Medium-dose radiotherapy optimization system for subtotal liver diffuse tumor thrombus liver cancer

Abstract

The present invention proposes a medium-dose radiotherapy optimization system suitable for subtotal liver diffuse tumor thrombus liver cancer, the system consists of the following modules: a data acquisition module; an image analysis module; an image processing module; a plan optimization module; the data acquisition module is used to obtain original radiotherapy plan data; the image analysis module is used to obtain original radiation dose distribution data from the original radiotherapy plan image; the image processing module is used to perform image processing on the original medical image to obtain a target medical image; the plan optimization module is used to optimize the original radiation dose distribution data in combination with the target medical image with the help of a plan optimization model to obtain a radiation dose distribution optimization result; based on the system, the radiation dose lines of the radiotherapy plan can be made denser and more reasonably distributed, meeting the requirements of actual application, thereby solving the problem that the international community is currently unable to perform radiotherapy on subtotal liver diffuse tumor thrombus liver cancer through existing radiotherapy plans.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a medium-dose radiotherapy optimization system suitable for subtotal liver diffuse tumor thrombus liver cancer. Background Art

[0002] Radiotherapy is one of the most important cancer treatment methods. For liver cancer that cannot be surgically removed, radiotherapy is the most effective local treatment. The implementation of radiotherapy depends on the radiotherapy plan, that is, the planning of the radiotherapy area and dose distribution. The rationality of the radiotherapy plan is crucial to the treatment effect and safety. At present, the design of radiotherapy plans relies on commercial software, and the radiotherapy plan is optimized by manually adjusting the optimization parameters and structure by physicists.

[0003] However, the existing means of generating radiotherapy plans rely on the experience of physicists and are susceptible to subjective influences, and the efficiency of generating optimized radiotherapy plans is low; when faced with specific cancer situations, it is also impossible to generate corresponding radiotherapy plans in a targeted manner.

[0004] In the face of Figure 1 As shown in the figure, diffuse tumor thrombus liver cancer involving nearly the entire liver (sub-entire liver) currently lacks effective treatment methods, and the commonly used treatment methods have extremely poor efficacy. For example, targeted therapy is prone to cause upper gastrointestinal bleeding and jaundice in such patients, while the efficacy of single-drug immunotherapy is only 10-15%, and interventional therapy is prone to liver failure and the efficacy is less than 20%. Liver cancer is sensitive to radiotherapy, but in the past, patients with such liver volume (normal residual liver area volume, Liver-GTV) less than 700ml and radiotherapy volume less than 300ml below 5Gy could not receive radiotherapy at home and abroad. Radiotherapy is highly effective for diffuse tumor thrombus liver cancer. If a reasonable treatment plan can be designed and optimized and implemented safely, it can significantly improve the effectiveness of treatment and prolong patient survival. However, the radiation dose lines of the radiotherapy plan generated by traditional methods are too sparse, far from meeting the international guidelines for liver radiotherapy dose limits, and cannot be used in actual scenarios. Summary of the invention

[0005] In view of the above-mentioned limitations, the present invention proposes a medium-dose radiotherapy optimization system for liver cancer with diffuse tumor thrombus invading subtotal liver, which is applied to the situation where the normal residual liver volume area (Liver-GTV) is less than 700ml and the volume of radiotherapy below 500cGy is less than 300ml; the radiotherapy plan is optimized through a deep learning algorithm model, and the usability of the radiotherapy plan generated by commercial software for patients with liver cancer with diffuse tumor thrombus invading subtotal liver and left and right lobes of the liver is improved by innovating dose limitation conditions and dose irradiation modes.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A medium-dose radiotherapy optimization system for subtotal liver diffuse tumor thrombus liver cancer, the system comprising the following modules:

[0008] Data acquisition module; image analysis module; image processing module; solution optimization module;

[0009] The data acquisition module is used to acquire the target radiotherapy plan and obtain the original radiotherapy plan data; the original radiotherapy plan data includes the original radiotherapy plan image and the original medical image;

[0010] The image analysis module is used to obtain original radiotherapy dose distribution data from the original radiotherapy plan image, and to perform key area detection from the target medical image to obtain core target area distribution data;

[0011] The image processing module is used to perform image processing on the original medical image to obtain a target medical image;

[0012] The scheme optimization module is used to optimize the original radiotherapy dose distribution data with the help of the plan optimization model in combination with the target medical image, obtain the radiation dose distribution optimization result, and generate the radiotherapy dose scheme.

