Method, device and equipment for automated identification of coronary artery origin anomaly and storage medium

Abstract

The application belongs to the technical field of medical image processing, and discloses a coronary artery origin anomaly automatic identification method, device, equipment and storage medium. The method comprises the following steps: determining an aorta segmentation binary image, a coronary artery multi-class segmentation image, an aortic sinus multi-class segmentation image and an aorta center line according to a coronary CTA image; intercepting the aortic sinus multi-class segmentation image to obtain a maximum aorta cross section; respectively performing inflation processing on a left coronary artery segmentation image and a right coronary artery segmentation image, and determining distances from a left centroid of a left connected domain and a right centroid of a right connected domain to the maximum aorta cross section as a first plane distance and a second plane distance based on the inflated left coronary artery segmentation image, the inflated right coronary artery segmentation image and the aorta segmentation binary image; and determining that the origin anomaly is a coronary artery high position when determining that the first plane distance or the second plane distance is greater than a preset distance. Through the above method, the coronary artery origin anomaly can be automatically identified.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical image processing, and particularly relates to a method and device for automatically identifying abnormal coronary artery origin, equipment and a storage medium. BACKGROUND

[0002] Cardiovascular disease has become one of the major diseases threatening human life safety. Doctors use blood vessel imaging technology to assist in the diagnosis of various blood vessel diseases. In clinical practice, the diagnosis of abnormal coronary artery origin of the cardiovascular system by doctors is very important. Abnormal coronary artery origin has various manifestations, from asymptomatic to angina pectoris, from myocardial infarction to heart failure, from arrhythmia to syncope and even sudden death, etc., all of which depend on whether it can cause myocardial ischemia and the degree of myocardial ischemia. In the traditional method, whether the coronary artery origin is abnormal is mainly determined by doctors manually judging the collected coronary angiography images, which has the problem of low determination efficiency. SUMMARY

[0003] The main purpose of the present application is to provide a method and device for automatically identifying abnormal coronary artery origin, equipment and a storage medium, which aims to solve the technical problem of low determination efficiency in the prior art.

[0004] To achieve the above purpose, the present application provides a method for automatically identifying abnormal coronary artery origin, which comprises the following steps:

[0005] Obtaining a coronary CTA image of a first coronary artery;

[0006] According to the coronary CTA image, determining a main aorta segmentation binary image, a coronary multi-class segmentation image and a main aorta sinus multi-class segmentation image of the first coronary artery, and extracting a main aorta centerline based on the main aorta segmentation binary image, wherein the coronary multi-class segmentation image includes a left coronary artery segmentation image and a right coronary artery segmentation image;

[0007] Respectively performing dilation processing on the left coronary artery segmentation image and the right coronary artery segmentation image to obtain a new left coronary artery segmentation image and a new right coronary artery segmentation image;

[0008] Using a target plane to intercept the main aorta sinus multi-class segmentation image to obtain a maximum aorta cross section, wherein the target plane is a plane perpendicular to the main aorta centerline;

[0009] Based on the new left coronary artery segmentation image, the new right coronary artery segmentation image and the main aorta segmentation binary image, determining a left connected domain and a right connected domain, and determining a left centroid of the left connected domain and a right centroid of the right connected domain;

[0010] determining a first plane distance as a distance from the left centroid to the maximum aortic cross section and a second plane distance as a distance from the right centroid to the maximum aortic cross section;

[0011] determining whether the first plane distance or the second plane distance is greater than a preset distance;

[0012] determining that the first coronary artery has an abnormal origin when it is determined that the first plane distance or the second plane distance is greater than the preset distance.

[0013] Optionally, the maximum aortic cross section includes a left sinus region, a right sinus region, and an innominate sinus region; wherein,

[0014] The determining whether the first plane distance or the second plane distance is greater than a preset distance further includes:

[0015] projecting the left centroid onto the maximum aortic cross section to obtain a left projection point when it is determined that the first plane distance or the second plane distance is not greater than the preset distance;

[0016] determining a first left distance, a second left distance, and a third left distance as distances from the left projection point to the left sinus region, the right sinus region, and the innominate sinus region, respectively;

[0017] determining whether the first left distance is a minimum distance among the first left distance, the second left distance, and the third left distance;

[0018] determining whether a minimum distance among the second left distance and the third left distance when the first left distance is not the minimum distance;

[0019] determining that the first coronary artery has an abnormal origin when it is determined that the minimum distance is the second left distance;

[0020] determining that the first coronary artery has an abnormal origin when it is determined that the minimum distance is the third left distance.

[0021] Optionally, the maximum aortic cross section includes a left sinus region, a right sinus region, and an innominate sinus region; wherein,

[0022] The determining whether the first plane distance or the second plane distance is greater than a preset distance further includes:

[0023] projecting the right centroid onto the maximum aortic cross section to obtain a right projection point when it is determined that the first plane distance or the second plane distance is not greater than the preset distance;

[0024] determining distances of the right projection point to the left sinus region, the right sinus region and the common sinus region as a first right distance, a second right distance and a third right distance respectively;

[0025] determining whether the second right distance is the minimum distance among the first right distance, the second right distance and the third right distance;

[0026] if not, determining the minimum distance among the first right distance and the third right distance;

[0027] if it is determined that the minimum distance is the first right distance, determining that the abnormal origin of the first coronary artery is that the right coronary originates from the left sinus;

[0028] if it is determined that the minimum distance is the third right distance, determining that the abnormal origin of the first coronary artery is that the right coronary originates from the common sinus.

