Tumor extraction method and related device

Abstract

The invention discloses a tumor extraction method and a related device, and the method comprises the steps: obtaining a reference image and a real-time motion feature of a patient, and the real-time motion feature is an image of motion fed back through a signal; utilizing a trained feature extraction model to extract boundary features of the reference image, and extracting boundary features of the real-time motion features; and taking the reference image and the boundary feature of the real-time motion feature as a reference, and extracting the feature contour of the real-time motion feature, the feature contour comprising at least one of a tumor and a normal tissue. According to the embodiment of the invention, the technical problem that tumor tracking is not timely in the prior art is solved.

Description

Technical Field

[0001] The present application relates to a tumor extraction method and related devices, belonging to the field of tumor monitoring and radiotherapy. Background Art

[0002] During radiotherapy, human tissues and organs undergo unpredictable deformation and displacement with the patient's heartbeat and breathing. The original features cannot truly reflect the changes in the tumor, and there will be certain differences before and after positioning and repositioning. If the onboard image features are used to reflect the movement pattern, there will be a large delay in image acquisition and fusion registration, which may lead to untimely tracking, thus affecting the relevant treatment effect and prognosis. Summary of the invention

[0003] In view of this, the present application provides a tumor extraction method and related devices, and the embodiments of the present application at least solve the technical problem of untimely tumor tracking in the related art.

[0004] The first aspect of an embodiment of the present application discloses a tumor extraction method, which includes: obtaining a reference image and real-time motion features of a patient; using a trained feature extraction model to extract boundary features of the reference image and boundary features of the real-time motion features; using the reference image and the boundary features of the real-time motion features as a reference, extracting a feature contour of the real-time motion features, wherein the feature contour includes at least one of: a tumor and normal tissue.

[0005] In some embodiments, the extracting the feature contour of the real-time motion feature with reference to the boundary features of the reference image and the real-time motion feature includes: registering the reference image and the real-time motion feature based on the boundary features in a rigid or elastic manner to obtain a registration result; optimizing the registration result based on a POWELL registration algorithm; pre-outlining a first position contour on the reference image, and outlining a second position contour on the real-time motion feature based on the registration result.

[0006] In some embodiments, the reference image includes one of: a positioning image; an image obtained by combining the positioning image with the corresponding first motion feature; a reset image; an image obtained by combining the reset image with the corresponding second motion feature; wherein the reference image and the real-time motion feature are in the same phase.

[0007] In some embodiments, the method further includes: when the patient's carrier moves to the positioning position, acquiring a positioning image including boundary features; when the patient's carrier moves to the reset position, acquiring a reset image including boundary features; determining whether the boundary features in the positioning image and the reset image are consistent; in response to system confirmation that the boundary features are inconsistent, and / or in response to manual confirmation that the boundary features are inconsistent, executing a motion control optimization strategy, wherein the motion control optimization strategy is to adjust the beam of the carrier and / or treatment device.

[0008] In some embodiments, the execution of the motion control optimization strategy includes: extracting a first contour boundary of the positioning image and a second contour boundary of the reset image, determining a reference point based on the first contour boundary, determining a matching point based on the second contour boundary, and the matching point is a point with the highest similarity to the reference point; determining the shortest distance between the reference point and the matching point, determining an adjustment direction of the vehicle based on the shortest distance, and executing movement along the adjustment direction.

[0009] In some embodiments, determining the reference point based on the first contour boundary and determining the matching point based on the second contour boundary include: detecting a plurality of first key points of the first contour boundary and a plurality of second key points of the second contour boundary; generating a first feature descriptor based on the first key point, the first feature descriptor being used to represent image information around the first key point; generating a second feature descriptor based on the second key point, the second feature descriptor being used to represent image information around the second key point; determining the best matching first feature descriptor and second feature descriptor according to a K nearest neighbor matching algorithm, using the first key point of the best matching first feature descriptor as a reference point, and using the second key point of the best matching second feature descriptor as a matching point.

[0010] In some embodiments, the real-time motion feature is an image of the motion fed back via a signal.

