Coronary artery scanning control method and system based on deep learning

Abstract

The invention relates to the field of coronary artery scanning, and discloses a coronary artery scanning control method and system based on deep learning, and the method comprises the steps: obtaining a diaphragm motion image, a left ventricular myocardial motion image, a right ventricular myocardial motion image and a cardiac apex motion image of a patient, and obtaining an arrhythmia monitoring index of the patient through ECG calibration; determining the diaphragm motion characteristics of the patient through the diaphragm motion image, determining the left ventricular myocardial motion characteristics of the patient through the left ventricular myocardial motion image, determining the right ventricular myocardial motion characteristics of the patient through the right ventricular myocardial motion image, and determining the cardiac apex motion characteristics of the patient through the cardiac apex motion image; and determining a coronary artery navigation mode of the patient according to the arrhythmia monitoring indexes, the diaphragm motion characteristics, the left ventricular myocardial motion characteristics, the right ventricular myocardial motion characteristics and the cardiac apex motion characteristics through a navigation prediction model. According to the method and the device, the MRCA examination accuracy and examination efficiency under the conditions of irregular respiration and arrhythmia are improved.

Description

Technical Field

[0001] The present application relates to the technical field of coronary artery scanning, and more specifically, to a coronary artery scanning control method and system based on deep learning. Background Art

[0002] In recent years, coronary magnetic resonance angiography (MRCA) is expected to replace coronary angiography (CAG) and coronary CT angiography (CTA), which have many contraindications and radiation, due to its advantages of no radiation, high soft tissue contrast, imaging through tissue's own signal contrast, and independence from the use of contrast agents. It has become a means of screening for coronary artery disease in people who are contraindicated with contrast agents, children, pregnant patients, etc., and is receiving increasing clinical attention.

[0003] At present, there is a method to improve the accuracy of MRCA examination by using diaphragm navigation. Diaphragm navigation is used to compensate for physiological movement. First, the movement of the patient's diaphragm can be monitored through navigation technology, and a user-controllable acceptance window can be established. Then, through diaphragm navigation, data acquisition can be selected to the relatively static part of the respiratory cycle, thereby allowing free breathing during the scan. There are many factors that affect the effect of MRCA examination, especially the adverse effects of irregular breathing and irregular heart rate on MRCA examination. When the patient breathes irregularly (breathing is fast and slow, shallow and deep) and has an irregular heartbeat during the examination, it is very likely to cause the scan to fail when using diaphragm navigation for MRCA examination. Summary of the invention

[0004] The purpose of this application is to provide a coronary artery scanning control method and system based on deep learning, which solves the technical problem of failure of MRCA examination scanning using diaphragm navigation when irregular breathing and arrhythmia occur, and achieves the technical effect of improving the accuracy and efficiency of MRCA examination under conditions of irregular breathing and arrhythmia.

[0005] A deep learning-based coronary artery scanning control method provided in an embodiment of the present application comprises: determining the patient's diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image respectively according to the film imaging of the patient's positive coronary position through the diaphragm top and myocardium, and obtaining the patient's arrhythmia monitoring index through ECG calibration; determining the patient's diaphragm motion characteristics through the diaphragm motion image, determining the patient's left ventricular myocardial motion characteristics through the left ventricular myocardial motion image, determining the patient's right ventricular myocardial motion characteristics through the right ventricular myocardial motion image, and determining the patient's apex motion characteristics through the apex motion image; determining the patient's coronary artery navigation mode through a navigation prediction model according to the arrhythmia monitoring index, diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apex motion characteristics; wherein the navigation modes include diaphragm navigation, left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation.

[0006] In a possible implementation, the method also includes: determining the diaphragm area through the patient's diaphragm motion image, determining the left ventricular myocardial area through the left ventricular myocardial motion image, determining the left ventricular myocardial position through the left ventricular myocardial motion image, determining the right ventricular myocardial position through the right ventricular myocardial motion image, and determining the apex position through the apex motion image; wherein the diaphragm motion image, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image respectively include multiple frames of static images in a time sequence within 2 to 3 respiratory motion cycles and multiple cardiac cycles; determining the patient's diaphragm motion characteristics through the diaphragm motion image, determining the patient's left ventricular myocardial motion characteristics through the left ventricular myocardial motion image, determining the patient's right ventricular myocardial motion characteristics through the right ventricular myocardial motion image, and determining the patient's apex motion characteristics through the apex motion image, including: determining the patient's diaphragm motion characteristics through the diaphragm area in the diaphragm motion image, determining the patient's left ventricular myocardial motion characteristics through the left ventricular myocardial area in the left ventricular myocardial motion image, determining the patient's right ventricular myocardial motion characteristics through the right ventricular myocardial position in the right ventricular myocardial motion image, and determining the patient's apex motion characteristics through the apex position in the apex motion image.