[0029] Optionally, the dilating the left coronary artery segmentation image and the right coronary artery segmentation image respectively to obtain a new left coronary artery segmentation image and a new right coronary artery segmentation image comprises:

[0030] dilating the left coronary artery segmentation image so that the intersection of the new left coronary artery segmentation image and the aorta segmentation binary image is not an empty set;

[0031] dilating the right coronary artery segmentation image so that the intersection of the new right coronary artery segmentation image and the aorta segmentation binary image is not an empty set.

[0032] Optionally, the intercepting the aortic sinus multi-class segmentation image by using a target plane to obtain a maximum aortic cross section, wherein the target plane is a plane perpendicular to the aortic center line, comprises:

[0033] intercepting the aortic sinus multi-class segmentation image by using a target plane to obtain a plurality of binary segmentations seg_N, wherein each seg_N corresponds to a coronary sinus cross-sectional area, wherein the target plane is a plane perpendicular to the aortic center line;

[0034] taking the seg_N corresponding to the maximum aortic cross-sectional area as the maximum aortic cross section.

[0035] Optionally, the aortic sinus multi-class segmentation image comprises a left sinus label, a right sinus label and a common sinus label; wherein,

[0036] after taking the seg_N corresponding to the maximum aortic cross-sectional area as the maximum aortic cross section, further comprising:

[0037] determining a left sinus region, a right sinus region and a common sinus region on the maximum aortic cross section according to the left sinus label, the right sinus label and the common sinus label.

[0038] Optionally, the extracting the aorta centerline based on the aorta segmentation binary graph comprises:

[0039] The aorta centerline is extracted based on the aorta segmentation binary graph through deep learning or image processing.

[0040] In addition, to achieve the above object, the application further provides an automatic identification device for coronary artery origin anomaly, which comprises:

[0041] An acquisition module is configured to acquire a coronary CTA image of a first coronary artery;

[0042] The acquisition module is configured to determine an aorta segmentation binary graph, a coronary multi-class segmentation graph and an aorta sinus multi-class segmentation graph of the first coronary artery according to the coronary CTA image, and extract an aorta centerline based on the aorta segmentation binary graph, wherein the coronary multi-class segmentation graph comprises a left coronary segmentation graph and a right coronary segmentation graph.

[0043] An expansion module is configured to perform expansion processing on the left coronary segmentation graph and the right coronary segmentation graph respectively to obtain a new left coronary segmentation graph and a new right coronary segmentation graph.

[0044] The determination module is further configured to intercept the aorta sinus multi-class segmentation graph using a target plane to obtain a maximum aorta cross section, wherein the target plane is a plane perpendicular to the aorta centerline.

[0045] The determination module is further configured to determine a left connected domain and a right connected domain based on the new left coronary segmentation graph, the new right coronary segmentation graph and the aorta segmentation binary graph, and determine a left centroid of the left connected domain and a right centroid of the right connected domain.

[0046] The determination module is further configured to determine a first plane distance of the left centroid to the maximum aorta cross section and a second plane distance of the right centroid to the maximum aorta cross section.

[0047] A judgment module is configured to judge whether the first plane distance or the second plane distance is greater than a preset distance.

[0048] The determination module is further configured to determine that the origin anomaly of the first coronary artery is coronary artery high position when it is judged that the first plane distance or the second plane distance is greater than the preset distance.

[0049] In addition, to achieve the above object, the application further provides an automatic identification device for coronary artery origin anomaly, which comprises a memory, a processor and an automatic identification program for coronary artery origin anomaly stored in the memory and executable on the processor, and the automatic identification program for coronary artery origin anomaly is configured to implement the steps of the automatic identification method for coronary artery origin anomaly.

[0050] In addition, to achieve the above object, the application further provides a storage medium, which stores an automatic identification program for coronary artery origin anomaly, and the automatic identification program for coronary artery origin anomaly implements the steps of the automatic identification method for coronary artery origin anomaly when executed by a processor.

[0051] The automatic identification method for coronary artery origin anomaly, device, equipment and storage medium provided by the application can automatically identify and determine the abnormal category of the coronary artery origin anomaly without manual identification of the collected coronary angiography image by a doctor, which can improve the identification efficiency and save medical resources. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 is a structural schematic diagram of the automatic identification device for coronary artery origin anomaly of the hardware running environment related to the embodiment scheme of the application;

[0053] Figure 2 is a flowchart of the first embodiment of the automatic identification method for coronary artery origin anomaly of the application.

[0054] Figure 3 Flowchart of a second embodiment of the method for automatically identifying abnormal coronary artery origin of the application;

[0055] Figure 4 Flowchart of a third embodiment of the method for automatically identifying abnormal coronary artery origin of the application;

[0056] Figure 5 Structural block diagram of a first embodiment of the device for automatically identifying abnormal coronary artery origin of the application.

[0057] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0058] It should be understood that the specific embodiments described herein are intended to be illustrative only and not limiting of the application.

[0059] Reference Figure 1 , Figure 1 Structural diagram of the device for automatically identifying abnormal coronary artery origin of the hardware running environment involved in the embodiment of the application.

[0060] As Figure 1 shown, the device for automatically identifying abnormal coronary artery origin can include a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to realize the connection and communication between the components. The user interface 1003 can include a display screen, an input unit such as a keyboard, and can also include a standard wired interface, a wireless interface. The network interface 1004 can optionally include a standard wired interface, a wireless interface (such as a wireless fidelity (Wi-Fi) interface). The memory 1005 can be a high-speed random access memory (RAM) memory, or a stable non-volatile memory (NVM), such as a disk memory. The memory 1005 can also be a storage device independent of the aforementioned processor 1001.

