CN112635024A - Automatic planning and designing system for radiotherapy and construction method thereof - Google Patents
Automatic planning and designing system for radiotherapy and construction method thereof Download PDFInfo
- Publication number
- CN112635024A CN112635024A CN202110257901.9A CN202110257901A CN112635024A CN 112635024 A CN112635024 A CN 112635024A CN 202110257901 A CN202110257901 A CN 202110257901A CN 112635024 A CN112635024 A CN 112635024A
- Authority
- CN
- China
- Prior art keywords
- neural network
- deep neural
- module
- radiotherapy
- dose
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/40—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to mechanical, radiation or invasive therapies, e.g. surgery, laser therapy, dialysis or acupuncture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Surgery (AREA)
- Urology & Nephrology (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Radiation-Therapy Devices (AREA)
Abstract
The invention discloses a radiotherapy automatic plan design system and a construction method thereof, relating to the field of radiotherapy plan systems and comprising a plan design auxiliary contour generation module, a prescription setting module, an addition radiation field module, a deep neural network dose prediction module and an optimization objective function generation and plan design module; the deep neural network dose prediction module is used for providing a reasonable dose design target for a reverse optimization process according to data obtained from the same disease species; after the deep neural network model is trained, the dose distribution condition of the radiotherapy patient can be predicted quickly within a few minutes, and the radiotherapy plan design is automatically carried out, so that the working efficiency of a radiotherapy doctor is effectively improved, and the formulation of a radiotherapy scheme of the patient is accelerated.
Description
Technical Field
The invention relates to the field of radiotherapy planning systems, in particular to a radiotherapy automatic planning design system and a construction method thereof.
Background
The tumor becomes a common high-incidence disease species in clinic, and the tumor incidence and the mortality of residents in China are improved year by year. Radiation Therapy (Radiation Therapy) is an important treatment means for primary tumor focus and tumor postoperation, and the treatment contribution rate of Radiation Therapy is higher than that of chemotherapy and surgical treatment.
Radiotherapy firstly needs to obtain the positioning CT of a patient, an oncologist outlines a target area, a dosimeter designs a radiotherapy plan, an accelerator execution file is generated, and then the accelerator execution file is guided into an accelerator system to treat the patient. Therefore, radiotherapy planning is an important step in the radiation therapy procedure. The radiation treatment planning design has the characteristics of patient customization and personalized planning design of a dossier. Patient customization means that the radiotherapy plan design is comprehensively designed according to the medical history, CT images, lesion ranges and treatment strategies of each radiotherapy patient, and has uniqueness; the individualization of the plan design of the dosimeter means that the design of the radiation treatment plan strongly depends on subjective factors such as previous experience, design thought and the like of the dosimeter, the dosimeter needs to make trial and error in a radiation treatment planning system repeatedly, the clinical goal is finally achieved, and the quality of the radiation treatment plan is directly related to the personal experience of the dosimeter.
Currently, the radiation therapy plan design also has the following problems: firstly, the generation of the radiotherapy plan needs to solve the large-scale dose calculation and optimization problem, the calculation time is long, and a large amount of manpower and material resources are occupied; secondly, the design of the radiation therapy plan requires the participation of the dosimeter in the whole process, and usually, the dosimeter needs to repeatedly try and error to adjust the plan design parameters so as to obtain a plan with higher quality. Also, the quality of the treatment plan depends largely on the personal experience of the dosimeter; thirdly, the learning and training period of the radiation treatment plan design is long, and the treatment plan design level of a dosimeter is difficult to improve in a short time, so that the design quality and consistency of the treatment plan are difficult to guarantee.
Currently, the clinical work of radiotherapy is mainly designed by a dosimeter, and the dosimeter usually takes several hours to design a radiotherapy plan manually according to different disease types. The radiation therapy planning design of a dosimeter needs to complete the steps of positioning information inspection, auxiliary contour generation, plan center point generation, radiation field addition, dose setting and grading, specified dose display line addition, reverse plan optimization and the like, and the reverse plan optimization process needs repeated trial and error of the dosimeter, so that a better result is finally obtained. The process of planning a radiation therapy plan is time-consuming and labor-consuming, the planning quality depends on the subjective experience of a dosimeter, and studies show that the planning result of different dosimeters is greatly different.
