CN113012118B

CN113012118B - Image processing method and image processing apparatus

Info

Publication number: CN113012118B
Application number: CN202110246915.0A
Authority: CN
Inventors: 张荣国; 夏晨; 张华杰; 张欢; 王少康; 陈宽
Original assignee: Infervision Medical Technology Co Ltd
Current assignee: Infervision Medical Technology Co Ltd
Priority date: 2021-03-05
Filing date: 2021-03-05
Publication date: 2022-04-19
Anticipated expiration: 2041-03-05
Also published as: CN113012118A

Abstract

The application provides an image processing method and an image processing device, wherein the image processing method comprises the following steps: determining lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the pulmonary flat-scan CT image based on the pulmonary flat-scan CT image; and determining reinforced judgment information corresponding to the lung flat-scan CT image based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data. And judging whether the lung has initial abnormal lesion or not by judging the strengthening state of the lesion area. Since no enhanced CT examination is required, the need for contrast media injection is avoided, thereby avoiding trauma and potential risks associated with contrast media injection. Meanwhile, radiation to the patient during CT examination is avoided, and the patient seeing cost is reduced.

Description

Image processing method and image processing apparatus

Technical Field

The present application relates to the field of medical image processing technologies, and in particular, to an image processing method, an image processing apparatus, an electronic device, and a computer-readable storage medium.

Background

CT examination is an important means for screening cancer and other diseases, and includes flat-scan CT examination and enhanced CT examination. The flat scanning CT examination is difficult to find the abnormal lesion at the initial stage of the focus, and the enhanced CT examination is needed to judge the abnormal lesion at the initial stage of the focus through the intensified phenomenon of the focus.

However, enhanced CT examinations require the patient to be injected with a contrast agent, and the human body may experience varying degrees of adverse effects on the contrast agent. Attention is paid to accurately examine the initial stage of lesion abnormality while avoiding the use of contrast media.

Disclosure of Invention

In view of the above, embodiments of the present application provide an image processing method, an image processing apparatus, an electronic device, and a computer-readable storage medium to solve the technical problem of how to accurately check an initial lesion of a lesion without using a contrast medium.

According to an aspect of the present application, an embodiment of the present application provides an image processing method, including: determining lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image based on the lung flat-scan CT image; and determining reinforced judgment information corresponding to the lung flat scanning CT image based on the flat scanning CT image, the focus segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data.

In one embodiment, the determining, based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data, enhanced determination information corresponding to the pulmonary flat-scan CT image includes: determining an image characteristic diagram, a focus and pulmonary vessel contour position relation characteristic diagram and a focus and pulmonary vessel contour position relation characteristic diagram based on the lung flat scanning CT image, the focus segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data; connecting the image feature map, the lesion and pulmonary vessel contour position relationship feature map and the lesion and pulmonary artery and vein contour position relationship feature map in series in a vector form to form a one-dimensional feature vector; and inputting the one-dimensional feature vector into an enhancement judgment module to generate the enhancement judgment information.

In one embodiment, the determining an image feature map, a lesion-to-pulmonary vessel contour position relationship feature map, and a pulmonary arteriovenous vessel contour position relationship feature map based on the lung flat scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data includes: determining flat-scan CT image block data, focus and pulmonary vessel contour position information block data, and focus and pulmonary artery and vein contour position information block data based on the lung flat-scan CT image, the focus segmentation data, the pulmonary vessel segmentation data, and the pulmonary artery and vein segmentation data; inputting the block data of the flat-scan CT image into a flat-scan CT image analysis module to generate the image feature map; inputting the data of the lesion and pulmonary vessel contour position information block into a lesion and pulmonary vessel contour position relation analysis module to generate a feature map of the lesion and pulmonary vessel contour position relation; and inputting the data of the lesion and pulmonary arteriovenous vessel contour position information block into a lesion and pulmonary arteriovenous vessel contour position relation analysis module to generate a lesion and pulmonary arteriovenous vessel contour position relation characteristic diagram.

In one embodiment, the determining flat-scan CT image block data, lesion and pulmonary vessel contour position information block data, lesion and pulmonary arteriovenous vessel contour position information block data based on the lung flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, the pulmonary arteriovenous vessel segmentation data comprises: determining the flat-scan CT image block data corresponding to a lesion region of the lung flat-scan CT image based on the lung flat-scan CT image; determining lesion segmentation block data, pulmonary vessel segmentation block data and pulmonary arteriovenous vessel segmentation block data corresponding to a lesion region of the lung flat scan CT image based on the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data; merging the lesion segmentation block data and the pulmonary vessel segmentation block data to obtain lesion and pulmonary vessel contour position information block data; and merging the focus segmentation block data and the pulmonary arteriovenous vessel segmentation block data to obtain the focus and pulmonary arteriovenous vessel contour position information block data.

In one embodiment, the determining the flat-scan CT image block data corresponding to a lesion region of the lung flat-scan CT image based on the lung flat-scan CT image comprises: intercepting flat-scan CT image block data with a preset size from the lung flat-scan CT image based on the coordinate of the focus region center; wherein the determining of lesion segmentation block data, pulmonary vessel segmentation block data, and pulmonary arteriovenous vessel segmentation block data corresponding to a lesion region of the lung flat scan CT image based on the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data includes: and based on the coordinates of the focus region center, intercepting flat scanning CT image block data, blood vessel segmentation block data and pulmonary arteriovenous blood vessel segmentation block data with preset sizes from the focus segmentation data, the pulmonary blood vessel segmentation data and the pulmonary arteriovenous blood vessel segmentation data.

In one embodiment, before the inputting the one-dimensional feature vector into the enhanced discrimination module, the image processing method further includes: determining a lung flat-scan CT image sample, a one-dimensional feature vector sample corresponding to the lung flat-scan CT image sample, and strengthening information of a lesion region corresponding to the lung flat-scan CT image sample; inputting the one-dimensional feature vector sample into an initial enhancement judging module to generate an enhancement judging information sample corresponding to the lung flat scanning CT image sample; and adjusting the network parameters of the initial enhancement judging module based on the enhancement judging information sample of the lesion region corresponding to the lung flat scan CT image sample and the enhancement judging information corresponding to the lung flat scan CT image sample.

In one embodiment, the determining a reinforced judgment information sample of a lesion region corresponding to the lung flat scan CT image sample includes: and determining a reinforced judgment information sample of the lesion region corresponding to the lung flat scan CT image sample based on the reinforced judgment information of the lesion region of the reinforced CT image sample corresponding to the lung flat scan CT image sample.

In one embodiment, the determining, based on the flat-scan lung CT image, lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the flat-scan lung CT image includes: inputting the lung flat scanning CT image into a lesion segmentation model to obtain the lesion segmentation data; inputting the lung flat scanning CT image into a blood vessel segmentation model to obtain the lung blood vessel segmentation data; and inputting the lung flat scanning CT image into an arteriovenous segmentation model to obtain the pulmonary arteriovenous vessel segmentation data.

In one embodiment, before the determining, based on the lung flat-scan CT image, lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image, the image processing method further includes: and carrying out normalization processing on the lung flat scanning CT image. Wherein, the determining of the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data corresponding to the pulmonary flat scan CT image based on the pulmonary flat scan CT image respectively comprises: and determining lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the normalized pulmonary flat-scan CT image based on the normalized pulmonary flat-scan CT image.