[0013] The system implements the radiotherapy plan optimization function according to the following steps:

[0014] Step 1: The data acquisition module obtains and reads the original radiation plan data to obtain the original radiation therapy plan data;

[0015] Step 2: The image analysis module performs radiation dose distribution detection on the original radiotherapy plan image to obtain original radiation dose distribution data;

[0016] Step 3: The image processing module performs image enhancement on the original medical image to obtain a target medical image;

[0017] Step 4: The image analysis module performs key area detection on the target medical image to obtain core target area distribution data and normal remaining liver area distribution data;

[0018] The core target area includes the GTV area, the PGTV area and the GTVboost area;

[0019] The core target area distribution data includes GTV distribution data, PGTV distribution data and GTVboost distribution area;

[0020] Step 5: The target medical image, original radiation dose distribution data, normal residual liver area distribution data and core target area distribution data are transmitted to the plan optimization module, and the radiation dose distribution optimization result is obtained with the help of the plan optimization model;

[0021] The planning optimization model is a model obtained by training with a deep learning algorithm.

[0022] Furthermore, the radiation dose distribution optimization result has the following characteristics:

[0023] The volume of the normal remaining liver area irradiated to less than 500 cGy is greater than 100 ml;

[0024] At least 95% of the volume of GTVboost received a total dose of 30 Gy or 35 Gy;

[0025] At least 95% of the GTV volume was irradiated to a total dose of 25 Gy;

[0026] At least 95% of the PGTV volume was irradiated with a total dose product of 20 Gy.

[0027] Compared with the prior art, the present invention has the following advantages:

[0028] Compared with directly using existing commercial software to design radiotherapy plans, the present invention optimizes the radiotherapy plan using a deep learning algorithm model, which can make the radiotherapy dose line of the radiotherapy plan denser and more reasonably distributed, meeting the requirements of actual safe application; and for specific tumor conditions, that is, diffuse tumor thrombus liver cancer invading the left and right lobes of the liver, the method of the present invention can generate a more targeted radiotherapy plan, ensuring that the normal remaining liver area is irradiated with a volume greater than 100 ml below 500 cGy, ensuring that at least 95% of the volume of the core target GTVboost should reach a dose of 30Gy / 6Gy / 5f or 35Gy / 7Gy / 5f, at least 95% of the volume of the GTV should reach a dose of 25Gy / 5Gy / 5f, and at least 95% of the volume of the PGTV should reach a dose of 20Gy / 4Gy / 5f; thereby ensuring that the radiotherapy plan can be safely implemented in clinical practice and safely combined with systemic treatment.

[0029] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the following specifically cites a preferred embodiment and describes it in detail with the accompanying drawings as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A schematic diagram of a subtotal liver diffuse tumor thrombus liver cancer provided by an embodiment of the present invention.

[0031] Figure 2 A structural diagram of a medium-dose radiotherapy optimization system for subtotal liver diffuse tumor thrombus liver cancer provided by an embodiment of the present invention.

[0032] Figure 3 This is an example diagram of an original radiotherapy planning image provided by an embodiment of the present invention.

[0033] Figure 4 A flowchart of a medium-dose radiotherapy optimization system for subtotal liver diffuse tumor thrombus liver cancer provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following is an explanation of the implementation of the present invention by specific specific embodiments. People familiar with the technology can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. In order to further understand the present invention, the present invention is further described in detail below in conjunction with the best embodiment.

[0035] The following are explanations and definitions of terms involved in the embodiments of the present invention:

[0036] GTV: tumor area, which simulates the positioning of the sub-whole liver diffuse tumor thrombus primary lesion and tumor thrombus area shown in CT and MR images.

