Method and device for identifying thoracic lymph nodes based on three-dimensional images

Abstract

The embodiment of the present invention provides a method and device for identifying chest lymph nodes based on three-dimensional images; the method includes: obtaining a chest three-dimensional image to be processed; determining all chest lymph nodes included in the chest three-dimensional image, the chest lymph nodes being used to identify the area where the chest lymph nodes are located; and determining all chest lymph nodes included in the chest three-dimensional image based on all chest lymph nodes and the chest three-dimensional image. The image recognition scheme provided in this embodiment effectively realizes the identification of chest lymph nodes within the limited search space and range corresponding to the chest lymph nodes, which not only effectively reduces other body areas with similar appearances, such as blood vessels, muscles, and soft tissues, etc., thereby effectively reducing the misidentification rate occurring at the above-mentioned similar appearances, but also can improve the accuracy of identifying chest lymph nodes, and at the same time has a high recall rate, thereby effectively improving the practicality of the method and facilitating its application in clinical practice.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method and device for identifying chest lymph nodes based on three-dimensional images. Background Art

[0002] Lymph node (LN) is an essential predictive or prognostic biomarker in radiology and oncology. For example, the detection and segmentation of chest LN is an essential step in cancer staging, treatment planning, and monitoring of disease progression in the thoracic region.

[0003] Specifically, for chest LN, the detection and segmentation of chest LN has been going on for more than ten years. At present, the recognition operation of chest LN mainly includes: extracting effective LN features in chest LN, learning effective LN features in combination with prior knowledge of organs or machine learning models, establishing statistical maps for chest lymph node detection, and then using the above-established statistical maps to process medical images to identify the lymph nodes included in the medical images.

[0004] Although the above method can simply identify lymph nodes, due to the changes in shape and size caused by the tumor, the contrast of LN in images such as computed tomography (CT) is low, which makes the accuracy and recall rate of the identified lymph nodes low. Therefore, it cannot be deployed in clinical practice. Summary of the invention

[0005] The embodiments of the present invention provide a method and device for identifying chest lymph nodes based on three-dimensional images, which can improve the accuracy and recall rate of identifying chest lymph nodes, ensure the practicability of the method, and can be deployed in clinical practice.

[0006] In a first aspect, an embodiment of the present invention provides a method for identifying chest lymph nodes based on three-dimensional images, comprising:

[0007] Acquire a three-dimensional chest image to be processed;

[0008] Determine all chest lymph nodes included in the chest three-dimensional image, wherein the chest lymph nodes are used to identify the area where the chest lymph nodes are located;

[0009] Based on all the chest lymph nodes and the three-dimensional chest image, all the chest lymph nodes included in the three-dimensional chest image are determined.

[0010] In a second aspect, an embodiment of the present invention provides a chest lymph node identification device based on three-dimensional images, comprising:

[0011] A first acquisition module is used to acquire a three-dimensional chest image to be processed;

[0012] A first determination module is used to determine all chest lymph nodes included in the chest three-dimensional image, and the chest lymph nodes are used to identify the area where the chest lymph nodes are located;

[0013] The first processing module is used to determine all chest lymph nodes included in the three-dimensional chest image based on all chest lymph nodes and the three-dimensional chest image.

[0014] In a third aspect, an embodiment of the present invention provides an electronic device, comprising: a memory, and a processor; wherein the memory is used to store one or more computer instructions, wherein when the one or more computer instructions are executed by the processor, the method for identifying chest lymph nodes based on three-dimensional images in the above-mentioned first aspect is implemented.

[0015] In a fourth aspect, an embodiment of the present invention provides a computer storage medium for storing a computer program, wherein the computer program enables a computer to implement the method for identifying chest lymph nodes based on three-dimensional images in the first aspect when executed.

[0016] In a fifth aspect, an embodiment of the present invention provides a computer program product, comprising: a computer program, which, when executed by a processor of an electronic device, enables the processor to execute the steps of the method for identifying chest lymph nodes based on three-dimensional images shown in the first aspect above.

[0017] In a sixth aspect, an embodiment of the present invention provides a method for identifying thoracic lymph nodes in a virtual reality scene, which is applied to a virtual reality device, and the method includes:

[0018] Obtain a three-dimensional image of the chest to be processed through a virtual reality device;

[0019] Determine all chest lymph nodes included in the chest three-dimensional image, wherein the chest lymph nodes are used to identify the area where the chest lymph nodes are located;

[0020] Based on all the chest lymph nodes and the three-dimensional chest image, all the chest lymph nodes included in the three-dimensional chest image are determined.

[0021] In a seventh aspect, an embodiment of the present invention provides a chest lymph node identification device in a virtual reality scene, which is applied to a virtual reality device, and the device includes:

[0022] A second acquisition module is used to acquire a three-dimensional chest image to be processed;

[0023] A second determination module is used to determine all chest lymph nodes included in the chest three-dimensional image, wherein the chest lymph nodes are used to identify the area where the chest lymph nodes are located;

[0024] The second processing module is used to determine all chest lymph nodes included in the three-dimensional chest image based on all chest lymph nodes and the three-dimensional chest image.

[0025] In an eighth aspect, an embodiment of the present invention provides an electronic device, comprising: a memory, and a processor; wherein the memory is used to store one or more computer instructions, wherein when the one or more computer instructions are executed by the processor, the method for identifying chest lymph nodes in a virtual reality scenario in the sixth aspect mentioned above is implemented.

[0026] In a ninth aspect, an embodiment of the present invention provides a computer storage medium for storing a computer program, wherein the computer program enables a computer to implement the method for identifying chest lymph nodes in a virtual reality scenario in the sixth aspect when executed.

[0027] In a tenth aspect, an embodiment of the present invention provides a computer program product, comprising: a computer program, which, when executed by a processor of an electronic device, enables the processor to execute the steps of the method for identifying chest lymph nodes in a virtual reality scenario in the sixth aspect.

[0028] The technical solution provided in this embodiment obtains a three-dimensional chest image to be processed, and then determines all chest lymph nodes included in the three-dimensional chest image. Since the chest lymph nodes are used to identify the area where the chest lymph nodes are located, all chest lymph nodes included in the three-dimensional chest image are determined based on all chest lymph nodes and the three-dimensional chest image, thereby effectively realizing the identification of chest lymph nodes within a limited search space and range corresponding to the chest lymph nodes. This not only effectively reduces other body areas with similar appearances, such as blood vessels, muscles, and soft tissues, etc., thereby effectively reducing the misidentification rate occurring at the above-mentioned similar appearances, but also improves the accuracy of identifying chest lymph nodes, and has a high recall rate, thereby effectively improving the practicability of the method, which is convenient for application in clinical practice and is conducive to market promotion and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0030] Figure 1 A schematic diagram of a scene of a method for identifying chest lymph nodes based on three-dimensional images provided by an embodiment of the present invention;

[0031] Figure 2 A schematic flow chart of a method for identifying thoracic lymph nodes based on three-dimensional images provided by an embodiment of the present invention;

[0032] Figure 3 A schematic diagram of a process for determining all chest lymph nodes included in the chest three-dimensional image based on all chest lymph nodes and the chest three-dimensional image provided by an embodiment of the present invention;

[0033] Figure 4 A schematic diagram of a process for determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on the at least three lymph node sets provided in an embodiment of the present invention;

[0034] Figure 5 A schematic diagram of the principle of a method for identifying chest lymph nodes based on three-dimensional images provided in an application embodiment of the present invention;

[0035] Figure 6 A schematic diagram of a flow chart of a method for identifying thoracic lymph nodes in a virtual reality scenario provided by an embodiment of the present invention;

[0036] Figure 7 A schematic diagram of the structure of a chest lymph node identification device based on three-dimensional images provided by an embodiment of the present invention;

[0037] Figure 8 For Figure 7 A schematic structural diagram of an electronic device corresponding to a chest lymph node identification device based on three-dimensional images provided in the illustrated embodiment;

[0038] Fig. 9 A schematic diagram of the structure of a chest lymph node identification device in a virtual reality scenario provided by an embodiment of the present invention;

[0039] Fig.10 For Fig. 9 The illustrated embodiment provides a schematic structural diagram of an electronic device corresponding to a chest lymph node identification device in a virtual reality scenario. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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 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 making creative work are protected by the present invention, but the case of including at least one is not excluded. It should be understood that the scope of the term "and / or" used in this article.

[0041] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms of "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. "Multiple" generally includes at least two. " is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0042] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0043] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a product or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such a product or system. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the product or system including the elements.

[0044] Definition of terms:

[0045] Lymph node (LN): It is an oval or bean-shaped lymphatic tissue body of varying sizes, gray-red in color when fresh. It runs along the course of lymphatic vessels and is connected to them.

[0046] Lymph node station (LN station): used to define the location or area where the lymph nodes are located.

[0047] Lymphatic station set: It is composed of two or more similar lymphatic stations and can also be called a "super station".

[0048] Computed Tomography (CT) images are images obtained by using precisely collimated X-ray beams, gamma rays, ultrasound waves, etc., together with extremely sensitive detectors to perform one-by-one cross-sectional scans around a certain part of the human body. It has the characteristics of fast scanning time and clear images, and can be used to detect a variety of diseases.

