Automatic segmentation and intelligent plan generation method and device suitable for self-adaptive radiotherapy of prostate cancer

Abstract

The invention relates to the technical field of computers, and discloses an automatic segmentation and intelligent plan generation method and device suitable for self-adaptive radiotherapy of prostate cancer. Comprising the following steps: acquiring medical image data of a treated object, clinical dose data of a target area and an organ at risk in a focus area of the treated object, and clinical treatment data of a preset priority level; performing automatic segmentation based on the medical image data to obtain target segmentation data; performing dose distribution prediction according to the target segmentation data to obtain initial dose distribution data, and performing iterative optimization on the initial dose distribution data by using clinical treatment data to obtain target dose distribution data of the treated object; and performing iterative optimization of a radiotherapy plan based on the target dose distribution data, and generating a target radiotherapy plan of the treated object. Therefore, on the basis of realizing accurate, efficient and automatic segmentation of the prostate cancer target region and the endangered organ, multi-round iterative optimization of dose distribution and the radiotherapy plan is carried out, and finally a high-quality target radiotherapy plan is obtained.

Description

Technical Field

[0001] This application relates to the field of computer technology, and particularly to an automatic segmentation and intelligent plan generation method and device suitable for adaptive radiotherapy of prostate cancer. Background Art

[0002] Globally, prostate cancer is one of the most common malignant tumors in men, and its incidence rate is increasing year by year. Radiotherapy is the main treatment method for prostate cancer, which runs through the entire treatment process of prostate cancer. However, there are still limitations in the related technologies for automatically generating radiotherapy plans, and an intelligent plan generation method that can improve the quality of prostate cancer radiotherapy plans is needed. Summary of the Invention

[0003] This application aims to solve at least one of the technical problems in the related technologies to some extent. For this purpose, this application proposes an automatic segmentation and intelligent plan generation method and device suitable for adaptive radiotherapy of prostate cancer. The main technical solutions adopted in this application include:

[0004] In a first aspect, an embodiment of this application provides an automatic segmentation and intelligent plan generation method suitable for adaptive radiotherapy of prostate cancer. The method includes: obtaining medical image data and clinical treatment data of the treated object; wherein, the medical image data is obtained by performing imaging examinations on the pelvic region of the treated object; the clinical treatment data includes clinical dose data and preset priority levels of the target area and critical organs in the lesion area of the treated object; performing automatic segmentation based on the medical image data to obtain target segmentation data; wherein, the target segmentation data is data used to reflect the distribution of the target area and critical organs in the lesion area of the treated object; predicting the dose distribution according to the target segmentation data to obtain initial dose distribution data; using the clinical dose data and preset priority levels to iteratively optimize the initial dose distribution data to obtain the target dose distribution data of the treated object; based on the target dose distribution data, performing iterative optimization of the radiotherapy plan to generate the target radiotherapy plan of the treated object; wherein, the target radiotherapy plan is used to describe the target dosimetric data of the target area and critical organs in the lesion area.

[0005] Second aspect, an automatic segmentation and intelligent plan generation device suitable for adaptive radiotherapy of prostate cancer provided by an embodiment of the present application includes: a data acquisition module, configured to acquire medical image data and clinical treatment data of a subject to be treated; wherein, the medical image data is obtained by performing imaging examination on the pelvic region of the subject to be treated; the clinical treatment data includes clinical dose data and preset priority levels of a target area and organs at risk in the lesion area of the subject to be treated; a delineation and segmentation module, configured to perform automatic segmentation based on the medical image data to obtain target segmentation data; wherein, the target segmentation data is data used to reflect the distribution of the target area and organs at risk in the lesion area of the subject to be treated; a dose prediction module, configured to perform dose distribution prediction according to the target segmentation data to obtain initial dose distribution data; a dose optimization module, configured to iteratively optimize the initial dose distribution data by using the clinical dose data and the preset priority levels to obtain the target dose distribution data of the subject to be treated; a plan generation module, configured to perform iterative optimization of a radiotherapy plan based on the target dose distribution data to generate a target radiotherapy plan for the subject to be treated; wherein, the target radiotherapy plan is used to describe the target dosimetric data of the target area and organs at risk in the lesion area.

[0006] Third aspect, the present application further provides a computer device including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the method in any one of the above are implemented.

[0007] Fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method in any one of the above are implemented.

[0008] Fifth aspect, the present invention provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps of the method in any one of the above are implemented.

[0009] In the above embodiments, based on the medical image data of the subject to be treated, automatic segmentation is performed to obtain target segmentation data reflecting the distribution of the target area and organs at risk in the lesion area of the subject to be treated. Then, dose distribution prediction is performed based on this to obtain initial dose distribution data, and iterative optimization is performed in combination with clinical dose data and clinical treatment data to obtain target dose distribution data, and finally a target radiotherapy plan is generated. Thus, on the basis of realizing accurate and efficient automatic segmentation of the prostate cancer target area and organs at risk, multiple rounds of iterative optimization of dose distribution and radiotherapy plan are performed, and finally a high-quality target radiotherapy plan is obtained. Description of the Drawings

[0010] To more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0011] Figure 1 FIG. is a flowchart of an automatic segmentation and intelligent plan generation method suitable for adaptive radiotherapy for prostate cancer according to an embodiment of the present application;

[0012] Figure 2 FIG. is a flowchart of a method for determining target dose distribution data according to an embodiment of the present application;

[0013] Figure 3 FIG. is a flowchart of a method for determining intermediate dose distribution data according to an embodiment of the present application;

[0014] Figure 4a FIG. is a flowchart of a method for determining intermediate dose distribution data according to another embodiment of the present application;

[0015] Figure 4b FIG. is a schematic diagram of the dose fall outside the target area according to another embodiment of the present application;

[0016] Figure 5 FIG. is a flowchart of a method for determining a target radiotherapy plan according to an embodiment of the present application;

[0017] Figure 6 FIG. is a flowchart of a method for determining a target radiotherapy plan according to another embodiment of the present application;

[0018] Figure 7 FIG. is a schematic comparison diagram of the segmentation results according to another embodiment of the present application;

[0019] Figure 8 FIG. is a structural block diagram of an automatic segmentation and intelligent plan generation device suitable for adaptive radiotherapy for prostate cancer according to an embodiment of the present application;

[0020] Figure 9 FIG. is an internal structure diagram of a computer device according to an embodiment of the present application. Specific Embodiments

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions in the embodiments of this application in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, rather than all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of protection of this application.

[0022] Prostate cancer is a common malignant tumor in men. Relevant data shows that in the 2024 Global Cancer Statistics, prostate cancer accounts for 29% of newly diagnosed cancers in American men. Radiotherapy is the main treatment method for prostate cancer. It should be understood that in the field of radiotherapy, there are several important concepts. The gross tumor volume (GTV) refers to the tumor visible on imaging; the clinical target volume (CTV) is, on the basis of the GTV, plus some subclinical lesions and possible infiltration ranges; the planning target volume (PTV) is the irradiation range formed by considering various random errors and systematic errors during the radiotherapy process. The treatment planning system (TPS) is a system used to formulate radiotherapy plans for radiotherapy patients. Through the built-in accelerator model, the incident angle and optimization function of radiotherapy are designed within the system to ensure that the tumor reaches the prescribed dose while ensuring that the surrounding normal organs do not exceed the limiting dose. Normal organs other than tumors are called organs at risk (OAR). The dose volume histogram (DVH) is a method / tool used to evaluate radiotherapy plans, and it can obtain data such as the dose volume and high dose of the target volume prescription dose, as well as the dose volume and average dose of the organs at risk. Among them, the development of intensity-modulated radiotherapy (IMRT) technology can enable the tumor to receive high-dose irradiation while minimizing the irradiated dose of the surrounding normal organs as much as possible. The key to achieving this goal lies in accurately delineating the target volume and the surrounding normal organs. Prostate cancer patients are usually relatively old (with an average age of 67 years), the radiotherapy cycle is relatively long (1-2 months), and the lesion location is between the bladder and the rectum. The filling degrees of the rectum and bladder are different during each radiotherapy, which may not only lead to tumor miss but also cause the intestine and bladder to receive more irradiation, resulting in radiotherapy gastrointestinal and urinary reactions. In severe cases, it may even lead to the interruption of radiotherapy and affect the treatment effect. Online adaptive radiotherapy (ART) technology can, based on the pre-treatment images of patients, online correct the radiotherapy plan through automatic segmentation and automatic planning to ensure that the tumor receives sufficient dose irradiation while reducing radiotherapy reactions.