[0037] PGTV: In order to take into account the tumor planning area for positioning errors and organ movement, PGTV is defined as the area obtained by expanding the GTV in the three-dimensional direction. Specifically, it is an area that is expanded 0.5 cm to 1.5 cm in the head-foot direction according to the respiratory movement amplitude shown on the 4D-CT image, and 0.5 cm in the other directions.

[0038] GTVboost is the hypoxic area within the tumor, which is defined as a 1-cm three-dimensional shrinkage of the GTV or a solid tumor area with strong activity shown on MR images.

[0039] Core target area: including GTV, PGTV and GTVboost areas.

[0040] The implementation of the present invention is described in detail below in conjunction with the aforementioned terms:

[0041] The invention is to provide a medium-dose radiotherapy optimization system suitable for subtotal liver diffuse tumor thrombus liver cancer, referring to Figure 2 , the system consists of the following modules:

[0042] Data acquisition module; image analysis module; image processing module; solution optimization module; system control module.

[0043] The data acquisition module is used to acquire the radiation plan and read the data to obtain original radiation therapy plan data; the original radiation therapy plan data includes original radiation therapy plan images and original medical images.

[0044] The image analysis module is used to obtain original radiation dose distribution data from the original radiation therapy plan image of the original radiation therapy plan data, and to obtain core target area distribution data by performing key area detection from the target medical image;

[0045] The original radiation therapy planning image reference Figure 3 As shown;

[0046] The radiation dose distribution data is composed of a number of radiation therapy doses and corresponding radiation position coordinate data; the radiation therapy dose is the total irradiation dose.

[0047] The image processing module is used to perform image processing on the original medical image to obtain the target medical image.

[0048] The scheme optimization module is used to optimize the original radiation dose distribution data by combining the target medical image with the plan optimization model, obtain the radiation dose distribution optimization result, and generate the radiotherapy dose scheme.

[0049] The system control module is used to call other modules of the system to implement the radiotherapy plan optimization function.

[0050] refer to Figure 4 The system implements the radiotherapy plan optimization function according to the following steps:

[0051] Step 1: The data acquisition module obtains and reads the original radiation plan data to obtain the original radiation therapy plan data;

[0052] Step 2: The image analysis module performs radiation dose distribution detection on the original radiotherapy plan image to obtain original radiation dose distribution data;

[0053] Step 3: The image processing module performs image enhancement on the original medical image to obtain a target medical image;

[0054] Step 4: The image analysis module performs key area detection on the target medical image to obtain core target area distribution data and normal remaining liver area distribution data;

[0055] The core target area includes the GTV area, the PGTV area and the GTVboost area;

[0056] The core target area distribution data includes GTV distribution data, PGTV distribution data and GTVboost distribution area;

[0057] Step 5: The target medical image, original radiation dose distribution data, normal residual liver area distribution data and core target area distribution data are transmitted to the plan optimization module, and the radiation dose distribution optimization result is obtained with the help of the plan optimization model;

[0058] The planning optimization model is a model obtained by training with a deep learning algorithm.

[0059] Furthermore, the radiation dose distribution optimization result has the following characteristics:

[0060] The volume of the normal remaining liver area irradiated to less than 500 cGy is greater than 100 ml;

[0061] At least 95% of the volume of GTVboost received a total dose of 30 Gy or 35 Gy;

[0062] At least 95% of the GTV volume was irradiated to a total dose of 25 Gy;

[0063] At least 95% of the PGTV volume received a total dose product of 20 Gy.

[0064] It can be understood that the planning optimization model will optimize the input data according to the characteristics of the above-mentioned radiation dose distribution optimization results.