[0049] In order to enable those skilled in the art to clearly understand the implementation principle and implementation effect of the technical solution in the embodiment of the present application, the relevant technology is described below:

[0050] The evaluation of lymph nodes (LN) is an essential predictive or prognostic biomarker in radiology and oncology. For example, the detection and segmentation of chest LN is an essential step in the monitoring of cancer staging, treatment planning and disease progression in the thoracic region. In current clinical practice, chest LN is visually identified, measured or depicted by computed tomography (CT), which is not only cumbersome, time-consuming and expensive, but also requires high professional knowledge.

[0051] Specifically, for chest LN, chest LN detection and segmentation have been developed for more than ten years. At present, the recognition operation of chest LN is mainly focused on extracting effective LN features, combining prior knowledge of organs or machine learning models to learn effective LN features, and establishing statistical maps for chest lymph node detection. Then, the above-established statistical maps are used to process medical images to identify the lymph nodes included in the medical images. Although the above-mentioned implementation methods of lymph nodes are being widely studied, the recognition performance and effect of lymph nodes have always been at a low sensitivity, accuracy and recall rate, and therefore cannot be deployed in clinical practice.

[0052] In addition, due to the low contrast between LN and surrounding anatomical tissues or structures, or due to the changes in shape and size caused by tumors, it is easy to cause visual confusion between lymph nodes and blood vessels or muscles. Although large LNs with a short axis greater than 10 mm are usually considered pathological target lesions to be evaluated, studies have shown that considering only large LNs cannot reliably predict malignant tumors.

[0053] For example, according to the recollections of 60%-80% of lung cancer patients, in the process of identifying large lymph nodes LNs (short axis ≥ 10mm) in each patient, a recall rate of 70.4% can be achieved, with 4 false positives (FPs) per patient, and in the process of identifying large and small lymph nodes LNs (1cm ≥ short axis ≥ 5mm) in each patient, a recall rate of 52.4% can be achieved, with 6 FPs per patient. As can be seen from the above, segmentation and identification of large and small lymph nodes has high clinical significance.

[0054] In order to solve the above technical problems, this embodiment provides a method and device for identifying chest lymph nodes based on three-dimensional images, wherein the execution subject of the method for identifying chest lymph nodes based on three-dimensional images is a device for identifying chest lymph nodes based on three-dimensional images, and the device for identifying chest lymph nodes based on three-dimensional images can be implemented as a tablet computer, a personal computer PC, a cluster server, a conventional server, a cloud server, a cloud host, a virtual center, and other devices capable of performing chest lymph node identification operations. In addition, the identification device can be communicatively connected to a three-dimensional image acquisition device. For details, refer to the attached Figure 1 As shown:

[0055] Among them, the three-dimensional image acquisition device can be any computing device with certain three-dimensional image acquisition and transmission capabilities. When implemented specifically, the three-dimensional image acquisition device can be implemented as an electronic computer tomography (Computed Tomography, referred to as CT) device, an ultrasound device, a magnetic resonance imaging (Magnetic Resonance Imaging, referred to as MRI) device, etc. In addition, the basic structure of the three-dimensional image acquisition device may include: at least one processor. The number of processors depends on the configuration and type of the three-dimensional image acquisition device. The three-dimensional image acquisition device may also include a memory, which may be volatile, such as RAM, or non-volatile, such as read-only memory (Read-Only Memory, referred to as ROM), flash memory, etc., or may include both types at the same time. The memory usually stores an operating system (Operating System, referred to as OS), one or more application programs, and may also store program data, etc. In addition to the processing unit and the memory, the three-dimensional image acquisition device also includes some basic configurations, such as a network card chip, an IO bus, a display component, and some peripheral devices, etc. Optionally, some peripheral devices may include, for example, a keyboard, a mouse, an input pen, a printer, etc. Other peripheral devices are well known in the art and will not be described in detail here.

[0056] The chest lymph node recognition device of three-dimensional image refers to a device that can provide chest lymph node recognition in a network virtual environment, and usually refers to a device that uses the network to perform information planning and chest lymph node recognition operations. In physical implementation, the chest lymph node recognition device can be any device that can provide computing services, respond to service requests, and perform processing, for example: it can be a cluster server, a conventional server, a cloud server, a cloud host, a virtual center, etc. The chest lymph node recognition device is mainly composed of a processor, a hard disk, a memory, a system bus, etc., which is similar to a general computer architecture.

[0057] In the above embodiment, the three-dimensional image acquisition device can be connected to the chest lymph node identification device through a network, and the network connection can be a wireless or wired network connection. If the three-dimensional image acquisition device and the chest lymph node identification device are connected by communication, the network standard of the mobile network can be any one of 2G (GSM), 2.5G (GPRS), 3G (WCDMA, TD-SCDMA, CDMA2000, UTMS), 4G (LTE), 4G+ (LTE+), WiMax, 5G, 6G, etc.

[0058] The three-dimensional image acquisition device is used to scan the user's chest to obtain a three-dimensional chest image. In specific implementation, the three-dimensional chest image may include any one of the following: a chest CT image, a chest ultrasound image, a chest magnetic resonance image, etc. After the three-dimensional chest image is obtained, the three-dimensional chest image can be actively or passively sent to a chest lymph node identification device.

[0059] The chest lymph node recognition device is connected to the three-dimensional image acquisition device in communication, and can actively or passively acquire the three-dimensional chest image through the three-dimensional image acquisition device, and then analyze and process the three-dimensional chest image to determine all the chest lymph nodes included in the three-dimensional chest image. The chest lymph nodes are used to identify the area where the chest lymph nodes are located. It can be understood that different chest lymph nodes can correspond to different areas. After all the chest lymph nodes are acquired, all the chest lymph nodes and the three-dimensional chest image can be analyzed and processed to determine all the chest lymph nodes included in the three-dimensional chest image, which effectively improves the accuracy of chest lymph node recognition and has a high recall rate.

[0060] After all the chest lymph nodes included in the chest three-dimensional image are determined, all the chest lymph nodes can be marked and displayed, so that the user can view all the chest lymph nodes more intuitively, so as to facilitate medical testing or diagnosis and identification operations.

[0061] The method and device for identifying chest lymph nodes based on three-dimensional images provided in the present embodiment obtain a chest three-dimensional image to be processed, and then determine all chest lymph nodes included in the chest three-dimensional image. Since the chest lymph nodes are used to identify the area where the chest lymph nodes are located, all chest lymph nodes included in the chest three-dimensional image are determined based on all chest lymph nodes and the chest three-dimensional image, thereby effectively realizing the identification of chest lymph nodes within the limited search space and range corresponding to the chest lymph nodes. This not only effectively reduces other body areas with similar appearances, such as blood vessels, muscles, and soft tissues, etc., thereby effectively reducing the misidentification rate occurring at the above-mentioned similar appearances, but also improves the accuracy of identifying chest lymph nodes, and at the same time has a high recall rate, effectively improving the practicability of the method, which is convenient for application in clinical practice and conducive to market promotion and application.

[0062] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the case where there is no conflict between the embodiments, the following embodiments and the features in the embodiments can be combined or separated with each other. In addition, the step sequence in the following method embodiments is only an example and not a strict limitation.

[0063] Figure 2 A schematic diagram of a flow chart of a method for identifying thoracic lymph nodes based on three-dimensional images provided by an embodiment of the present invention; Figure 2 As shown, this embodiment provides a method for identifying chest lymph nodes based on three-dimensional images. The executor of the method may be a device for identifying chest lymph nodes based on three-dimensional images. The device for identifying chest lymph nodes based on three-dimensional images may be implemented as software or a combination of software and hardware. Specifically, when the device for identifying chest lymph nodes based on three-dimensional images is implemented as hardware, it may be various electronic devices capable of performing data processing operations, including but not limited to tablet computers, personal computers PCs, servers, and the like. When the device for identifying chest lymph nodes based on three-dimensional images is implemented as software, it may be installed in the electronic devices listed above. Based on the above-mentioned device for identifying chest lymph nodes based on three-dimensional images, the method for identifying chest lymph nodes based on three-dimensional images in this embodiment may include the following steps:

[0064] Step S201: Acquire a three-dimensional chest image to be processed.

[0065] Step S202: determining all chest lymph nodes included in the three-dimensional chest image, where the chest lymph nodes are used to identify the area where the chest lymph nodes are located.

[0066] Step S203: Based on all the chest lymph nodes and the three-dimensional chest image, all the chest lymph nodes included in the three-dimensional chest image are determined.

[0067] The specific implementation process and effects of each of the above steps are described in detail below:

[0068] Step S201: Acquire a three-dimensional chest image to be processed.

[0069] Among them, when there is a need to identify chest lymph nodes, the chest lymph node identification device can obtain a chest three-dimensional image to be processed. The above-mentioned chest three-dimensional image to be processed is a chest scan image that requires a lymph node identification operation. The chest three-dimensional image can have different formats. Technical personnel in this field can configure the chest three-dimensional image according to specific application scenarios or application requirements. For example: the chest three-dimensional image can be a chest CT image, and the chest CT image can be obtained by scanning the user's chest through a CT scanning device; the chest three-dimensional image can be a chest ultrasound image, at this time, the chest ultrasound image can be obtained by scanning the user's chest through an ultrasound device; the chest three-dimensional image can be a chest magnetic resonance image, at this time, the chest magnetic resonance image can be obtained by scanning the user's chest through a magnetic resonance imaging device.