[0007] In another possible implementation, the method further includes: obtaining a first arrhythmia monitoring index of the patient through an ECG signal within a first time period; when the first arrhythmia monitoring index is greater than or equal to a preset arrhythmia monitoring index, obtaining a coronary transmyocardial cine imaging of the patient within a second time period after the first time period, and determining the patient's left ventricular myocardial motion image, right ventricular myocardial motion image, and apical motion image, respectively; within a third time period after the second time period, determining the patient's left ventricular myocardial motion characteristics through the left ventricular myocardial motion image, determining the patient's right ventricular myocardial motion characteristics through the right ventricular myocardial motion image, and determining the patient's apical motion characteristics through the apical motion image; determining the patient's coronary artery navigation mode through a navigation prediction model according to the left ventricular myocardial motion characteristics, the right ventricular myocardial motion characteristics, and the apical motion characteristics; wherein the navigation modes include left ventricular myocardial navigation, right ventricular myocardial navigation, and apical navigation.

[0008] In another possible implementation, the method also includes: determining the diastolic time period of the cardiac cycle through the ECG signal during the second time period and the third time period; and determining the area and monitoring time window of the monitoring windows in the left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation through the navigation window prediction model according to the ECG signal, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image during the diastolic time period.

[0009] In another possible implementation, the navigation window prediction model includes an input layer, a CNN layer, an LSTM layer, a merging layer, a fully connected layer and an output layer, wherein the input layer is used to receive the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, the CNN layer is used to receive the input of the input layer and extract the spatial features and local features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, the LSTM layer is used to receive the input of the input layer and extract the time series features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, the merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging, the fully connected layer is used to receive the output of the merging layer and input it to the output layer, and the output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, the right ventricular myocardial navigation and the apex navigation.

[0010] In another possible implementation, the input layer is used to receive ECG signals, left ventricular myocardial motion images, right ventricular myocardial motion images and apical motion images, the LSTM layer is also used to receive input from the input layer and extract time series features of the ECG signals, left ventricular myocardial motion images, right ventricular myocardial motion images and apical motion images, when the first arrhythmia monitoring index is greater than or equal to the preset arrhythmia monitoring index, the LSTM layer is used to determine the arrhythmia monitoring features based on the ECG signal, the LSTM layer is also used to extract time series features of the left ventricular myocardial motion images, right ventricular myocardial motion images and apical motion images, the merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging, the fully connected layer is used to receive the output of the merging layer and input it to the output layer, and the output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, right ventricular myocardial navigation and apical navigation.

[0011] In another possible implementation, it also includes a front LSTM unit, which is used to receive input from the input layer and determine arrhythmia monitoring features based on the ECG signal. The front LSTM unit inputs the arrhythmia monitoring features into the LSTM layer. The LSTM layer is used to extract time series features of arrhythmia monitoring features, left ventricular myocardial motion images, right ventricular myocardial motion images, and apex motion images. The merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging. The fully connected layer is used to receive the output of the merging layer and input it into the output layer. The output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, right ventricular myocardial navigation, and apex navigation.

[0012] In another possible implementation, the method also includes: determining the area of ​​the left ventricular myocardial navigation monitoring window and the monitoring time window according to the ECG signal and the left ventricular myocardial motion image through the navigation window prediction model, and determining the window area of ​​the area of ​​the monitoring window; when the window area is smaller than the preset window area, or the duration of the monitoring time window is smaller than the preset duration, determining the area and monitoring time window of the right ventricular myocardial navigation and apex navigation monitoring windows according to the ECG signal, the right ventricular myocardial motion image and the apex motion image through the navigation window prediction model.

[0013] In another possible implementation, the method also includes: when the window area of ​​the monitoring window determined according to the ECG signal, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image is smaller than the preset window area, or the duration of the monitoring time window is smaller than the preset duration, the ECG signal of the patient is reacquired after adjusting the position of the ECG patch, and the area of ​​the monitoring window and the monitoring time window in the left ventricular myocardial navigation are determined according to the reacquired ECG signal and the left ventricular myocardial motion image through the navigation window prediction model.

[0014] An embodiment of the present application also provides a deep learning-based coronary artery scanning control system, comprising a unit for executing the deep learning-based coronary artery scanning control method as described in any one of the above items.