In one embodiment, the normalizing the lung flat scan CT image comprises: and normalizing the pixel distance in the coordinate vector direction in the lung flat scanning CT image to a preset value.

According to another aspect of the present application, an embodiment of the present application provides an image processing apparatus, wherein the segmentation data determining module is configured to determine, based on a flat-scan lung CT image, lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the flat-scan lung CT image; and an enhanced judgment information determining module configured to determine enhanced judgment information corresponding to the lung flat-scan CT image based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data.

According to another aspect of the present application, an embodiment of the present application provides an electronic device, including: a processor; a memory; and computer program instructions stored in the memory, which when executed by the processor, cause the processor to perform the image processing method as claimed in any one of the preceding claims.

According to yet another aspect of the present application, an embodiment of the present application provides a computer-readable storage medium having stored thereon computer program instructions, which, when executed by a processor, cause the processor to perform the image processing method of any one of the above.

According to the image processing method provided by the embodiment of the application, focus segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the pulmonary flat scan CT image are determined based on the pulmonary flat scan CT image, and enhanced judgment information corresponding to the pulmonary flat scan CT image is determined based on the flat scan CT image, the focus segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data. And judging whether the lung has initial abnormal lesion or not according to the strengthening state of the lesion area. Since no enhanced CT examination is required, the need for contrast media injection is avoided, thereby avoiding trauma and potential risks associated with contrast media injection. Meanwhile, radiation to the patient during CT examination is avoided, and the patient seeing cost is reduced.

Drawings

The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.

Fig. 1 is a schematic view of a scenario applicable to the embodiment of the present application.

Fig. 2 is a schematic view of another scenario applicable to the embodiment of the present application.

Fig. 3 is a schematic flowchart illustrating an image processing method according to an embodiment of the present application.

Fig. 4 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application.

Fig. 5 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application.

Fig. 6 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application.

Fig. 7 is a schematic flow chart illustrating a process of obtaining enhanced determination information corresponding to a flat-scan lung CT image based on flat-scan CT image block data, lesion and pulmonary artery and vein contour position information block data, and lesion and pulmonary artery and vein contour position information block data according to an embodiment of the present application.

Fig. 8 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application.

Fig. 9 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application.

Fig. 10 is a flowchart illustrating an image processing method according to yet another embodiment of the present application.

Fig. 11 is a flowchart illustrating an image processing method according to another embodiment of the present application.

Fig. 12 is a schematic structural diagram of an image processing apparatus according to an embodiment of the present application.

Fig. 13 is a schematic structural diagram of an image processing apparatus according to yet another embodiment of the present application.

Fig. 14 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

Computed Tomography (CT) mainly uses precisely collimated X-ray beams, gamma rays, ultrasonic waves, and the like, and performs cross-section scanning one by one around a certain part of a human body together with a detector with extremely high sensitivity.

The CT examination comprises flat-scan CT examination and enhanced CT examination, the common CT examination directly carried out on an instrument without injecting a contrast medium is the flat-scan CT examination, and compared with the flat-scan CT examination, the enhanced CT examination has the greatest difference that the contrast medium is injected to a patient before the enhanced CT examination is carried out so as to clearly display blood vessels, so that a doctor can more easily know the condition of lesion of a focus and the diagnosis accuracy is improved.

The principle of diagnosing diseases by flat scan CT examination is as follows: the attenuation value of the measuring radiation after passing through different tissues of a human body is different due to different refractive indexes of different tissues, the professional term of the attenuation value of the reaction radiation is density, and the CT value is a measuring Unit for measuring the density of a certain local tissue or organ of the human body and is generally called Hounsfield Unit (Hu). The same texture should be of the same density and different textures should be of different densities. For a flat scan CT picture of the lung, the density of the lung without the focus is relatively uniform, and the density of the lung with the focus is different from that of other normal tissues.

The focus can be found by CT flat scanning, but the primary abnormal lesion of the focus is not easy to be found by flat scanning CT examination (the abnormal lesion of the focus indicates that the focus is malignant with high probability). Since the lesion grows in the normal lung tissue organ, the lesion is small in number and small in size in the early stage, and when the flat scanning CT scanning is adopted, the early stage abnormal lesion of the lesion is not easy to be found, such as small hepatic cyst, hepatic hemangioma, small liver cancer and the like. At this time, an enhanced CT examination is required, which injects a contrast medium into a patient, and since the lesion of the abnormal lesion has a richer blood supply, the contrast medium is more concentrated in the lesion of the abnormal lesion, and on the enhanced CT image, the blood vessel at the lesion of the abnormal lesion is clearer, and the lesion of the abnormal lesion is enhanced (i.e., the local contrast at the lesion of the abnormal lesion in the enhanced CT image is increased), so that the lesion is discovered at an early stage and a scientific and reasonable treatment is performed as soon as possible.

However, the contrast agent is required to be injected into a patient for enhancing the CT examination, and the contrast agent is a contrast agent containing iodine, so that adverse reactions of different degrees may occur to the human body while providing high-quality diagnostic images for the examination, the adverse reactions commonly include fever, accelerated heartbeat and the like, and even severe respiratory cardiac arrest, anaphylactic shock and the like may occur. Therefore, there is a need to solve the technical problems of the prior art that the injection of contrast medium causes trauma to the patient and has potential risks when performing the enhanced CT examination.

In view of the above technical problems, the present application provides an image processing method, which is used for processing a flat-scan CT image, so as to determine an initial abnormal lesion in a lesion area based on the flat-scan CT image, instead of determining the initial abnormal lesion in the lesion area by using an enhanced CT examination, thereby avoiding the injection of a contrast medium, and thus avoiding the trauma and the potential risk of the injection of the contrast medium. Meanwhile, radiation to the patient during CT examination is avoided, and the patient seeing cost is reduced.

Exemplary application scenarios

Fig. 1 is a schematic view of a scenario applicable to the embodiment of the present application. As shown in fig. 1, a scenario to which the embodiment of the present application is applied includes a server 1 and an image capturing device 2, where there is a communication connection relationship between the server 1 and the image capturing device 2.

Specifically, the image acquisition device 2 is configured to acquire a lung flat-scan CT image, and the server 1 is configured to determine, based on the lung flat-scan CT image acquired by the image acquisition device 2, lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image, and then determine, based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data, enhancement judgment information corresponding to the lung flat-scan CT image, where the enhancement judgment information can judge an enhancement state of a lesion region. That is, the scene implements an image processing method. Since the scene shown in fig. 1 implements the image processing method by using the server 1, the scene not only can improve the adaptability of the scene, but also can effectively reduce the calculation amount of the image acquisition device 2.

It should be noted that the present disclosure is also applicable to another scenario. Fig. 2 is a schematic view of another scenario applicable to the embodiment of the present application. Specifically, the scene includes an image processing device 3, wherein the image processing device 3 includes an image acquisition module 31 and a calculation module 32, and a communication connection relationship exists between the image acquisition module 301 and the calculation module 302.