[0070] Specifically, in the specific implementation method of acquiring the chest three-dimensional image to be processed, in some instances, the chest lymph node identification device is communicatively connected to a three-dimensional image acquisition device, and the three-dimensional image acquisition device scans the user's chest, thereby obtaining a chest three-dimensional image. It is understandable that chest three-dimensional images of different formats can be obtained by different three-dimensional image acquisition devices. After obtaining the chest three-dimensional image, the chest lymph node identification device can actively or passively obtain the chest three-dimensional image through the three-dimensional image acquisition device. In other instances, the chest three-dimensional image can be stored in a preset area or a third device. In this case, the chest three-dimensional image can be obtained by accessing the preset area or the third device.

[0071] Of course, the method of acquiring the three-dimensional chest image is not limited to the implementation method exemplified above. Those skilled in the art may also adopt other methods to acquire the three-dimensional chest image, as long as it can ensure that the chest lymph node identification device can stably acquire the three-dimensional chest image, which will not be elaborated here.

[0072] Step S202: determining all chest lymph nodes included in the three-dimensional chest image, where the chest lymph nodes are used to identify the area where the chest lymph nodes are located.

[0073] For lymph nodes, since almost all lymph nodes LN are located in the corresponding lymph nodes, the above-mentioned lymph nodes can be areas defined according to key anatomical organs or identifiers, and LNs in different LN stations usually have different contexts (or upper and lower boundary contents) and show different degrees of recognition uncertainty. For example, different lymph nodes usually have different contexts (or upper and lower boundary contents). Therefore, when performing recognition operations on different lymph nodes, different degrees of recognition uncertainty are shown. For example, there is usually a relatively clear boundary between chest lymph node 2 and chest lymph node 4, while chest lymph node 7 and chest lymph node 8 are more easily confused with adjacent blood vessels or esophagus. Therefore, when identifying lymph nodes, it is very meaningful to limit the recognition operation of lymph nodes to the range corresponding to the lymph node station. Since the search and recognition space area for identifying lymph nodes is limited, other identity areas with similar local appearances are effectively reduced, such as blood vessels, muscles, and soft tissues, etc., and the false alarm rate FPs generated at the above-mentioned similar local appearances can be reduced.

[0074] Based on the above statements, it can be known that in order to improve the quality and effect of identifying chest lymph nodes, after obtaining the chest three-dimensional image, the chest three-dimensional image can be analyzed and processed to determine all chest lymph nodes included in the chest three-dimensional image. The obtained chest lymph nodes are used to identify the area where the chest lymph nodes are located, and different chest lymph nodes can correspond to the same or different chest lymph nodes. In some examples, for any user, the number of all chest lymph nodes is 14. At this time, all chest lymph nodes can include: chest lymph node 1 to chest lymph node 14, and different chest lymph nodes can correspond to different location information or area ranges.

[0075] In some instances, all chest lymph nodes may be determined based on preset chest reference organs. In this case, determining all chest lymph nodes included in a three-dimensional chest image may include: determining a chest reference organ based on the three-dimensional chest image, wherein the chest reference organ is used to assist in determining the chest lymph nodes; and determining all chest lymph nodes included in the three-dimensional chest image based on the chest reference organ and the three-dimensional chest image.

[0076] Among them, in order to improve the quality and effect of identifying the chest lymph node, after obtaining the chest three-dimensional image, the chest three-dimensional image can be analyzed and processed to determine the chest reference organ, which can be pre-defined or configured to assist in determining the main organ of the chest lymph node, which can specifically include at least one of the following: esophagus, aortic arch, ascending aorta, heart, spine, sternum. Specifically, since the chest reference organ accounts for a large proportion of the chest three-dimensional image, and different chest reference organs can correspond to different size features, shape features, position features, upper and lower boundary (context) features, etc., therefore, the size features, shape features, position features, upper and lower boundary (context) features and other organ features of the preset organ can be determined based on the chest three-dimensional image, and then the organ features and the preset standard organ features (corresponding to the standard organ type) are analyzed and compared to determine the standard organ type corresponding to the organ features, so as to determine the organ type of the preset organ, for example, the preset organ can be determined to be the esophagus or the aortic arch, etc., so that the chest reference organ can be effectively accurately identified and determined.

[0077] After the chest reference organ is determined, the chest reference organ and the chest three-dimensional image can be analyzed and processed to determine all the chest lymph nodes included in the chest three-dimensional image. In some examples, based on the chest reference organ and the chest three-dimensional image, determining all the chest lymph nodes included in the chest three-dimensional image can include: obtaining a network model for determining the chest lymph nodes, inputting the chest reference organ and the chest three-dimensional image into the network model, and obtaining all the chest lymph nodes output by the network model. In other examples, the chest reference organ can be used as reference information, and then the chest three-dimensional image can be analyzed and processed to determine all the chest lymph nodes included in the chest three-dimensional image, thereby effectively ensuring the quality and effect of identifying all the chest lymph nodes.

[0078] In other instances, all chest lymph nodes can be determined based on a pre-trained machine learning model. In this case, determining all chest lymph nodes included in a three-dimensional chest image can include: obtaining a machine learning model for analyzing and processing the three-dimensional chest image, and inputting the three-dimensional chest image into the machine learning model, so as to obtain all chest lymph nodes output by the machine learning model.

[0079] Step S203: Based on all the chest lymph nodes and the three-dimensional chest image, all the chest lymph nodes included in the three-dimensional chest image are determined.

[0080] Since the chest lymph stations are used to identify the area where the chest lymph nodes are located, after all the chest lymph stations are acquired, all the lymph stations and the chest three-dimensional image can be analyzed and processed. Specifically, all the chest lymph stations and the chest three-dimensional image can be analyzed and processed using a preset algorithm, preset rules or a pre-trained machine learning model to determine all the chest lymph nodes included in the chest three-dimensional image. All the chest lymph stations include all the chest lymph stations located in each chest lymph station. This effectively limits the chest lymph node recognition operation to the area corresponding to the chest lymph station, which is beneficial to improve the accuracy and recall rate of chest lymph node recognition.

[0081] The method for identifying chest lymph nodes based on three-dimensional images provided in the present embodiment obtains a chest three-dimensional image to be processed, and then determines all chest lymph nodes included in the chest three-dimensional image. Since the chest lymph nodes are used to identify the area where the chest lymph nodes are located, all chest lymph nodes included in the chest three-dimensional image are determined based on all chest lymph nodes and the chest three-dimensional image, thereby effectively identifying the chest lymph nodes in the search space and range corresponding to the chest lymph nodes. This not only effectively reduces other body areas with similar appearances, such as blood vessels, muscles, and soft tissues, etc., thereby effectively reducing the misidentification rate occurring at the above-mentioned similar appearances, but also improves the accuracy of identifying the chest lymph nodes. At the same time, it has a high recall rate, effectively improves the practicability of the method, is easy to apply in clinical practice, and is conducive to market promotion and application.

[0082] Figure 3 The present invention provides a flowchart of determining all chest lymph nodes included in a chest three-dimensional image based on all chest lymph nodes and a chest three-dimensional image; based on the above embodiment, refer to the attached Figure 3 As shown, since there are multiple chest lymph nodes, and some of the multiple chest lymph nodes have the same or similar characteristics, when determining all chest lymph nodes based on the multiple chest lymph nodes and the chest three-dimensional image, all chest lymph nodes may be first divided to obtain a lymph node set, and then all chest lymph nodes may be determined based on the lymph node set and the chest three-dimensional image. Specifically, in this embodiment, determining all chest lymph nodes included in the chest three-dimensional image based on all chest lymph nodes and the chest three-dimensional image may include:

[0083] Step S301: Divide all chest lymph nodes to obtain at least three lymph node sets, each of which includes at least two chest lymph nodes whose similarity is greater than or equal to a preset threshold.

[0084] For all chest lymph nodes, since there are multiple chest lymph nodes, some of the multiple chest lymph nodes have the same or similar features or characteristics, and some chest lymph nodes have completely different or dissimilar features or characteristics. For the same or similar chest lymph nodes, the same or similar rules or principles can be used to identify the chest lymph nodes in the above chest lymph nodes, and for the different or dissimilar chest lymph nodes, different or dissimilar rules or principles can be used to identify the chest lymph nodes in the above chest lymph nodes. This can not only ensure the accuracy of identifying the chest lymph nodes, but also improve the efficiency of identifying the chest lymph nodes. Therefore, after obtaining all chest lymph nodes, all chest lymph nodes can be divided, so that at least three lymph node sets can be obtained, wherein each lymph node set includes at least two chest lymph nodes whose similarity is greater than or equal to a preset threshold.

[0085] In some instances, when obtaining at least three sets of lymph nodes, a preset rule, a preset algorithm, or a machine learning model can be used to perform a division operation on all chest lymph nodes. At this time, dividing all chest lymph nodes to obtain at least three sets of lymph nodes may include: obtaining a preset algorithm, a preset rule, or a machine learning model for dividing all chest lymph nodes, and dividing all chest lymph nodes using the preset algorithm, the preset rule, or the machine learning model to obtain at least three sets of lymph nodes.

[0086] In other instances, in addition to using preset rules, preset algorithms or machine learning models to divide all chest lymph nodes, characteristic information of each chest lymph node can also be extracted, and then all chest lymph nodes can be divided based on the characteristic information. At this time, dividing all chest lymph nodes to obtain at least three lymph node sets can include: obtaining lymph node features corresponding to all chest lymph nodes; dividing all chest lymph nodes based on the lymph node features corresponding to all chest lymph nodes to obtain at least three lymph node sets.