Besides the design of artificial radiotherapy plan, there is part of software to realize the auxiliary function of plan design. The methods used by these software are mainly pareto optima and parameter tuning. The multi-objective optimization module of Raystation corporation is based on the pareto optimization method to weigh the weight of each target in the radiation therapy plan design inverse optimization process, and provides multiple radiation therapy plans at the same time, but does not give optimal recommendations. An automatic planning module of the Pinnacle company defines a reverse optimization parameter set in advance, optimizes the CT image and the target region delineation data which need planning design by using the corresponding parameter set, and adjusts the data. Therefore, the existing automatic radiotherapy plan design auxiliary software cannot realize the customized design of a specific patient, cannot perform the customized design according to the experience and the rule of a radiotherapy unit, and cannot ensure that the plan design quality can reach a higher level.
Disclosure of Invention
The invention aims to: in order to solve the existing problems, the invention provides a radiation therapy automatic planning design system and a construction method thereof, the invention predicts the dose based on a deep neural network, and then performs the radiation therapy automatic planning design, all the steps of operation are completed by using a code set of an application program interface instruction of the radiation therapy planning system based on a special radiation therapy planning system, manual operation is not needed, and the automatic planning design can be realized.
The technical scheme adopted by the invention is as follows:
an automatic planning and designing system for radiotherapy comprises a planning and designing auxiliary contour generation module, a prescription setting module, an additional radiation field module, a deep neural network dose prediction module and an optimization objective function generation and planning and designing module;
the plan design auxiliary contour generation module is used for generating an auxiliary contour to carry out target optimization in a reverse optimization stage in order to meet the requirement of clinical radiotherapy plan design;
the prescription setting module is used for carrying out prescription setting in a radiation treatment planning system according to a radiation treatment prescription of an oncologist;
the field adding module is used for adding a field in the radiotherapy planning system;
the deep neural network dose prediction module is used for providing a reasonable dose design target for a reverse optimization process according to data obtained from the same disease species;
the optimization objective function generation and plan design module is used for reverse optimization of radiotherapy plan design and optimization is carried out according to the deep neural network dose prediction module.
Preferably, the deep neural network dose prediction module extracts abstract features from the input data through a plurality of hidden layers and performs prediction by the output layer according to the extracted features. Neural networks are composed of neurons and connections between neurons. The neural network is divided into an input layer, a hidden layer and an output layer. Compared with a shallow neural network, the deep neural network has the advantages that the depth is reflected in the fact that the number of hidden layers is larger, the connection mode is more flexible and complex, stronger nonlinear expression capability is achieved, more essential features can be extracted from an input image, and therefore prediction with higher accuracy is achieved.
Preferably, the deep neural network dose prediction module uses a variety of deep neural network frameworks, including U-Net, DeepLabv3 +.
A construction method based on a radiotherapy automatic planning design system comprises the following steps:
step 1: generating a plan design auxiliary contour;
step 2: setting a prescription;
and step 3: adding the wild;
and 4, step 4: deep neural network dose prediction: the method comprises the steps of data acquisition, model training and application of a trained deep neural network dose prediction model to a prediction object;
and 5: and (4) optimizing the generation of an objective function and planning and designing. Preferably, the deep neural network dose prediction is mainly divided into three steps, which are respectively:
s1: data acquisition: acquiring a radiation therapy plan CT image, a radiation therapy prescription, a target area, organs at risk and dose distribution data of a patient; dividing data into two parts, namely a training set and a verification set; the training set is used for training a deep neural network model, and the verification set is used for quantitatively evaluating the segmentation effect of the trained model;
s2: model training: preprocessing a training set based on a deep neural network dose prediction model, wherein the preprocessing comprises the steps of normalizing, randomly zooming, rotating and translating CT data according to a window level to achieve the purpose of amplifying a training sample, and finally training a deep neural network segmentation model according to the training sample obtained by amplification;
s3: and applying the trained deep neural network dose prediction model to the prediction object to obtain the dose distribution of the prediction object.