Specifically, the image acquisition module 31 in the image processing device 3 is configured to acquire a lung flat-scan CT image, the calculation module 32 in the image processing device 3 is configured to determine, based on the lung flat-scan CT image acquired by the image acquisition module 31, lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image, and then determine, based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data, reinforcement judgment information corresponding to the lung flat-scan CT image, where the reinforcement judgment information can judge a reinforcement state of a lesion region. That is, the scene implements an image processing method. Since the scene shown in fig. 2 implements an image processing method using the image processing apparatus 3, data transmission operations with a server or other related devices are not required, and thus the scene can ensure real-time performance of the image processing method.

Exemplary image processing method

Fig. 3 is a schematic flowchart illustrating an image processing method according to an embodiment of the present application. As shown in fig. 3, the image processing method includes the following steps.

Step 101: and determining lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image based on the lung flat-scan CT image.

Specifically, the flat-scan CT image of the lung is an image obtained by performing a flat-scan CT examination on the chest of a human body. Lesion segmentation data is data that characterizes the location and contour of a lesion region. The pulmonary vessel segmentation data is used as data characterizing the position and contour of the pulmonary vessels. The pulmonary arteriovenous vessel segmentation data is used for characterizing the position and contour data of pulmonary artery vessels and pulmonary vein vessels. The lesion segmentation data corresponding to the lung horizontal scanning CT image is determined based on the lung horizontal scanning CT image, the lung horizontal scanning CT image can be input into a trained neural network model to obtain the lesion segmentation data, or the lesion segmentation data can be directly obtained by manually marking on the horizontal scanning CT image. The specific determination method of the lesion segmentation data is not limited in the present application as long as the lesion segmentation data corresponding to the lung flat-scan CT image is determined based on the lung flat-scan CT image.

The lung vessel segmentation data corresponding to the lung flat-scan CT image is determined based on the lung flat-scan CT image, and the lung arteriovenous vessel segmentation data is determined based on the lung flat-scan CT image, which are similar to the lesion segmentation data corresponding to the lung flat-scan CT image determined based on the lung flat-scan CT image, and are not repeated herein.

It should be noted that the pulmonary blood vessel refers to all blood vessels located in the lung. Pulmonary arteriovenous vessels include pulmonary arterial vessels as well as pulmonary venous vessels.

Step 102: and determining reinforced judgment information corresponding to the lung flat-scan CT image based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data.

Specifically, the enhancement state of the lesion region includes an enhanced state and an unenhanced state. The reinforcement judgment information is a judgment result of whether the lesion area is reinforced. The focus area is in a strengthened state, which indicates that the focus area has richer blood supply, indicates that the focus area has abnormal lesion, and indicates that the focus has high probability of malignancy. The lesion area is in an unreinforced state, which indicates that the blood flow of the lesion area is normal, and indicates that no abnormal lesion exists in the lesion area, indicating that the lesion is benign with a high probability.

Since the lesion segmentation data is used for representing the position and the contour of a lesion region, the pulmonary vessel segmentation data is used for representing the position and the contour of a pulmonary blood vessel, and the pulmonary arteriovenous blood vessel segmentation data is used for representing the position and the contour of a pulmonary artery blood vessel and a pulmonary vein blood vessel. Based on the information carried by the flat scanning CT image, the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data, the relationship between the lesion and the pulmonary vessel and the relationship between the lesion and the pulmonary vein can be better mastered, and the blood flow supply condition of the lesion area is judged, so that the strengthening state of the lesion area is judged.

In the embodiment of the application, lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image are determined based on the lung flat-scan CT image, and determination information is determined based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data to determine the enhancement state of a lesion region. And judging whether the lung has initial abnormal lesion or not according to the strengthening state of the lesion area. Since no enhanced CT examination is required, the need for contrast media injection is avoided, thereby avoiding trauma and potential risks associated with contrast media injection. Meanwhile, radiation to the patient during CT examination is avoided, and the patient seeing cost is reduced.

Fig. 4 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application. As shown in fig. 4, the step of determining the enhanced judgment information corresponding to the lung flat scan CT image based on the flat scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data includes the following steps.

Step 2021: determining an image characteristic diagram, a focus and pulmonary vessel contour position relation characteristic diagram and a focus and pulmonary artery and vein contour position relation characteristic diagram based on the lung flat scanning CT image, focus segmentation data, pulmonary vessel segmentation data and pulmonary artery and vein segmentation data.

Specifically, feature extraction is carried out on a lung flat scanning CT image to obtain an image feature map; and (3) taking into consideration the position relationship between the focus and the pulmonary vessel contour and the position relationship between the focus and the pulmonary arteriovenous vessel contour, and performing feature extraction and combination on the focus segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data to obtain a focus and pulmonary vessel contour position relationship feature map and a focus and pulmonary arteriovenous vessel contour position relationship feature map. And (4) extracting and combining features of different dimensions of different images, and reflecting the features of the focus and the blood vessel from different directions.

Step 2022: and connecting the image characteristic diagram, the position relation characteristic diagram of the lesion and the pulmonary vessel contour and the position relation characteristic diagram of the lesion and the pulmonary artery and vein contour in series in a vector form to form a one-dimensional characteristic vector.

Specifically, feature graphs of different dimensions are connected in series in a vector form to form a one-dimensional feature vector, so that feature information of different dimensions is accurately transmitted, and more accurate enhanced judgment information is obtained subsequently.

Step 2023: and inputting the one-dimensional feature vector into an enhanced judgment module to generate enhanced judgment information.

Specifically, the enhancement discrimination module is trained in advance, and has a function of discriminating whether the focus region is enhanced by inputting the one-dimensional feature vector, so as to judge whether the focus region has abnormal lesions.

In the embodiment of the application, feature extraction is respectively performed on a lung flat-scan CT image, focus segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data, and different-dimension feature maps of an image feature map, a focus and pulmonary vessel contour position relationship feature map and a focus and pulmonary arteriovenous vessel contour position relationship feature map are obtained based on the combination of a focus and pulmonary vessel contour position relationship and a focus and pulmonary arteriovenous vessel contour position relationship, the different-dimension feature maps are connected in series in a vector form to form a one-dimensional feature vector, the one-dimensional feature vector is input into an enhancement judging module, whether a focus region is enhanced or not is judged, and therefore whether abnormal lesions exist in the focus region or not is judged. By the method, whether abnormal lesions exist in the lesion area can be judged, and the contrast medium can be prevented from being injected without enhanced CT (computed tomography) examination, so that the trauma and the potential risk of the contrast medium injection are avoided.

Fig. 5 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application. As shown in fig. 5, the steps of determining an image feature map, a lesion and pulmonary vessel contour position relationship feature map, and a lesion and pulmonary artery and vein contour position relationship feature map based on the lung flat scan CT image, lesion segmentation data, pulmonary vessel segmentation data, and pulmonary artery and vein segmentation data include the following steps.

Step 30211: and determining flat-scan CT image block data, focus and pulmonary blood vessel contour position information block data and focus and pulmonary arteriovenous blood vessel contour position information block data based on the lung flat-scan CT image, focus segmentation data, pulmonary blood vessel segmentation data and pulmonary arteriovenous blood vessel segmentation data.

Specifically, the block data of the flat-scan CT image is an image block cut out from the flat-scan CT image with the lesion area as the center. The focus and lung blood vessel contour position information block data is combined with two image blocks which are cut in focus segmentation data and lung blood vessel segmentation data by taking a focus as a center, and the block data is used for representing the position relation of the focus and the lung blood vessel contour. The focus and pulmonary arteriovenous vessel contour position information block data is combined with two image blocks which are cut in focus segmentation data and pulmonary arteriovenous vessel segmentation data by taking the focus as a center, and the block data is used for representing the position relation of the focus and the pulmonary arteriovenous vessel contour.