[0087] For the chest lymph node, since the chest lymph node is used to identify the area where the chest lymph nodes are located, different chest lymph nodes are used to identify different areas where the chest lymph nodes are located, and the above-mentioned different areas may correspond to different position features, shape features, and surrounding context features (or upper and lower boundary features), etc., it can be seen from the above that the same or similar chest lymph nodes may have the same or similar lymph node features, and different chest lymph nodes also have different lymph node features. Therefore, in order to accurately obtain at least three lymph node sets, after all chest lymph nodes are obtained, the lymph node features corresponding to all chest lymph nodes can be obtained. The lymph node features may include at least one of the following: position features, area features (area shape, area size), surrounding context features (or upper and lower boundary features), etc. Specifically, the lymph node features corresponding to all chest lymph nodes can be obtained by analyzing and processing all chest lymph nodes through a preset feature extraction algorithm or a preset feature extraction model.

[0088] After acquiring the lymph node features corresponding to all the chest lymph nodes, the lymph node features corresponding to all the chest lymph nodes can be analyzed and processed, so as to divide all the chest lymph nodes based on the analysis and processing results to obtain at least three lymph node sets. In some examples, dividing all the chest lymph nodes based on the lymph node features corresponding to all the chest lymph nodes to obtain at least three lymph node sets may include: obtaining the similarity between any two chest lymph nodes based on the lymph node features corresponding to all the chest lymph nodes; dividing all the chest lymph nodes based on the similarity between any two chest lymph nodes to obtain at least three lymph node sets, and the similarity between any two chest lymph nodes in the same lymph node set is greater than or equal to a preset threshold.

[0089] After obtaining the lymph node features corresponding to all the chest lymph nodes, the lymph node features corresponding to any two chest lymph nodes can be processed to obtain the similarity between any two chest lymph nodes. In some examples, the similarity between any two chest lymph nodes can be determined by the Euclidean distance or cosine distance between the lymph node features corresponding to any two chest lymph nodes. After obtaining the similarity between any two chest lymph nodes, all chest lymph nodes can be divided based on the similarity obtained above, so as to obtain at least three lymph node sets. Specifically, the similarity between any two chest lymph nodes in the same lymph node set is greater than or equal to a preset threshold, which can be 80%, 90%, 95% or 99%, etc., which is a pre-configured minimum threshold for evaluating whether two chest lymph nodes are the same or similar chest lymph nodes.

[0090] For example, when all chest lymph nodes include: lymph node 1, lymph node 2, lymph node 3, lymph node 4, lymph node 5, lymph node 6, lymph node 7, lymph node 8, lymph node 9, lymph node 10, lymph node 11, lymph node 12, lymph node 13, and lymph node 14; dividing all chest lymph nodes based on the similarity between any two chest lymph nodes to obtain at least three lymph node sets may include: when the similarity between lymph node 1, lymph node 2, lymph node 3 and lymph node 4 is greater than or equal to a preset threshold, it means that lymph node 1 and lymph node 2, lymph node 3 and lymph node 4 belong to the same or similar chest lymph nodes, and then based on the similarity between any two chest lymph nodes, lymph node 1, lymph node 2, lymph node 3, and lymph node 4 in all chest lymph nodes can be divided into a lymph node set to obtain a first lymph node set (super station 1).

[0091] Similarly, when the similarity between lymph node 5, lymph node 6, lymph node 7, lymph node 8, and lymph node 9 is greater than or equal to a preset threshold, it means that lymph node 5 and lymph node 6, lymph node 7, lymph node 8, and lymph node 9 belong to the same or similar chest lymph node stations, and then based on the similarity between any two chest lymph nodes, lymph node 5, lymph node 6, lymph node 7, lymph node 8, and lymph node 9 in all chest lymph nodes are divided into a lymph node set to obtain a second lymph node set (super station 2). When the similarity between lymph node 10, lymph node 11, lymph node 12, lymph node 13, and lymph node 14 is greater than or equal to a preset threshold, it means that lymph node 10 and lymph node 11, lymph node 12, lymph node 13, and lymph node 14 belong to the same or similar chest lymph node stations, and then based on the similarity between any two chest lymph nodes, lymph node 10, lymph node 11, lymph node 12, lymph node 13, and lymph node 14 in all chest lymph nodes are divided into a lymph node set to obtain a third lymph node set (super station 3).

[0092] From the above, it can be seen that when the number of chest lymph stations is 14, all chest lymph stations can be divided into three lymph station sets. The first lymph station set (super station 1) can include chest lymph station 1 to chest lymph station 4, the second lymph station set (super station 2) can include chest lymph station 5 to chest lymph station 9, and the third lymph station set (super station 3) can include chest lymph station 10 to chest lymph station 14.

[0093] It is understandable that when dividing the chest lymph nodes, not only three lymph node sets can be obtained, but also four lymph node sets or five lymph node sets, etc. can be obtained. Technical personnel in this field can obtain a specific number of at least three lymph node sets according to specific application scenarios or application requirements.

[0094] In some other instances, after obtaining the lymph station set, since the lymph station set is used to identify the area where the chest lymph nodes are located, it is necessary to perform an identification operation on the chest lymph nodes based on the lymph station set. At this time, in order to avoid the situation where some chest lymph nodes are located outside the lymph station set due to under-segmentation of the lymph station set, after obtaining at least three lymph station sets, the area corresponding to the lymph station set can be adjusted. Specifically, the method in this embodiment may also include: obtaining preset parameters for adjusting the area corresponding to the lymph station set; adjusting the area corresponding to the lymph station set based on the preset parameters, and obtaining an adjusted lymph station set corresponding to the lymph station set, the area range corresponding to the adjusted lymph station set is larger than the area range corresponding to the lymph station set.

[0095] In order to adjust the area where the lymph node set is located, the preset parameters for adjusting the area corresponding to the lymph node set can be obtained first. The preset parameters can be 15mm, 20mm, 10mm or 18mm, etc., wherein the parameter values ​​corresponding to the preset parameters can be configured according to specific application scenarios or application requirements. In addition, the present embodiment does not limit the specific method for obtaining the preset parameters. In some instances, the preset parameters can be pre-configured parameters stored in the preset area. In this case, the preset parameters can be obtained by accessing the preset area. In other instances, the preset parameters can be obtained based on the user's configuration operation. In this case, obtaining the preset parameters for adjusting the area corresponding to the lymph node set can include: displaying a display interface for interactive operation with the user, obtaining the parameter configuration operation input by the user through the display interface, and obtaining the preset parameters for adjusting the area corresponding to the lymph node set based on the parameter configuration operation. The preset parameters are mainly used to expand the area corresponding to the lymph node set.

[0096] After obtaining the preset parameters, the area corresponding to the lymph node set can be adjusted based on the preset parameters, so that the adjusted lymph node set corresponding to the lymph node set can be obtained. The area range corresponding to the obtained adjusted lymph node set is larger than the area range corresponding to the lymph node set. Specifically, the preset parameters can be used as the sphere expansion radius, and the sphere area range corresponding to the lymph node set can be expanded with the sphere expansion radius, so that the adjusted lymph node set can be obtained. Alternatively, the preset parameters can be used as the cube expansion side length, and the cube expansion operation can be performed on the area range corresponding to the lymph node set with the cube expansion side length, so that the adjusted lymph node set can be obtained. In the above manner, the expansion operation of the area range corresponding to the lymph node set is effectively realized by the preset parameters, so that all chest lymph nodes can be effectively included in the area corresponding to the lymph node set, so as to improve the accuracy and reliability of identifying all chest lymph nodes.

[0097] Step S302: Based on at least three lymph node sets and the three-dimensional chest image, all chest lymph nodes included in the three-dimensional chest image are determined.

[0098] After acquiring at least three lymph node sets, the at least three lymph node sets and the three-dimensional chest image can be analyzed and processed to determine all the chest lymph nodes included in the three-dimensional chest image. In some examples, the at least three lymph node sets and the three-dimensional chest image can be processed based on a preset algorithm or a pre-trained machine learning model to determine all the chest lymph nodes included in the three-dimensional chest image.

[0099] In other examples, since lymph nodes may include large-sized lymph nodes and small-sized lymph nodes, large-sized lymph nodes generally have different texture patterns (e.g., calcification / necrosis) and shapes (e.g., tree / star) relative to small-sized lymph nodes, it is beneficial to identify lymph nodes by different lymph nodes and lymph node size types. At this time, based on at least three lymph node sets and the three-dimensional chest image, determining all chest lymph nodes included in the three-dimensional chest image may include: based on at least three lymph node sets, determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image, the short axis size of the first type of lymph nodes is greater than or equal to a preset threshold, and the short axis size of the second type of lymph nodes is less than the preset threshold; based on the first type of lymph nodes, the second type of lymph nodes and the three-dimensional chest image, determining all chest lymph nodes included in the three-dimensional chest image.

[0100] After acquiring at least three lymph node sets, the at least three lymph node sets may be analyzed and processed to determine the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image, the first type of lymph nodes may be called large-sized lymph nodes, and the second type of lymph nodes may be called small-sized lymph nodes. In some examples, determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on at least three lymph node sets may include: determining each lymph node and the short-axis size corresponding to each lymph node based on the at least three lymph node sets; and then determining the first type of lymph nodes and the second type of lymph nodes based on the short-axis size.