[0015] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0016] An embodiment of the present application provides a computer program product. When the computer program product runs on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0017] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0018] An embodiment of the present application provides a coronary artery scanning control method based on deep learning, the method comprising: determining the patient's diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image respectively according to the patient's positive coronary position film imaging through the diaphragm top and myocardium, and obtaining the patient's arrhythmia monitoring index through ECG calibration; determining the patient's diaphragm motion characteristics through the diaphragm motion image, determining the patient's left ventricular myocardial motion characteristics through the left ventricular myocardial motion image, determining the patient's right ventricular myocardial motion characteristics through the right ventricular myocardial motion image, and determining the patient's apex motion characteristics through the apex motion image; determining the patient's coronary artery navigation mode through a navigation prediction model according to the arrhythmia monitoring index, diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apex motion characteristics; wherein the navigation modes include diaphragm navigation, left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation. The deep learning-based coronary artery scanning control method in the embodiment of the present application can determine the navigation mode through the diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics when working, which can avoid the problem of diaphragm navigation failure caused by arrhythmia, and directly improve the accuracy of coronary artery scanning through myocardial navigation, thereby improving the accuracy and efficiency of coronary artery scanning. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0020] Figure 1A schematic flow chart of a first deep learning-based coronary artery scanning control method provided in an embodiment of the present application;

[0021] Figure 2 A schematic diagram of the working process of a coronary artery scanning control method based on deep learning provided in an embodiment of the present application;

[0022] Figure 3 A schematic diagram of the positions of navigation windows corresponding to different navigation modes in an embodiment of the present application;

[0023] Figure 4 Schematic diagram of right coronary artery imaging corresponding to different navigation methods in the embodiment of the present application;

[0024] Figure 5 A schematic flow chart of a second deep learning-based coronary artery scanning control method provided in an embodiment of the present application;

[0025] Figure 6 A schematic diagram of the structure of a navigation window prediction model provided in an embodiment of the present application;

[0026] Figure 7 A schematic flow chart of a third deep learning-based coronary artery scanning control method provided in an embodiment of the present application;

[0027] Figure 8 A schematic diagram of the logical structure of a deep learning-based coronary artery scanning control system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0029] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0030] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0031] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0032] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0033] If the patient has irregular breathing (fast or slow, shallow or deep breathing) or arrhythmia during the examination, it is very likely to cause scan failure during MRCA examination using diaphragm navigation.

[0034] Based on the above reasons, an embodiment of the present application provides a coronary artery scanning control method based on deep learning, the method comprising: determining the patient's diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image respectively according to the patient's positive coronary position film imaging through the diaphragm top and myocardium, and obtaining the patient's arrhythmia monitoring index through ECG calibration; determining the patient's diaphragm motion characteristics through the diaphragm motion image, determining the patient's left ventricular myocardial motion characteristics through the left ventricular myocardial motion image, determining the patient's right ventricular myocardial motion characteristics through the right ventricular myocardial motion image, and determining the patient's apex motion characteristics through the apex motion image; determining the patient's coronary artery navigation mode through the navigation prediction model according to the arrhythmia monitoring indicators, diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apex motion characteristics; wherein the navigation modes include diaphragm navigation, left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation. The deep learning-based coronary artery scanning control method in the embodiment of the present application can determine the navigation mode through the diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics when working, which can avoid the problem of diaphragm navigation failure caused by arrhythmia, and directly improve the accuracy of coronary artery scanning through myocardial navigation, thereby improving the accuracy and efficiency of coronary artery scanning.

[0035] In some scenarios, a deep learning-based coronary artery scanning control method of an embodiment of the present application can be applied to coronary artery scanning, and can determine the navigation mode through diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics, thereby improving the accuracy and efficiency of coronary artery scanning.

[0036] The following is a detailed description of a coronary artery scanning control method based on deep learning provided in an embodiment of the present application with reference to specific examples.

[0037] Figure 1 A schematic flow chart of a first deep learning-based coronary artery scanning control method provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method includes S110 to S120, and S110 to S120 are described in detail below.

[0038] S110, based on the patient's coronal position film imaging through the diaphragm top and myocardium, determine the patient's diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image respectively, and obtain the patient's arrhythmia monitoring index through ECG calibration.

[0039] In an embodiment of the present application, a movie image of the patient's coronal position penetrating the diaphragm top and the myocardium can be first obtained. The movie image is a dynamic image, and then the navigation position can be identified based on the movie image of the patient's coronal position penetrating the diaphragm top and the myocardium.

[0040] Exemplarily, when obtaining cine imaging of the patient's coronal position through the diaphragmatic dome and myocardium, cine imaging of the diaphragmatic dome and myocardium can be obtained through MRI imaging. First, the diaphragmatic dome and myocardium can be quickly scanned to determine the exact location of the region of interest. At the same time, a suitable imaging sequence can be selected, such as a fast gradient echo (FGRE) or a balanced steady-state free precession (SSFP) sequence, to obtain dynamic images of the heart and diaphragm.

[0041] After obtaining the patient's coronal cine imaging of the diaphragm apex and myocardium, the patient's diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image can be determined respectively based on the patient's coronary cine imaging of the diaphragm apex and myocardium, and the patient's arrhythmia monitoring indicators can be obtained through ECG calibration.