Step 30212: and inputting the block data of the flat-scan CT image into a flat-scan CT image analysis module to generate an image characteristic map.

Specifically, the flat-scan CT image analysis module is also a feature analysis and extraction module composed of a neural network model, and performs 3d convolution operation and 3d pooling operation on the flat-scan CT image block data to obtain an image feature map.

Step 30213: and inputting the data of the lesion and pulmonary vessel contour position information block into a lesion and pulmonary vessel contour position relation analysis module to generate a lesion and pulmonary vessel contour position relation characteristic diagram.

Specifically, the lesion and pulmonary vessel contour position relation analysis module is also a feature analysis extraction module formed by a neural network model, and performs 3d convolution operation and 3d pooling operation on lesion and pulmonary vessel contour position information block data to obtain a lesion and pulmonary vessel contour position relation feature map.

Step 30214: and inputting the data of the lesion and pulmonary arteriovenous vessel contour position information block into a lesion and pulmonary arteriovenous vessel contour position relation analysis module to generate a lesion and pulmonary arteriovenous vessel contour position relation characteristic diagram.

Specifically, the lesion and pulmonary arteriovenous vessel contour position relation analysis module is also a feature analysis extraction module formed by a neural network model, and performs 3d convolution operation and 3d pooling operation on lesion and pulmonary arteriovenous vessel contour position information block data to obtain a lesion and pulmonary arteriovenous vessel contour position relation feature map.

In the embodiment of the application, image blocks are intercepted from a lung flat scanning CT image, focus segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data, and are combined based on the focus-pulmonary vessel contour position relationship and the focus-pulmonary arteriovenous vessel contour position relationship to obtain flat scanning CT image block data, focus-pulmonary vessel contour position information block data and focus-pulmonary arteriovenous vessel contour position information block data, and the focus-pulmonary arteriovenous vessel contour position information block data, which are respectively input into corresponding analysis modules to be subjected to feature extraction, so that an image feature map, a focus-pulmonary vessel contour position relationship feature map and a focus-pulmonary arteriovenous vessel contour position relationship feature map are obtained.

The feature analysis and extraction module may employ neural network models such as RNN, CNN, and transform.

Fig. 6 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application. As shown in fig. 6, the step of determining flat-scan CT image block data, lesion and pulmonary vessel contour position information block data, lesion and pulmonary arteriovenous vessel contour position information block data based on the lung flat-scan CT image, lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data includes the following steps.

Step 402111: flat-scan CT image block data corresponding to a lesion region of the lung flat-scan CT image is determined based on the lung flat-scan CT image.

Specifically, the flat-scan CT image block data that is determined based on the lung flat-scan CT image to correspond to the lesion region of the lung flat-scan CT image may be an image block that is cut out in the lung flat-scan CT image based on the entire region contour line of the lesion region, or an image block that is cut out in the lung flat-scan CT image based on the coordinates at the center of the lesion region. The method and the device do not specifically limit how to obtain the block data of the flat-scan CT image corresponding to the lesion region of the lung flat-scan CT image.

Step 402112: and determining lesion segmentation block data, pulmonary vessel segmentation block data and pulmonary arteriovenous vessel segmentation block data corresponding to a lesion region of the lung flat scan CT image based on the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data.

Specifically, the lesion segmentation block data may be extracted from the lesion segmentation data, the pulmonary blood vessel segmentation data, and the pulmonary arteriovenous blood vessel segmentation data based on the entire region contour line of the lesion region, or based on the coordinates of the center of the lesion region, and the pulmonary blood vessel segmentation block data and the pulmonary arteriovenous blood vessel segmentation block data are not described herein again.

Step 402113: and combining the lesion segmentation block data and the pulmonary vessel segmentation block data to obtain lesion and pulmonary vessel contour position information block data.

Step 402114: and combining the focus segmentation block data and the pulmonary arteriovenous vessel segmentation block data to obtain focus and pulmonary arteriovenous vessel contour position information block data.

In the embodiment of the application, flat-scan CT image block data corresponding to a lesion area of a lung flat-scan CT image are determined based on the lung flat-scan CT image; and determining focus segmentation block data corresponding to a focus region of the lung flat scan CT image, pulmonary vessel segmentation block data and pulmonary arteriovenous vessel segmentation block data based on the focus segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data, and combining the focus segmentation block data and the pulmonary vessel segmentation block data to obtain focus and pulmonary vessel contour position information block data for representing the position relationship of the focus and the pulmonary vessel contour. And merging the focus segmentation block data and the pulmonary arteriovenous vessel segmentation block data to obtain focus and pulmonary arteriovenous vessel contour position information block data used for representing the position relationship of the focus and the pulmonary arteriovenous vessel contour.

In one embodiment, the step of determining the flat-scan CT image patch data corresponding to a lesion region of the lung flat-scan CT image based on the lung flat-scan CT image comprises: and intercepting flat-scan CT image block data with a preset size from the lung flat-scan CT image based on the coordinates of the center of the lesion area.

Specifically, according to a first coordinate of a focus region center in focus segmentation data, an image block with a preset size is intercepted at a coordinate corresponding to the first coordinate in a flat-scan CT medical image, and CT image block data are obtained.

Determining lesion segmentation block data corresponding to a lesion region of the lung flat scan CT image based on lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data, wherein the steps of the lesion segmentation block data and the pulmonary arteriovenous vessel segmentation block data comprise: based on the coordinates of the focus region center, flat scanning CT image block data, blood vessel segmentation block data and pulmonary arteriovenous blood vessel segmentation block data with preset sizes are intercepted from focus segmentation data, pulmonary blood vessel segmentation data and pulmonary arteriovenous blood vessel segmentation data.

Specifically, image blocks with preset sizes are intercepted at coordinates corresponding to first coordinates in the focus segmentation data, and focus segmentation block data are obtained; intercepting an image block with a preset size at a coordinate corresponding to the first coordinate in the pulmonary vessel segmentation data to obtain pulmonary vessel segmentation block data; and intercepting image blocks with preset sizes at the coordinates corresponding to the first coordinates in the pulmonary arteriovenous vessel segmentation data, and pulmonary arteriovenous vessel segmentation block data.

In the embodiment of the present application, based on the coordinates of the center of the lesion region, lesion segmentation block data, pulmonary vessel segmentation block data, and pulmonary arteriovenous vessel segmentation block data are respectively cut out from the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data. And combining the lesion segmentation block data and the pulmonary vessel segmentation block data to obtain lesion and pulmonary vessel contour position information block data used for representing the relationship between the lesion and the pulmonary vessel contour position. And merging the focus segmentation block data and the pulmonary arteriovenous vessel segmentation block data to obtain focus and pulmonary arteriovenous vessel contour position information block data used for representing the position relationship of the focus and the pulmonary arteriovenous vessel contour.