[0101] Since the above-mentioned large-sized lymph nodes and small-sized lymph nodes are the identification results of stratifying the lymph nodes by the short-axis size of the lymph nodes, in order to improve the accuracy and reliability of identifying all chest lymph nodes included in the chest three-dimensional image, after obtaining the first-category lymph nodes and the second-category lymph nodes, the first-category lymph nodes, the second-category lymph nodes and the chest three-dimensional image can be further processed to determine all chest lymph nodes included in the chest three-dimensional image.

[0102] In some instances, determining all chest lymph nodes included in the chest three-dimensional image based on the first category of lymph nodes, the second category of lymph nodes, and the chest three-dimensional image may include: acquiring a first model for processing the first category of lymph nodes and the chest three-dimensional image and a second model for processing the second category of lymph nodes and the chest three-dimensional image; processing the first category of lymph nodes and the chest three-dimensional image using the first model to obtain a first chest lymph node portion; processing the second category of lymph nodes and the chest three-dimensional image using the second model to obtain a second chest lymph node portion; and aggregating the first chest lymph node portion and the second chest lymph node portion to obtain all chest lymph nodes included in the chest three-dimensional image.

[0103] In other instances, based on the first type of lymph nodes, the second type of lymph nodes and the three-dimensional chest image, determining all the chest lymph nodes included in the three-dimensional chest image may include: using the first type of lymph nodes and the second type of lymph nodes as reference information, re-identifying the three-dimensional chest image to obtain false positive lymph nodes included in the first type of lymph nodes and the second type of lymph nodes; removing the false positive lymph nodes in the first type of lymph nodes and the second type of lymph nodes to obtain all the chest lymph nodes included in the three-dimensional chest image.

[0104] In the process of chest lymph node identification on chest three-dimensional images, there are often a certain number of false-positive lymph nodes. Therefore, in order to reduce the number of false-positive lymph nodes, after obtaining the first type of lymph nodes and the second type of lymph nodes, the first type of lymph nodes and the second type of lymph nodes can be used as reference information to perform a re-identification operation on the chest three-dimensional image, that is, a targeted lymph node re-identification operation is performed on the chest three-dimensional image to obtain the false-positive lymph nodes included in the first type of lymph nodes and the second type of lymph nodes, and then the false-positive lymph nodes in the first type of lymph nodes and the second type of lymph nodes can be removed, so that all chest lymph nodes including a smaller number of false-positive lymph nodes can be obtained, thereby effectively improving the accuracy and recall rate of chest lymph nodes.

[0105] In this embodiment, by dividing all chest lymph nodes to obtain at least three lymph node sets, and then determining all chest lymph nodes included in the chest three-dimensional image based on the at least three lymph node sets and the chest three-dimensional image, the chest lymph node identification operation is effectively realized from the dimension of the lymph node set. Since the same lymph node set includes the same or similar chest lymph nodes, and different lymph node sets include different or dissimilar chest lymph nodes, the accuracy and recall rate of chest lymph node identification can be improved, further improving the practicality of the method.

[0106] Figure 4 A schematic diagram of a process for determining the first type of lymph nodes and the second type of lymph nodes included in a three-dimensional chest image based on at least three lymph node sets provided in an embodiment of the present invention; based on the above embodiment, refer to the attached Figure 4 As shown, this embodiment provides an implementation method for determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image by means of lymph node image features. Specifically, based on at least three lymph node sets, determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image may include:

[0107] Step S401: Obtaining lymph node image features corresponding to at least three lymph node station sets.

[0108] Among them, in order to accurately identify the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image, after obtaining at least three lymph station sets, the at least three lymph station sets can be analyzed and processed to obtain lymph node image features corresponding to each of the at least three lymph station sets. In some instances, obtaining lymph node image features corresponding to each of the at least three lymph station sets can include: obtaining a preset algorithm or machine learning model for performing feature extraction operations on at least three lymph station sets, and then using the preset algorithm or machine learning model to separately process the at least three lymph station sets, so that the lymph node image features corresponding to each of the at least three lymph station sets can be obtained.

[0109] In other instances, since different lymph node sets often correspond to different lymph node image features, in order to ensure the accuracy and reliability of acquiring lymph node image features, acquiring the set image features corresponding to each of at least three lymph node sets in this embodiment may include: acquiring the pre-trained lymph node feature coding models corresponding to each of at least three lymph node sets; using the lymph node feature coding models corresponding to each of at least three lymph node sets to process the corresponding lymph node sets respectively, and obtain the lymph node image features corresponding to each of the at least three lymph node sets.

[0110] For example, at least three lymph station sets include: lymph station set 1, lymph station set 2 and lymph station set 3, and then pre-trained lymph station feature coding models corresponding to the above at least three lymph station sets are obtained, for example: lymph station set 1 corresponds to lymph node feature coding model 1, lymph station set 2 corresponds to lymph node feature coding model 2, and lymph station set 3 corresponds to lymph node feature coding model 3, and then lymph station set 1 can be processed using lymph node feature coding model 1 to obtain lymph node image features corresponding to lymph station set 1; similarly, lymph station set 2 can be processed using lymph node feature coding model 2 to obtain lymph node image features corresponding to lymph station set 2; lymph station set 3 can be processed using lymph node feature coding model 3 to obtain lymph node image features corresponding to lymph station set 3, thereby effectively ensuring the quality and effect of acquiring lymph node image features.

[0111] It should be noted that the model parameters corresponding to the above-mentioned different lymph node feature coding models corresponding to different lymph node station sets are different. At this time, in order to ensure the accurate and reliable acquisition of lymph node image features corresponding to different lymph node station sets, when training the above-mentioned lymph node feature coding models, training operations can be performed on the regional images corresponding to at least three lymph node station sets respectively, so as to ensure the processing performance and effect of each lymph node feature coding model.

[0112] Step S402: Based on all lymph node image features, determine the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image.

[0113] After all the lymph node image features are acquired, all the lymph node image features may be processed to determine the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image. In some examples, based on all the lymph node image features, determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image may include: determining the short axis size corresponding to each lymph node based on all the lymph node image features; determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on the short axis size corresponding to each lymph node.

[0114] In other instances, determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on all lymph node image features may include: obtaining a pre-trained first decoding model for identifying the first type of lymph nodes and a second decoding model for identifying the second type of lymph nodes; using the first decoding model to process all lymph node image features to obtain the first type of lymph nodes included in the three-dimensional chest image; and using the second decoding model to process all lymph node image features to obtain the second type of lymph nodes included in the three-dimensional chest image.

[0115] Among them, a first decoding model for identifying the first type of lymph nodes and a second decoding model for identifying the second type of lymph nodes are pre-trained, and the model parameters corresponding to the first decoding model and the second decoding model are different. In order to accurately obtain the first type of lymph nodes and the second type of lymph nodes, the first decoding model and the second decoding model can be obtained first, and then the first decoding model can be used to process all lymph node image features, so that the first type of lymph nodes included in the three-dimensional chest image can be obtained, that is, the large-sized lymph nodes can be effectively identified. Similarly, the second decoding model can be used to process all lymph node image features, so that the second type of lymph nodes included in the three-dimensional chest image can be obtained, that is, the small-sized lymph nodes can be effectively identified.

[0116] In this embodiment, by acquiring the lymph node image features corresponding to each of at least three lymph station sets, and then determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on all the lymph node image features, the first type of lymph nodes and the second type of lymph nodes of different sizes are effectively recognized, thereby improving the accuracy and reliability of the recognition of all chest lymph nodes in the three-dimensional chest image based on the first type of lymph nodes and the second type of lymph nodes, and further improving the practicability of the method.

[0117] In specific use, take the chest CT image as the chest three-dimensional image, the number of chest lymph nodes is 14, and the number of at least three lymph node sets (super nodes) is 3 as an example, refer to the attached Figure 5 As shown, this application embodiment provides a method for identifying chest lymph nodes based on CT images. The method accurately identifies lymph nodes through LN stations and size-aware dimensions. Specifically, the method may include the following steps:

[0118] Step 1: Obtain a chest CT image, which is a three-dimensional image.

[0119] Step 2: Process the chest CT image to obtain 14 chest lymph nodes included in the chest CT image, each chest lymph node is used to define the area where the chest lymph nodes are located.

[0120] Step 3: Divide the 14 thoracic lymph nodes to obtain three super nodes, super node 1 includes thoracic lymph node 1 to thoracic lymph node 4, super node 2 includes thoracic lymph node 5 to thoracic lymph node 9, and super node 3 includes thoracic lymph node 10 to thoracic lymph node 14.

[0121] Specifically, in the process of segmenting and stratifying LN stations, the prior knowledge of LN stations can be used to identify a group of 14 chest lymph nodes based on chest CT image segmentation. In order to ensure the accuracy and reliability of chest lymph node identification, the key reference organ-guided LN station segmentation model can be used to identify chest LN stations. Among them, the key reference organs can include 6, including esophagus, aortic arch, ascending aorta, heart, spine and sternum.

[0122] Specifically, during training, N training data can be expressed as Among them, X n represents the input chest CT image, Indicates the image area corresponding to the key reference organ, Indicates the thoracic lymph node station, represents the set of thoracic lymph nodes (superstations), and then let C K and C S They represent the number of key reference organs and the number of LN stations respectively. At this time, the key reference organs are obtained by identifying the chest CT image using the following formula (1), and the LN stations included in the chest CT image can be identified using the following formula (2).