[0042] Exemplarily, when determining the patient's diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image, the positions of the diaphragm, left ventricular myocardium, right ventricular myocardium and apex can be determined in the diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image through an image recognition model, and then the diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image can be obtained.

[0043] In order to obtain arrhythmia monitoring indicators, ECG calibration can be used to obtain the patient's arrhythmia monitoring indicators, and then the arrhythmia monitoring indicators can be used to preliminarily determine whether diaphragm navigation or myocardial navigation should be used.

[0044] S120, determining the patient's diaphragm motion characteristics through the diaphragm motion image, determining the patient's left ventricular myocardial motion characteristics through the left ventricular myocardial motion image, determining the patient's right ventricular myocardial motion characteristics through the right ventricular myocardial motion image, and determining the patient's apex motion characteristics through the apex motion image. Determine the patient's coronary artery navigation mode through the navigation prediction model according to the arrhythmia monitoring index, diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics, and apex motion characteristics. The navigation modes include diaphragm navigation, left ventricular myocardial navigation, right ventricular myocardial navigation, and apex navigation.

[0045] After obtaining the diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apical motion image, the patient's diaphragm motion characteristics can be determined through the diaphragm motion image, the patient's left ventricular myocardial motion characteristics can be determined through the left ventricular myocardial motion image, the patient's right ventricular myocardial motion characteristics can be determined through the right ventricular myocardial motion image, and the patient's apical motion characteristics can be determined through the apical motion image. Then, the reliability of coronary artery navigation of the diaphragm, left ventricular myocardium, right ventricular myocardium and apex can be determined respectively according to the diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics, and then the method of coronary artery navigation can be determined according to the reliability of coronary artery navigation at different positions.

[0046] Figure 2 A schematic diagram of the working process of a coronary artery scanning control method based on deep learning provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, after obtaining the diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics, the patient's coronary artery navigation mode can be determined through the navigation prediction model according to arrhythmia monitoring indicators, diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics, and the patient's coronary artery navigation mode can be determined scientifically.

[0047] When determining the coronary artery navigation mode, the navigation mode may include diaphragm navigation, left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation, wherein diaphragm navigation is navigation through the diaphragm, left ventricular myocardial navigation is navigation through the position of the left ventricular myocardium, right ventricular myocardial navigation is navigation through the position of the right ventricular myocardium, and apex navigation is navigation through the position of the apex.

[0048] Figure 3 Schematic diagram of the positions of navigation windows corresponding to different navigation modes in the embodiment of the present application. The positions of the four different navigation windows are as follows: Figure 3 As shown, DN represents diaphragm navigation; LVN represents left myocardial navigation, RVN represents right myocardial navigation, and AN represents apical navigation. Figure 3 The navigation windows corresponding to the different navigation modes in the image can be determined according to the arrhythmia monitoring index, the diaphragm movement characteristics, the left ventricular myocardial movement characteristics, the right ventricular myocardial movement characteristics and the apex movement characteristics. Figure 3 The rectangular boxes in the figure indicate the positions of the four navigation windows, and the arrows are used to indicate the positions of the right coronary artery under the four navigations.

[0049] Figure 4 is a schematic diagram of right coronary artery imaging corresponding to different navigation methods in the embodiment of the present application, Figure 4 The middle image shows the right coronary artery of a 32-year-old male obtained according to left myocardial navigation LVN, diaphragm navigation DN, apex navigation AN and right myocardial navigation RVN.

[0050] The beneficial effect of the above-mentioned implementation method is that the navigation mode is determined by the diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics, which can avoid the problem of diaphragm navigation failure caused by arrhythmia, and directly improve the accuracy of coronary artery scanning through myocardial navigation, thereby improving the accuracy and efficiency of coronary artery scanning.

[0051] In some implementations, the method further includes: determining the diaphragm area through the patient's diaphragm motion image, determining the left ventricular myocardium area through the left ventricular myocardium motion image, determining the left ventricular myocardium position through the left ventricular myocardium motion image, determining the right ventricular myocardium position through the right ventricular myocardium motion image, and determining the apex position through the apex motion image. The diaphragm motion image, the left ventricular myocardium motion image, the right ventricular myocardium motion image, and the apex motion image respectively include multiple static images in a time sequence within 2 to 3 respiratory motion cycles and multiple cardiac cycles.

[0052] When identifying motion features at different positions, the position area of ​​the motion features can be first determined through image recognition. First, the diaphragm area can be determined through the patient's diaphragm motion image, and the left ventricular myocardial area can be determined through the left ventricular myocardial motion image. The left ventricular myocardial position can be determined through the left ventricular myocardial motion image, the right ventricular myocardial position can be determined through the right ventricular myocardial motion image, and the apex position can be determined through the apex motion image. Subsequently, the motion features of the diaphragm area, left ventricular myocardial area, right ventricular myocardial position and apex position can be identified based on the diaphragm area, left ventricular myocardial area, right ventricular myocardial position and apex position.