[0011] The second aspect of an embodiment of the present application discloses a tumor extraction device, which includes: an acquisition module, used to acquire a reference image and real-time motion features of a patient; a first extraction module, used to extract boundary features of the reference image and the boundary features of the real-time motion features using a trained feature extraction model; and a second extraction module, used to extract a feature contour of the real-time motion features with reference to the reference image and the boundary features of the real-time motion features, wherein the feature contour includes at least one of: a tumor and normal tissue.

[0012] The third aspect of an embodiment of the present application discloses a radiotherapy device, including a tumor extraction device, a patient information management module, a data transceiver module and a radiotherapy beam module of the above-mentioned embodiment, wherein the patient information management module is used to record patient information, the data transceiver module is used to receive and send image data, and the radiotherapy beam module is used to dynamically adjust the beam irradiated on the patient based on the movement of the tumor.

[0013] A fourth aspect of the embodiments of the present application discloses a computer-readable storage medium, which includes a stored program, wherein when the program is run, the processor of the device where the program is located is controlled to execute the tumor extraction method of the above embodiment.

[0014] The fifth aspect of an embodiment of the present application discloses a server, which includes: one or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors execute the tumor extraction method of the above embodiment.

[0015] Compared with the prior art, the embodiments of the present application have the following beneficial effects: The embodiments of the present application provide a tumor extraction method and related devices. The tumor extraction method includes: obtaining a reference image and real-time motion features of a patient, wherein the real-time motion features are images that move through signal feedback; using a trained feature extraction model to extract boundary features of the reference image and the boundary features of the real-time motion features; using the reference image and the boundary features of the real-time motion features as references, extracting feature contours of the real-time motion features, wherein the feature contours include at least one of: tumors and normal tissues. This embodiment solves the problem of untimely tumor tracking in the prior art by establishing a new feature relationship related to tumor motion and using a motion monitoring device with a higher feedback speed instead of a real-time imaging device with a slower feedback speed to accurately simulate the movement of the tumor. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0017] Figure 1 A schematic diagram of a tumor extraction method provided in an embodiment of the present application.

[0018] Figure 2 A flowchart of a delineation operation provided in an embodiment of the present application.

[0019] Figure 3 A schematic diagram of a process of carrier adjustment provided in an embodiment of the present application.

[0020] Figure 4 A flowchart of a point selection operation provided in an embodiment of the present application.

[0021] Figure 5 A schematic diagram of the structure of a tumor extraction device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only 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 this field without creative work should fall within the scope of protection of the present invention.

[0023] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] When managing organ movement during radiotherapy, the movement relationship of relevant feature information (including markers, body surface, gold dots, inhalation and exhalation volume, etc.) is usually obtained through external instruments during positioning to track and simulate the movement of organs, and CT images of different phases are collected accordingly. During repositioning treatment, the same feature information and phase as positioning are first collected, and then treatment is performed according to the time point selected by the planning system. However, during the positioning and repositioning process, system errors, manual errors, and changes in the patient's body state may cause large changes in the feature values ​​of markers, body surface, and respiratory volume, and they cannot be accurately matched. Secondly, the human body structure is non-rigid, the movement states of different tissues and organs are not synchronized, and the movement, heartbeat, and breathing of tumors are also complex. These factors do not follow strict periodicity and linearity. This will cause the tissues and organs of the human body to undergo unpredictable deformation and displacement with the patient's heartbeat and breathing during radiotherapy. The original features cannot truly reflect the changes in the tumor, and there will be certain differences before and after positioning and repositioning. If image features are used to reflect the law of movement, image acquisition and fusion registration will have a large delay, which may lead to untimely tracking, thereby affecting the relevant treatment effect and prognosis.

[0025] In view of the above problems, this embodiment provides the following solutions: Embodiment 1:

[0026] Figure 1 A schematic diagram of a tumor extraction method provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method may include the following steps: 101 obtains reference images and real-time motion characteristics of the patient.

[0027] 102 uses the trained feature extraction model to extract boundary features of the reference image and to extract boundary features of the real-time motion features.