Fig. 7 is a schematic flow chart illustrating a process of obtaining enhanced determination information corresponding to a flat-scan lung CT image based on flat-scan CT image block data, lesion and pulmonary artery and vein contour position information block data, and lesion and pulmonary artery and vein contour position information block data according to an embodiment of the present application. As shown in fig. 7, the image processing method includes: respectively inputting the flat-scan CT image block data, the focus and pulmonary vessel contour position information block data and the focus and pulmonary arteriovenous vessel contour position information block data into a corresponding flat-scan CT image analysis module, a focus and pulmonary vessel contour position relation analysis module and a focus and pulmonary arteriovenous vessel contour position relation analysis module, and respectively extracting features to obtain an image feature map, a focus and pulmonary vessel contour position relation feature map and a focus and pulmonary arteriovenous vessel contour position relation feature map. And connecting the image characteristic diagram, the position relation characteristic diagram of the lesion and the pulmonary vessel contour and the position relation characteristic diagram of the lesion and the pulmonary artery and vein contour in series in a vector form to form a one-dimensional characteristic vector. And inputting the one-dimensional characteristic vector into an enhancement judging module to obtain enhancement judging information and judging whether the focus area is enhanced or not so as to judge whether abnormal lesions exist in the focus area or not.

Fig. 8 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application. As shown in fig. 8, before inputting the one-dimensional feature vector to the enhanced discrimination module, the image processing method further includes the following steps.

Step 5001: determining a lung flat-scan CT image sample, a one-dimensional characteristic vector sample corresponding to the lung flat-scan CT image sample, and a reinforced judgment information sample of a lesion region corresponding to the lung flat-scan CT image sample.

Specifically, the method for acquiring the one-dimensional feature vector sample corresponding to the lung flat scan CT image sample may be as follows. And determining a lesion segmentation data sample, a pulmonary blood vessel segmentation data sample and a pulmonary arteriovenous blood vessel segmentation data sample corresponding to the lung flat-scan CT image sample based on the lung flat-scan CT image sample. Intercepting a flat-scan CT image block data sample from a lung flat-scan CT image sample based on the position of a focus region; based on the position of the focus area, respectively cutting focus segmentation block data samples, pulmonary vessel segmentation block data samples and pulmonary arteriovenous vessel segmentation block data samples from the focus segmentation data samples, the pulmonary vessel segmentation data samples and the pulmonary arteriovenous vessel segmentation data samples. Merging the focus segmentation block data sample and the pulmonary vessel segmentation block data sample to obtain a focus and pulmonary vessel contour position information block data sample; and merging the focus segmentation block data sample and the pulmonary arteriovenous vessel segmentation block data sample to obtain a focus and pulmonary arteriovenous vessel contour position information block data sample. Respectively inputting a flat scanning CT image block data sample, a focus and pulmonary vessel contour position information block data sample and a focus and pulmonary arteriovenous vessel contour position information block data sample into a flat scanning CT image analysis module, a focus and pulmonary vessel contour position relation analysis module and a focus and pulmonary arteriovenous vessel contour position relation analysis module for feature extraction to obtain an image feature map sample, a focus and pulmonary vessel contour position relation feature map sample and a focus and pulmonary arteriovenous vessel contour position relation feature map sample. And serially connecting the image characteristic map sample, the lesion and pulmonary vessel contour position relation characteristic map sample and the lesion and pulmonary artery and vein contour position relation characteristic map sample in a vector form to form a one-dimensional characteristic vector sample.

The reinforced judgment information sample of the lesion region corresponding to the lung flat-scan CT image sample is the reinforced judgment information of the lesion region marked on the lung flat-scan CT image sample, and includes a judgment information sample in which the lesion region is reinforced and a judgment information sample in which the lesion region is not reinforced. In supervised learning, a reinforced judgment information sample of a focus region in the lung flat-scan CT sample needs to be obtained, and the specific obtaining means may be to obtain the reinforced judgment information sample of the focus region corresponding to the lung flat-scan CT image sample according to the reinforced judgment information of the focus region in the lung reinforced CT image corresponding to the lung flat-scan CT image sample. As long as the enhanced judgment information sample of the lesion region corresponding to the lung flat scan CT image sample can be obtained, the specific obtaining means is not limited in the embodiment of the present application.

Step 5002: and inputting the one-dimensional characteristic vector sample into an initial enhancement judging module to generate enhancement judging information corresponding to the lung flat-scan CT image sample.

Specifically, the supervised learning training initial enhancement discrimination module inputs the one-dimensional feature vector sample into the initial enhancement discrimination module, and trains the initial enhancement discrimination module to enable the initial enhancement discrimination module to learn how to judge the enhancement state of the oven area.

Illustratively, the enhanced discrimination module may be implemented using a classifier.

As further examples, the classifier may be a SVM, Random Forest or a fully connected layer classifier.

Step 5003: and adjusting the network parameters of the initial enhancement judging module based on the enhancement judging information sample of the lesion region corresponding to the lung flat-scan CT image sample and the enhancement judging information corresponding to the lung flat-scan CT image sample.

Specifically, the enhanced judgment information sample of the lesion area corresponding to the lung flat-scan CT image sample is an input reference value, the judgment information corresponding to the lung flat-scan CT image sample is an output value, and the network parameter of the initial enhanced judgment module needs to be adjusted if the output value is different from the input reference value. And adjusting the network parameters of the initial enhancement judging module based on the enhancement judging information sample of the lesion area corresponding to the lung flat-scan CT image sample and the judging information corresponding to the lung flat-scan CT image sample until the loss result of the enhancement judging information of the lesion area corresponding to the lung flat-scan CT image sample and the enhancement judging information corresponding to the lung flat-scan CT image sample is within a threshold value, and stopping adjusting the parameters of the initial enhancement judging module.

It should be noted that the threshold is determined according to a specific training scenario of the enhanced discrimination module, and a specific numerical value of the threshold is not limited in the embodiment of the present application.

In the embodiment of the application, the initial enhancement judging module is trained through the method, so that the initial enhancement judging module has the capability of inputting the one-dimensional feature vector and outputting the enhancement state of the focus area. The one-dimensional feature vector is input into the trained initial enhancement judging module to obtain the enhancement state of the focus area, and whether the focus area is enhanced or not is judged, so that whether abnormal lesions exist in the focus area or not is judged.

Fig. 9 is a schematic flowchart illustrating an image processing method according to another embodiment of the present application. As shown in fig. 9, the step of determining the information sample for enhancing the lesion area corresponding to the lung flat scan CT image sample includes the following steps.

Step 60011: and acquiring a reinforced judgment information sample of the lesion region corresponding to the lung flat-scan CT image sample based on the reinforced judgment information of the lesion region on the reinforced CT image sample corresponding to the lung flat-scan CT image sample.

In particular, the enhanced CT image sample may be obtained from a database of enhanced CT images of historical patients, as long as the lung plan scan CT image sample and the enhanced CT image sample correspond to one patient.

In the embodiment of the application, a reinforced judgment information sample of a lesion area corresponding to the lung flat scan CT image sample is obtained according to the reinforced judgment information of the lesion area in the lung reinforced CT image corresponding to the lung flat scan CT image. If the focus region in the lung enhanced CT image corresponding to the lung flat scan CT image is in a strengthened state, the strengthened judgment information sample of the focus region is a strengthened judgment information sample of the focus region. If the focus region in the lung enhanced CT image corresponding to the lung flat scan CT image is in an unreinforced state, the enhanced judgment information of the focus region is a judgment information sample that the focus region is not enhanced.