[0053] Exemplarily, when acquiring the diaphragm motion image, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, the diaphragm motion image, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image respectively include multiple frames of static images in a time sequence within 2 to 3 respiratory motion cycles and multiple cardiac cycles.

[0054] In the above-mentioned S120, the patient's diaphragm motion characteristics are determined by the diaphragm motion image, the patient's left ventricular myocardial motion characteristics are determined by the left ventricular myocardial motion image, the patient's right ventricular myocardial motion characteristics are determined by the right ventricular myocardial motion image, and the patient's apical motion characteristics are determined by the apical motion image, including: determining the patient's diaphragm motion characteristics by the diaphragm area in the diaphragm motion image, determining the patient's left ventricular myocardial motion characteristics by the left ventricular myocardial area in the left ventricular myocardial motion image, determining the patient's right ventricular myocardial motion characteristics by the right ventricular myocardial position in the right ventricular myocardial motion image, and determining the patient's apical motion characteristics by the apex position in the apical motion image.

[0055] In an embodiment of the present application, the patient's diaphragm movement characteristics can be determined by the diaphragm area in the diaphragm motion image, the patient's left ventricular myocardial motion characteristics can be determined by the left ventricular myocardial area in the left ventricular myocardial motion image, the patient's right ventricular myocardial motion characteristics can be determined by the right ventricular myocardial position in the right ventricular myocardial motion image, and the patient's apex motion characteristics can be determined by the apex position in the apex motion image, and then the navigation mode can be determined by the motion characteristics of different positions.

[0056] Exemplarily, when determining the left ventricular myocardial motion characteristics corresponding to the left ventricular myocardial area, the multiple frames of the left ventricular myocardial motion image can be positioned first, and then the left ventricular myocardial area in the multiple frames of the left ventricular myocardial motion image can be determined, and then the left ventricular myocardial motion characteristics can be determined based on the left ventricular myocardial area in the left ventricular myocardial motion image.

[0057] The beneficial effect of the above-mentioned implementation method is that the diaphragm area is determined by the patient's diaphragm motion image, and the left ventricular myocardial area is determined by the left ventricular myocardial motion image, the left ventricular myocardial position is determined by the left ventricular myocardial motion image, the right ventricular myocardial position is determined by the right ventricular myocardial motion image, and the apex position is determined by the apex motion image, thereby identifying the motion characteristics of different positions, which can improve the efficiency of determining the navigation method.

[0058] Figure 5 A schematic flow chart of a second deep learning-based coronary artery scanning control method provided in an embodiment of the present application is shown in FIG. Figure 5 As shown, the above method further includes S210 to S220, and S210 to S220 are described in detail below.

[0059] S210. Acquire a first arrhythmia monitoring index of the patient through the ECG signal within a first time period. When the first arrhythmia monitoring index is greater than or equal to a preset arrhythmia monitoring index, acquire a coronary transmyocardial cine imaging of the patient within a second time period after the first time period to respectively determine the patient's left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image.

[0060] When determining the navigation method, the patient's first arrhythmia monitoring index can be obtained through the ECG signal within a first time period. The first arrhythmia monitoring index represents the patient's heart rate state. Since arrhythmia may cause diaphragm navigation to fail in coronary artery scanning, the first arrhythmia monitoring index can be used to determine whether navigation through the diaphragm is possible.

[0061] Exemplarily, the first time period may be 30 s, 60 s or 90 s.

[0062] When judging, when the first arrhythmia monitoring index is greater than or equal to the preset arrhythmia monitoring index, it indicates that the patient has arrhythmia. At this time, diaphragm navigation can be avoided, that is, navigation can be performed by determining the optimal myocardial navigation position to improve the effect of coronary scanning.

[0063] When determining the myocardial navigation position, a coronary film image of the patient's myocardium can be obtained in a second time period after the first time period to respectively determine the patient's left ventricular myocardial motion image, right ventricular myocardial motion image and apical motion image, and then the optimal myocardial navigation position can be determined based on the left ventricular myocardial motion image, right ventricular myocardial motion image and apical motion image.

[0064] Exemplarily, the second time period may be 30s, 60s or 90s.

[0065] S220. In a third time period after the second time period, determine the patient's left ventricular myocardial motion characteristics through the left ventricular myocardial motion image, determine the patient's right ventricular myocardial motion characteristics through the right ventricular myocardial motion image, and determine the patient's apex motion characteristics through the apex motion image. Determine the patient's coronary artery navigation mode through the navigation prediction model according to the left ventricular myocardial motion characteristics, the right ventricular myocardial motion characteristics, and the apex motion characteristics. The navigation modes include left ventricular myocardial navigation, right ventricular myocardial navigation, and apex navigation.