[0028] 103 extracts a feature contour of the real-time motion feature by taking the reference image and the boundary feature of the real-time motion feature as reference.

[0029] It is understandable that the reference image and the real-time motion feature include the target tumor area. The real-time motion feature can be in the form of a motion curve or a morphological image. The feature contour includes at least one of: a tumor and normal tissue. It should be noted that when 103 is extracted, the tumor and / or normal tissue near the tumor may be extracted to show the feature contour to be extracted.

[0030] It should be noted that the boundary features of the reference image are mainly related to the feature morphology, and the boundary features of the real-time motion features may be related to the feature morphology or to the periodic feature points of the motion curve.

[0031] In this embodiment, the accelerator control system moves the vehicle carrying the patient to the predetermined treatment position according to the coordinate values ​​provided by the planning system. Once the treatment position is reached, the motion monitoring device records the real-time motion characteristics of the target tumor area. The reference image is a non-real-time image, which can be a positioning image, a resetting image, or an image from a previous treatment.

[0032] It is worth noting that this embodiment accurately simulates the movement of the tumor by establishing a new characteristic relationship related to the movement of the tumor and using a motion monitoring device with a higher feedback speed instead of a real-time imaging device with a slower feedback speed.

[0033] In one embodiment, the server obtains a reference image and real-time motion features of the patient, and uses a trained feature extraction model to extract boundary features of the reference image and boundary features of the real-time motion features. The server uses the reference image and the boundary features of the real-time motion features as a reference to extract feature contours of the real-time motion features and displays them on a medical display device.

[0034] In one embodiment, the server obtains a reference image and real-time motion features of the patient, and uses a trained feature extraction model to extract boundary features of the reference image and boundary features of the real-time motion features. The doctor uses a medical display device to extract and display the feature contours of the real-time motion features with reference to the reference image and the boundary features of the real-time motion features.

[0035] The above-mentioned server also includes an uninterruptible voltage-stabilized power supply, which is used to prevent data loss caused by abnormal shutdown of computer equipment in the event of an emergency power outage.

[0036] In one embodiment, the imaging module obtains a reference image and real-time motion features of the patient. The imaging module uses a trained feature extraction model to extract boundary features of the reference image and boundary features of the real-time motion features. The imaging module uses the reference image and the boundary features of the real-time motion features as a reference to extract feature contours of the real-time motion features and outputs them to a motion monitoring device and / or an accelerator control system.

[0037] The above-mentioned imaging module is applied to a sports management system, which also includes a patient information management module and a data transceiver module. The patient information management module is used for recording patient information, and the data transceiver module is used for receiving and sending imaging data and plan data.

[0038] In one embodiment, before 102, if the reference image is a low-resolution image, the low-resolution image is converted into a high-resolution image based on the trained SRGAN model; the reference image is enhanced based on the CLAHE algorithm and the Retinex algorithm. The same is true for the real-time motion feature.

[0039] In a possible embodiment, the reference image is a real-time image captured by an onboard imaging device.

[0040] The above-mentioned carrier is used to support and position the patient to ensure that the patient maintains a stable position during treatment, such as a treatment bed or a bracket.

[0041] The accelerator control system mentioned above is responsible for controlling the operation of the particle accelerator, ensuring that the particles are accelerated with the correct energy and direction. In addition, the controller in which it is located controls the movement of the vehicle to ensure that the patient's position matches the planning system.

[0042] The above-mentioned airborne imaging equipment can be a device for obtaining 2D or 3D images such as CT, CBCT, KV, MV, MR, PET, etc., which is used to obtain images of patients and help doctors determine the location of tumors.

[0043] In some embodiments, the above-mentioned motion monitoring equipment can be a multi-modal device that can provide rapid feedback through signals, such as ultrasound, body surface optical information, breath-hold system, marker monitoring, abdominal pressure sensor monitoring, etc., and is not limited to a single device.