[0123]

[0124]

[0125] Among them, f *() is used to identify the deep learning network or network model that can identify key reference organs and LN stacks, and is used to represent network functions, W * are the preset network parameters. For the predicted segmented image, Y K (j) used to identify the key reference organs, Y S (j) Used to identify LN stations obtained based on key reference organs and CT images.

[0126] After obtaining the super-station, in order to avoid the possibility that the lymph nodes are not included in the area corresponding to the divided super-station, after obtaining the super-station, each super-station can be expanded. Specifically, an area can be expanded with a radius of 15 mm to obtain the expanded super-station. This can effectively reduce the situation where the super-station is under-segmented, and ensure the accurate and reliable identification of the chest lymph nodes.

[0127] Step 4: Determine the encoders used to analyze and process the super stations, wherein the number of encoders corresponds to the number of super stations.

[0128] Among them, in the process of identifying chest lymph nodes, multiple encoders corresponding to super stations can be pre-trained, and the above encoders can be obtained by learning and training based on the image areas corresponding to the super stations. Specifically, the extended super station can be used for binary mapping to "shield" the CT image covered by the super station (abbreviated as mCT), and then the pixel values ​​of other areas are set to a constant of -1024 (null value), so that the mCT images corresponding to the three super stations can be obtained. The pixel values ​​of the areas outside the super stations in the mCT images are all constants (null values), and then the learning and training operations can be performed based on the mCT images corresponding to the above three super stations, so as to obtain the encoders corresponding to the three super stations.

[0129] Step 5: Obtain the image area corresponding to each super station, input the image area corresponding to each super station into the corresponding encoder, and obtain the image features corresponding to each super station.

[0130] For example: obtain image area 1 corresponding to super station 1, input image area 1 into encoder 1, and obtain image feature 1 corresponding to super station 1. Similarly, obtain image area 2 corresponding to super station 2, input image area 2 into encoder 2, and obtain image feature 2 corresponding to super station 2; obtain image area 3 corresponding to super station 3, input image area 3 into encoder 3, and obtain image feature 3 corresponding to super station 3.

[0131] It should be noted that the training data and input data of the three encoders are independent, for example: super station 1 corresponds to encoder 1, super station 2 corresponds to encoder 2, and super station 3 corresponds to encoder 3. For encoder 1, its output data can be the image area mCT corresponding to the expanded super station 1 (i.e., chest lymph station 1 to chest lymph station 4), that is, Similarly, for encoder 2, its output data may be the image area mCT corresponding to the expanded super station 2 (i.e., chest lymph station 5 to chest lymph station 9), that is, For encoder 3, its output data may be the image area mCT corresponding to the expanded super station 3 (ie, chest lymph node station 10 to chest lymph node station 14), that is, Above Represents the grouped LN station image area. From the above, it can be seen that each encoder needs to perform processing operations on the targeted super station to obtain image features.

[0132]

[0133] in, Encoder for analyzing and processing the super station, Indicates the image area mCT corresponding to each super station, Indicates super station, Used to identify the super station. are the preset network parameters. Represents the image feature map corresponding to the super station obtained after passing through the encoder.

[0134] Step 6: Obtain pre-trained decoder 1 for identifying large-sized lymph nodes and decoder 2 for identifying small-sized lymph nodes, use decoder 1 to process all lymph node image features to obtain large-sized lymph nodes included in the chest CT image; use decoder 2 to process all lymph node image features to obtain small-sized lymph nodes included in the chest CT image.

[0135] Specifically, considering that large-sized LNs usually produce different texture patterns (e.g., calcification / necrosis) or shapes (e.g., tree / star) compared with small-sized lymph nodes LN, two decoders are introduced to learn LN features of specific sizes, that is, decoder 1 learns LN features of large-sized lymph nodes, and decoder 2 learns LN features of small-sized lymph nodes, where large-sized lymph nodes may refer to lymph nodes with a short axis size greater than or equal to 1 cm, and small-sized lymph nodes may refer to lymph nodes with a short axis size less than 1 cm and greater than or equal to 5 mm.

[0136] Among them, when training the decoder, each decoder uses the relevant large-size LN label or small-size LN label for supervised learning. After obtaining the trained decoder, the decoder can be used to identify the large-size lymph nodes and the small-size lymph nodes. Specifically, for decoder 1 and decoder 2, the input of each decoder is the image feature map output by the encoder from the three super stations. In order to facilitate decoder 1 and decoder 2 to process all the image feature maps corresponding to the three super stations, all the image feature maps corresponding to at least three super stations can be cascaded to obtain cascaded image features. In some instances, the cascaded image features can be obtained by splicing all the image feature maps. The cascaded image features

[0137] After obtaining the cascade image features, the decoder 1 and the decoder 2 can be used to process the obtained cascade image features, that is, to process them through the following formula:

[0138]

[0139] in, represents the predicted image of the large-size lymph node output by decoder 1, The predicted map of the small-sized lymph nodes output by the decoder 2 is represented, thereby effectively achieving the accurate reliability of identifying the large-sized lymph nodes and the small-sized lymph nodes.

[0140] Step 7: After the large-sized lymph nodes and the small-sized lymph nodes are obtained, the large-sized lymph nodes, the small-sized lymph nodes and the chest CT image may be fused and post-processed, so that all the chest lymph nodes included in the chest CT image may be obtained.

[0141] After obtaining the large-sized lymph nodes and the small-sized lymph nodes, the output feature maps of each decoder (including the output feature maps corresponding to the large-sized lymph nodes and the output feature maps corresponding to the small-sized lymph nodes) can be combined with the chest CT image and then input into a simple fusion module, wherein the fusion module can be a module for identifying chest lymph nodes created by two nnU-Net (an adaptive medical image segmentation framework based on U-Net) network layers. Specifically, To obtain all the thoracic lymph nodes, Used to represent thoracic lymph nodes, f * () is used to represent a fusion module, which is used to identify all chest lymph nodes included in the chest CT image, W() is a preset network parameter, X represents the chest CT image, Used to indicate the image features corresponding to large lymph nodes. Used to represent the imaging features corresponding to small-sized lymph nodes.

[0142] Specifically, The specific implementation principle is as follows: using large-sized lymph nodes and small-sized lymph nodes as reference information, re-identifying the chest CT image to obtain the false-positive lymph nodes included in the large-sized lymph nodes and the small-sized lymph nodes; removing the false-positive lymph nodes in the large-sized lymph nodes and the small-sized lymph nodes to obtain all the chest lymph nodes included in the chest CT image, thereby effectively realizing a targeted re-identification operation of the lymph nodes in the chest CT image, which can effectively reduce the false-positive lymph nodes included in the identification results. At present, the number of false-positive lymph nodes in the prior art is 6. Through the technical solution in this embodiment, the number of false-positive lymph nodes in all lymph nodes is 4, thereby improving the accuracy and effect of lymph node identification.

[0143] The technical solution provided in this application embodiment proposes a novel LN station-specific and size-aware LN partitioning framework by utilizing prior knowledge of LN stations and learning LN size variance. To achieve this goal, the technical solution first uses a robust deep station model to segment chest LN stations 1 to 14. Specifically, according to the background of the LN stations and the clinical experience of the physician, the 14 LN stations can be stratified and divided into 3 super-stations, namely LN stations 1-4 (super-station 1 or upper layer), LN stations 5-9 (super-station 2 or lower layer) and LN stations 10-14 (super-station 3 or lung area). On this basis, by designing a new deep segmentation network with multiple encoding paths (corresponding to encoder 1, encoder 2, encoder 3, etc.), each deep segmentation network focuses on learning the LN features in a specific super station. Next, in order to explicitly learn the size differences of LNs, two decoding branches (corresponding to decoder 1 and decoder 2) are adopted. The above decoder 1 and decoder 2 focus on the dimensions of small-sized lymph nodes and large-sized lymph nodes for feature learning operations, respectively. Then, decoder 1 and decoder 2 can identify large-sized lymph nodes and small-sized lymph nodes, and a fusion module can be used to merge large-sized lymph nodes, small-sized lymph nodes and chest CT images, so that all chest lymph nodes included in the chest CT images can be accurately obtained.

[0144] Through experimental evaluation, a 4-fold cross-validation was performed on a public LN dataset of 89 lung cancer patients and more than 2,000 recently annotated LNs (≥5 mm). When the technical solution in this embodiment was used to identify chest lymph nodes, the recognition accuracy of chest lymph nodes reached an average of 74.2% (an increase of 9.9% compared to the prior art) and a detection recall rate of 72.0% (an increase of 15.6% compared to the prior art). There were approximately 4 in each patient (a decrease of 1.9 compared to the prior art). False positives (fp-pw), significantly improved segmentation and detection accuracy compared to previous guided LN segmentation / detection methods. In addition, in an external test dataset of 57 users with esophageal cancer with 360 thoracic LNs (≥5 mm markers), the accuracy was approximately 70.4% (an increase of 13.8%) and the recall rate was approximately 70.2% (an increase of 17.0%) compared to related technologies. There were approximately 4.4fp-pw per user, which effectively proved that the technical solution has good universality and versatility, making this technical solution an important step towards reliable and automated LN segmentation.