[0061] Those skilled in the art can understand that Figure 1 the structure shown in the foregoing embodiments does not constitute a limitation on the device for automatically identifying abnormal coronary artery origin, and can include more or fewer components than the illustrated components, or combine certain components, or different component arrangements.

[0062] As shown in Figure 1 The memory 1005 as a storage medium can include an operating system, a network communication module, a user interface module, and an automatic identification program of coronary artery origin anomaly.

[0063] In Figure 1 The network interface 1004 is mainly used for data communication with a network server, and the user interface 1003 is mainly used for data interaction with a user. The processor 1001 and the memory 1005 in the automatic identification device of coronary artery origin anomaly can be arranged in the automatic identification device of coronary artery origin anomaly. The automatic identification device of coronary artery origin anomaly calls the automatic identification program of coronary artery origin anomaly stored in the memory 1005 through the processor 1001, and executes the automatic identification method of coronary artery origin anomaly provided in the embodiments of the present application.

[0064] Based on the above hardware structure, the automatic identification method of coronary artery origin anomaly is proposed.

[0065] Referring to Figure 2 , Figure 2 is a flowchart of a first embodiment of the automatic identification method of coronary artery origin anomaly.

[0066] In this embodiment, the automatic identification method of coronary artery origin anomaly includes the following steps:

[0067] Step S10: Obtain a coronary CTA image of a first coronary artery.

[0068] It should be noted that the execution subject of the present embodiment can be a computing service device with data processing, network communication and program running functions, such as a mobile phone, a tablet computer, a personal computer, etc., or an electronic device or an automatic identification device of coronary artery origin anomaly capable of realizing the above functions. In the following, the automatic identification device of coronary artery origin anomaly is taken as an example to describe the present embodiment and the following embodiments.

[0069] It should be noted that the coronary CTA image is obtained by CT angiography, which is a very important part of CT clinical application. Due to the poor natural contrast of coronary vessels and their background soft tissue, it is often difficult to display coronary vessels by conventional CT plain scan. When performing CTA examination, contrast agent needs to be introduced to change the image contrast of coronary vessels and background tissue, so as to highlight the coronary vessels. The coronary CTA image is a 3D image with good three-dimensional information. Through three-dimensional reconstruction of the coronary CTA image, the three-dimensional spatial information of the coronary vessels can be intuitively and stereoscopically presented. In addition, 3DVR, vascular cross section, 3D MIP, surface CPR, line CPR, slice graph and CT-FFR functional information can also be derived from the CTA image.

[0070] It should be noted that the abnormal origin of coronary artery can be high position of coronary artery (i.e. left coronary artery entry point or right coronary artery entry point is far away from the aortic sinus), or left coronary artery not originating from the left sinus, or right coronary artery not originating from the right sinus.

[0071] Step S20: determining an aorta segmentation binary graph, a coronary multi-class segmentation graph and an aortic sinus multi-class segmentation graph of the first coronary artery according to the coronary CTA image, and extracting an aorta centerline based on the aorta segmentation binary graph, wherein the coronary multi-class segmentation graph includes a left coronary segmentation graph and a right coronary segmentation graph.

[0072] It should be noted that the aortic sinus refers to the outward bulging of the arterial wall opposite to the aortic valve, and the inner cavity between the valve and the aortic wall.

[0073] In a specific implementation, the coronary CTA image can be input into the first deep convolutional network to obtain the main coronary segmentation binary graph. The first deep convolutional network is a trained first deep convolutional network. The training process of the trained first deep convolutional network is as follows: a large number of coronary CTA images are input into the initial first deep convolutional network to obtain a plurality of predicted aorta segmentation binary graphs. The manually labeled aorta segmentation binary graph is compared with the predicted aorta segmentation binary graph, and the difference obtained by the comparison is fed back to the initial first deep convolutional neural network, so as to iteratively update the network parameters of the initial first deep convolutional network, thereby obtaining the trained first deep convolutional network. The first deep convolutional network can segment the aorta according to the coronary CTA image, thereby obtaining the aorta segmentation binary graph.

[0074] In a specific implementation, the coronary CTA image can be input into the second deep convolutional network to obtain a coronary multi-class segmentation map. The second deep convolutional network is a trained second deep convolutional network. The training process of the trained second deep convolutional network is as follows: a large number of coronary CTA images are input into an initial second deep convolutional network to obtain a plurality of predicted coronary multi-class segmentation maps. The predicted coronary multi-class segmentation map includes a background prediction label, a left coronary prediction label, and a right coronary prediction label. The manually annotated coronary multi-class segmentation map includes a manually annotated background, a manually annotated left coronary, and a manually annotated right coronary. The manually annotated coronary multi-class segmentation map is continuously compared with the predicted coronary multi-class segmentation map. The differences obtained by the comparison are continuously fed back to the initial second deep convolutional network, so as to iteratively update the network parameters of the initial second deep convolutional network, and obtain the trained second deep convolutional network. The second deep convolutional network can be used to perform coronary segmentation based on the coronary CTA image, so as to obtain the coronary multi-class segmentation map.