[0044] The above-mentioned planning system is an important auxiliary device used to design and calculate dose distribution in modern radiotherapy. It uses a dedicated computer to input the patient's body cross-section image collected on the image together with the physical parameters of radiotherapy (such as energy, irradiation field size, irradiation distance, various corrections, wedge plates, tissue compensation, etc.), and displays the isodose distribution curve after processing, and then adjusts the physical conditions until the optimal dose distribution is obtained.

[0045] Specifically, the main functions of the planning system include: Image Processing: Ability to import and process cross-sectional images of patient bodies from different imaging techniques.

[0046] Dose calculation: Based on the input treatment parameters, the absorbed dose distribution in the patient is calculated and optimized to achieve the treatment goals.

[0047] Treatment plan design: Design a variety of radiotherapy plans including three-dimensional radiotherapy (3D-CRT) and intensity-modulated radiotherapy (IMRT).

[0048] Image fusion registration: Fusion of images of different modalities.

[0049] The above-mentioned feature extraction model can be a segmentation neural network based on deep learning, or a detection neural network, or a combination of the two, which is not limited here. Segmentation neural network: used to accurately segment various regions and provide the contour and shape information of the tumor. For example, Transformer, U-Net or CNN. Detection neural network: monitors the position changes of the tumor over time, especially in dynamic images (such as videos or time series images). The detection network can track the movement and changes of the tumor. For example, LSTM and YOLO. Exemplarily, the image segmentation network model for feature extraction on the image is trained using feature annotations such as tissue information, bone information, vertebral information and nerve information in the pre-collected CT images.

[0050] It is worth noting that the present embodiment can use the trained first feature extraction model to extract the reference image and the real-time motion feature, or can use the trained second feature extraction model and the third feature extraction model to extract the reference image and the real-time motion feature respectively, which is not limited here. In addition, the real-time motion feature can be simulated in the form of a curve.

[0051] In one embodiment, extracting the feature contour of the real-time motion feature with reference to the boundary features of the reference image and the real-time motion feature may include: a doctor compares a first boundary feature of the reference image with a second boundary feature of the real-time motion feature through a medical display device, draws a first contour of the first boundary feature, draws a second contour of the second boundary feature, and then corrects the first contour according to the second contour, to finally obtain the feature contour of the real-time motion feature.

[0052] In some embodiments, Figure 2 As shown, extracting the feature contour of the real-time motion feature by taking the reference image and the boundary feature of the real-time motion feature as reference may include: 201 aligns the reference image and the real-time motion feature based on the boundary feature in a rigid or elastic registration manner to obtain a registration result.

[0053] In this step, the registration may be automatically performed based on the internal markers of the bed, or may be automatically performed based on the fusion registration of the patient's images, which is not limited here.

[0054] 202 Optimize the registration result based on the POWELL registration algorithm.

[0055] In this step, the algorithm optimizes the registration result by iteratively searching for the optimal parameters to minimize the registration error.

[0056] 203 pre-outlines a first position contour on the reference image, and outlines a second position contour on the real-time motion feature based on the registration result.

[0057] In this step, contour drawing can be completed based on contour drawing software.

[0058] In some embodiments, the reference image may include one of: 1) Position the image.

[0059] 2) An image obtained by combining the positioning image with the corresponding first motion feature.

[0060] 3) Reset the image.

[0061] 4) An image obtained by combining the reset image with the corresponding second motion feature.

[0062] The reference image and the real-time motion feature are in the same phase. For example, the abdominal tumor CT image at the end of expiration of the patient's respiratory cycle is called, and the B-ultrasound image of the corresponding cycle is acquired in real time.

[0063] Exemplarily, the real-time motion feature and the first motion feature or the second motion feature may be of the same type or may be different in type, and are preferably morphological feature images.

[0064] Exemplarily, the above combination is multimodal fusion.

[0065] In some embodiments, the method may further include: when the patient's carrier moves to the positioning position, acquiring a positioning image including boundary features; when the patient's carrier moves to the reset position, acquiring a reset image including boundary features; determining whether the boundary features in the positioning image and the reset image are consistent; in response to system confirmation that the boundary features are inconsistent, and / or in response to manual confirmation that the boundary features are inconsistent, executing a motion control optimization strategy, wherein the motion control optimization strategy is to adjust the beam of the carrier and / or treatment device.