[0145] Figure 6 A schematic diagram of a flow chart of a method for identifying thoracic lymph nodes in a virtual reality scenario provided by an embodiment of the present invention; Figure 6 As shown, this embodiment provides a method for identifying thoracic lymph nodes in a virtual reality scenario. The identification method can be applied to virtual reality equipment. The execution subject of the identification method can be a thoracic lymph node identification device in a virtual reality scenario. The thoracic lymph node identification device in the virtual reality scenario can be implemented as software or a combination of software and hardware. Specifically, when the thoracic lymph node identification device in the virtual reality scenario is implemented as hardware, it can be various virtual reality devices with data processing operations, that is, the identification method can be applied to virtual reality devices. It should be noted that the above-mentioned virtual reality devices include but are not limited to head-mounted display devices (Headset Mount Device, HMD) in the fields of augmented reality (AR), virtual reality (VR) or mixed reality (MR). When the thoracic lymph node identification device in the virtual reality scenario is implemented as software, it can be installed in the electronic devices listed above. Based on the above-mentioned thoracic lymph node identification device in the virtual reality scenario, the thoracic lymph node identification method in the virtual reality scenario in this embodiment can include the following steps:

[0146] Step S601: Acquire a three-dimensional chest image to be processed.

[0147] Step S602: determining all chest lymph nodes included in the three-dimensional chest image, where the chest lymph nodes are used to identify the area where the chest lymph nodes are located.

[0148] Step S603: Based on all the chest lymph nodes and the three-dimensional chest image, all the chest lymph nodes included in the three-dimensional chest image are determined.

[0149] Among them, the specific implementation method and implementation effect of steps S601 to S603 in this embodiment are similar to the specific implementation method and implementation effect of the above-mentioned steps S201 to S203. Please refer to the above-mentioned statements for details, and no further details will be given here.

[0150] In other examples, after all the chest lymph nodes included in the three-dimensional chest image are determined, all the determined chest lymph nodes can be marked and displayed through a virtual reality device, so that the user can view and operate all the determined chest lymph nodes through the virtual reality device.

[0151] It should be noted that the method in this embodiment may also include the above Figure 1-Figure 5 For the method of the embodiment shown in the figure, the part not described in detail in this embodiment can be referred to Figure 1-Figure 5 The implementation process and technical effects of this technical solution refer to Figure 1-Figure 5 The description in the illustrated embodiment will not be repeated here.

[0152] The method for identifying chest lymph nodes in a virtual reality scenario provided by the present embodiment obtains a chest three-dimensional image to be processed, and then determines all chest lymph nodes included in the chest three-dimensional image. Since the chest lymph nodes are used to identify the area where the chest lymph nodes are located, all chest lymph nodes included in the chest three-dimensional image are determined based on all chest lymph nodes and the chest three-dimensional image, thereby effectively realizing the identification of chest lymph nodes in a limited search space and range in a virtual reality scenario. This not only effectively reduces other body areas with similar appearances, such as blood vessels, muscles, and soft tissues, etc., thereby effectively reducing the misidentification rate occurring at the above-mentioned similar appearances, but also improves the accuracy of identifying chest lymph nodes, and at the same time has a high recall rate, effectively improving the practicability of the method, and is conducive to market promotion and application.

[0153] Figure 7 A schematic diagram of the structure of a chest lymph node identification device based on three-dimensional images provided by an embodiment of the present invention; Figure 7 As shown, this embodiment provides a chest lymph node recognition device based on three-dimensional images, and the chest lymph node recognition device based on three-dimensional images is used to perform the above Figure 2The method for identifying chest lymph nodes based on three-dimensional images shown in the figure, specifically, the device for identifying chest lymph nodes based on three-dimensional images may include:

[0154] A first acquisition module 11 is used to acquire a three-dimensional chest image to be processed;

[0155] A first determination module 12 is used to determine all chest lymph nodes included in the chest three-dimensional image, where the chest lymph nodes are used to identify the area where the chest lymph nodes are located;

[0156] The first processing module 13 is used to determine all chest lymph nodes included in the three-dimensional chest image based on all chest lymph nodes and the three-dimensional chest image.

[0157] In some examples, when the first determination module 12 determines all chest lymph nodes included in the chest three-dimensional image, the first determination module 12 is used to perform: determining chest reference organs based on the chest three-dimensional image, the chest reference organs are used to assist in determining the chest lymph nodes; and determining all chest lymph nodes included in the chest three-dimensional image based on the chest reference organs and the chest three-dimensional image.

[0158] In some instances, when the first processing module 13 determines all chest lymph nodes included in the chest three-dimensional image based on all chest lymph nodes and the chest three-dimensional image, the first processing module 13 is used to perform: dividing all chest lymph nodes to obtain at least three lymph node sets, each lymph node set including at least two chest lymph nodes whose similarity is greater than or equal to a preset threshold; and determining all chest lymph nodes included in the chest three-dimensional image based on the at least three lymph node sets and the chest three-dimensional image.

[0159] In some instances, when the first processing module 13 divides all chest lymph nodes to obtain at least three lymph node sets, the first processing module 13 is used to execute: obtaining lymph node features corresponding to all chest lymph nodes; dividing all chest lymph nodes based on the lymph node features corresponding to all chest lymph nodes to obtain at least three lymph node sets.

[0160] In some instances, when the first processing module 13 divides all chest lymph nodes based on the lymph node characteristics corresponding to each of the chest lymph nodes to obtain at least three lymph node sets, the first processing module 13 is used to execute: obtaining the similarity between any two chest lymph nodes based on the lymph node characteristics corresponding to each of the chest lymph nodes; dividing all chest lymph nodes based on the similarity between any two chest lymph nodes to obtain at least three lymph node sets, and the similarity between any two chest lymph nodes in the same lymph node set is greater than or equal to a preset threshold.

[0161] In some instances, when the first processing module 13 determines all chest lymph nodes included in the three-dimensional chest image based on at least three lymph station sets and the three-dimensional chest image, the first processing module 13 is used to perform: determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on at least three lymph station sets, the short axis size of the first type of lymph nodes is greater than or equal to a preset threshold, and the short axis size of the second type of lymph nodes is less than the preset threshold; determining all chest lymph nodes included in the three-dimensional chest image based on the first type of lymph nodes, the second type of lymph nodes and the three-dimensional chest image.

[0162] In some instances, when the first processing module 13 determines the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on at least three lymph station sets, the first processing module 13 is used to execute: obtaining lymph node image features corresponding to each of the at least three lymph station sets; and determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on all the lymph node image features.

[0163] In some instances, when the first processing module 13 obtains the collective image features corresponding to at least three lymph node sets, the first processing module 13 is used to execute: obtaining the pre-trained lymph node feature coding models corresponding to at least three lymph node sets; using the lymph node feature coding models corresponding to at least three lymph node sets to process the corresponding lymph node sets respectively, to obtain the lymph node image features corresponding to at least three lymph node sets.

[0164] In some instances, when the first processing module 13 determines the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on all lymph node image features, the first processing module 13 is used to execute: obtaining a pre-trained first decoding model for identifying the first type of lymph nodes and a second decoding model for identifying the second type of lymph nodes; using the first decoding model to process all lymph node image features to obtain the first type of lymph nodes included in the three-dimensional chest image; using the second decoding model to process all lymph node image features to obtain the second type of lymph nodes included in the three-dimensional chest image.

[0165] In some instances, when the first processing module 13 determines all chest lymph nodes included in the three-dimensional chest image based on the first type of lymph nodes, the second type of lymph nodes and the three-dimensional chest image, the first processing module 13 is used to execute: re-identify the three-dimensional chest image using the first type of lymph nodes and the second type of lymph nodes as reference information to obtain false-positive lymph nodes included in the first type of lymph nodes and the second type of lymph nodes; remove the false-positive lymph nodes in the first type of lymph nodes and the second type of lymph nodes to obtain all chest lymph nodes included in the three-dimensional chest image.

[0166] In some examples, after obtaining at least three lymph node station sets, the first acquisition module 11 and the first processing module 13 in this embodiment are used to perform the following steps:

[0167] A first acquisition module 11 is used to acquire preset parameters for adjusting the area corresponding to the lymph node station set;

[0168] The first processing module 13 is used to adjust the area corresponding to the lymph node set based on preset parameters to obtain an adjusted lymph node set corresponding to the lymph node set, and the area range corresponding to the adjusted lymph node set is larger than the area range corresponding to the lymph node set.

[0169] Figure 7 The device shown can perform Figure 1-Figure 5 For the method of the embodiment shown in the figure, the part not described in detail in this embodiment can be referred to Figure 1-Figure 5 The implementation process and technical effects of this technical solution refer to Figure 1-Figure 5 The description in the illustrated embodiment will not be repeated here.

[0170] In one possible design, Figure 7 The structure of the thoracic lymph node identification device based on three-dimensional images can be implemented as an electronic device, which can be a tablet computer, a personal computer PC, a server, or other devices. Figure 8 As shown, the electronic device may include: a first processor 21 and a first memory 22. The first memory 22 is used to store the corresponding electronic device to execute the above Figure 1 - In the embodiment shown in FIG. 5 , the program of the method for identifying thoracic lymph nodes based on three-dimensional images, the first processor 21 is configured to execute the program stored in the first memory 22 .

[0171] The program includes one or more computer instructions, wherein when the one or more computer instructions are executed by the first processor 21, the following steps can be implemented:

[0172] Acquire a three-dimensional chest image to be processed;

[0173] Determine all thoracic lymph nodes included in the 3D chest image, where the thoracic lymph nodes are used to identify the area where the thoracic lymph nodes are located;

[0174] Based on all the chest lymph nodes and the three-dimensional chest image, all the chest lymph nodes included in the three-dimensional chest image are determined.