Fig. 10 is a flowchart illustrating an image processing method according to yet another embodiment of the present application. As shown in fig. 10, the step of determining lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image based on the lung flat-scan CT image includes the following steps.

Step 7011: and inputting the lung flat scanning CT image into a lesion segmentation model to obtain lesion segmentation data.

The focus segmentation model is a trained neural network model, and the neural network model is trained through supervised learning, so that the neural network model has the function of inputting lung flat scanning CT images and outputting focus segmentation data.

Step 7012: and inputting the lung flat scanning CT image into a blood vessel segmentation model to obtain lung blood vessel segmentation data.

The blood vessel segmentation model is a trained neural network model, and the neural network model is trained through supervised learning, so that the neural network model has the function of inputting a lung flat scanning CT image and outputting lung blood vessel segmentation data.

Step 7013: and inputting the lung flat scanning CT image into an arteriovenous segmentation model to obtain pulmonary arteriovenous vessel segmentation data.

The arteriovenous segmentation model is a trained neural network model, and the neural network model is trained through supervised learning, so that the neural network model has the function of inputting a lung flat scanning CT image and outputting pulmonary arteriovenous vessel segmentation data. The lesion segmentation model, the blood vessel segmentation model and the arteriovenous segmentation model can be obtained by training at least one of nerve network models such as MobilenetV2, Fast-RCNN, R-FCN, YOLO and SSD.

In the embodiment of the application, the lesion segmentation data, the pulmonary blood vessel segmentation data and the pulmonary arteriovenous vessel segmentation data are obtained by respectively inputting the lung flat scan CT image into the lesion segmentation model, the blood vessel segmentation model and the arteriovenous segmentation model.

Fig. 11 is a flowchart illustrating an image processing method according to another embodiment of the present application. As shown in fig. 11, before determining lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image based on the lung flat-scan CT image, the image processing method further includes the following steps.

Step 804: and carrying out normalization processing on the lung flat scanning CT image.

In addition, in the embodiment of the present application, the step of determining lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image based on the lung flat-scan CT image includes the following steps.

Step 801: and determining lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the normalized pulmonary flat-scan CT image based on the normalized pulmonary flat-scan CT image.

Specifically, the lung flat-scan CT image is normalized to obtain a normalized flat-scan CT image; based on the normalized lung flat scan CT image, determining focus segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data; the specifications of the lung flat scanning CT image, the lesion segmentation data, the lung blood vessel segmentation data and the lung arteriovenous blood vessel segmentation data are consistent, and convenience is brought to subsequent feature extraction.

In the embodiment, the lung flat-scan CT images are normalized in consideration of the difference of flat-scan CT images obtained by CT instruments of different models, so that interference factors are eliminated in subsequent processing, and information such as focuses is guaranteed to be accurately transmitted.

In one embodiment, determining pulmonary arteriovenous vessel segmentation data corresponding to the pulmonary flat scan CT image based on the normalized pulmonary flat scan CT image comprises: and normalizing the pixel distance in the coordinate vector direction in the flat-scan CT medical image to a preset value.

Specifically, for convenience of calculation, coordinates are established in the flat-scan CT medical image, which are typically XYZ three-dimensional coordinate vectors. The pixel distance in the coordinate vector direction in the flat-scan CT medical image is normalized to a preset value, which may be that the inter-pixel distance in the X, Y, Z direction is normalized to 1mm, or the inter-pixel distance in the X, Y, Z direction is normalized to 5mm, and the like. The preset value is set according to a specific application scene, and the specific numerical value of the preset value is not limited in the embodiment of the application.

Exemplary information processing apparatus

Fig. 12 is a schematic structural diagram of an image processing apparatus according to an embodiment of the present application. As shown in fig. 12, the image processing apparatus 100 includes a segmentation data determining module 101 configured to determine lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to a lung flat-scan CT image based on the lung flat-scan CT image; and an enhanced judgment information determining module 102 configured to determine enhanced judgment information corresponding to the lung flat-scan CT image based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data

In this embodiment of the application, the segmentation data determining module 101 determines lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image based on the lung flat-scan CT image, and the determination information determining module 102 determines determination information based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data, and determines a reinforcement state of a lesion region. Since no enhanced CT examination is required, the need for contrast media injection is avoided, thereby avoiding trauma and potential risks associated with contrast media injection. Meanwhile, radiation to the patient during CT examination is avoided, and the patient seeing cost is reduced.

Fig. 13 is a schematic structural diagram of an image processing apparatus according to yet another embodiment of the present application. As shown in fig. 13, the enhancement judgment information determination module 102 includes a feature extraction unit 1021: the method comprises the steps of configuring a focus segmentation data, a lung flat scanning CT image, a pulmonary vessel segmentation data and a pulmonary arteriovenous vessel segmentation data, determining an image characteristic diagram, a focus and pulmonary vessel contour position relation characteristic diagram and a focus and pulmonary arteriovenous vessel contour position relation characteristic diagram; the feature concatenation unit 1022 is configured to concatenate the image feature map, the lesion and pulmonary vessel contour position relationship feature map, and the lesion and pulmonary arteriovenous vessel contour position relationship feature map in a vector form to form a one-dimensional feature vector; and a judging unit 1023 configured to input the one-dimensional feature vector to the enhancement judging module to generate enhancement judging information.

In one embodiment, as shown in fig. 13, the feature extraction unit 1021 includes a block data determination subunit 10211 configured to determine flat-scan CT image block data, lesion and pulmonary vessel contour position information block data, lesion and pulmonary arteriovenous vessel contour position information block data, based on the lung flat-scan CT image, lesion segmentation data, pulmonary vessel segmentation data, pulmonary arteriovenous vessel segmentation data; flat scan CT image analysis subunit 10212: analyzing the block data of the flat scanning CT image to generate an image characteristic map; a lesion and pulmonary vessel contour position relationship analysis subunit 10213 configured to analyze data based on the lesion and pulmonary vessel contour position information block, and generate a feature map of the lesion and pulmonary vessel contour position relationship; and a lesion and pulmonary arteriovenous vessel contour position relationship analysis subunit 10214 configured to analyze lesion and pulmonary arteriovenous vessel contour position information block data, and generate a lesion and pulmonary arteriovenous vessel contour position relationship feature map.

In one embodiment, as shown in fig. 13, the block data determination subunit 10211 further includes: a first block data determination subunit 102111 configured to determine, based on the lung flat-scan CT image, flat-scan CT image block data corresponding to a lesion region of the lung flat-scan CT image; a second block data determination subunit 102112 configured to determine lesion segmentation block data, pulmonary vessel segmentation block data, and pulmonary arteriovenous vessel segmentation block data corresponding to a lesion region of the lung flat-scan CT image based on the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data; a first block data merging subunit 102113 configured to merge lesion segmentation block data and pulmonary vessel segmentation block data to obtain lesion and pulmonary vessel contour position information block data; a second block data merging subunit 102114 configured to merge the lesion segmentation block data and the pulmonary arteriovenous vessel segmentation block data to obtain lesion and pulmonary arteriovenous vessel contour position information block data.