[0066] In this embodiment, the positioning image and the resetting image contain delineation information, which includes delineation of the normal tissue area near the tumor and the region of interest of the specific organ area. The system achieves consistency evaluation by calculating the similarity or difference between features, such as using correlation analysis in image processing, difference measurement (such as Euclidean distance or structural similarity index), etc. If the result of the automatic evaluation is uncertain or needs further verification, it can be manually confirmed by the doctor to determine whether the boundary information is consistent.

[0067] Furthermore, the positioning image and the resetting image are obtained before the reference image and the real-time motion feature are obtained.

[0068] Alternatively, in response to the system confirming that the boundary information is inconsistent, and / or in response to manual confirmation that the boundary information is inconsistent, the treatment plan and the motion control strategy can be adjusted. Adjusting the treatment plan includes: using the originally planned dose to load onto the onboard imaging device to obtain the patient's real-time dose distribution. Based on this dose distribution and the doctor's opinion, the treatment plan can be automatically adjusted, or the original plan can be directly executed. The motion control strategy is the adjustment of the motion monitoring device.

[0069] Optionally, after positioning is completed, CT scans of different phases, delineation information, treatment plans, doses, and phases during treatment are obtained from the planning system.

[0070] Optionally, during resetting, the onboard imaging device is used to obtain real-time images and images of different phases, and the motion monitoring device is used to collect new characteristic images. The system will receive these image data and characteristic images at the same time.

[0071] The above-mentioned delineation information is automatically, semi-automatically or fully manually identified and marked in the medical image data.

[0072] The above motion control optimization strategy is used to make adjustments based on the boundary information difference to minimize the difference, thereby achieving accurate positioning of the patient's radiotherapy position.

[0073] The above-mentioned carrier can move in any direction on the set surface along the X-axis and the Y-axis.

[0074] Furthermore, if Figure 3 As shown, the execution of the motion control optimization strategy may include: 301 extracting the first contour boundary of the positioning image and the second contour boundary of the reset image, determining a reference point based on the first contour boundary, and determining a matching point based on the second contour boundary, wherein the matching point is a point with the highest similarity to the reference point; 302 determining the shortest distance between the reference point and the matching point, determining an adjustment direction of the vehicle based on the shortest distance, and executing movement along the adjustment direction.

[0075] In some embodiments, determining reference points based on the first contour boundary and determining matching points based on the second contour boundary may include: determining reference points and matching points in response to a user's selection operation based on the first contour boundary and the second contour boundary, and the same applies to the extraction operation.

[0076] In other embodiments, Figure 4As shown, the determining of reference points based on the first contour boundary and the determining of matching points based on the second contour boundary may include: 401 detecting a plurality of first key points of the first contour boundary and a plurality of second key points of the second contour boundary; 402 generating a first feature descriptor based on the first key point, the first feature descriptor being used to represent image information around the first key point; 403 generating a second feature descriptor based on the second key point, the second feature descriptor being used to represent image information around the second key point; 404 determining the best matching first feature descriptor and second feature descriptor according to a K nearest neighbor matching algorithm, taking the first key point of the best matching first feature descriptor as a reference point, and taking the second key point of the best matching second feature descriptor as a matching point.

[0077] The above-mentioned key points refer to points with obvious features, such as edges and corners, which can be obtained by algorithms based on traditional computer vision (such as Harris corner detection) or by algorithms based on deep learning (such as the Transformer model), and are not limited here.

[0078] The above-mentioned feature descriptor describes the image information in the local area where the key point is located, which may include gradient and texture information, and may be obtained by the SIFT algorithm or the SURF algorithm, which is not limited here.

[0079] The K-nearest neighbor (KNN) matching algorithm described above can be based on a simple Euclidean distance or a more complex distance metric, such as the Mahalanobis distance, which takes into account the covariance structure of the data. The KNN algorithm predicts the label of the query point based on the labels of these neighbors by finding the K training samples closest to the query point.