[0175] Furthermore, the first processor 21 is also used to execute the aforementioned Figure 1-Figure 5 All or part of the steps in the illustrated embodiments.

[0176] The structure of the electronic device may further include a first communication interface 23 for the electronic device to communicate with other devices or a communication network.

[0177] In addition, an embodiment of the present invention provides a computer storage medium for storing computer software instructions used by electronic devices, which includes instructions for executing the above Figure 1-Figure 5 The procedures involved in the method for identifying chest lymph nodes based on three-dimensional images in the method embodiment are shown.

[0178] In addition, an embodiment of the present invention provides a computer program product, including: a computer program, when the computer program is executed by a processor of an electronic device, the processor executes the above Figure 1-Figure 5 The steps in the method for identifying chest lymph nodes based on three-dimensional images are shown.

[0179] Fig. 9 FIG. 1 is a schematic diagram of a structure of a chest lymph node identification device in a virtual reality scene provided by an embodiment of the present invention; Fig. 9 As shown, this embodiment provides a chest lymph node identification device in a virtual reality scene, which is applied to a virtual reality device. The chest lymph node identification device in the virtual reality scene can perform Figure 6 The method for identifying chest lymph nodes in the virtual reality scene shown, specifically, the device for identifying chest lymph nodes in the virtual reality scene may include:

[0180] A second acquisition module 31 is used to acquire a three-dimensional chest image to be processed;

[0181] A second determination module 32 is used to determine all chest lymph nodes included in the chest three-dimensional image, where the chest lymph nodes are used to identify the area where the chest lymph nodes are located;

[0182] The second processing module 33 is used to determine all chest lymph nodes included in the three-dimensional chest image based on all chest lymph nodes and the three-dimensional chest image.

[0183] Fig. 9 The device shown can perform Figure 6 For the method of the embodiment shown in the figure, the part not described in detail in this embodiment can be referred to Figure 6 The implementation process and technical effects of this technical solution refer to Figure 6 The description in the illustrated embodiment will not be repeated here.

[0184] In one possible design, Fig. 9The structure of the thoracic lymph node identification device in the virtual reality scene shown can be implemented as an electronic device, which can be various devices such as a head-mounted display device (HMD) in the field of augmented reality (AR), virtual reality (VR) or mixed reality (MR). Fig.10 As shown, the electronic device may include: a second processor 41 and a second memory 42. The second memory 42 is used to store the corresponding electronic device executing the above Figure 6 In the illustrated embodiment, the second processor 41 is configured to execute the program stored in the second memory 42 .

[0185] The program includes one or more computer instructions, wherein when the one or more computer instructions are executed by the second processor 41, the following steps can be implemented:

[0186] Acquire a three-dimensional chest image to be processed;

[0187] Determine all thoracic lymph nodes included in the 3D chest image, where the thoracic lymph nodes are used to identify the area where the thoracic lymph nodes are located;

[0188] Based on all the chest lymph nodes and the three-dimensional chest image, all the chest lymph nodes included in the three-dimensional chest image are determined.

[0189] Furthermore, the second processor 41 is also used to execute the aforementioned Figure 6 All or part of the steps in the illustrated embodiments.

[0190] The structure of the electronic device may further include a second communication interface 43 for the electronic device to communicate with other devices or a communication network.

[0191] In addition, an embodiment of the present invention provides a computer storage medium for storing computer software instructions used by electronic devices, which includes instructions for executing the above Figure 6 The procedures involved in the method for identifying chest lymph nodes in a virtual reality scene in the method embodiment shown.

[0192] In addition, an embodiment of the present invention provides a computer program product, including: a computer program, when the computer program is executed by a processor of an electronic device, the processor executes the above Figure 6 The steps in the method for identifying chest lymph nodes in the virtual reality scene are shown.

[0193] The device embodiments described above are merely illustrative, wherein 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, i.e., they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0194] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by adding a necessary general hardware platform, and of course can also be implemented by combining hardware and software. Based on such an understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a computer product, and the present invention can be in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0195] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable device to produce a machine, so that the instructions executed by the processor of the computer or other programmable device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0196] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable device to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0197] These computer program instructions may also be loaded onto a computer or other programmable device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1The steps for the functions specified in one or more boxes.

[0198] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0199] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0200] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0201] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying thoracic lymph nodes based on three-dimensional images. It is characterized in that include: Acquire a three-dimensional chest image to be processed; Determine all chest lymph nodes included in the chest three-dimensional image, wherein the chest lymph nodes are used to identify the area where the chest lymph nodes are located; Dividing all chest lymph nodes to obtain at least three lymph node sets, each lymph node set including at least two chest lymph nodes having a similarity greater than or equal to a preset threshold; Based on at least three lymph node sets and the three-dimensional chest image, all chest lymph nodes included in the three-dimensional chest image are determined.

2. The method according to claim 1, It is characterized in that Determine all thoracic lymph nodes included in the 3D chest image, including: Determine a chest reference organ based on the chest three-dimensional image, wherein the chest reference organ is used to assist in determining a chest lymph node station; Based on the chest reference organ and the three-dimensional chest image, all chest lymph nodes included in the three-dimensional chest image are determined.

3. The method according to claim 1, It is characterized in that All thoracic lymph nodes are divided into at least three sets, including: Obtain the lymph node characteristics corresponding to all chest lymph nodes; All the thoracic lymph nodes are divided based on the lymph node characteristics corresponding to each of the thoracic lymph nodes to obtain at least three lymph node sets.

4. The method according to claim 3, It is characterized in that All thoracic lymph nodes are divided based on their corresponding lymph node characteristics to obtain at least three lymph node sets, including: Based on the lymph node features corresponding to all the thoracic lymph nodes, the similarity between any two thoracic lymph nodes is obtained; All chest lymph nodes are divided based on the similarity between any two chest lymph nodes to obtain at least three lymph node sets, and the similarity between any two chest lymph nodes in the same lymph node set is greater than or equal to a preset threshold.

5. The method according to claim 1, It is characterized in that Based on at least three lymph node sets and the three-dimensional chest image, all chest lymph nodes included in the three-dimensional chest image are determined, including: Based on the at least three lymph node sets, determining a first type of lymph nodes and a second type of lymph nodes included in the three-dimensional chest image, wherein the short axis size of the first type of lymph nodes is greater than or equal to a preset threshold, and the short axis size of the second type of lymph nodes is less than the preset threshold; Based on the first type of lymph nodes, the second type of lymph nodes and the three-dimensional chest image, all chest lymph nodes included in the three-dimensional chest image are determined.

6. The method according to claim 5, It is characterized in that Determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image based on the at least three lymph node sets includes: Obtaining lymph node image features corresponding to at least three lymph node station sets; Based on all the lymph node image features, the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image are determined.

7. The method according to claim 6, It is characterized in that Obtain the image features corresponding to at least three lymph node sets, including: Obtaining lymph node feature encoding models corresponding to at least three pre-trained lymph node station sets; The corresponding lymph node sets are processed respectively using the lymph node feature coding models corresponding to the at least three lymph node sets to obtain the lymph node image features corresponding to the at least three lymph node sets.

8. The method according to claim 6, It is characterized in that Based on all the lymph node image features, determining the first type of lymph nodes and the second type of lymph nodes included in the three-dimensional chest image includes: Obtaining a pre-trained first decoding model for identifying a first type of lymph node and a second decoding model for identifying a second type of lymph node; Using the first decoding model to process all lymph node image features to obtain a first type of lymph nodes included in the three-dimensional chest image; The second decoding model is used to process all lymph node image features to obtain the second type of lymph nodes included in the three-dimensional chest image.

9. The method according to claim 5, It is characterized in that Based on the first type of lymph nodes, the second type of lymph nodes and the three-dimensional chest image, all chest lymph nodes included in the three-dimensional chest image are determined, including: Re-identifying the chest three-dimensional image using the first type of lymph nodes and the second type of lymph nodes as reference information to obtain false positive lymph nodes included in the first type of lymph nodes and the second type of lymph nodes; False positive lymph nodes in the first category of lymph nodes and the second category of lymph nodes are removed to obtain all chest lymph nodes included in the three-dimensional chest image.

10. The method according to any one of claims 3 to 9, It is characterized in that After obtaining at least three lymph node station sets, the method further includes: Obtaining preset parameters for adjusting the area corresponding to the lymph node station set; The area corresponding to the lymph node set is adjusted based on the preset parameters to obtain an adjusted lymph node set corresponding to the lymph node set, and the area range corresponding to the adjusted lymph node set is larger than the area range corresponding to the lymph node set.

11. A method for identifying chest lymph nodes in a virtual reality scenario. It is characterized in that Applied to virtual reality equipment, the method comprises: Acquire a three-dimensional chest image to be processed; Determine all chest lymph nodes included in the chest three-dimensional image, wherein the chest lymph nodes are used to identify the area where the chest lymph nodes are located; Dividing all chest lymph nodes to obtain at least three lymph node sets, each lymph node set including at least two chest lymph nodes having a similarity greater than or equal to a preset threshold; Based on at least three lymph node sets and the three-dimensional chest image, all chest lymph nodes included in the three-dimensional chest image are determined.

12. An electronic device, It is characterized in that include: A memory and a processor; wherein the memory is used to store one or more computer instructions, wherein when the one or more computer instructions are executed by the processor, the method for identifying chest lymph nodes based on three-dimensional images as described in any one of claims 1 to 10 is implemented.

Images