Preferably, the data obtained in step S1 is divided into two parts, a training set and a validation set.
Compared with the prior art, the invention has the beneficial effects that:
1) the invention automatically carries out radiotherapy plan design based on the neural network dose model, and the process does not need to manually extract features or set parameters;
2) after the deep neural network model is trained, the dose distribution condition of the radiotherapy patient can be predicted quickly within a few minutes, and the radiotherapy plan design is automatically carried out, so that the working efficiency of a radiotherapy doctor is effectively improved, and the formulation of a radiotherapy scheme of the patient is accelerated.
Drawings
FIG. 1 is a flow chart of the system of the present invention.
Detailed Description
The present invention will be described in further detail in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the detailed description and specific examples, while indicating the preferred embodiment of the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention.
The invention predicts the dose based on the deep neural network, and then carries out the automatic planning design of the radiotherapy, and based on a special radiotherapy planning system, all the steps of operation are completed by using a code set of an application program interface instruction of the radiotherapy planning system, and the automatic planning design can be realized without manual operation.
As shown in fig. 1, an automatic planning and designing system for radiotherapy based on deep neural network dose prediction includes a planning and designing auxiliary contour generation module, a prescription setting module, an additional radiation field module, a deep neural network dose prediction module, and an optimization objective function generation and planning and designing module;
the plan design auxiliary contour generation module is used for generating an auxiliary contour to perform target optimization in a reverse optimization stage in order to meet the requirement of clinical radiotherapy plan design, and the details are shown in table 1;
the prescription setting module is used for carrying out prescription setting in a radiation treatment planning system according to a radiation treatment prescription of an oncologist;
the field adding module is used for adding a field in the radiotherapy planning system;
the deep neural network dose prediction module is used for providing a reasonable dose design target for a reverse optimization process according to data obtained from the same disease species;
the optimization objective function generation and plan design module is used for reverse optimization of radiotherapy plan design and optimization is carried out according to the deep neural network dose prediction module.
TABLE 1 auxiliary outline Generation Table for planning design
The deep neural network dosage prediction mainly comprises three steps, namely:
s1: data acquisition: acquiring a radiation therapy plan CT image, a radiation therapy prescription, a target area, organs at risk and dose distribution data of a patient; dividing data into two parts, namely a training set and a verification set; the training set is used for training a deep neural network model, and the verification set is used for quantitatively evaluating the segmentation effect of the trained model;
s2: model training: preprocessing a training set based on a deep neural network dose prediction model, wherein the preprocessing comprises the steps of normalizing, randomly zooming, rotating and translating CT data according to a window level to achieve the purpose of amplifying a training sample, and finally training a deep neural network segmentation model according to the training sample obtained by amplification;
s3: and applying the trained deep neural network dose prediction model to the prediction object to obtain the dose distribution of the prediction object.
Wherein, the data obtained in step S1 is processed according to the following formula 4:1 is divided into two parts, a training set and a validation set.
The deep neural network dose prediction module extracts abstract features from input data through a plurality of hidden layers and predicts the abstract features according to the extracted features by the output layer. Neural networks are composed of neurons and connections between neurons. The neural network is divided into an input layer, a hidden layer and an output layer. Compared with a shallow neural network, the deep neural network has the advantages that the depth is reflected in the fact that the number of hidden layers is larger, the connection mode is more flexible and complex, stronger nonlinear expression capability is achieved, more essential features can be extracted from an input image, and therefore prediction with higher accuracy is achieved.
Wherein the deep neural network dose prediction module uses a plurality of deep neural network frameworks including U-Net, DeepLabv3 +.
Wherein, the reverse optimization process is divided into two rounds, and the first round is carried out according to the target function in the table 2; under the condition that other conditions are not changed, the second round is set
. The final result can be obtained.
TABLE 2 radiation treatment planning design inverse optimization objective function List
The above-mentioned embodiments only express the specific embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for those skilled in the art, without departing from the technical idea of the present application, several changes and modifications can be made, which are all within the protection scope of the present application.