[0023] In current clinical practice, there are still many problems in the delineation of radiotherapy target areas and organs at risk for prostate cancer. For experienced urological radiotherapy doctors, it takes about three hours to manually segment the target areas and organs at risk, while doctors with less experience have difficulty accurately defining the scope of the target areas and organs at risk. Although artificial intelligence technology is helpful in the delineation of organs at risk and can significantly shorten the manual delineation time, due to the great difficulty in delineating organs at risk around pelvic tumors, the segmentation effects of the small intestine and colon are still not ideal, and it takes dozens of minutes to modify. Moreover, the segmentation of the target area mainly still relies on manual operation. These situations not only affect the work efficiency of radiotherapy doctors, but for second- and third-tier hospitals, they are also important factors restricting the development of prostate cancer radiotherapy. At the same time, prostate cancer radiotherapy plans involve many target areas and organs at risk. The positional relationship between the target area and many surrounding organs at risk determines that the complexity of the radiotherapy plan is relatively high. Therefore, the design of prostate cancer radiotherapy plans is a time-consuming and labor-intensive task, and the plan quality is easily affected by the personal experience of the radiotherapy plan designers.

[0024] In addition, although online adaptive radiotherapy can timely correct the radiotherapy plan based on the patient's image on the same day, during the implementation process, the patient is fixed on the treatment bed. Even if the planning target volume (PTV) is expanded by 5 mm, as time goes by, the organ movement of the prostate will intensify. To ensure that the treatment does not miss the target, it may be necessary to expand the boundary of the planning target volume by a larger margin. Moreover, the workload of delineating the target area and surrounding organs at risk is large, the manual design of radiotherapy plans takes a long time, and the plan quality based on personal experience varies. These are all key technical problems that urgently need to be solved in the clinical application practice of online adaptive radiotherapy. Therefore, accurately and quickly automatically segmenting the target area and organs at risk of prostate cancer and realizing intelligent prostate cancer plans are of great significance. The accurate delineation of automatic segmentation of prostate cancer and the realization of automatic plans can not only improve the work efficiency of radiotherapy doctors and plan designers, improve the quality of radiotherapy plans, but also promote the homogenization of the radiotherapy quality of prostate cancer in second- and third-tier hospitals. It can also shorten the time of online adaptive radiotherapy, which is beneficial to the application and popularization of online adaptive radiotherapy technology in clinical practice, so that more prostate cancer patients can benefit.

[0025] Based on this, according to the embodiments of the present application, an embodiment of an automatic segmentation and intelligent plan generation method suitable for prostate cancer adaptive radiotherapy is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0026] In this embodiment, an automatic segmentation and intelligent plan generation method suitable for prostate cancer adaptive radiotherapy is provided. Figure 1It is a flowchart of an automatic segmentation and intelligent plan generation method suitable for adaptive radiotherapy of prostate cancer according to an embodiment of the present application. As Figure 1 shown, the process includes the following steps:

[0027] S110. Obtain the medical image data and clinical treatment data of the subject to be treated.

[0028] Among them, the medical image data is the visual information data about the internal structure and function of the human body obtained by performing imaging examinations on the pelvic part of the subject to be treated. Exemplarily, if the subject to be treated is a patient diagnosed with prostate cancer and in need of radiotherapy, then the medical image data can be CT images, ultrasound images, MRI images, etc. of the lesion site obtained by various medical imaging techniques.

[0029] It should be understood that in the process of clinical practice, the clinical treatment data of the treated subjects are also important bases for medical decision-making and clinical research. These clinical treatment data can be formulated by experienced physicians, and their contents cover various aspects such as the basic information of patients, symptom manifestations, examination and test results, treatment plans and implementation conditions, and records of disease condition changes. It should be noted that the clinical treatment data are treatment target values formulated based on tumor radiotherapy guidelines and clinical practice experience, belonging to pre-radiotherapy data in the radiotherapy plan design stage, and are used to guide the design of treatment plans, rather than representing the result data of actual treatment. Specifically, the clinical treatment data can include the clinical dose data and preset priority levels of the target area and critical organs in the lesion area of the treated subject. Among them, the clinical dose data of the target area can include data such as the requirements for the prescription dose of the target area, the requirements for dose coverage of the target area, and the dose uniformity index; the clinical dose data of the critical organs can include data such as the maximum dose limit, average dose limit, and volume limit at different dose levels of the critical organs. The preset priority level can be a hierarchical system set for the importance and sensitivity of different target areas and critical organs to guide dose optimization and adjustment of optimization goals. Exemplarily, the preset priority level can divide the optimization directions of different target areas and critical organs into multiple levels based on clinical needs and treatment goals, and different levels correspond to different optimization goals and dose handling methods. Further, according to different preset priority levels, different treatment goals are set for each target area and critical organ in radiotherapy. Exemplarily, for high-priority target areas, emphasis may be placed on ensuring that they can obtain sufficient doses; for low-priority critical organs, dose coverage can be sacrificed to a certain extent to ensure the treatment effect of high-priority areas. Or for high-priority target areas and critical organ areas, their processing order may be prior to that of low-priority lesion areas to give priority to ensuring the safety threshold of high-risk areas and perform multi-stage optimization process control; or, when there are mutual exclusions among multiple optimization strategies for a certain lesion area, the conflicting strategies can be arbitrated according to the preset level of the lesion area to maintain the effectiveness of high-priority constraint conditions. S120. Automatically segment based on medical image data to obtain target segmentation data.

[0030] Among them, the target segmentation data is data used to reflect the distribution of the target area and organs at risk in the lesion area of the treated object. Specifically, after obtaining the medical image data of the treated object, an area of interest can be segmented by means of a pre-trained segmentation model, and then the target segmentation data can be obtained. Exemplarily, first, a batch of medical image data of prostate cancer patients with areas of interest marked by professional doctors can be collected, and these data can be divided into a training set, a validation set, and a test set according to a certain ratio. After the data division, preprocessing operations such as cropping and normalization are performed on these data, and at the same time, the diversity of the training set data is enhanced by means such as rotation. Next, the segmentation model is built using a deep learning framework, and relevant parameters such as the batch size and loss function are set. In the model training stage, the input image will generate a segmentation result through the forward propagation of the model. This result is compared with the true label to calculate the loss value, and then the backpropagation algorithm is used to update the model weights. During this process, the validation set and the test set are also used to evaluate the model performance in order to adjust and optimize the model. Finally, in the actual application scenario, the medical image data is input into this trained segmentation model, and the model can automatically perform the segmentation operation, thereby obtaining the target segmentation data reflecting the distribution of the target area and organs at risk in the lesion area of the treated object.

[0031] S130. Perform dose distribution prediction based on the target segmentation data to obtain initial dose distribution data.

[0032] It should be understood that the dose distribution data is also three-dimensional dose distribution data, which refers to data that can quantitatively describe and evaluate the dose distribution of radiotherapy, mainly describing the distribution pattern of the dose in the entire treatment volume, covering information such as the uniformity and gradient of the dose. Among them, the initial dose distribution data can refer to data that reflects the preliminary spatial distribution prediction result of the radiotherapy dose in this area based on the distribution of the target area and organs at risk in the lesion area of the treated object. Exemplarily, the initial dose distribution data may contain a preliminary estimate of the expected dose received by the target area and organs at risk, such as information related to the dose magnitude and dose coverage range that may be received in different areas.

[0033] Specifically, the initial dose distribution data is predicted based on the target segmentation data, which can also be achieved by using a pre-trained dose distribution prediction model. Exemplarily, first, the actual radiotherapy dose distribution data corresponding to the target segmentation data is collected, and after annotation and arrangement, it is divided into a training set, a validation set, and a test set according to a certain ratio. Then, a suitable deep learning model (such as a convolutional neural network, etc.) is selected and the architecture is designed. Then, the training parameters are set, the loss function is defined, the training set data is input into the model for forward propagation, the parameters are updated using backpropagation and optimization algorithms, the performance is evaluated using the validation set during training to adjust the model, and the model is evaluated using the test set. Finally, the target segmentation data to be predicted is input into this trained dose distribution prediction model, and the model calculates and outputs to obtain the initial dose distribution data.

[0034] S140. Iteratively optimize the initial dose distribution data using the clinical dose data and the preset priority level to obtain the target dose distribution data of the subject to be treated.