In one embodiment, the first block data determining subunit 102111 is further configured to cut out a flat-scan CT image block data of a preset size in the lung flat-scan CT image based on the coordinates of the lesion region center, and the second block data determining subunit 102112 is further configured to cut out a flat-scan CT image block data, a vessel segmentation block data, a pulmonary arteriovenous vessel segmentation block data of a preset size in the lesion segmentation data, the pulmonary vessel segmentation data, the pulmonary arteriovenous vessel segmentation data based on the coordinates of the lesion region center.

In a further embodiment, the first block data determination subunit 102111 is integrated with the second block data determination subunit 102112.

In a further embodiment, the first block data merge sub-unit 102113 is integrated with the second block data merge sub-unit 102114.

In one embodiment, as shown in fig. 13, the image processing apparatus 100 further includes a training module 103, and the training module 103 includes: the sample determination unit 1031 is configured to determine a lung flat-scan CT image sample, a one-dimensional feature vector sample corresponding to the lung flat-scan CT image sample, and a reinforced judgment information sample of a lesion region corresponding to the lung flat-scan CT image sample; the enhancement judgment information generation unit 1032 inputs the one-dimensional feature vector sample into the initial enhancement judgment module to generate enhancement judgment information corresponding to the lung flat scan CT image sample; an adjusting unit 1033 configured to adjust a network parameter of the initial enhancement discrimination module based on the marking information and the judgment information corresponding to the lung flat scan CT image sample.

In a further embodiment, the enhanced discrimination module may be implemented using a classifier.

In yet a further embodiment, the classifier may be an SVM, Random Forest or a fully connected layer classifier.

In one embodiment, as shown in fig. 13, the sample determination unit 1031 is further configured to obtain an enhanced determination information sample of a lesion region corresponding to the lung flat scan CT image sample based on the enhanced determination information of the lesion region on the enhanced CT image sample corresponding to the lung flat scan CT image sample.

In one embodiment, as shown in fig. 13, the segmentation data determination module 101 includes: a lesion segmentation data acquisition unit 1011 configured to input the lung flat scan CT image into a lesion segmentation model to obtain lesion segmentation data; a pulmonary vessel segmentation data acquisition unit 1012 configured to input the lung flat scan CT image into a vessel segmentation model to obtain pulmonary vessel segmentation data; a pulmonary arteriovenous vessel segmentation data acquisition unit 1013 configured to input the pulmonary flat scan CT image into an arteriovenous segmentation model to obtain pulmonary arteriovenous vessel segmentation data.

In one embodiment, as shown in fig. 13, the image processing apparatus 100 further includes a preprocessing module 104 configured to perform a normalization process on the lung flat-scan CT image, wherein the segmentation data determining module 101 is further configured to determine lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image based on the normalized lung flat-scan CT image.

In one embodiment, as shown in fig. 13, the preprocessing module 104 is further configured to normalize the pixel distance in the direction of the coordinate vector in the lung swept-flat CT image to a preset value.

It should be understood that the segmentation data determination module 101, the enhanced judgment information determination module 102, the training module 103, the preprocessing module 104, the lesion segmentation data acquisition unit 1011, the pulmonary vessel segmentation data acquisition unit 1012, and the pulmonary arteriovenous vessel segmentation data acquisition unit 1013 in the segmentation data determination module 101, the feature extraction unit 1021, the feature concatenation unit 1022, and the judgment unit 1023 in the enhanced judgment information determination module 102, the sample determination unit 1031, the enhanced judgment information generation unit 1032, and the adjustment unit 1033 in the training module 103, the block data determination subunit 10211, the flat scan CT image analysis subunit 10212, the lesion-to-pulmonary vessel contour position relationship analysis subunit 10213, and the lesion-to-pulmonary arteriovenous vessel contour position relationship analysis subunit 10214 in the feature extraction unit 1021, the first block data determination subunit 102111 in the determination subunit 10211, the enhanced judgment information determination module 102, the training module 103, and the segmentation data acquisition unit 101, the pulmonary vessel segmentation data acquisition unit 1012, the enhanced judgment information acquisition unit 1012, and the adjustment unit 1033 provided in fig. 11 to fig. 12, The operations and functions of the second block data determining subunit 102112, the first block data merging subunit 102113, and the second block data merging subunit 102114 may refer to the image processing methods provided in fig. 3 to 11, and are not described herein again to avoid redundancy.

Exemplary electronic device

Fig. 14 is a schematic structural diagram of an electronic device according to an embodiment of the present application. As shown in fig. 14, the electronic device 200 includes one or more processors 210 and memory 220.

The processor 210 may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 200 to perform desired functions.

Memory 220 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, Random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, Read Only Memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium and executed by the processor 210 to implement the image processing methods of the various embodiments of the present application described above and/or other desired functions.

In one example, the electronic device 200 may further include: an input device 230 and an output device 240, which are interconnected by a bus system and/or other form of connection mechanism (not shown).

For example, the input device 230 may be the above-mentioned apparatus for performing flat-scan CT detection.

The output means 240 may output various information, information for judging the enhancement status of the lesion area, and the like to the outside, and the output device 240 may include, for example, a display, a printer, and a communication network and a remote output device connected thereto, and the like.

Of course, for the sake of simplicity, only some of the components related to the present application in the electronic apparatus 200 are shown in fig. 14, and components such as a bus, an input/output interface, and the like are omitted. In addition, electronic device 200 may include any other suitable components depending on the particular application.

Exemplary computer program product and computer-readable storage Medium

In addition to the above-described methods and apparatus, embodiments of the present application may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the image processing method according to various embodiments of the present application described in the "example line image processing method" section of this specification, above.

The computer program product may write program code for carrying out operations for embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.

Furthermore, embodiments of the present application may also be a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, cause the processor to perform the steps in the image processing method according to various embodiments of the present application described in the "image processing method" section above in this specification.

The computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application.

The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.

Claims

1. An image processing method, comprising:

determining lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat-scan CT image based on the lung flat-scan CT image; and

determining reinforced judgment information corresponding to the lung flat scanning CT image based on the flat scanning CT image, the focus segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data;

wherein the determining of the reinforced judgment information corresponding to the lung flat scan CT image based on the flat scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data includes:

determining an image characteristic diagram, a focus and pulmonary vessel contour position relation characteristic diagram and a focus and pulmonary vessel contour position relation characteristic diagram based on the lung flat scanning CT image, the focus segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data;

connecting the image feature map, the lesion and pulmonary vessel contour position relationship feature map and the lesion and pulmonary artery and vein contour position relationship feature map in series in a vector form to form a one-dimensional feature vector;

inputting the one-dimensional feature vector into an enhancement judgment module to generate the enhancement judgment information;

wherein the determining an image feature map, a lesion and pulmonary vessel contour position relationship feature map, a lesion and pulmonary artery and vein contour position relationship feature map based on the lung flat scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary artery and vein segmentation data includes:

determining flat-scan CT image block data, focus and pulmonary vessel contour position information block data, and focus and pulmonary artery and vein contour position information block data based on the lung flat-scan CT image, the focus segmentation data, the pulmonary vessel segmentation data, and the pulmonary artery and vein segmentation data;

inputting the block data of the flat-scan CT image into a flat-scan CT image analysis module to generate the image feature map;

inputting the data of the lesion and pulmonary vessel contour position information block into a lesion and pulmonary vessel contour position relation analysis module to generate a feature map of the lesion and pulmonary vessel contour position relation; and

inputting the data of the lesion and pulmonary arteriovenous vessel contour position information block into a lesion and pulmonary arteriovenous vessel contour position relationship analysis module to generate a lesion and pulmonary arteriovenous vessel contour position relationship feature map;

before the one-dimensional feature vector is input into an enhanced discrimination module, the method further includes:

determining a lung flat-scan CT image sample, a one-dimensional feature vector sample corresponding to the lung flat-scan CT image sample, and a reinforced judgment information sample of a lesion region corresponding to the lung flat-scan CT image sample;

inputting the one-dimensional feature vector sample into an initial enhancement judging module to generate enhancement judging information corresponding to the lung flat scanning CT image sample; and

and adjusting the network parameters of the initial enhancement judging module based on the enhancement judging information sample of the lesion region corresponding to the lung flat-scan CT image sample and the enhancement judging information corresponding to the lung flat-scan CT image sample.