[0066] In the third time period after the second time period, the patient's left ventricular myocardial motion characteristics can be determined by the left ventricular myocardial motion image, the patient's right ventricular myocardial motion characteristics can be determined by the right ventricular myocardial motion image, and the patient's apical motion characteristics can be determined by the apical motion image, and then the position of the myocardial navigation can be determined based on the left ventricular myocardial motion characteristics, the right ventricular myocardial motion characteristics and the apical motion characteristics.

[0067] After obtaining the left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics, the patient's coronary artery navigation method can be determined through a navigation prediction model based on the left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics.

[0068] When determining the patient's coronary artery navigation mode based on the left ventricular myocardial motion characteristics, the right ventricular myocardial motion characteristics and the apex motion characteristics, the specific navigation modes include left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation.

[0069] The beneficial effect of the above-mentioned implementation method is that by determining whether diaphragm navigation can be performed within the first time period, when it is determined that diaphragm navigation cannot be performed, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image are acquired within the second time period, and the optimal myocardial navigation position is determined, thereby avoiding redundant calculations and processing when diaphragm navigation can be performed, thereby improving the efficiency of coronary scanning.

[0070] The beneficial effect of the above-mentioned implementation method is that the left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics are determined in the third time period after the second time period, and the patient's coronary artery navigation method is further determined. There is no need to collect myocardial motion images in the third time period, which can avoid the patient's long-term collection of myocardial motion images from being affected by radiation.

[0071] In some implementations, the method further includes: determining the diastolic period of the cardiac cycle by using the ECG signal during the second period and the third period, and determining the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, the right ventricular myocardial navigation and the apex navigation by using the navigation window prediction model according to the ECG signal, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image during the diastolic period.

[0072] When specifically determining the navigation position, the diastole period of the cardiac cycle can be determined through ECG signals in the second and third time periods. The myocardial movement amplitude in the diastole period is small, and the optimal myocardial navigation position can be determined based on the myocardial movement characteristics in the diastole period.

[0073] When subsequently determining the patient's coronary artery navigation mode based on the left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics, the navigation window prediction model can be used within the diastolic time period to determine the area and monitoring time window of the monitoring windows in the left ventricular myocardial navigation, right ventricular myocardial navigation and apical navigation based on the ECG signal, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apical motion image. The left ventricular myocardial motion image, the right ventricular myocardial motion image and the apical motion image are used to determine the area of ​​the monitoring window in the navigation, and the ECG signal is used to determine the monitoring time window, so as to facilitate subsequent coronary artery scanning navigation.

[0074] When determining the area and monitoring time window of the monitoring window in left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation through the navigation window prediction model, the left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apex motion characteristics in the above-mentioned S110 to S120 are represented and calculated through the intermediate matrix, thereby being able to accurately and efficiently determine the area and monitoring time window of the monitoring window in the navigation.

[0075] The beneficial effect of the above implementation method is that the navigation window prediction model can accurately and efficiently determine the area and monitoring time window of the monitoring window in left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation, thereby improving the efficiency and accuracy of myocardial navigation.

[0076] In some implementations, the navigation window prediction model includes an input layer, a CNN layer, an LSTM layer, a merging layer, a fully connected layer and an output layer, the input layer is used to receive the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, the CNN layer is used to receive the input of the input layer and extract the spatial features and local features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, the LSTM layer is used to receive the input of the input layer and extract the time series features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, the merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging, the fully connected layer is used to receive the output of the merging layer and input it to the output layer, and the output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, the right ventricular myocardial navigation and the apex navigation.

[0077] Figure 6 A schematic diagram of the structure of a navigation window prediction model provided in an embodiment of the present application is shown in FIG. Figure 6As shown, the navigation window prediction model includes an input layer, a CNN layer, an LSTM layer, a merging layer, a fully connected layer and an output layer. The input layer, the CNN layer, the LSTM layer, the merging layer, the fully connected layer and the output layer cooperate with each other to realize the recognition of the monitoring window area and the monitoring time window.

[0078] like Figure 6 As shown, the input layer is used to receive the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, so that the left ventricular myocardial motion image is converted into a processable matrix form.

[0079] like Figure 6 As shown, the CNN layer is used to receive the input of the input layer and extract the spatial features and local features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, and then determine the area of ​​the monitoring window based on the spatial features and local features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image.

[0080] like Figure 6 As shown, the LSTM layer is used to receive the input of the input layer and extract the time series features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, and then determine the left ventricular myocardial motion features, the right ventricular myocardial motion features, the apex motion features and the monitoring time window according to the time series features.

[0081] like Figure 6 As shown, the merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging. The fully connected layer is used to receive the output of the merging layer and input it into the output layer. The fully connected layer may include multiple fully connected layers to achieve the fusion of deep features.

