving along the second dividing line where the starting point is located toward the edge of the tumor target area by a second preset distance to obtain a new position point; The closed area obtained by connecting two adjacent new position points is determined as the adjusted push target area; The prediction process and the judgment process are repeated until the predicted dose indexes of the tumor and organs at risk do not meet the plan requirements, and the boost target area that can meet the plan requirements before adjustment is determined as the boost target area for radiotherapy of the patient.

10. The storage medium according to any one of claims 7 to 9, characterized in that: When the program stored on the storage medium is executed by the processor, the following steps are also implemented: When the dose indexes of the tumor and the organ at risk meet the planning requirements, a structure set containing the corresponding boost target volume, the tumor target volume and the region where the organ at risk is located in the image is transmitted to the radiotherapy planning system; Alternatively, when the dose indicators of the tumor and the organs at risk predicted based on the boost target area before adjustment do not meet the planning requirements and the adjusted boost target area is not completely within the range of the tumor target area, the structure set containing the boost target area before adjustment, the tumor target area and the area where the organs at risk are located in the image is transmitted to the radiotherapy planning system.

Images