[0065] Furthermore, after obtaining the radiation dose distribution optimization result, a radiation therapy dose plan will be generated according to the preset number of irradiation times and the radiation dose distribution optimization result;

[0066] The radiotherapy dosage plan includes a number of radiation position coordinates and the radiotherapy dosage corresponding to the position;

[0067] The radiotherapy dose is composed of three parameters: total irradiation dose, single irradiation dose, and number of irradiations; the format is “total irradiation dose / single irradiation dose / number of irradiations”;

[0068] In the radiotherapy dose plan, the value of the total irradiation dose parameter corresponds to the radiotherapy dose in the radiation dose distribution optimization result;

[0069] The value of the irradiation times parameter is the preset irradiation times;

[0070] The single irradiation dose is the ratio of the total irradiation dose to the number of irradiations.

[0071] The preset number of irradiation times is 5f (f is the unit of irradiation times).

[0072] It is understood that a radiation therapy dose regimen should have the following characteristics:

[0073] The volume of the normal remaining liver area irradiated to less than 500 cGy is greater than 100 ml;

[0074] At least 95% of the volume of GTVboost should reach a dose of 30Gy / 6Gy / 5f or 35Gy / 7Gy / 5f;

[0075] At least 95% of the volume of the GTV should receive a dose of 25Gy / 5Gy / 5f;

[0076] At least 95% of the volume of the PGTV should receive a dose of 20 Gy / 4 Gy / 5f.

[0077] As an embodiment, step 2 specifically includes:

[0078] Step 21, performing image preprocessing on the original radiotherapy plan image to obtain a first image to be identified;

[0079] The image preprocessing specifically includes: performing image filtering on the original radiotherapy planning image; converting the color space of the original radiotherapy planning image into the HSV space;

[0080] Step 22, using a threshold segmentation method to segment the first image to be identified according to preset radiation therapy dose distribution color features to obtain a plurality of dose distribution areas;

[0081] Step 23, using a connected region analysis algorithm to find a continuous dose distribution region;

[0082] Step 24: extract edge information of each dose distribution area with the help of edge detection algorithm, and obtain original radiotherapy dose distribution data after screening all edge information.

[0083] As an embodiment, performing image enhancement in step 3 specifically includes:

[0084] Step 31, obtaining the radiation position corresponding to the minimum radiation therapy dose from the original radiation therapy dose distribution data to obtain the minimum dose area;

[0085] Step 32, intercepting a first key image area in the original medical image; the first key image area is a rectangular area, the first key image area should include the minimum dose area, and the distance between the edge of the first key image area and the minimum dose area is a preset retention length;

[0086] Step 33: Use an image enhancement algorithm to perform image enhancement on the first key image area to obtain a target medical image.

[0087] As an embodiment, the key area detection in step 4 is implemented by the following steps:

[0088] Step 41, identifying the normal residual liver area and the GTV area from the target medical image by means of a target recognition algorithm, and obtaining normal residual liver area distribution data and GTV distribution data;

[0089] Step 42: Expand the GTV area outward to obtain the PGTV area and obtain the PGTV distribution data;

[0090] Step 43: Based on the GTV area shrinking inwards by a preset shrinking distance to the GTVboost area, GTVboost distribution data is obtained.

[0091] Furthermore, the target recognition algorithm is implemented by a target detection method based on deep learning, and can be implemented by any one of the YOLO algorithm, the SSD algorithm, and the Faster-CNN algorithm. The above algorithms are all mature technical solutions, and those skilled in the art can implement them smoothly based on the description of the above embodiments, and will not be repeated here.

[0092] The specific methods of expanding the GTV area to obtain the PGTV area include:

[0093] Step 421, obtaining a 4D-CT image and obtaining a respiratory motion amplitude value therefrom;

[0094] Step 422, expand the GTV area outward in the head-to-foot direction, and the expansion distance is the respiratory movement amplitude value; at the same time, expand outward in other directions except the head-to-foot direction, and the expansion distance is the preset expansion distance to obtain the PGTV area.

[0095] Furthermore, the range of the respiratory movement amplitude value is 0.5-1.5 cm; the preset outward expansion distance is 0.5 cm; and the preset inward contraction distance is 1 cm.