[0082] like Figure 6 As shown, the output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, the right ventricular myocardial navigation and the apex navigation, so that the area and monitoring time window of the monitoring window can be directly obtained to realize efficient recognition of the myocardial navigation mode.

[0083] The beneficial effect of the above-mentioned implementation method is that by fusing the spatial features, local features and time series features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image through the CNN layer, the LSTM layer and the merging layer, the area of ​​the monitoring window and the monitoring time window can be directly obtained, thereby realizing efficient identification of the myocardial navigation mode.

[0084] In some implementations, the input layer is used to receive ECG signals, left ventricular myocardial motion images, right ventricular myocardial motion images, and apical motion images. The LSTM layer is also used to receive input from the input layer and extract time series features of the ECG signals, left ventricular myocardial motion images, right ventricular myocardial motion images, and apical motion images. When the first arrhythmia monitoring index is greater than or equal to a preset arrhythmia monitoring index, the LSTM layer is used to determine the arrhythmia monitoring feature based on the ECG signal. The LSTM layer is also used to extract time series features of the left ventricular myocardial motion image, right ventricular myocardial motion image, and apical motion image. The merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging. The fully connected layer is used to receive the output of the merging layer and input it to the output layer. The output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, right ventricular myocardial navigation, and apical navigation.

[0085] When working, in the embodiment of the present application, the input layer is used to receive ECG signals, left ventricular myocardial motion images, right ventricular myocardial motion images and apical motion images, and the LSTM layer is also used to receive inputs from the input layer and extract time series features of the ECG signals, left ventricular myocardial motion images, right ventricular myocardial motion images and apical motion images, thereby improving the accuracy of the monitoring time window in combination with the ECG signal.

[0086] When the first arrhythmia monitoring indicator is greater than or equal to the preset arrhythmia monitoring indicator, the LSTM layer is used to determine the arrhythmia monitoring feature based on the ECG signal. The arrhythmia monitoring feature characterizes the change characteristics of the heart rate, thereby improving the accuracy of determining the monitoring time window based on the arrhythmia monitoring feature.

[0087] In an embodiment of the present application, the LSTM layer is also used to extract the time series characteristics of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, thereby combining the arrhythmia monitoring characteristics, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image to improve the accuracy of determining the monitoring time window.

[0088] Similarly, the merging layer in this implementation is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging, the fully connected layer is used to receive the output of the merging layer and input it to the output layer, and the output layer is used to output the area and monitoring time window of the monitoring window in left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation.

[0089] The beneficial effect of the above implementation is that it can improve the accuracy of determining the monitoring time window by combining the time series characteristics of arrhythmia monitoring characteristics, left ventricular myocardial motion images, right ventricular myocardial motion images and apex motion images.

[0090] In some implementations, the above-mentioned navigation window prediction model also includes a front LSTM unit, which is used to receive input from the input layer and determine the arrhythmia monitoring features based on the ECG signal. The front LSTM unit inputs the arrhythmia monitoring features into the LSTM layer. The LSTM layer is used to extract the time series features of the arrhythmia monitoring features, the left ventricular myocardial motion images, the right ventricular myocardial motion images and the apex motion images. The merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging. The fully connected layer is used to receive the output of the merging layer and input it into the output layer. The output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, the right ventricular myocardial navigation and the apex navigation.

[0091] The above-mentioned navigation window prediction model also includes a front LSTM unit, which is used to receive input from the input layer and determine the arrhythmia monitoring characteristics according to the ECG signal, thereby improving the accuracy of determining the monitoring time window according to the arrhythmia monitoring characteristics.

[0092] When determining the area of ​​the monitoring window and the monitoring time window, the arrhythmia monitoring features can be input into the LSTM layer through the front LSTM unit. The LSTM layer is used to extract the time series features of the arrhythmia monitoring features, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image, thereby improving the accuracy of determining the area of ​​the monitoring window and the monitoring time window based on the time series features of the arrhythmia monitoring features, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image.

[0093] When determining the area of ​​the monitoring window and the monitoring time window, the merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging. The fully connected layer is used to receive the output of the merging layer and input it into the output layer. The output layer is used to output the area of ​​the monitoring window and the monitoring time window in left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation.

[0094] The beneficial effect of the above-mentioned implementation method is that by receiving the input of the input layer through the front LSTM unit and determining the arrhythmia monitoring characteristics according to the ECG signal, the accuracy of determining the monitoring time window according to the arrhythmia monitoring characteristics can be improved through the front LSTM unit, thereby improving the accuracy of determining the myocardial navigation method.

[0095] Figure 7 A schematic diagram of a flow chart of a third deep learning-based coronary artery scanning control method provided in an embodiment of the present application, such as Figure 7 As shown, the above method also includes S310 to S320, and S310 to S320 are described in detail below.