[0075] In a specific implementation, the coronary CTA image can be input into the third deep convolutional network to obtain an aortic sinus multi-class segmentation map. The third deep convolutional network is a trained third deep convolutional network. The training process of the trained third deep convolutional network is as follows: a large number of coronary CTA images are input into an initial third deep convolutional network to obtain a plurality of predicted aortic sinus multi-class segmentation maps. The predicted coronary multi-class segmentation map includes a left sinus prediction label, a right sinus prediction label, and an innominate sinus prediction label. The manually annotated aortic sinus multi-class segmentation map includes a manually annotated left sinus, a manually annotated right sinus, and a manually annotated innominate sinus. The manually annotated aortic sinus multi-class segmentation map is continuously compared with the predicted aortic sinus multi-class segmentation map. The differences obtained by the comparison are continuously fed back to the initial third deep convolutional network, so as to iteratively update the network parameters of the initial third deep convolutional network, and obtain the trained third deep convolutional network. The third deep convolutional network can be used to perform aortic sinus segmentation based on the coronary CTA image, so as to obtain the aortic sinus multi-class segmentation map.

[0076] In an embodiment, the extracting the aortic centerline based on the aortic segmentation binary map comprises:

[0077] The aortic centerline is extracted based on the aortic segmentation binary map through deep learning or image processing.

[0078] Step S30: respectively performing dilation processing on the left coronary segmentation map and the right coronary segmentation map to obtain a new left coronary segmentation map and a new right coronary segmentation map.

[0079] In an embodiment, the respectively performing dilation processing on the left coronary segmentation map and the right coronary segmentation map to obtain a new left coronary segmentation map and a new right coronary segmentation map comprises:

[0080] performing dilation processing on the left coronary artery segmentation map, so that the intersection of the new left coronary artery segmentation map and the aorta segmentation binary image is not an empty set;

[0081] performing dilation processing on the right coronary artery segmentation map, so that the intersection of the new right coronary artery segmentation map and the aorta segmentation binary image is not an empty set.

[0082] It should be noted that the dilation processing is an image processing operation method, and the dilation processing on the left coronary artery segmentation map means that the boundary points of the left coronary artery in the coronary artery multi-class segmentation map are expanded, that is, all background points in contact with the left coronary artery are merged into the left coronary artery, so that the boundary of the left coronary artery expands outward, and the left coronary artery segmentation map after the dilation processing is obtained; the dilation processing on the right coronary artery segmentation map means that the boundary points of the right coronary artery in the coronary artery multi-class segmentation map are expanded, that is, all background points in contact with the right coronary artery are merged into the right coronary artery, so that the boundary of the right coronary artery expands outward, and the right coronary artery segmentation map after the dilation processing is obtained.

[0083] Step S40: performing cutting on the aortic sinus multi-class segmentation map by using a target plane to obtain a maximum aortic cross section, wherein the target plane is a plane perpendicular to the aortic center line.

[0084] In an embodiment, the performing cutting on the aortic sinus multi-class segmentation map by using a target plane to obtain a maximum aortic cross section, wherein the target plane is a plane perpendicular to the aortic center line, comprises:

[0085] performing cutting on the aortic sinus multi-class segmentation map by using a target plane to obtain a plurality of binary segmentations seg_N, wherein each seg_N corresponds to a coronary sinus cross section area, and the target plane is a plane perpendicular to the aortic center line;

[0086] taking the seg_N corresponding to the maximum aortic cross section area as the maximum aortic cross section.

[0087] It should be noted that the number of target planes is not limited to one, so that the aortic sinus can be cut by using a target plane to obtain a plurality of cut binary segmentations seg_N, and each binary segmentation seg_N corresponds to an area seg_S. The plane corresponding to the maximum seg_S can be taken as the maximum aortic cross section.

[0088] In an embodiment, the aortic sinus multi-class segmentation map comprises a left sinus label, a right sinus label, and an innominate sinus label; wherein,

[0089] After the taking the seg_N corresponding to the maximum aortic cross section area as the maximum aortic cross section, the method further comprises:

[0090] According to the left sinus tag, the right sinus tag, and the common sinus tag, a left sinus region, a right sinus region, and a common sinus region on the maximum aortic cross section are determined.

[0091] It should be noted that the left sinus region, the right sinus region, and the common sinus region on the maximum aortic cross section can be determined according to the aortic sinus multi-class segmentation map including the left sinus tag, the right sinus tag, and the common sinus tag.

[0092] Step S50: Based on the new left coronary segmentation map, the new right coronary segmentation map, and the aortic segmentation binary map, a left connected domain and a right connected domain are determined, and a left centroid of the left connected domain and a right centroid of the right connected domain are determined.

[0093] It should be noted that the left connected domain refers to a region where the new left coronary segmentation map intersects with the aortic segmentation binary map, and the right connected domain refers to a region where the new right coronary segmentation map intersects with the aortic segmentation binary map.

[0094] Step S60: A distance from the left centroid to the maximum aortic cross section is determined as a first plane distance, and a distance from the right centroid to the maximum aortic cross section is determined as a second plane distance.

[0095] Step S70: It is determined whether the first plane distance or the second plane distance is greater than a preset distance.

[0096] It should be noted that the preset distance is set in advance.

[0097] Step S80: When it is determined that the first plane distance or the second plane distance is greater than the preset distance, it is determined that the origin of the first coronary artery is abnormal, that is, the coronary artery is high.

[0098] It should be noted that the coronary artery origin abnormality can be coronary artery high (that is, the left coronary entry point or the right coronary entry point is at a position far from the aortic sinus).