[0096] As an embodiment, the planning optimization model is trained by the following steps:

[0097] Step 501: Obtain a model training data set;

[0098] Step 502: Perform model training using an adversarial training method with the help of a generative adversarial network structure; the generative adversarial network structure is composed of a generator and a discriminator;

[0099] The adversarial training method comprises:

[0100] Step 5021, initializing the generative adversarial network structure;

[0101] Step 5022: fix the parameters of the discriminator and train the generator so that the generator reaches the first training goal;

[0102] Step 5023: fix the parameters of the generator and train the discriminator so that the discriminator reaches the second training goal;

[0103] Step 5024: Evaluate the model performance at this time. If the model performance does not meet the preset model performance condition, return to step 5022;

[0104] Step 503: When the performance of the trained model meets the preset model performance conditions, a trained planning optimization model is obtained.

[0105] Furthermore, the model training data set is composed of a number of image optimization data pairs; the image optimization data pairs are the radiological image data to be optimized and the corresponding optimized radiological image;

[0106] The radiological image data to be optimized includes radiological medical images and corresponding radiation dose distribution data;

[0107] The radiation dose distribution data is obtained through the method of steps 21-24 based on the radiation therapy plan data corresponding to the radiation medical image.

[0108] Furthermore, the generator of the generative adversarial network structure consists of an encoding unit and a decoding unit.

[0109] The encoding unit is composed of the following structures connected in series: a first encoding convolution layer, a second encoding convolution layer, a first encoding maximum pooling layer, a third encoding convolution layer, a fourth encoding convolution layer, a second encoding maximum pooling layer, a fifth encoding convolution layer, a sixth encoding convolution layer, a third encoding maximum pooling layer, a seventh encoding convolution layer, an eighth encoding convolution layer, and a fourth encoding maximum pooling layer.

[0110] The input data size of the encoding unit is [H, W, C1+C2], where H and W are the height and width of the radiological medical image, respectively; C1 is the number of channels of the radiological medical image, and C2 is the number of channels of the radiation dose distribution data.

[0111] The first coding convolution layer, the second coding convolution layer, the third coding convolution layer, the fourth coding convolution layer, the fifth coding convolution layer, the sixth coding convolution layer, the seventh coding convolution layer, and the eighth coding convolution layer are all 3*3 convolution layers.

[0112] The first coded maximum pooling layer, the second coded maximum pooling layer, the third coded maximum pooling layer, and the fourth coded maximum pooling layer are all MaxPool layers with a pooling size of 2*2.

[0113] The decoding unit is composed of the following structure: a first deconvolution layer, a first merging layer, a first decoding convolution layer, a second decoding convolution layer, a first maximum anti-pooling layer, a second deconvolution layer, a second merging layer, a third decoding convolution layer, a fourth decoding convolution layer, a second maximum anti-pooling layer, a third deconvolution layer, a third merging layer, a fifth decoding convolution layer, a sixth decoding convolution layer, a third maximum anti-pooling layer, a seventh decoding convolution layer, and an eighth decoding convolution layer.

[0114] Among them, the input of the first deconvolution layer is the output result of the encoding unit, the input of the first merging layer is the first deconvolution layer and the eighth encoding convolution layer; the input of the first decoding convolution layer is the output of the first merging layer; the first decoding convolution layer, the second decoding convolution layer, the first maximum anti-pooling layer and the second deconvolution layer are connected in series;

[0115] The input of the second merging layer is the second deconvolution layer and the sixth encoding convolution layer; the input of the third decoding convolution layer is the output of the second merging layer; the third decoding convolution layer, the fourth decoding convolution layer, the second maximum anti-pooling layer and the third deconvolution layer are connected in series;

[0116] The input of the third merging layer is the third deconvolution layer and the fourth encoding convolution layer; the input of the fifth decoding convolution layer is the output of the third merging layer; the fifth decoding convolution layer, the sixth decoding convolution layer, the third maximum depooling layer, the seventh decoding convolution layer, and the eighth decoding convolution layer are connected in series.