[0080] As an optional implementation, the KNN algorithm is used to find the most similar K matches for the feature descriptor in the first image (set as des1) in the feature descriptor in the second image (set as des2). Usually, the K value is set to 2, that is, the two closest matches for each feature descriptor are found. In order to screen out good matches, a ratio test is used. Specifically, for each match, the ratio of the nearest neighbor distance to the next nearest neighbor distance is calculated. If this ratio is less than a preset threshold (usually less than 1), the match is considered valid.

[0081] Embodiment 2:

[0082] Figure 5 A schematic diagram of the structure of a tumor extraction device provided in an embodiment of the present application is shown in FIG. Figure 5 As shown, the device may include the following modules: The acquisition module 501 is used to acquire the patient's reference image and real-time motion characteristics.

[0083] The first extraction module 502 is used to extract the boundary features of the reference image and the boundary features of the real-time motion features by using the trained feature extraction model.

[0084] The second extraction module 503 is used to extract the feature contour of the real-time motion feature by taking the reference image and the boundary feature of the real-time motion feature as reference.

[0085] In some embodiments, the radiotherapy device includes the tumor extraction device of the above embodiment and a radiotherapy beam module, wherein the radiotherapy beam module is used to dynamically adjust the beam irradiated on the patient based on the movement of the tumor.

[0086] It is worth noting that this embodiment improves the accuracy of irradiation, increases the irradiation dose to the target area, reduces the irradiation dose to normal tissue, and improves the treatment effect by establishing a new characteristic relationship related to tumor motion and using a motion monitoring device with a higher feedback speed instead of a real-time imaging device with a slower feedback speed.

[0087] Embodiment 3:

[0088] An embodiment of the present application further provides a server, comprising: a memory storing an executable program; and a processor for running the program, wherein the method in each embodiment of the present invention is executed when the program is running.

[0089] The above-mentioned memory may refer to a device inside a computer for storing data and programs, and may include memory, hard disk, etc., wherein the memory may be used to temporarily store running programs and data, the hard disk may be used to store programs and data for a long time, and the memory may be used to enable the computer to read and write data, and execute programs; the above-mentioned processor may be responsible for executing instructions in computer programs and performing data processing, and may be responsible for controlling and executing various operations, including arithmetic operations, logical operations, data transmission, etc.

[0090] Embodiment 4:

[0091] An embodiment of the present application further provides a computer-readable storage medium, which includes a stored executable program, wherein when the executable program is running, the device where the computer-readable storage medium is located is controlled to execute the methods in various embodiments of the present invention.

[0092] The above-mentioned computer storage medium may refer to a medium in a computer memory used to store certain discontinuous physical quantities. Computer storage media mainly include semiconductors, magnetic cores, magnetic drums, magnetic tapes, laser disks, etc.; the stored program included in the computer-readable storage medium may be a set of instructions that can be recognized and executed by a computer, running on an electronic computer, and is an information tool that meets certain needs of people.

[0093] Embodiment 5:

[0094] An embodiment of the present application further provides a computer program product, including a computer program, which implements the methods in various embodiments of the present invention when executed by a processor.

[0095] The above-mentioned computer program product may refer to a software program that has been written, tested and released and can be run on a computer or other device. The computer program product may include an application, an operating system, tool software, etc., which is used to implement specific functions or solve specific problems.

[0096] Embodiment 6:

[0097] An embodiment of the present application further provides a computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium is used to store a computer program, and when the computer program is executed by a processor, the method in each embodiment of the present invention is implemented.

[0098] The above-mentioned non-volatile computer-readable storage medium may refer to a medium for storing data. The non-volatile computer-readable storage medium can keep the data from being lost when the power is off, and can be used to store long-term data, such as operating systems, applications, and user files. The non-volatile storage medium may include hard disk drives, solid-state drives, optical disks, and flash memory storage devices, etc.

[0099] Embodiment 7:

[0100] The embodiments of the present application further provide a computer program, which implements the methods in the above-mentioned embodiments of the present invention when executed by a processor.