[0099] In a specific implementation, whether the coronary artery origin is high can be determined by judging the distance from the left centroid to the maximum aortic cross section or the distance from the right centroid to the maximum aortic cross section.

[0100] The embodiment obtains a coronary CTA image of a first coronary artery; determines an aorta segmentation binary image, a coronary multi-class segmentation image and an aortic sinus multi-class segmentation image of the first coronary artery according to the coronary CTA image, and extracts an aorta center line based on the aorta segmentation binary image, wherein the coronary multi-class segmentation image comprises a left coronary segmentation image and a right coronary segmentation image; performs inflation processing on the left coronary segmentation image and the right coronary segmentation image respectively to obtain a new left coronary segmentation image and a new right coronary segmentation image; intercepts the aortic sinus multi-class segmentation image by using a target plane to obtain a maximum aorta cross section, wherein the target plane is a plane perpendicular to the aorta center line; determines a left connected domain and a right connected domain based on the new left coronary segmentation image, the new right coronary segmentation image and the aorta segmentation binary image, and determines a left centroid of the left connected domain and a right centroid of the right connected domain; determines a first plane distance as a distance from the left centroid to the maximum aorta cross section, and determines a second plane distance as a distance from the right centroid to the maximum aorta cross section; determines that the origin abnormality of the first coronary artery is coronary high position when it is judged that the first plane distance or the second plane distance is greater than a preset distance. In the above manner, the coronary origin abnormality does not need to be determined by a doctor manually on the collected coronary angiography image, the coronary origin abnormality can be automatically identified and the abnormal category of the coronary origin abnormality can be determined, which not only improves the determination efficiency, but also saves medical resources.

[0101] Reference Figure 3 , Figure 3 The flowchart of a second embodiment of the automatic identification method of the coronary origin abnormality is shown.

[0102] Based on the first embodiment, the maximum aorta cross section comprises a left sinus region, a right sinus region and an anonymous sinus region; wherein,

[0103] The automatic identification method of the coronary origin abnormality of the embodiment further comprises the following steps after the step S80:

[0104] Step S801: When it is judged that the first plane distance or the second plane distance is not greater than a preset distance, the left projection point is projected onto the maximum aorta cross section to obtain a new left projection point.

[0105] It should be noted that the left projection point is a projection point of the left centroid projected on the maximum aorta cross section.

[0106] Step S802: The distances of the left projection point to the left sinus region, the right sinus region and the anonymous sinus region are determined as a first left distance, a second left distance and a third left distance respectively.

[0107] Step S803: judging whether the first left distance is the minimum distance among the first left distance, the second left distance and the third left distance.

[0108] It can be understood that when the first left distance is the minimum distance among the first left distance, the second left distance and the third left distance, it can be indicated that the left coronary originates from the left sinus, i.e. the origin of the left coronary artery is normal, and when the first left distance is not the minimum distance among the first left distance, the second left distance and the third left distance, the left coronary may originate from the right sinus or the common sinus, i.e. the origin of the left coronary artery is abnormal.

[0109] Step S804: if not, judging the minimum distance among the second left distance and the third left distance.

[0110] Step S805: if it is judged that the minimum distance is the second left distance, determining that the abnormal origin of the first coronary artery is that the left coronary originates from the right sinus.

[0111] Step S806: if it is judged that the minimum distance is the third left distance, determining that the abnormal origin of the first coronary artery is that the left coronary originates from the common sinus.

[0112] In the embodiment, not only whether the origin of the coronary artery is abnormal can be automatically determined, but also the origin sinus of the left coronary with abnormal origin can be accurately determined when the left coronary with abnormal origin is determined.

[0113] Reference Figure 4 , Figure 4 is a flowchart of a second embodiment of the automatic identification method of the abnormal origin of the coronary artery.

[0114] Based on the first embodiment, the maximum aortic section includes a left sinus region, a right sinus region and a common sinus region; wherein,

[0115] The automatic identification method of the abnormal origin of the coronary artery in the embodiment further includes, after the step S80:

[0116] Step S901: when it is judged that the first plane distance or the second plane distance is not greater than a preset distance, projecting the right projection point onto the maximum aortic section to obtain a right projection point.

[0117] It should be noted that the right projection point is a projection point of the right centroid projected on the maximum aortic section.

[0118] Step S902: respectively determining distances of the right projection point to the left sinus region, the right sinus region and the common sinus region as a first right distance, a second right distance and a third right distance.

[0119] Step S903: judging whether the second right distance is the minimum distance among the first right distance, the second right distance and the third right distance.

[0120] It can be understood that when the second right distance is the minimum distance among the first right distance, the second right distance and the third right distance, it can be indicated that the right originates from the right sinus, that is, the origin of the right coronary artery is normal, and when the second right distance is not the minimum distance among the first right distance, the second right distance and the third right distance, the right coronary artery can originate from the left sinus or the innominate sinus, that is, the origin of the right coronary artery is abnormal.

[0121] Step S904: if not, judging the minimum distance among the first right distance and the third right distance.

[0122] Step S905: if it is judged that the minimum distance is the first right distance, it is determined that the abnormal origin of the first coronary artery is that the right coronary originates from the left sinus.

[0123] Step S906: if it is judged that the minimum distance is the third right distance, it is determined that the abnormal origin of the first coronary artery is that the right coronary originates from the innominate sinus.

[0124] In the embodiment, not only can the coronary artery origin be automatically judged to be abnormal or not, but also when the right coronary origin is determined to be abnormal, the origin sinus of the right coronary with abnormal origin can be accurately determined.