[0035] It should be clarified again that the clinical dose data here is the dose requirement data of the target area and the organs at risk in the lesion area of the subject to be treated, determined based on the clinical treatment data. The initial dose distribution data is a three-dimensional dose distribution data obtained by predicting the dose distribution based on the target segmentation data using a deep learning model, which is a preliminary estimate of the expected dose received by the target area and the organs at risk.

[0036] Among them, the target dose distribution data can refer to the ideal dose distribution result that can meet the clinical treatment requirements obtained after a series of control strategy optimization processes in the radiotherapy plan for prostate cancer. Specifically, to obtain the target dose distribution data of the subject to be treated, first, the optimization objectives and strategies need to be clarified. The clinical dose data included in the clinical treatment data can be used to obtain specific objectives such as the maximum dose limit, average dose limit, and volume limit at different dose levels of the target area and the organs at risk. Then, combined with the preset priority level, the optimization order and trade-off principle are clarified. For example, when the dose coverage priority of the target area is higher than the dose limit of the organs at risk, the dose requirement of the target area should be prioritized during the optimization process, and the dose limit of the organs at risk should be satisfied as much as possible. Further, taking the initial dose distribution data as the starting point, iterative calculations are performed according to the selected optimization algorithm and parameters. In each iteration, according to the optimization objectives and strategies, the dose distribution parameters are adjusted, such as adjusting the weights, angles, and shapes of the irradiation fields, to change the dose distribution. Finally, after each iteration, an evaluation index can be used to judge whether the dose distribution obtained in the current round meets the target requirements. Once the target requirements are met or the maximum number of iterations is reached, the iteration stops, and the dose distribution data obtained at this time is the target dose distribution data.

[0037] S150. Based on the target dose distribution data, perform iterative optimization of the radiotherapy plan to generate the target radiotherapy plan for the subject to be treated.

[0038] Among them, the target radiotherapy plan can be the target dosimetric data for describing the target area and organs at risk in the lesion area. Exemplarily, the target radiotherapy plan covers a series of information data from the setting of radiotherapy equipment parameters to the selection of irradiation methods during the entire radiotherapy process. Its function is to ensure that the target area can accurately receive the radiation dose specified by the target dose distribution data while maximizing the protection of organs at risk. Specifically, first, an initial radiotherapy plan can be constructed based on the target dose distribution data to determine some key elements in the radiotherapy plan, such as basic parameters like ray energy and dose rate. Then, taking this initial radiotherapy plan as the core basis, perform evaluation and iterative optimization. After each iteration ends, modify the variable parameters in the radiotherapy plan, and then restart a new round of evaluation and optimization. Repeat this process until the radiotherapy plan meets all preset criteria and reaches the various indicators required clinically, that is, generate the target radiotherapy plan applicable to the subject to be treated.

[0039] In the above embodiment, based on the medical image data of the subject to be treated, perform automatic segmentation to obtain the target segmentation data reflecting the distribution of the target area and organs at risk in the lesion area of the subject to be treated. Then, based on this, perform dose distribution prediction to obtain the initial dose distribution data, and perform iterative optimization by combining clinical dose data and clinical treatment data to obtain the target dose distribution data, and finally generate the target radiotherapy plan. Thus, based on the accurate and efficient automatic segmentation of the prostate cancer target area and organs at risk, multiple rounds of iterative optimization of the dose distribution and radiotherapy plan are performed, and finally a high-quality target radiotherapy plan is obtained.

[0040] In some embodiments, please refer to Appendix Figure 2 , and use clinical dose data and preset priority levels to perform iterative optimization on the initial dose distribution data to obtain the target dose distribution data of the subject to be treated, including:

[0041] S210. Obtain multiple preset optimization strategies.

[0042] Among them, the preset optimization strategy can refer to a series of methods preset to meet different clinical needs and goals during the generation of the radiotherapy plan, which are used to improve the quality of the radiotherapy plan and ensure the treatment effect. Specifically, these preset optimization strategies can be formulated based on clinical experience and research. Exemplarily, the preset optimization strategy can be a priority control strategy that sets specific dose targets for different target areas and organs at risk in accordance with the preset priority order; it can also be a target coverage rate strategy that adjusts parameters such as the prescription dose of a specific target area to improve the target coverage rate; it can also be a target dose conformity strategy that controls the conformity degree of the target area prescription dose, etc.

[0043] S220. According to the execution order of each of multiple preset optimization strategies, use the clinical dose data and the preset priority levels to sequentially adjust the initial dose distribution data, so as to obtain the intermediate dose distribution data of the subject to be treated.

[0044] It should be understood that the radiotherapy process involves numerous target areas and organs at risk. The dose requirements for each part are different, and there may be mutual restrictions between different clinical needs and goals. Therefore, in order to meet complex clinical needs and avoid conflicts between multiple optimization strategies, it is necessary to adjust the initial dose distribution data obtained based on the deep learning model according to the execution order of each of multiple preset optimization strategies. Specifically, first, based on the clinical treatment data, clearly define the specific content of the clinical dose data, including but not limited to the current dose values and dose distribution ranges of each target area and organ at risk and other detailed information. At the same time, it is also necessary to accurately obtain the preset priority levels from the clinical treatment data, so that when performing optimization subsequently, using these preset priority levels as a guide, different optimization target setting methods are adopted for different lesion sites. Exemplarily, if the preset priority level of a certain organ at risk is clinical priority, then the clinical dose data in the clinical treatment data is used as its optimization target during optimization; if the preset priority level of another organ at risk is prediction priority, then the initial dose distribution data obtained based on the deep learning model is used as its optimization target during optimization. Further, according to the optimization targets corresponding to each lesion site, according to the execution order of each of multiple preset optimization strategies, the corresponding dose parts in the initial dose distribution data are preliminarily adjusted to obtain the intermediate dose distribution data of the subject to be treated.

[0045] S230. Use the intermediate dose distribution data as the initial dose distribution data, and repeat the above iterative optimization process until the target dose distribution data of the subject to be treated is obtained.

[0046] Specifically, regard the intermediate dose distribution data obtained from the previous round of optimization as the initial dose distribution data for the new round of optimization. Again, according to the preset priority levels, according to the execution order of each of multiple preset optimization strategies, use the initial dose distribution data of the new round to adjust the doses of each target area and organ at risk. Further, after each iterative optimization, a comprehensive evaluation can also be performed on the newly obtained intermediate dose distribution data, and the evaluation result is compared with the preset evaluation criteria. If the evaluation result meets the preset evaluation criteria, it can be determined that the prediction target has been achieved. At this time, stop the iterative optimization process, and determine the iterative result of this round as the target dose distribution data of the subject to be treated. Correspondingly, if the evaluation result does not meet the preset evaluation criteria, continue to use the current intermediate dose distribution data as the new initial dose distribution data, repeat the above steps of iterative optimization and evaluation, and perform iterative optimization again.

[0047] It can be understood that the preset evaluation criteria can be in various forms. Exemplarily, the preset evaluation criteria can be the maximum number of iteration values. Once the number of iterations reaches this set value, regardless of whether the final result meets the accuracy requirements, the iterative process will terminate. The preset evaluation criteria can also be various clinical evaluation indicators. When all the clinical evaluation indicators of the iterative result reach the expected standards, it is considered that the target dose distribution data of the treated object has been obtained, and the iterative optimization process is stopped.

[0048] In the above embodiment, by using the clinical dose data and the preset priority levels to iteratively optimize the initial dose distribution data, the quality and treatment effect of the radiotherapy plan are effectively improved. At the same time, during the radiotherapy process, with the help of multiple preset optimization strategies formulated based on clinical experience and research, the dose is adjusted sequentially, and multiple iterative optimizations are carried out and judged according to the preset evaluation criteria to ensure that the finally obtained target dose distribution data is more accurate and reliable, thereby improving the accuracy and safety of radiotherapy, reducing the damage to normal tissues, and at the same time improving the formulation efficiency of the radiotherapy plan, which helps to achieve the precise and standardized treatment of diseases such as prostate cancer.

[0049] In some embodiments, the multiple preset optimization strategies include a priority control strategy and an auxiliary adjustment strategy, and the execution order of the priority control strategy is earlier than that of the auxiliary adjustment strategy. According to the respective execution orders of the multiple preset optimization strategies, the initial dose distribution data is optimized by using the clinical dose data and the preset priority levels to obtain the intermediate dose distribution data of the treated object. Please refer to the appendix Figure 3 , including:

[0050] S310. Through the priority control strategy, the initial dose distribution data of the lesion area is adjusted by using the clinical dose data and the preset priority levels to obtain the first dose distribution data that matches the preset priority levels.