[0101] The above-mentioned computer program may refer to a collection of instructions used to tell a computer to perform a specific task or operation. A computer program may be written by a programmer using a specific programming language and may include algorithms, data structures, logic, and control flows. A computer program may be used for a variety of purposes, including application software, operating systems, and the like.

[0102] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0103] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0104] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0105] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0106] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0107] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A tumor extraction method, characterized in that: include: Acquiring a reference image and real-time motion characteristics of a patient, wherein the real-time motion characteristics are images of motion fed back by signals; Extracting boundary features of the reference image and the boundary features of the real-time motion features using a trained feature extraction model; Taking the reference image and the boundary features of the real-time motion feature as reference, a feature contour of the real-time motion feature is extracted, wherein the feature contour includes at least one of: a tumor and a normal tissue.

2. The method according to claim 1, characterized in that The step of extracting the feature contour of the real-time motion feature by taking the reference image and the boundary feature of the real-time motion feature as reference comprises: Registering the reference image and the real-time motion feature based on the boundary feature in a rigid or elastic registration manner to obtain a registration result; Optimizing the registration result based on the POWELL registration algorithm; A first position contour is pre-drawn on the reference image, and a second position contour is drawn on the real-time motion feature based on the registration result.

3. The method according to claim 1 or 2, characterized in that: The reference image includes one of: Positioning images; An image obtained by combining the positioning image with the corresponding first motion feature; Reset image; An image obtained by combining the reset image with the corresponding second motion feature; The reference image and the real-time motion feature are in the same phase.

4. The method according to claim 1, characterized in that: The method further comprises: When the patient's carrier moves to the positioning position, acquiring a positioning image including boundary features; When the patient carrier moves to the reset position, acquiring a reset image including boundary features; Determining whether the boundary features in the positioning image and the resetting image are consistent; In response to the system confirming that the boundary features are inconsistent, and / or in response to manual confirmation that the boundary features are inconsistent, a motion control optimization strategy is executed, wherein the motion control optimization strategy is an adjustment of the beam of the carrier and / or treatment device.

5. The method according to claim 4, characterized in that ,The execution of motion control optimization strategy includes: Extracting a first contour boundary of the positioning image and a second contour boundary of the restoration image, determining a reference point based on the first contour boundary, and determining a matching point based on the second contour boundary, wherein the matching point is a point with the highest similarity to the reference point; The shortest distance between the reference point and the matching point is determined, an adjustment direction of the carrier is determined based on the shortest distance, and movement along the adjustment direction is performed.

6. The method according to claim 5, characterized in that The determining of the reference point based on the first contour boundary and the determining of the matching point based on the second contour boundary comprises: Detecting a plurality of first key points of the first contour boundary and a plurality of second key points of the second contour boundary; generating a first feature descriptor based on the first key point, where the first feature descriptor is used to represent image information around the first key point; generating a second feature descriptor based on the second key point, where the second feature descriptor is used to represent image information around the second key point; The best matching first feature descriptor and second feature descriptor are determined according to the K nearest neighbor matching algorithm, the first key point of the best matching first feature descriptor is used as a reference point, and the second key point of the best matching second feature descriptor is used as a matching point.

7. A tumor extraction device, characterized in that: include: An acquisition module, used to acquire a reference image and real-time motion characteristics of a patient, wherein the real-time motion characteristics are images of motion through signal feedback; A first extraction module, used to extract the boundary features of the reference image and the boundary features of the real-time motion features by using a trained feature extraction model; The second extraction module is used to extract the feature contour of the real-time motion feature with reference to the reference image and the boundary feature of the real-time motion feature, wherein the feature contour includes at least one of: a tumor and normal tissue.

8. A radiotherapy system, characterized in that: It comprises the tumor extraction device as claimed in claim 7, a patient information management module, a data transceiver module and a radiotherapy beam module, wherein the patient information management module is used to record patient information, the data transceiver module is used to receive and send image data, and the radiotherapy beam module is used to dynamically adjust the beam irradiated on the patient based on the movement of the tumor.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the method according to any one of claims 1 to 6 is executed in a processor of a device where the program is controlled.

10. A server, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 6.

Images