[0125] In addition, the embodiment of the present application also provides a storage medium, wherein the storage medium stores a coronary artery origin abnormality automatic identification program, and the coronary artery origin abnormality automatic identification program is executed by a processor to realize the steps of the coronary artery origin abnormality automatic identification method as described above.

[0126] Reference Figure 5 , Figure 5 is a structural block diagram of a first embodiment of the coronary artery origin abnormality automatic identification device of the present application.

[0127] As Figure 5 shown, the coronary artery origin abnormality automatic identification device provided by the embodiment of the present application comprises:

[0128] An acquisition module 10 is configured to acquire a coronary CTA image of a first coronary artery;

[0129] A determination module 20 is configured to determine an aorta segmentation binary image, a coronary multi-class segmentation image and an aortic sinus multi-class segmentation image of the first coronary artery according to the coronary CTA image, and extract an aorta center line based on the aorta segmentation binary image, wherein the coronary multi-class segmentation image comprises a left coronary artery segmentation image and a right coronary artery segmentation image.

[0130] An expansion module 30 is configured to expand the left coronary artery segmentation image and the right coronary artery segmentation image respectively to obtain a new left coronary artery segmentation image and a new right coronary artery segmentation image;

[0131] The determination module 20 is further configured to intercept the aortic sinus multi-class segmentation image by using a target plane to obtain a maximum aortic cross section, wherein the target plane is a plane perpendicular to the aortic centerline;

[0132] The determination module 20 is further configured to determine a left connected domain and a right connected domain based on the new left coronary artery segmentation image, the new right coronary artery segmentation image and the aortic artery segmentation binary image, and determine a left centroid of the left connected domain and a right centroid of the right connected domain;

[0133] The determination module 20 is further configured to determine a first plane distance of a distance from the left centroid to the maximum aortic cross section and a second plane distance of a distance from the right centroid to the maximum aortic cross section;

[0134] A judgment module 40 is configured to judge whether the first plane distance or the second plane distance is greater than a preset distance;

[0135] The determination module 20 is further configured to determine that the origin of the first coronary artery is abnormal as a high coronary artery when it is judged that the first plane distance or the second plane distance is greater than the preset distance.

[0136] It should be understood that the above is only an example, and the technical solutions of the present application do not constitute any limitation. In specific applications, those skilled in the art can set it according to the needs, and the present application does not limit it.

[0137] The embodiment obtains a coronary CTA image of a first coronary artery; determines an aorta segmentation binary image, a coronary multi-class segmentation image and an aortic sinus multi-class segmentation image of the first coronary artery according to the coronary CTA image, and extracts an aorta centerline based on the aorta segmentation binary image, wherein the coronary multi-class segmentation image comprises a left coronary segmentation image and a right coronary segmentation image; performs dilation processing on the left coronary segmentation image and the right coronary segmentation image respectively to obtain a new left coronary segmentation image and a new right coronary segmentation image; intercepts the aortic sinus multi-class segmentation image by using a target plane to obtain a maximum aorta cross section, wherein the target plane is a plane perpendicular to the aorta centerline; determines a left connected domain and a right connected domain based on the new left coronary segmentation image, the new right coronary segmentation image and the aorta segmentation binary image, and determines a left centroid of the left connected domain and a right centroid of the right connected domain; determines a first plane distance as a distance from the left centroid to the maximum aorta cross section, and determines a second plane distance as a distance from the right centroid to the maximum aorta cross section; determines whether the first plane distance or the second plane distance is greater than a preset distance; when it is determined that the first plane distance or the second plane distance is greater than the preset distance, determines that an origin abnormality of the first coronary artery is coronary artery high position. In the above manner, the coronary origin abnormality determination on the collected coronary angiography image does not need to be performed by a doctor manually, the abnormal category of the coronary origin abnormality can be automatically identified and determined, the determination efficiency can be improved, and medical resources can be saved.

[0138] In an embodiment, the maximum aorta cross section comprises a left sinus region, a right sinus region and an innominate sinus region; wherein,

[0139] The determination module 40 is further configured to:

[0140] When it is determined that the first plane distance or the second plane distance is not greater than the preset distance, project the left centroid onto the maximum aorta cross section to obtain a left projection point;

[0141] Determine distances from the left projection point to the left sinus region, the right sinus region and the innominate sinus region as a first left distance, a second left distance and a third left distance respectively;

[0142] Determine whether the first left distance is a minimum distance among the first left distance, the second left distance and the third left distance;

[0143] If not, determine a minimum distance among the second left distance and the third left distance;

[0144] If it is determined that the minimum distance is the second left distance, determine that the origin abnormality of the first coronary artery is that the left coronary originates from the right sinus;

[0145] If the minimum distance is determined as the third left distance, it is determined that the abnormal origin of the first coronary artery is that the left coronary originates from the aortic sinus.

[0146] In an embodiment, the maximum aortic cross section comprises a left sinus region, a right sinus region, and an aortic sinus region; wherein,

[0147] The determining module 40 is further configured to:

[0148] When the first plane distance or the second plane distance is determined as not greater than a preset distance, the right centroid is projected onto the maximum aortic cross section to obtain a right projection point;

[0149] The distances of the right projection point to the left sinus region, the right sinus region, and the aortic sinus region are determined as a first right distance, a second right distance, and a third right distance, respectively;

[0150] It is determined whether the second right distance is the minimum distance among the first right distance, the second right distance, and the third right distance;

[0151] If not, it is determined that the minimum distance is among the first right distance and the third right distance;

[0152] If the minimum distance is determined as the first right distance, it is determined that the abnormal origin of the first coronary artery is that the right coronary originates from the left sinus;

[0153] If the minimum distance is determined as the third right distance, it is determined that the abnormal origin of the first coronary artery is that the right coronary originates from the aortic sinus.