[0051] Among them, the priority control strategy can refer to a strategy for specifically adjusting the dose distributions of different target areas and organs at risk in the lesion area based on the preset priority levels in the clinical treatment data. Exemplarily, as shown in Table 1 below, the preset priority levels of all parts of the lesion area can be divided into three levels: 0, 1, and ≥2. Through the priority control strategy, different optimization goals are set for the lesion parts of different levels.

[0052] Table 1 Priority control strategy

[0053]

[0054]

[0055] Furthermore, after determining the optimization objectives for different lesion sites based on the priority control strategy, the initial dose distribution data of the lesion area can be optimized and adjusted. Exemplarily, taking the small intestine, one of the organs at risk, as an example, its priority is set to 1, and based on clinical experience, its maximum dose is limited to less than or equal to 5000.00 cGy. If the clinical dose data in its clinical treatment data is less than this limit, but the initial dose distribution data predicted by the deep learning model shows that its current maximum dose exceeds this limit, in this case, according to the priority control strategy, the clinical dose data should be used as the predicted target value of the maximum dose, and the radiotherapy parameters should be adjusted to make the dose distribution of the small intestine approach the predicted target value, and finally the first dose distribution data matching the preset priority level is obtained.

[0056] S320. Through the auxiliary adjustment strategy, the first dose distribution data is further adjusted to obtain the intermediate dose distribution data.

[0057] Among them, the auxiliary adjustment strategy can refer to an optimization strategy that performs a series of supplementary adjustments on the first dose distribution data during the radiotherapy plan optimization process. Exemplarily, the auxiliary adjustment strategy can include the internal target dose control strategy and the dose reduction strategy outside the target, etc., which are used to optimize and adjust the deficiencies or factors not fully considered after the adjustment of the priority control strategy.

[0058] It should be noted that compared with the priority control strategy that determines the general direction of the optimization objective, the auxiliary adjustment strategy is more inclined to modify and adjust some details of the first dose distribution data to obtain more accurate, safer and more effective intermediate dose distribution data. Therefore, the execution order of the priority control strategy is earlier than that of the auxiliary adjustment strategy. Further, after obtaining the first dose distribution data, through a series of auxiliary adjustment strategies, various complex situations during radiotherapy are comprehensively considered, such as the uniformity of the target area dose, the dose limit of the organs at risk, the conformality of the dose distribution, etc., and the details of the first dose distribution data are further supplemented and adjusted, and finally the intermediate dose distribution data can be obtained.

[0059] In the above embodiments, by sequentially executing the priority control strategy and the auxiliary adjustment strategy, the initial dose distribution data is optimized, significantly improving the quality and accuracy of the radiotherapy plan. Among them, the priority control strategy sets clear optimization goals for different target areas and organs at risk, enabling the dose distribution to initially meet the basic clinical requirements and laying a foundation for subsequent optimization. The auxiliary adjustment strategy further takes into account various complex factors in radiotherapy and makes complementary adjustments to the deficiencies existing after the priority control strategy is adjusted. It can not only more accurately meet the dose requirements of each target area and organ at risk, reduce the radiation damage to normal tissues, but also improve the overall safety and effectiveness of the radiotherapy plan, enhance the radiotherapy effect, and contribute to the precise treatment of diseases such as prostate cancer during radiotherapy.

[0060] In some embodiments, the auxiliary adjustment strategy includes a target coverage rate strategy, a target dose conformity strategy, and a dose fall-off strategy outside the target area, and the execution order of the auxiliary adjustment strategy is sequentially executed in the order of the target coverage rate strategy, the target dose conformity strategy, and the dose fall-off strategy outside the target area.

[0061] It should be understood that the sequential execution of the target coverage rate strategy, the target dose conformity strategy, and the dose fall-off strategy outside the target area in the auxiliary adjustment strategy is to ensure the safety and effectiveness of the radiotherapy plan. Specifically, the target coverage rate is the basis of radiotherapy. If the target area cannot be effectively covered, the subsequent optimization of dose conformity and protection of organs at risk will lose its meaning. Only by first ensuring sufficient target coverage rate can the dose conformity be further optimized to make the irradiation dose more accurately fit the shape of the target area and improve the lethality to tumors. After both of these are ensured, the dose of organs at risk outside the target area is controlled through the dose fall-off strategy outside the target area to avoid excessive radiation to normal tissues and reduce the side effects of radiotherapy, thereby achieving the purpose of overall optimizing the radiotherapy plan.

[0062] Based on this, through the auxiliary adjustment strategy, the first dose distribution data is further adjusted to obtain intermediate dose distribution data. Please refer to the appendix Figure 4a , including:

[0063] S410. Adjust the target coverage rate of the first dose distribution data through the target coverage rate strategy to obtain the second dose distribution data.

[0064] Among them, the target coverage rate strategy can refer to an optimization strategy that, during the radiotherapy plan optimization process, adjusts relevant parameters to irradiate the target area with sufficient dose to improve the coverage degree of the target area by the prescribed dose. Specifically, first, based on clinical experience and research, determine the dose coverage rate requirements for different target areas, analyze the actual coverage rate of the corresponding target areas in the first dose distribution data, and find the areas that do not meet the requirements. Further, according to the positions and dose conditions of these areas, adjust the parameters of the first dose distribution data to obtain the second dose distribution data that meets the target coverage rate requirements. Exemplarily, if it is found that the dose in a certain area of the PTV (planned target volume) is insufficient, the intensity of the incident rays from the corresponding direction of this area can be appropriately increased to increase the dose of this area, thereby improving the overall coverage rate of the target area and finally obtaining the second dose distribution data.

[0065] S420. Adjust the dose field shape of the second dose distribution data through the target dose conformity strategy to obtain the third dose distribution data.

[0066] Among them, the target dose conformity strategy can refer to an optimization strategy used to control the conformity degree of the target area prescription dose, so that the distribution shape of the irradiation dose is as close as possible to the shape of the target area. Specifically, also based on clinical experience and research, determine the target dose conformity index. Then, based on the target volume and irradiated volume, calculate to determine the current dose conformity index. Further, based on the target dose conformity strategy, measure the difference between the current dose conformity index and the target dose conformity index, adjust the dose distribution of the second dose distribution data, improve the accuracy of radiotherapy, reduce unnecessary irradiation of normal tissues, and obtain the third dose distribution data. Exemplarily, the target dose conformity index can be determined with reference to the following formula:

[0067] CI current =TV RI *TV RI / (V RI *TV)

[0068] In the formula, TV RI represents the target volume, that is, the volume of the planned target volume (PTV); V RI represents the irradiated volume, that is, the actually irradiated volume; CI current represents the current dose conformity index, which is used to reflect the conformity degree of the current dose distribution to the target area. The closer it is to 1, the more conformable the dose distribution is, that is, the closer the irradiated volume is to the target volume.

[0069] Further, the process of adjustment using the target dose conformity strategy can be referred to the following formula:

[0070] f CIobj =w(CIcurrent -CI obj ) 2

[0071] wherein, f CIobj represents the target region dose conformity objective function, which is used to measure the difference between the current dose conformity index (CI) and the target dose conformity index; w represents the weight factor, which is used to adjust the weight of the objective function; CI current represents the current dose conformity index; CI obj represents the target dose conformity index, which is used to reflect the conformity degree of the dose distribution in the ideal situation with the target region.

[0072] S430. Adjust the dose distribution of the organs at risk outside the target region in the third dose distribution data through the dose fall-off strategy outside the target region to obtain the intermediate dose distribution data.

[0073] Among them, the dose fall-off strategy outside the target region may refer to an optimization strategy for controlling the dose distribution outside the target region in radiotherapy plan optimization, so that the dose decreases at a reasonable rate outside the target region, and reduces the irradiation dose of the organs at risk and normal tissues outside the target region. Specifically, if there are important organs at risk near the target region, the parameters such as the starting distance of dose fall-off, the dose value at different distances, and the fall-off rate can be adjusted through the dose fall-off strategy outside the target region to optimize the third dose distribution data, ensure that the dose received by these organs at risk is within the safe range, and at the same time maintain the dose of the target region, so as to achieve effective control of the dose outside the target region.