[0117] The first deconvolution layer, the second deconvolution layer, and the third deconvolution layer are all 3*3 deconvolution layers;

[0118] The first decoding convolution layer, the second decoding convolution layer, the third decoding convolution layer, the fourth decoding convolution layer, the fifth decoding convolution layer, the sixth decoding convolution layer, the seventh decoding convolution layer, and the eighth decoding convolution layer are all 3*3 convolution layers;

[0119] The first maximum anti-pooling layer, the second maximum anti-pooling layer, and the third maximum anti-pooling layer are all 2*2 maximum anti-pooling layers.

[0120] Furthermore, the first training objective is to minimize the value of the mean square error loss function; and the second training objective is to minimize the value of the cross entropy loss function.

[0121] As an embodiment, the system further includes a system optimization module, which is used to evaluate the radiation dose distribution optimization result in combination with the actual radiation therapy plan and optimize the radiation therapy plan optimization function.

[0122] The system optimization module optimizes the radiotherapy plan optimization function including:

[0123] Step 61, obtaining an actual radiation therapy plan;

[0124] Step 62, comparing the difference in radiation dose distribution data between the actual radiation therapy plan and the corresponding radiation dose distribution optimization result, and calculating the similarity of the results;

[0125] Step 63: If the result similarity is greater than a preset result similarity threshold, the radiation dose distribution optimization result is considered valid; otherwise, the radiation dose distribution optimization result is invalid;

[0126] Step 64, counting the invalid ratio of the radiation dose distribution optimization results, and calculating the optimization result failure rate;

[0127] Step 65: When the failure rate of the optimization result is greater than a preset optimization failure threshold, the actual radiotherapy plan is used to retrain the plan optimization model.

[0128] As an embodiment, the method of the present invention may be implemented in software and / or a combination of software and hardware, for example, may be implemented using an application specific integrated circuit (ASIC), a general purpose computer or any other similar hardware device.

[0129] The method described in the present invention can be implemented in the form of a software program, and the software program can be executed by a processor to implement the steps or functions described above. Similarly, the software program (including related data structures) can be stored in a computer-readable recording medium, such as a RAM memory, a magnetic or optical drive or a floppy disk and the like.

[0130] In addition, some steps or functions of the method described in the present invention may be implemented by hardware, for example, as a circuit that cooperates with a processor to execute each step or function.

[0131] In addition, a part of the method described in the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present application through the operation of the computer. The program instruction for calling the method described in the present invention may be stored in a fixed or removable recording medium, and / or transmitted through a data stream in a broadcast or other signal-bearing medium, and / or stored in a working memory of a computer device that runs according to the program instruction.

[0132] As an embodiment, the present invention also provides a device, which includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to run the methods and / or technical solutions based on the aforementioned multiple embodiments.

[0133] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present application.

[0134] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0135] In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0136] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with this profession can make some changes or modifications to equivalent embodiments of equivalent changes using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of the technical solution of the present invention.

Claims

1. Applicable to the medium-dose radiotherapy optimization system for subtotal liver diffuse tumor thrombus liver cancer, characterized by: The system consists of the following modules: Data acquisition module; image analysis module; image processing module; solution optimization module; The data acquisition module is used to acquire the target radiotherapy plan and obtain the original radiotherapy plan data; the original radiotherapy plan data includes the original radiotherapy plan image and the original medical image; The image analysis module is used to obtain original radiotherapy dose distribution data from the original radiotherapy plan image, and to perform key area detection from the target medical image to obtain core target area distribution data; The image processing module is used to perform image processing on the original medical image to obtain a target medical image; The scheme optimization module is used to optimize the original radiotherapy dose distribution data by combining the plan optimization model with the target medical image, obtain the radiation dose distribution optimization result, and generate the radiotherapy dose scheme; The radiation therapy dose regimen should have the following characteristics: The volume of the normal remaining liver area irradiated to less than 500 cGy is greater than 100 ml; At least 95% of the volume of GTVboost should reach a dose of 30Gy / 6Gy / 5f or 35Gy / 7Gy / 5f; At least 95% of the volume of the GTV should receive a dose of 25Gy / 5Gy / 5f; At least 95% of the volume of the PGTV should receive a dose of 20 Gy / 4 Gy / 5f; The system realizes the function of optimizing the radiotherapy plan, including the following steps: the image analysis module detects the radiation dose distribution of the original radiotherapy plan image to obtain the original radiation dose distribution data; specifically, the following steps are included: Step 21, performing image preprocessing on the original radiotherapy plan image to obtain a first image to be identified; The image preprocessing specifically includes: performing image filtering on the original radiotherapy planning image; converting the color space of the original radiotherapy planning image into the HSV space; Step 22, using a threshold segmentation method to segment the first image to be identified according to preset radiation dose distribution color features to obtain a plurality of dose distribution areas; Step 23, using a connected region analysis algorithm to find a continuous dose distribution region; Step 24: extract edge information of each dose distribution area with the help of edge detection algorithm, and obtain original radiation dose distribution data after screening all edge information.