Claims (6)
1. An automatic planning and designing system for radiotherapy is characterized by comprising a planning and designing auxiliary contour generation module, a prescription setting module, an additional radiation field module, a deep neural network dose prediction module and an optimization objective function generation and planning and designing module;
the plan design auxiliary contour generation module is used for generating an auxiliary contour to carry out target optimization in a reverse optimization stage in order to meet the requirement of clinical radiotherapy plan design;
the prescription setting module is used for carrying out prescription setting in a radiation treatment planning system according to a radiation treatment prescription of an oncologist;
the field adding module is used for adding a field in the radiotherapy planning system;
the deep neural network dose prediction module is used for providing a reasonable dose design target for a reverse optimization process according to data obtained from the same disease species;
the optimization objective function generation and plan design module is used for reverse optimization of radiotherapy plan design and optimization is carried out according to the deep neural network dose prediction module.
2. An automated radiation therapy planning system according to claim 1 and wherein said deep neural network dose prediction module uses a deep neural network framework comprising U-Net or deep labv3 +.
3. The automated planning system for radiation therapy of claim 1, wherein the inverse optimization of the parameters set for radiation therapy is performed using the results of the deep neural network dose prediction module.
4. The construction method based on the radiotherapy automatic planning and designing system is characterized by comprising the following steps of:
step 1: generating a plan design auxiliary contour;
step 2: setting a prescription;
and step 3: adding the wild;
and 4, step 4: deep neural network dose prediction: the method comprises the steps of data acquisition, model training and application of a trained deep neural network dose prediction model to a prediction object;
and 5: and (4) optimizing the generation of an objective function and planning and designing.
5. The method as claimed in claim 4, wherein the deep neural network dose prediction is mainly divided into three steps, which are:
s1: data acquisition: acquiring a radiation therapy plan CT image, a radiation therapy prescription, a target area, organs at risk and dose distribution data of a patient; dividing data into two parts, namely a training set and a verification set; the training set is used for training a deep neural network model, and the verification set is used for quantitatively evaluating the segmentation effect of the trained model;
s2: model training: preprocessing a training set based on a deep neural network dose prediction model, wherein the preprocessing comprises the steps of normalizing, randomly zooming, rotating and translating CT data according to a window level to achieve the purpose of amplifying a training sample, and finally training a deep neural network segmentation model according to the training sample obtained by amplification;
s3: and applying the trained deep neural network dose prediction model to the prediction object to obtain the dose distribution of the prediction object.
6. The method of claim 5, wherein the data obtained in step S1 is divided into training set and verification set at 4: 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110257901.9A CN112635024A (en) | 2021-03-10 | 2021-03-10 | Automatic planning and designing system for radiotherapy and construction method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110257901.9A CN112635024A (en) | 2021-03-10 | 2021-03-10 | Automatic planning and designing system for radiotherapy and construction method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112635024A true CN112635024A (en) | 2021-04-09 |
Family
ID=75297794
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110257901.9A Pending CN112635024A (en) | 2021-03-10 | 2021-03-10 | Automatic planning and designing system for radiotherapy and construction method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112635024A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113517072A (en) * | 2021-09-13 | 2021-10-19 | 四川大学 | Prediction method of VMAT radiotherapy plan based on deep neural network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1438602A (en) * | 2003-03-17 | 2003-08-27 | 吴大可 | Radiotherapy system of automatic computing prescription dosage |