[0074] Exemplarily, first, the key parameters of the dose fall-off outside the target region can be determined according to clinical experience and radiotherapy requirements. Please refer to Figure 4b , in the figure, the horizontal axis represents the distance from the center of the target region, and the vertical axis represents the percentage of the dose (Rx). Base dose is the base dose within the target region, which is located at 100 on the vertical axis in the figure and represents 100% of the prescribed dose. Start dose is the starting dose at the edge of the target region, which is located at 105 on the vertical axis in the figure, meaning slightly higher than 100% of the prescribed dose. End dose is the final dose at a certain distance, which is located at 50 on the vertical axis in the figure and represents 50% of the prescribed dose. distStart refers to the distance from the edge of the target region where the dose starts to fall off, which is shown as a specific distance value in the figure and is the starting point where the dose starts to decrease rapidly. Lower Fall-off Rate represents the curve with a slower dose fall-off, which means that at a farther distance from the center of the target region, the dose gradually decreases, but the decrease rate is slower. Higher Fall-off Rate is the curve with a faster dose fall-off, indicating that at a closer distance from the center of the target region, the dose decreases rapidly.

[0075] Furthermore, the dose distribution of organs at risk outside the target area is determined in the third dose distribution data, and its dose distribution is adjusted based on the above parameters. For example, if the dose in an area close to the target area is found to be too high, the dose in that area can be rapidly reduced according to the Higher Fall-off Rate by adjusting the intensity distribution of the radiation or adding shielding measures. For areas farther from the target area, the dose is adjusted according to the Lower Fall-off Rate to ensure that the dose drops reasonably outside the target area. Finally, after the third dose distribution data is adjusted, the intermediate dose distribution data that effectively protects the organs at risk can be obtained.

[0076] In the above-mentioned embodiment, by sequentially executing a series of auxiliary adjustment strategies, namely, the target coverage strategy, the target dose conformal strategy, and the dose drop strategy outside the target area, the quality and safety of the radiotherapy plan are significantly improved, and better treatment effects are achieved for radiotherapy of diseases such as prostate cancer. The target coverage strategy ensures that the target area receives sufficient dose irradiation, improves the coverage of the target area by the prescribed dose, and fundamentally guarantees the lethality of radiotherapy to tumors. The target dose conformal strategy makes the distribution shape of the irradiation dose highly consistent with the target area by accurately calculating and adjusting the dose conformal index, further improving the accuracy of radiotherapy, reducing unnecessary irradiation of normal tissues, and enhancing the effect of attacking tumors while protecting normal tissues. The dose drop strategy outside the target area focuses on the protection of organs at risk outside the target area. By reasonably adjusting the key parameters of the dose drop, the dose distribution of organs at risk outside the target area is effectively controlled to avoid excessive radiation to normal tissues and reduce the side effects of radiotherapy. This series of strategies work synergistically, and the resulting intermediate dose distribution data not only ensures effective treatment of tumors but also minimizes damage to normal tissues. It optimizes the overall radiotherapy plan, improves the safety and effectiveness of radiotherapy, and helps enhance patients' treatment experience and recovery outcomes.

[0077] In some embodiments, please refer to the attached Figure 5 Based on the target dose distribution data, the radiotherapy plan is iteratively optimized to generate the target radiotherapy plan for the patient, including:

[0078] S510 : Determine an initial radiotherapy plan based on the target dose distribution data.

[0079] Among them, the initial radiotherapy plan can refer to the radiotherapy plan preliminarily formulated based on the target dose distribution data. This plan has preliminarily set the basic parameters of radiotherapy according to the target dose distribution data, but has not been further optimized and adjusted, and may not fully meet the clinical requirements. Specifically, after obtaining the target dose distribution data, appropriate radiotherapy equipment and radiotherapy techniques can be selected based on information such as the dose distribution of the target area and critical organs included therein, and the type and energy of the radiation can be determined. Further, according to the shape, position, and size of the target area, information such as the number, size, and direction of the irradiation fields can be planned.

[0080] S520. Optimize and adjust the maximum dose in the lesion area of the initial radiotherapy plan through the global maximum dose control strategy to obtain an intermediate radiotherapy plan.

[0081] Among them, the global maximum dose control strategy can refer to the strategy of controlling and optimizing the maximum dose that may appear in the entire treatment area in the radiotherapy plan, which is used to ensure that during the treatment process, the dose at any point does not exceed the preset safety threshold, thereby avoiding excessive damage to normal tissues, and at the same time ensuring that the tumor target area can receive sufficient dose irradiation to achieve the balance between treatment effect and safety.

[0082] Specifically, after obtaining the initial radiotherapy plan, through the global maximum dose control strategy, based on the maximum dose threshold allowed in the entire treatment area specified clinically, analyze the dose distribution of the lesion area and the surrounding normal tissues in the initial radiotherapy plan, and find out the position and value where the maximum dose may appear. If it is found that the maximum dose at a certain place exceeds the preset maximum dose threshold, then correspondingly optimize and adjust the dose in the corresponding area of the initial radiotherapy plan to reduce the maximum dose value to be within the safe range, and finally obtain an intermediate radiotherapy plan.

[0083] S530. Take the intermediate radiotherapy plan as the initial radiotherapy plan and repeat the above iterative optimization process until the final radiotherapy plan of the treated object is obtained.

[0084] Specifically, after obtaining the intermediate radiotherapy plan, the intermediate radiotherapy plan can also be regarded as a new initial radiotherapy plan, and the global maximum dose control strategy can be applied again for optimization, and this process is continuously repeated. In each iteration, the dose distribution is further optimized to adjust the parameters of the radiotherapy plan. At the same time, continuously evaluate the results after each iterative optimization. When the maximum dose is stable below the safe threshold in the entire treatment area, and the target area dose and critical organ dose both meet the clinical requirements, the iteration can be stopped. At this time, the obtained radiotherapy plan can be used as the final radiotherapy plan.

[0085] S540. Determine the target radiotherapy plan of the treated object according to the final radiotherapy plan.

[0086] Specifically, after obtaining the final radiotherapy plan, a comprehensive evaluation can be carried out on the final radiotherapy plan to check whether the dose coverage of the target area is sufficient and uniform, whether the dose of the organs at risk is within an acceptable range, and whether the overall dose distribution conforms to clinical specifications and treatment goals. When it is confirmed that there is no error, the final radiotherapy plan can be determined as the target radiotherapy plan for guiding the actual radiotherapy treatment process of the treated object.

[0087] In the above embodiment, the global maximum dose control strategy is used to strictly control the maximum dose of the entire treatment area during the optimization process, avoiding excessive radiation damage to normal tissues, while ensuring that the tumor target area receives sufficient dose irradiation, balancing the treatment effect and safety. At the same time, multiple iterative optimizations are also carried out to continuously adjust the radiotherapy plan parameters, gradually approaching the requirements, making the dose distribution more accurate and reasonable, until the maximum dose is stabilized below the safety threshold, and finally generating a target radiotherapy plan that highly fits the actual needs of the treated object, effectively improving the safety and effectiveness of radiotherapy.

[0088] In some embodiments, please refer to the appendix Figure 6 , and determine the target radiotherapy plan of the treated object according to the final radiotherapy plan, including:

[0089] S610. Optimize and adjust the uniformity of the dose distribution in the lesion area of the final radiotherapy plan based on the dose normalization strategy to obtain an optimized radiotherapy plan.

[0090] Among them, the dose normalization strategy can refer to a strategy used to adjust and standardize the dose distribution in the lesion area during the radiotherapy plan optimization process, so that on the one hand, it can meet the uniformity requirements, and at the same time, it can also take into account the integrity of the dose coverage of the target area. Specifically, a unified dose standard or reference value can be set in advance based on clinical indicators, and a detailed analysis of the dose distribution in the lesion area of the final radiotherapy plan can be carried out. According to the analysis results, the dose normalization strategy is used to optimize and adjust the dose distribution. If it is found that the dose in a certain area is too high after analysis, the dose of this area can be reduced by reducing the corresponding ray weight of this area, adjusting the incident angle of the ray or reducing the irradiation time, etc.; if the dose in a certain area is too low, the opposite measures are taken to increase the dose of this area to ensure that the adjusted dose distribution is more uniform.

[0091] Optionally, after completing the optimization and adjustment, the dose distribution in the lesion area can be evaluated again to check whether the requirements of dose normalization are met. Exemplarily, the uniformity of the dose distribution can be measured by calculating indicators such as the dose uniformity index. If the evaluation result is not ideal, continue to adjust until the dose normalization standard is met, thereby obtaining an optimized radiotherapy plan.