[0154] In an embodiment, the expanding module 30 is further configured to:

[0155] The left coronary artery segmentation graph is dilated to make the intersection of the new left coronary artery segmentation graph and the aortic artery segmentation binary image not an empty set;

[0156] The right coronary artery segmentation graph is dilated to make the intersection of the new right coronary artery segmentation graph and the aortic artery segmentation binary image not an empty set.

[0157] In an embodiment, the determining module 20 is further configured to:

[0158] The aortic sinus multi-class segmentation graph is intercepted by a target plane to obtain a plurality of binary segmentations seg_N, wherein each seg_N corresponds to a coronary sinus cross-sectional area, and the target plane is a plane perpendicular to the aortic center line;

[0159] The seg_N corresponding to the maximum aortic cross-sectional area is taken as the maximum aortic cross section.

[0160] In an embodiment, the aortic sinus multi-class segmentation map comprises a left sinus label, a right sinus label, and an innominate sinus label; wherein,

[0161] The determination module 20 is further configured to:

[0162] determine, according to the left sinus label, the right sinus label, and the innominate sinus label, a left sinus region, a right sinus region, and an innominate sinus region on the maximum aortic cross section.

[0163] In an embodiment, the determination module 20 is further configured to:

[0164] extract an aortic centerline based on the aortic segmentation binary map by deep learning or image processing.

[0165] It should be noted that the above-described workflow is merely illustrative and does not limit the scope of protection of the present application. In actual applications, a person skilled in the art can select part or all of the above-described workflow to achieve the purpose of the present embodiment, and the present application is not limited thereto.

[0166] In addition, technical details not described in detail in the present embodiment can be found in the automatic identification method of coronary artery origin abnormalities provided by any embodiment of the present application, and will not be described here.

[0167] In addition, it should be noted that in this document, the terms "comprise", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system comprising a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or system. Without more limitations, the element defined by the statement "comprises a" does not exclude the presence of another identical element in the process, method, article or system comprising the element.

[0168] The above-mentioned embodiment numbers of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0169] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and the necessary general hardware platform, of course, can also be through hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application essentially or say the part of the prior art contribution can be embodied in the form of software products, the computer software product is stored in a storage medium (such as read only memory (Read Only Memory, ROM) / RAM, disk, optical disk), including a number of instructions to make a terminal device (may be a mobile phone, computer, server, or network equipment, etc.) executes the method described in various embodiments of the present application.

[0170] The above is only the preferred embodiment of the present application, not therefore limit the patent scope of the present application, any equivalent structure or equivalent process transformation using the content of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for automated identification of anomalous coronary origin, comprising: The automatic identification method of the abnormal coronary artery origin comprises: obtaining a coronary CTA image of a first coronary artery; determining an aorta segmentation binary image, a coronary multi-class segmentation image and an aortic sinus multi-class segmentation image of the first coronary artery according to the coronary CTA image, and extracting an aorta centerline based on the aorta segmentation binary image, wherein the coronary multi-class segmentation image comprises a left coronary segmentation image and a right coronary segmentation image; respectively performing inflation processing on the left coronary segmentation image and the right coronary segmentation image to obtain a new left coronary segmentation image and a new right coronary segmentation image; obtaining a maximum aorta cross section by intercepting the aortic sinus multi-class segmentation image using a target plane, wherein the target plane is a plane perpendicular to the aorta centerline; determining a left connected domain and a right connected domain based on the new left coronary segmentation image, the new right coronary segmentation image and the aorta segmentation binary image, and determining a left centroid of the left connected domain and a right centroid of the right connected domain; determining a first plane distance as a distance from the left centroid to the maximum aorta cross section, and determining a second plane distance as a distance from the right centroid to the maximum aorta cross section; determining whether the first plane distance or the second plane distance is greater than a preset distance; when it is determined that the first plane distance or the second plane distance is greater than the preset distance, determining that the origin of the first coronary artery is abnormal as coronary artery high position; the maximum aorta cross section comprises a left sinus region, a right sinus region and an innominate sinus region; wherein after the determination of whether the first plane distance or the second plane distance is greater than the preset distance, the method further comprises: when it is determined that the first plane distance or the second plane distance is not greater than the preset distance, projecting the left centroid onto the maximum aorta cross section to obtain a left projection point; respectively determining a first left distance, a second left distance and a third left distance as distances from the left projection point to the left sinus region, the right sinus region and the innominate sinus region; determining whether the first left distance is the minimum distance among the first left distance, the second left distance and the third left distance; if not, determining the minimum distance among the second left distance and the third left distance; if it is determined that the minimum distance is the second left distance, determining that the origin of the first coronary artery is abnormal as left coronary originating from the right sinus; if it is determined that the minimum distance is the third left distance, determining that the origin of the first coronary artery is abnormal as left coronary originating from the innominate sinus; the method of obtaining the maximum aorta cross section by intercepting the aortic sinus multi-class segmentation image using the target plane, wherein the target plane is a plane perpendicular to the aorta centerline, comprises: obtaining a plurality of binary segmentation seg_N by intercepting the aortic sinus multi-class segmentation image using the target plane, wherein each seg_N corresponds to a coronary sinus cross section area, and the target plane is a plane perpendicular to the aorta centerline; taking the seg_N corresponding to the maximum aorta cross section area as the maximum aorta cross section.