2. The image processing method according to claim 1, wherein said determining flat-scan CT image block data, lesion and pulmonary vessel contour position information block data, lesion and pulmonary arteriovenous vessel contour position information block data based on the lung flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, the pulmonary arteriovenous vessel segmentation data comprises:

determining the flat-scan CT image block data corresponding to a lesion region of the lung flat-scan CT image based on the lung flat-scan CT image;

determining lesion segmentation block data, pulmonary vessel segmentation block data and pulmonary arteriovenous vessel segmentation block data corresponding to a lesion region of the lung flat scan CT image based on the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data;

merging the lesion segmentation block data and the pulmonary vessel segmentation block data to obtain lesion and pulmonary vessel contour position information block data; and

and merging the focus segmentation block data and the pulmonary arteriovenous vessel segmentation block data to obtain the focus and pulmonary arteriovenous vessel contour position information block data.

3. The image processing method according to claim 2, wherein the determining the flat-scan CT image block data corresponding to a lesion region of the lung flat-scan CT image based on the lung flat-scan CT image includes:

intercepting flat-scan CT image block data with a preset size from the lung flat-scan CT image based on the coordinate of the focus region center;

wherein the determining of lesion segmentation block data, pulmonary vessel segmentation block data, and pulmonary arteriovenous vessel segmentation block data corresponding to a lesion region of the lung flat scan CT image based on the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data includes:

and based on the coordinates of the focus region center, intercepting flat scanning CT image block data, blood vessel segmentation block data and pulmonary arteriovenous blood vessel segmentation block data with preset sizes from the focus segmentation data, the pulmonary blood vessel segmentation data and the pulmonary arteriovenous blood vessel segmentation data.

4. The image processing method according to claim 1, wherein the determining the enhancement judgment information sample of the lesion region corresponding to the lung flat scan CT image sample comprises:

and determining a reinforced judgment information sample of the lesion region corresponding to the lung flat scan CT image sample based on the reinforced judgment information of the lesion region of the reinforced CT image sample corresponding to the lung flat scan CT image sample.

5. The image processing method according to any one of claims 1 to 4, wherein the determining of lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat scan CT image based on the lung flat scan CT image comprises:

inputting the lung flat scanning CT image into a lesion segmentation model to obtain the lesion segmentation data;

inputting the lung flat scanning CT image into a blood vessel segmentation model to obtain the lung blood vessel segmentation data; and

and inputting the lung flat scanning CT image into an arteriovenous segmentation model to obtain the pulmonary arteriovenous vessel segmentation data.

6. The image processing method according to any one of claims 1 to 4, wherein before the determining, based on the lung scout CT image, lesion segmentation data, pulmonary vessel segmentation data, and pulmonary arteriovenous vessel segmentation data corresponding to the lung scout CT image, the method further comprises:

carrying out normalization processing on the lung flat scanning CT image;

the method for determining the lesion segmentation data, the pulmonary vessel segmentation data and the pulmonary arteriovenous vessel segmentation data corresponding to the pulmonary flat scan CT image respectively based on the pulmonary flat scan CT image comprises the following steps:

and determining lesion segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the normalized pulmonary flat-scan CT image based on the normalized pulmonary flat-scan CT image.

7. The image processing method of claim 6, wherein the normalizing the lung scout CT image comprises:

and normalizing the pixel distance in the coordinate vector direction in the lung flat scanning CT medical image to a preset value.

8. An image processing apparatus characterized by comprising:

the segmentation data determination module is configured to determine focus segmentation data, pulmonary vessel segmentation data and pulmonary arteriovenous vessel segmentation data corresponding to the lung flat scan CT image based on the lung flat scan CT image; and

an enhanced judgment information determining module configured to determine enhanced judgment information corresponding to the lung flat-scan CT image based on the flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary arteriovenous vessel segmentation data;

wherein, the strengthening judgment information determining module comprises:

a feature extraction unit, configured to determine an image feature map, a lesion and pulmonary artery and vein contour position relationship feature map, and a lesion and pulmonary artery and vein contour position relationship feature map based on the lung flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary artery and vein segmentation data;

the characteristic series connection unit is configured to serially connect the image characteristic diagram, the lesion and pulmonary vessel contour position relation characteristic diagram and the lesion and pulmonary arteriovenous vessel contour position relation characteristic diagram in a vector form to form a one-dimensional characteristic vector;

the judging unit is configured to input the one-dimensional feature vector into an enhancement judging module to generate the enhancement judging information;

wherein the feature extraction unit further includes:

a block data determination subunit configured to determine flat-scan CT image block data, lesion and pulmonary artery and vein contour position information block data, based on the lung flat-scan CT image, the lesion segmentation data, the pulmonary vessel segmentation data, and the pulmonary artery and vein segmentation data;

a flat-scan CT image analysis subunit configured to input the flat-scan CT image block data to a flat-scan CT image analysis module to generate the image feature map;

the lesion and pulmonary vessel contour position relation analysis subunit is configured to input data of the lesion and pulmonary vessel contour position information block into a lesion and pulmonary vessel contour position relation analysis module to generate a feature map of the lesion and pulmonary vessel contour position relation;

the lesion and pulmonary arteriovenous blood vessel contour position relation analysis subunit is configured to input lesion and pulmonary arteriovenous blood vessel contour position information block data into a lesion and pulmonary arteriovenous blood vessel contour position relation analysis module so as to generate a lesion and pulmonary arteriovenous blood vessel contour position relation characteristic diagram;

wherein the image processing apparatus further comprises a training module, the training module further comprising:

the sample determining unit is configured to determine a lung flat-scan CT image sample, a one-dimensional feature vector sample corresponding to the lung flat-scan CT image sample, and a reinforced judgment information sample of a lesion region corresponding to the lung flat-scan CT image sample;

the reinforced judgment information generating unit is configured to input the one-dimensional feature vector sample into an initial reinforced judgment module so as to generate reinforced judgment information corresponding to the lung flat scan CT image sample;

and the adjusting unit is configured to adjust the network parameters of the initial enhancement judging module based on the enhancement judging information sample of the lesion region corresponding to the lung flat-scan CT image sample and the enhancement judging information corresponding to the lung flat-scan CT image sample.

9. An electronic device, comprising:

a processor; and

a memory having stored therein computer program instructions which, when executed by the processor, cause the processor to perform the image processing method of any of claims 1 to 7.

10. A computer readable storage medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to carry out the image processing method of any one of claims 1 to 7.