[0092] S620. Determine the target radiotherapy plan of the treated object according to the optimized radiotherapy plan.

[0093] Specifically, after obtaining the optimized radiotherapy plan, a comprehensive inspection of the optimized radiotherapy plan can be carried out, including aspects such as the dose coverage of the target area, whether the dose limits of the organs at risk are met, the feasibility and safety of the radiotherapy plan, etc. Check whether the target area is completely covered by the prescribed dose and the dose distribution is uniform; whether the dose of the organs at risk is within the safe range and will not cause serious damage to normal tissues, etc. Further, after comprehensive inspection and evaluation, if the optimized radiotherapy plan meets all clinical requirements and standards and no problems or potential risks are found, the optimized radiotherapy plan can be determined as the target radiotherapy plan for the treated object to guide the actual radiotherapy treatment process of the treated object.

[0094] Exemplarily, the comparison result between the target radiotherapy plan of the treated object obtained through the above automatic planning and the manual radiotherapy plan of the treated object actually manually planned by an experienced physician in clinical practice can be referred to as shown in the following dose advantage table:

[0095] Table 2 Dose Advantage Table

[0096]

[0097]

[0098] The various evaluation indicators of the target radiotherapy plan and the manual radiotherapy plan in multiple lesion regions and organs at risk are compared in detail in the table. Among them, the target volume-related indicators include PTV45Gy, PTV65Gy, and PTV67.5Gy, which represent the Planning Target Volume at different prescription doses respectively; HI refers to the Homogeneity Index, which is used to measure the degree of dose distribution uniformity within the target volume; CI is the Conformity Index, which reflects the degree of fit between the irradiation dose distribution and the target volume shape; coverage represents the coverage rate, that is, the proportion of the volume of the target volume actually receiving the prescription dose in the entire target volume. The indicators related to organs at risk involve multiple parts, such as the Bladder, Rectum, Small Bowel, Colons, Anal Canal, Penile Bulb, Spinal Cord, and Pelvic Bone. These indicators include the percentage of the volume of each organ at risk receiving different doses of irradiation in the total volume of the corresponding part, as well as the maximum dose Dmax (Gy) and the average dose Dmean (Gy), which are used to evaluate the degree of impact of radiotherapy on these organs. In addition, the t-value / z-value in the table is used to measure the significance of the difference between the target radiotherapy plan and the manual radiotherapy plan in the corresponding evaluation indicators, while the p-value further indicates whether this difference is significant.

[0099] Furthermore, it can be clearly seen from the results in the table that the dose distribution similarity between the target radiotherapy plan and the manual radiotherapy plan is relatively high, the differences in some key clinical dose parameters are small, and they are all within the clinically acceptable range. And in some cases, the effect of the target radiotherapy plan is better than the original clinical manual radiotherapy plan. At the same time, the DVH (Dose-Volume Histogram) curves of PTV (Planning Target Volume) and OARs (Organs at Risk) in the two groups of plans are also relatively consistent to a large extent, with a high degree of coincidence. In addition, the average calculation time required to generate the target radiotherapy plan is only 8 minutes and 51 seconds, which is far less than the generation time of the manual radiotherapy plan, significantly improving the efficiency of radiotherapy plan formulation.

[0100] In the above embodiments, through the dose normalization strategy, the dose distribution in the lesion area of the final radiotherapy plan is analyzed in detail, and precise regulation is implemented for areas with too high or too low doses to obtain an optimized radiotherapy plan. After obtaining the optimized radiotherapy plan, it is comprehensively evaluated again to ensure that the plan can meet clinical requirements and there are no potential risks, so as to finally determine the target radiotherapy plan. This target radiotherapy plan can not only accurately guide the actual radiotherapy process of the treated object, significantly improve the accuracy of radiotherapy, but also minimize the risk of damage to normal tissues, effectively enhance the radiotherapy effect, and improve the treatment experience of patients.

[0101] In some embodiments, automatic segmentation is performed based on medical image data to obtain target segmentation data, including:

[0102] Automatically delineate the medical image data based on the 3D U-Net model to obtain the target segmentation data.

[0103] It should be understood that the U-Net model is an image segmentation model based on convolutional neural network (CNN), and the 3D U-Net model is developed based on the U-Net model and can also process input images of any size. Specifically, it consists of two parts: an encoder (downsampling path) and a decoder (upsampling path), forming a U-shaped structure. Among them, the encoder part extracts high-level features of the image by gradually applying convolutional layers and pooling layers, and the size of the feature map will gradually decrease. Exemplarily, its encoder can be composed of four convolutional blocks, each convolutional block contains two 3x3 convolutional layers and a ReLU activation function, and a 2x2 max pooling layer is used for downsampling after each convolutional block. For the decoder part, the spatial resolution of the image is gradually restored through upsampling operations, and it combines the feature maps of the encoder part, and fuses low-level and high-level features through skip connections. Exemplarily, its decoder can also be composed of four upsampling blocks, each upsampling block contains a transposed convolutional layer, two 3x3 convolutional layers and a ReLU activation function. The upsampling block concatenates the output of the previous layer and the output of the corresponding encoder layer, that is, performs skip connection, so that the model can make full use of the detailed information in the encoding process during the decoding process, thereby improving the segmentation accuracy. In addition, there is a bottleneck layer as a bridge part between the encoder and the decoder. This bottleneck layer contains two 3x3 convolutional layers and a ReLU activation function. The model starts from the input image, passes through the encoder, bottleneck layer and decoder in sequence, and the final convolutional layer is a 1x1 convolutional layer, whose function is to adjust the number of output channels to the required number, and finally output the segmentation result.

[0104] It is understandable that after determining the framework of the segmentation model, the model needs to be trained. Specifically, medical image data with regions of interest marked by professional doctors can be used as samples, and the training set, validation set, and test set are divided in a ratio of 8:1:1. Then, image preprocessing is performed on the medical image dataset, including cropping the images to a unified size, normalizing the pixel values of the images, and increasing the data diversity through operations such as rotation, flipping, translation, and scaling. During the training process, the 3D U-Net model can be trained using the deep learning framework PyTorch. Set its batch size to 4, and use the cross-entropy loss function as its loss function, and use the Momentum optimizer (momentum factor is 0.9, weight decay coefficient is 0.0005) as the optimization function. Set the initial learning rate to 0.003, the upper limit of training rounds to 1000, and the number of training iterations epoch to 30 times, and use the cosine annealing strategy to dynamically adjust the learning rate. At the same time, the validation set is used to evaluate the model performance, and the segmentation effect is measured by calculating indicators such as the Dice coefficient and the Intersection over Union (IoU), and the model with the best performance on the validation set is used as the segmentation model. In each iteration, the input image undergoes forward propagation through the UNet network to generate a segmentation result, and then the loss function value is calculated by comparing it with the true label. The gradient is calculated through the backpropagation algorithm and the model weights are updated. At the same time, data augmentation techniques are used to generate new training samples to improve the training efficiency and stability. Correspondingly, after the training is completed, the final performance of the model can also be evaluated on the test set to ensure its effectiveness in practical applications. Exemplarily, please refer to Figure 7 , in the figure, the dashed part is the segmentation data output by the segmentation model, and the solid part is the region of interest marked by the physician. It can be seen from the figure that the segmentation range and accuracy of the target area by this segmentation model are much higher than manual marking, and the delineation of critical organs is also very comprehensive and the accuracy is also relatively high.

[0105] It is understandable that based on the above training process, the trained segmentation model can be used for medical image segmentation tasks, that is, automatically segmenting the medical image data of the treated object, and then obtaining the target segmentation data reflecting the distribution of the target area and critical organs in the lesion area of the treated object.

[0106] Furthermore, based on the target segmentation data, dose distribution prediction is performed to obtain initial dose distribution data, including:

[0107] Based on the HD U-Net model, dose distribution prediction is performed on the target segmentation data to obtain initial dose distribution data.