2. The method of claim 1, wherein, after the determination of whether the first plane distance or the second plane distance is greater than the preset distance, the method further comprises: projecting the right centroid onto the largest aortic section to obtain a right projection point when it is determined that the first plane distance or the second plane distance is not greater than a preset distance; determining distances from the right projection point to the left sinus region, the right sinus region, and the innominate sinus region as a first right distance, a second right distance, and a third right distance, respectively; determining whether the second right distance is the smallest distance among the first right distance, the second right distance, and the third right distance; if not, determining the smallest distance among the first right distance and the third right distance; if it is determined that the smallest distance is the first right distance, determining that the origin abnormality of the first coronary artery is that the right coronary artery originates from the left sinus; if it is determined that the smallest distance is the third right distance, determining that the origin abnormality of the first coronary artery is that the right coronary artery originates from the innominate sinus.

3. The method of claim 1, wherein, The dilation processing of the left coronary artery segmentation graph and the right coronary artery segmentation graph respectively to obtain a new left coronary artery segmentation graph and a new right coronary artery segmentation graph includes: dilation processing of the left coronary artery segmentation graph to make the intersection of the new left coronary artery segmentation graph and the aortic segmentation binary image not an empty set; dilation processing of the right coronary artery segmentation graph to make the intersection of the new right coronary artery segmentation graph and the aortic segmentation binary image not an empty set.

4. The method of claim 1, wherein, The aortic sinus multi-class segmentation graph includes a left sinus label, a right sinus label, and an innominate sinus label; wherein after the seg_N corresponding to the largest aortic section area is taken as the largest aortic section, the method further includes: determining, according to the left sinus label, the right sinus label, and the innominate sinus label, a left sinus region, a right sinus region, and an innominate sinus region on the largest aortic section.

5. The method of claim 1, wherein, The extraction of the aortic centerline based on the aortic segmentation binary image includes: extracting the aortic centerline based on the aortic segmentation binary image through deep learning or image processing.

6. An automated device for identifying anomalous coronary origins, comprising: The automatic identification device for coronary artery origin abnormality includes: an acquisition module configured to acquire a coronary CTA image of a first coronary artery; a determination module configured to determine, according to the coronary CTA image, an aortic segmentation binary image, a coronary multi-class segmentation graph, and an aortic sinus multi-class segmentation graph of the first coronary artery, and extract an aortic centerline based on the aortic segmentation binary image, wherein the coronary multi-class segmentation graph includes a left coronary artery segmentation graph and a right coronary artery segmentation graph; a dilation module configured to perform dilation processing on the left coronary artery segmentation graph and the right coronary artery segmentation graph respectively to obtain a new left coronary artery segmentation graph and a new right coronary artery segmentation graph; the determination module is further configured to cut the aortic sinus multi-class segmentation graph using a target plane to obtain a largest aortic section, wherein the target plane is a plane perpendicular to the aortic centerline; the determination module is further configured to determine a left connected domain and a right connected domain based on the new left coronary artery segmentation graph, the new right coronary artery segmentation graph, and the aortic segmentation binary image, and determine a left centroid of the left connected domain and a right centroid of the right connected domain; The determination module is further configured to determine a first plane distance of the left centroid to the maximum aortic cross section and a second plane distance of the right centroid to the maximum aortic cross section; The determination module is further configured to determine, when it is determined that the first plane distance or the second plane distance is greater than the preset distance, that the abnormal origin of the first coronary artery is a coronary artery high position. The maximum aortic cross section includes a left sinus region, a right sinus region, and an innominate sinus region; wherein The determination module is further configured to: When it is determined that the first plane distance or the second plane distance is not greater than the preset distance, project the left centroid onto the maximum aortic cross section to obtain a left projection point; Determine a first left distance, a second left distance, and a third left distance of the left projection point to the left sinus region, the right sinus region, and the innominate sinus region, respectively; Determine whether the first left distance is the minimum distance among the first left distance, the second left distance, and the third left distance; If not, determine the minimum distance among the second left distance and the third left distance; If it is determined that the minimum distance is the second left distance, determine that the abnormal origin of the first coronary artery is that the left coronary artery originates from the right sinus; If it is determined that the minimum distance is the third left distance, determine that the abnormal origin of the first coronary artery is that the left coronary artery originates from the innominate sinus. The determination module is further configured to: Use a target plane to intercept the aortic sinus multi-class segmentation map to obtain a plurality of binary segmentation seg_N, wherein each seg_N corresponds to a coronary sinus cross-sectional area, and the target plane is a plane perpendicular to the aortic centerline; Take the seg_N corresponding to the maximum aortic cross-sectional area as the maximum aortic cross section. The device includes a memory, a processor, and a coronary artery origin abnormality automatic identification program stored on the memory and executable on the processor, and the coronary artery origin abnormality automatic identification program is configured to implement the steps of the coronary artery origin abnormality automatic identification method according to any one of claims 1 to 5.

7. An automated identification device of anomalous coronary origin, characterized by, The storage medium stores a coronary artery origin abnormality automatic identification program, and the coronary artery origin abnormality automatic identification program implements the steps of the coronary artery origin abnormality automatic identification method according to any one of claims 1 to 5 when executed by a processor.

8. A storage medium, characterized by ​

Images