2. The system according to claim 1, characterized in that The system implements the radiotherapy plan optimization function according to the following steps: Step 1: The data acquisition module obtains and reads the original radiotherapy plan data to obtain the original radiotherapy plan data; Step 2: The image analysis module performs radiation dose distribution detection on the original radiotherapy plan image to obtain original radiation dose distribution data; Step 3: The image processing module performs image enhancement on the original medical image to obtain a target medical image; Step 4: The image analysis module performs key area detection on the target medical image to obtain core target area distribution data and normal remaining liver area distribution data; The core target area includes the GTV area, the PGTV area and the GTVboost area; the core target area distribution data includes the GTV distribution data, the PGTV distribution data and the GTVboost distribution area; Step 5: The target medical image and the original radiation dose distribution data are transferred to the plan optimization module, and the radiation dose distribution optimization result is obtained with the help of the plan optimization model; The planning optimization model is a model obtained by training with a deep learning algorithm.

3. The system according to claim 2, characterized in that Image enhancement in step 3 specifically includes: Step 31, obtaining the position corresponding to the minimum radiation dose from the original radiation therapy dose distribution data to obtain the minimum dose area; Step 32: intercepting a first key image region in the original medical image; Step 33: Use an image enhancement algorithm to perform image enhancement on the first key image area to obtain a target medical image.

4. The system according to claim 3, characterized in that The first key image area is a rectangular area, the first key image area should include the minimum dose area, and the distance between the edge of the first key image area and the minimum dose area is a preset retention length.

5. The system according to claim 2, characterized in that Step 4 specifically includes: Step 41, identifying the GTV region from the target medical image with the aid of a target recognition algorithm, and obtaining GTV distribution data; Step 42: Expand the GTV area outward to obtain the PGTV area and obtain the PGTV distribution data; Step 43: Based on the GTV area shrinking inward by a preset shrinking distance, a GTVboost area is obtained to obtain GTVboost distribution data; wherein the preset shrinking distance is 1 cm.

6. The system according to claim 1, characterized in that The planning optimization model is trained by the following steps: Step 501: Obtain a model training data set; Step 502: Perform model training using an adversarial training method with the help of a generative adversarial network structure; Step 503: When the performance of the trained model meets the preset model performance conditions, a trained planning optimization model is obtained.

7. The system according to claim 6, characterized in that The model training data set is composed of a number of image optimization data pairs; the image optimization data pairs are the radiological image data to be optimized and the corresponding optimized radiotherapy images; The radiological image data to be optimized include radiotherapy medical images and corresponding radiotherapy dose distribution data; The radiotherapy dose distribution data is obtained through the method of steps 21-24 based on the radiotherapy plan data corresponding to the radiotherapy medical image.

8. The system according to claim 6, characterized in that The generative adversarial network structure consists of a generator and a discriminator; The generator of the generative adversarial network structure consists of an encoding unit and a decoding unit.

9. The system according to claim 1, characterized in that The system further comprises a system optimization module, which is used to evaluate the radiation dose distribution optimization result in combination with the actual radiation therapy plan and optimize the radiation therapy plan optimization function.

Images