[0108] Among them, the HD U-Net model is also developed on the basis of the U-Net model. Others can give full play to the global and local information processing capabilities of U-Net and integrate the more efficient feature propagation and reuse advantages of Dense Net. In the HD U-Net model, dense convolution is used to effectively combine the rectified linear unit (ReLU) for quasi-convolution operations. Dense downsampling is performed through strided convolution and ReLU to calculate a new feature set with half the resolution. Subsequently, max pooling is performed on the previous feature set, and the previous feature sets are concatenated to form a new feature set. Finally, upsampling, convolution, and ReLU are used for the U-Net upsampling operation to concatenate the feature sets on the other side of the "U" shape structure. For each dense operation, a growth rate is defined to determine the number of new features generated in the convolution step. During this process, the HD U-Net model makes full use of the hierarchical structure, that is, there are different resolution levels between each max pooling or upsampling operation in U-Net. The convolutional layers are densely connected within each hierarchical structure but not connected during the upsampling operation between different levels of U-Net. This structural design can capture and process information at different scales more accurately. It can be understood that although the prediction time of the HD U-Net architecture is relatively slightly longer and requires more random access memory of the graphics processor, it has better anti-overfitting ability when modeling the dose. Exemplarily, the neural network model based on 3D-UNet also consists of an encoder and a decoder. Both the encoder and the decoder adopt modules containing convolutional layers and attention mechanisms (CAD blocks). In the encoding stage, the number of convolutional filters in each module is 32, 64, 128, 256, and 512 respectively. After passing through the Max Pooling layer, the size of the feature map is reduced to half of the original. During the decoding process, the upsampling operation doubles the size of the feature map. Except for the convolutional filter size of the output layer being 1×1×1, the sizes of the remaining convolutional filters are all 3×3×3. In addition, a skip connection mechanism is adopted between the encoder and the decoder to fuse the shallow image information and the deep semantic information. By combining the Group Norm layer and the He_normal method, the generalization ability of this model has been significantly improved. The input of this model is three-dimensional contour information (i.e., segmentation data), and the output is three-dimensional dose field information (i.e., dose distribution data).

[0109] Similarly, after determining the framework of the dose prediction model, the model also needs to be trained. First, the clinical treatment data of each patient can be sampled in advance, and at the same time, the contour information of the planned target volume (PTV) and organs at risk (OAR) of each patient is collected. These contour data will be used as the input of the HD U-Net model. The actual dose distribution in the patient's body is used as the ground truth for training to guide the learning of the model. To ensure that the data can be effectively processed by the dose prediction model, preprocessing operations such as normalization and resizing can also be performed on the data. Further, during the training process, the Adam optimizer can be used for training, with the initial learning rate set to 1e-4, the total number of training epochs set to 1000, and the weight decay coefficient to 1e-4. These parameter settings help the model adjust the weights more reasonably during training, effectively avoid overfitting, and improve the generalization ability of the model. At the same time, the mean squared error (MSE) is selected as the loss function to measure the gap between the dose distribution predicted by the model and the actual dose distribution (ground truth). By continuously minimizing this loss function, the prediction result of the model is gradually made to approach the true dose distribution. In the specific training implementation, Python 3.8 can be used as the programming language, Pytorch 1.13 as the deep learning framework, and the training task is executed on the NVIDIA RTX 3090 GPU running the Linux system. With the powerful computing power of the GPU, the training speed of the model is greatly accelerated.

[0110] It should be noted that this dose prediction algorithm using deep learning mainly trains the model for dose control of organs at risk in radical prostate cancer, and most of the target prescription doses are still predicted and optimized based on the actual situation. Therefore, this model can be applied to multi-prescription plans.

[0111] Further, based on the above training process, the trained HD U-Net model can be used to input the processed target segmentation data into the trained HD U-Net model. The model will perform a series of operations on the input data through the encoder and decoder, such as feature extraction, downsampling, upsampling, and feature fusion, so as to predict the dose distribution and finally output the predicted three-dimensional dose field, that is, the initial dose distribution data is obtained.

[0112] In the above embodiments, the automatic segmentation of medical image data based on the 3D U-Net model can accurately perform automatic delineation of medical images, greatly improving the segmentation range and accuracy of the target area, and comprehensively and accurately delineating the organs at risk. It not only improves the automatic segmentation accuracy of the pelvic organs at risk, but also greatly improves the work efficiency of radiotherapy doctors. At the same time, the proposed automatic segmentation technology for the target area fills the gap in the automatic segmentation of the prostate cancer target area and helps to homogenize the radiotherapy of prostate cancer. In the dose distribution prediction stage, the dose distribution is predicted based on the HD U-Net model, which integrates the advantages of U-Net and Dense Net, accurately captures information at different scales using a hierarchical structure, and can quickly and accurately output the initial dose distribution data according to the target segmentation data, providing a precise and efficient basis for the formulation of radiotherapy plans. It strongly promotes the development of radiotherapy technology towards precision and intelligence, helps the application and popularization of online adaptive radiotherapy technology in the radiotherapy of prostate cancer, improves the radiotherapy accuracy, and reduces radiotherapy reactions.

[0113] It should be understood that although the steps in the above flowchart are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order limit for the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in the above flowchart may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0114] The embodiments of this specification also provide an automatic segmentation and intelligent plan generation device 800 suitable for the adaptive radiotherapy of prostate cancer, as Figure 8 shown, including: a data acquisition module 810, a delineation and segmentation module 820, a dose prediction module 830, a dose optimization module 840, and a plan generation module 850, where:

[0115] The data acquisition module 810 is used to acquire the medical image data and clinical treatment data of the treated object; among them, the medical image data is obtained by performing imaging examinations on the pelvic part of the treated object; the clinical treatment data includes the clinical dose data and preset priority levels of the target area and the organs at risk in the lesion area of the treated object.

[0116] The delineation and segmentation module 820 is used to perform automatic segmentation based on the medical image data to obtain target segmentation data; among them, the target segmentation data is data used to reflect the distribution of the target area and the organs at risk in the lesion area of the treated object.

[0117] The dose prediction module 830 is configured to predict a dose distribution based on the target segmentation data to obtain initial dose distribution data.

[0118] The dose optimization module 840 is configured to iteratively optimize the initial dose distribution data by using clinical dose data and a preset priority level to obtain the target dose distribution data of the subject to be treated.

[0119] The treatment plan generation module 850 is configured to iteratively optimize a radiotherapy plan based on the target dose distribution data to generate a target radiotherapy plan for the subject to be treated; wherein, the target radiotherapy plan is used to describe the target dosimetric data of the target area and the organs at risk in the lesion area.

[0120] In some embodiments, the dose optimization module 840 is further configured to obtain a plurality of preset optimization strategies; adjust the initial dose distribution data in sequence by using the clinical dose data and the preset priority level according to the execution order of the plurality of preset optimization strategies respectively to obtain intermediate dose distribution data of the subject to be treated; use the intermediate dose distribution data as the initial dose distribution data, and repeat the above iterative optimization process until the target dose distribution data of the subject to be treated is obtained.

[0121] In some embodiments, the plurality of preset optimization strategies include a priority control strategy and an auxiliary adjustment strategy, and the execution order of the priority control strategy is earlier than that of the auxiliary adjustment strategy. The dose optimization module 840 is further configured to adjust the initial dose distribution data of the lesion area by using the clinical dose data and the preset priority level through the priority control strategy to obtain a first dose distribution data matching the preset priority level; continue to adjust the first dose distribution data through the auxiliary adjustment strategy to obtain intermediate dose distribution data.

[0122] In some embodiments, the auxiliary adjustment strategy includes a target coverage rate strategy, a target dose conformity strategy, and a dose fall-off strategy outside the target area, and the execution order of the auxiliary adjustment strategy is sequentially executed in the order of the target coverage rate strategy, the target dose conformity strategy, and the dose fall-off strategy outside the target area. The dose optimization module 840 is further configured to adjust the target coverage rate of the first dose distribution data through the target coverage rate strategy to obtain a second dose distribution data; adjust the dose field shape of the second dose distribution data through the target dose conformity strategy to obtain a third dose distribution data; adjust the dose distribution of the organs at risk outside the target area in the third dose distribution data through the dose fall-off strategy outside the target area to obtain intermediate dose distribution data.

[0123] In some embodiments, the plan generation module 850 is further configured to determine an initial radiotherapy plan based on the target dose distribution data; optimize and adjust the maximum dose in the lesion area of the initial radiotherapy plan through a global maximum dose control strategy to obtain an intermediate radiotherapy plan; use the intermediate radiotherapy plan as the initial radiotherapy plan, and repeat the above iterative optimization process until the final radiotherapy plan of the treated subject is obtained; and determine the target radiotherapy plan of the treated subject according to the final radiotherapy plan.

[0124] In some embodiments, the plan generation module 850 is further configured to optimize and adjust the uniformity of the dose distribution in the lesion area of the final radiotherapy plan based on a dose normalization strategy to obtain an optimized radiotherapy plan; and determine the target radiotherapy plan of the treated subject according to the optimized radiotherapy plan.