| CN105224822A (en) * | 2015-11-19 | 2016-01-06 | 四川大学华西医院 | Carcinoma of the rectum VMAT based on Pinnacle planning system plans automatic generation method |
| CN108364678A (en) * | 2018-02-10 | 2018-08-03 | 福建省肿瘤医院（福建省肿瘤研究所、福建省癌症防治中心） | A kind of nasopharyngeal carcinoma radiotherapy plan the Automation Design method based on machine learning |
| CN110197709A (en) * | 2019-05-29 | 2019-09-03 | 广州瑞多思医疗科技有限公司 | A kind of 3-dimensional dose prediction technique based on deep learning Yu priori plan |
-
2021
- 2021-03-10 CN CN202110257901.9A patent/CN112635024A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1438602A (en) * | 2003-03-17 | 2003-08-27 | 吴大可 | Radiotherapy system of automatic computing prescription dosage |
| CN105224822A (en) * | 2015-11-19 | 2016-01-06 | 四川大学华西医院 | Carcinoma of the rectum VMAT based on Pinnacle planning system plans automatic generation method |
| CN108364678A (en) * | 2018-02-10 | 2018-08-03 | 福建省肿瘤医院（福建省肿瘤研究所、福建省癌症防治中心） | A kind of nasopharyngeal carcinoma radiotherapy plan the Automation Design method based on machine learning |
| CN110197709A (en) * | 2019-05-29 | 2019-09-03 | 广州瑞多思医疗科技有限公司 | A kind of 3-dimensional dose prediction technique based on deep learning Yu priori plan |
Non-Patent Citations (2)
| Title |
|---|
| 周解平: "基于三维残差与U-Net混合网络的调强放疗自动计划方法研究", 《中国博士学位论文全文数据库 医药卫生科技辑》 * |
| 魏小娜等: "医学影像人工智能辅助诊断的样本增广方法", 《计算机应用》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113517072A (en) * | 2021-09-13 | 2021-10-19 | 四川大学 | Prediction method of VMAT radiotherapy plan based on deep neural network |
| CN113517072B (en) * | 2021-09-13 | 2022-01-28 | 四川大学 | VMAT radiotherapy plan prediction device based on deep neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110197709B (en) | Three-dimensional dose prediction method based on deep learning and priori planning | |
| CN108717866B (en) | Method, device, equipment and storage medium for predicting radiotherapy plan dose distribution | |
| CN110415785A (en) | The method and system of artificial intelligence guidance radiotherapy planning | |
| CN110211664B (en) | System for automatically designing radiotherapy scheme based on machine learning | |
| US8976929B2 (en) | Automatic generation of patient-specific radiation therapy planning parameters | |
| CN107545137A (en) | Tumour radiotherapy plan intelligent optimization method | |
| US20230128148A1 (en) | Standardized Artificial Intelligence Automatic Radiation Therapy Planning Method and System | |
| CN110841205B (en) | Accurate dose verification device for tumor patients | |
| CN112581475B (en) | Method for predicting gamma passing rate of radiotherapy plan | |
| CN109166613A (en) | Radiotherapy treatment planning assessment system and method based on machine learning | |
| US10076673B2 (en) | Interactive dose gradient based optimization technique to control IMRT delivery complexity | |
| CN109771843B (en) | Cloud radiotherapy plan evaluation method and device and storage medium | |
| US11679274B2 (en) | Artificial intelligence modeling for radiation therapy dose distribution analysis | |
| CN110349665A (en) | Carcinoma of the rectum radiotherapy planning the Automation Design method based on deep learning | |
| CN109979564A (en) | A kind of intelligence radiotherapy planning method, equipment and storage medium | |
| CN110404184A (en) | A kind of method and system of measuring and calculating radiotherapy roentgen dose X distribution and dose objective function | |
| CN111462916A (en) | Prediction method and device for radiotherapy plan organ-at-risk dose volume histogram | |
| Lee et al. | Mixed integer programming approaches to treatment planning for brachytherapy–application to permanent prostate implants | |
| CN110021399A (en) | A kind of automatic design method of radiotherapy treatment planning | |
| CN112635024A (en) | Automatic planning and designing system for radiotherapy and construction method thereof | |
| CN109785962B (en) | Method and system for intelligently predicting radiation field parameters in radiotherapy based on learning model | |
| CN116570848B (en) | Radiotherapy clinical auxiliary decision-making system based on automatic sketching | |
| CN109248385B (en) | Radiation therapy plan optimization system based on Monte Carlo tree search | |
| Aleef et al. | Rapid treatment planning for low-dose-rate prostate brachytherapy with TP-GAN | |
| CN113178242B (en) | Automatic plan optimization system based on coupled generation countermeasure network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210409 |