[0125] In some embodiments, the delineation and segmentation module 820 is further configured to automatically delineate the medical image data based on a 3D U-Net model to obtain target segmentation data.

[0126] In some embodiments, the dose prediction module 830 is further configured to predict the dose distribution of the target segmentation data based on an HD U-Net model to obtain initial dose distribution data.

[0127] For the specific limitations of an automatic segmentation and intelligent plan generation device suitable for adaptive radiotherapy for prostate cancer, reference may be made to the limitations of an automatic segmentation and intelligent plan generation method suitable for adaptive radiotherapy for prostate cancer described above, which will not be elaborated here. Each module in the above-mentioned automatic segmentation and intelligent plan generation device suitable for adaptive radiotherapy for prostate cancer can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of the computer device in the form of hardware or be independent of it, or be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0128] An automatic segmentation and intelligent plan generation device suitable for adaptive radiotherapy for prostate cancer in this embodiment is presented in the form of functional units. Here, the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0129] An embodiment of the present application further provides a computer device, which may be a terminal, and its internal structure diagram may be as Figure 9As shown. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements an automatic segmentation and intelligent plan generation method suitable for adaptive radiotherapy of prostate cancer. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball, or touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0130] Those skilled in the art can understand that Figure 9 the structure shown in is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0131] The embodiment of this application also provides a computer-readable storage medium. The method according to the embodiment of this application can be implemented in hardware, firmware, or be implemented as computer code that can be recorded on a storage medium, or be implemented as computer code that is originally stored in a remote storage medium or a non-transitory machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disc, a read-only memory, a random access memory, etc.; further, the storage medium can also include a combination of the above-mentioned types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code, and when the software or computer code is accessed and executed, it implements the method shown in the above embodiment.

[0132] The embodiment of this application provides a computer program product. The computer program product includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method of any embodiment of this application.

[0133] An automatic segmentation and intelligent plan generation method and device suitable for adaptive radiotherapy of prostate cancer illustrated in the above embodiments can be specifically implemented by a computer chip or entity, or by a product with a certain function. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, or a combination of any of these devices. For the convenience of description, when describing the above device, it is divided into various units according to functions and described separately. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0134] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0135] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate computer-implemented processing. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0136] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of these features. In the description of the present application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise specifically defined.

[0137] It should also be noted that the term "comprise", "include" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or apparatus that comprises a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or apparatus. Without further limitation, an element defined by the phrase "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or apparatus that comprises the element.

[0138] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. Since it is basically similar to the method embodiment, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiment.

[0139] The above are only the embodiments of the present application and are not used to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

[0140] Although the embodiments of the present application are described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations fall within the scope defined by the appended claims.

Claims

1. An automatic segmentation and intelligent plan generation method suitable for adaptive radiotherapy of prostate cancer, characterized in that, The method comprises: Acquiring medical imaging data and clinical treatment data of the subject; wherein the medical imaging data is obtained by performing an imaging examination of the pelvic region of the subject; and the clinical treatment data includes clinical dose data and preset priority levels of target areas and organs at risk in the lesion area of the subject; Automatically segmenting the medical image data to obtain target segmentation data; wherein the target segmentation data is used to reflect the distribution of target areas and organs at risk in the lesion area of the subject; Performing dose distribution prediction based on the target segmentation data to obtain initial dose distribution data; Iteratively optimizing the initial dose distribution data using the clinical dose data and the preset priority level to obtain target dose distribution data for the subject; Based on the target dose distribution data, an iterative optimization of the radiotherapy plan is performed to generate a target radiotherapy plan for the subject; wherein the target radiotherapy plan is used to describe the target dose data of the target area and the organ at risk in the lesion area.

2. The method according to claim 1, wherein The iteratively optimizing the initial dose distribution data using the clinical dose data and the preset priority level to obtain target dose distribution data of the treated subject includes: Get multiple preset optimization strategies; adjusting the initial dose distribution data sequentially using the clinical dose data and the preset priority levels according to the execution order of each of the plurality of preset optimization strategies to obtain intermediate dose distribution data of the subject; The intermediate dose distribution data is used as the initial dose distribution data, and the above-mentioned iterative optimization process is repeated until the target dose distribution data of the treated object is obtained.

3. The method according to claim 2, characterized in that The plurality of preset optimization strategies include a priority control strategy and an auxiliary adjustment strategy, and the execution order of the priority control strategy is earlier than the execution order of the auxiliary adjustment strategy; The optimizing the initial dose distribution data using the clinical dose data and the preset priority levels according to the execution order of the plurality of preset optimization strategies to obtain the intermediate dose distribution data of the treated subject includes: By using the priority control strategy, the initial dose distribution data of the lesion area is adjusted using the clinical dose data and the preset priority level to obtain first dose distribution data that matches the preset priority level; The first dose distribution data is further adjusted through the auxiliary adjustment strategy to obtain the intermediate dose distribution data.

4. The method according to claim 3, characterized in that The auxiliary adjustment strategy includes a target coverage strategy, a target dose conformal strategy, and a dose drop strategy outside the target area, and the execution order of the auxiliary adjustment strategy is the target coverage strategy, the target dose conformal strategy, and the dose drop strategy outside the target area. The step of further adjusting the first dose distribution data by using the auxiliary adjustment strategy to obtain the intermediate dose distribution data includes: Adjusting the target coverage of the first dose distribution data by using the target coverage strategy to obtain second dose distribution data; Adjust the dose field shape of the second dose distribution data through the target area dose conformation strategy to obtain the third dose distribution data; Adjust the dose distribution of the organs at risk outside the target area in the third dose distribution data through the dose fall-off strategy outside the target area to obtain the intermediate dose distribution data.

5. The method according to claim 1, characterized in that, The iterative optimization of the radiotherapy plan based on the target dose distribution data to generate the target radiotherapy plan for the treated subject includes: Determine the initial radiotherapy plan based on the target dose distribution data; Optimize and adjust the maximum dose in the lesion area of the initial radiotherapy plan through the global maximum dose control strategy to obtain the intermediate radiotherapy plan; Use the intermediate radiotherapy plan as the initial radiotherapy plan and repeat the above iterative optimization process until the final radiotherapy plan for the treated subject is obtained; Determine the target radiotherapy plan for the treated subject according to the final radiotherapy plan.

6. The method according to claim 5, characterized in that The determining the target radiotherapy plan for the treated subject according to the final radiotherapy plan includes: Optimize and adjust the uniformity of the dose distribution in the lesion area of the final radiotherapy plan based on the dose normalization strategy to obtain the optimized radiotherapy plan; Determine the target radiotherapy plan for the treated subject according to the optimized radiotherapy plan.

7. The method according to claim 1, wherein The automatically segmenting based on the medical image data to obtain the target segmentation data includes: Automatically delineate the medical image data based on the 3D U-Net model to obtain the target segmentation data; The predicting the dose distribution according to the target segmentation data to obtain the initial dose distribution data includes: Predict the dose distribution of the target segmentation data based on the HD U-Net model to obtain the initial dose distribution data.

8. An automatic segmentation and intelligent plan generation device for adaptive radiotherapy of prostate cancer, characterized in that: The device includes: A data acquisition module for acquiring medical image data and clinical treatment data of a treated subject; wherein, the medical image data is obtained by performing an imaging examination on the pelvic part of the treated subject; the clinical treatment data includes the clinical dose data and the preset priority levels of the target area and the organs at risk in the lesion area of the treated subject; A delineation and segmentation module for automatically segmenting based on the medical image data to obtain target segmentation data; wherein, the target segmentation data is data for reflecting the distribution of the target area and the organs at risk in the lesion area of the treated subject; A dose prediction module for predicting the dose distribution according to the target segmentation data to obtain the initial dose distribution data; A dose optimization module for iteratively optimizing the initial dose distribution data by using the clinical dose data and the preset priority levels to obtain the target dose distribution data of the treated subject; A plan generation module for iteratively optimizing the radiotherapy plan based on the target dose distribution data to generate the target radiotherapy plan for the treated subject; wherein, the target radiotherapy plan is used to describe the target dosimetric data of the target area and the organs at risk in the lesion area.

9. A computer device, characterized in that, Includes: A memory and a processor, which are communicatively connected to each other. Computer instructions are stored in the memory, and the processor executes the computer instructions to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Computer instructions are stored on the computer-readable storage medium, and the computer instructions are used to cause a computer to perform the method according to any one of claims 1 to 7.

Images