[0096] S310, determining the region and monitoring time window of the left ventricular myocardium navigation monitoring window and determining the window area of ​​the monitoring window region according to the ECG signal and the left ventricular myocardial motion image through the navigation window prediction model.

[0097] A study on the effect of navigator placement on free-breathing 3D acquisition and slice tracking with a navigation bar length of 40 mm and a gate window of 3.5-4.5 on the myocardium showed that there was almost no difference in scanning time or image quality scores when placed on the right semi-diaphragm and left ventricle. Therefore, in the embodiment of the present application, the navigation window prediction model can be preferentially used to determine the area and monitoring time window of the left ventricular myocardial navigation monitoring window based on the ECG signal and the left ventricular myocardial motion image, thereby determining the parameters of the left ventricular myocardial navigation.

[0098] After obtaining the region and monitoring time window of the left ventricular myocardium navigation monitoring window, the window area of ​​the region of the monitoring window can be determined, and then the navigation quality can be determined according to the window area of ​​the region of the monitoring window.

[0099] S320. When the window area is smaller than the preset window area, or the duration of the monitoring time window is smaller than the preset duration, the navigation window prediction model is used to determine the area and monitoring time window of the right ventricular myocardial navigation and apex navigation monitoring windows according to ECG signals, right ventricular myocardial motion images and apex motion images.

[0100] After obtaining the window area, when the window area is smaller than the preset window area, or the duration of the monitoring time window is smaller than the preset duration, it means that the area of ​​the navigation window is too small and the duration of the monitoring time window is too short, which may lead to poor navigation effect through the left ventricular myocardium. At this time, the navigation window prediction model can be used to determine the area and monitoring time window of the right ventricular myocardium navigation and apex navigation monitoring windows according to ECG signals, right ventricular myocardial motion images and apex motion images.

[0101] The beneficial effect of the above implementation method is that the navigation window prediction model can be preferentially used to determine the area and monitoring time window of the left ventricular myocardial navigation monitoring window according to the ECG signal and the left ventricular myocardial motion image, and then determine the parameters of the left ventricular myocardial navigation, thereby reducing the amount of calculation and improving the calculation efficiency of the myocardial navigation parameters.

[0102] The beneficial effect of the above-mentioned implementation method is that when the left ventricular myocardial navigation parameters are not good, the area and monitoring time window of the right ventricular myocardial navigation and apex navigation monitoring windows can be determined through ECG signals, right ventricular myocardial motion images and apex motion images, and the myocardial navigation parameters that meet the requirements can be gradually determined, thereby improving the accuracy of the myocardial navigation parameters.

[0103] In some implementations, the above method also includes: when the window area of ​​the monitoring window determined according to the ECG signal, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image is smaller than the preset window area, or the duration of the monitoring time window is smaller than the preset duration, the ECG signal of the patient is reacquired after adjusting the position of the ECG patch, and the area of ​​the monitoring window and the monitoring time window in the left ventricular myocardial navigation are determined according to the reacquired ECG signal and the left ventricular myocardial motion image through the navigation window prediction model.

[0104] When the window area of ​​the monitoring window determined based on the ECG signal, left ventricular myocardial motion image, right ventricular myocardial motion image and apical motion image is smaller than the preset window area, or the duration of the monitoring time window is less than the preset duration, it means that the obtained navigation parameters are not good, which may lead to navigation failure. At this time, the position of the ECG patch can be adjusted to reacquire the patient's ECG signal, and then the navigation parameters can be determined based on the reacquired patient's ECG signal to ensure the navigation effect.

[0105] After obtaining the reacquired ECG signal of the patient, the navigation window prediction model can be used to determine the area and monitoring time window of the left ventricular myocardial navigation according to the reacquired ECG signal and the left ventricular myocardial motion image, thereby determining the left visual myocardial navigation parameters.

[0106] The beneficial effect of the above implementation is that when the navigation parameters are not good, the ECG signal of the patient is reacquired after adjusting the position of the ECG patch, and the navigation parameters can be determined according to the reacquired ECG signal of the patient to ensure the navigation effect.

[0107] An embodiment of the present application also provides a deep learning-based coronary artery scanning control system, comprising a unit for executing the deep learning-based coronary artery scanning control method as described in any one of the above items.

[0108] Figure 8 A schematic diagram of the logical structure of a coronary artery scanning control system based on deep learning is provided in one embodiment of the present application, such as Figure 8 As shown, the system 1 of this embodiment includes a processing unit 11, a storage unit 12 and a transceiver unit 13. The processing unit 11 is used to process data, the storage unit 12 is used to store data, and the transceiver unit 13 is used to send and receive data. The processing unit 11, the storage unit 12 and the transceiver unit 13 cooperate with each other to implement the above method. The beneficial effects of the embodiment of the present application have been described in the above method and will not be repeated here.

[0109] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0110] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0111] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0112] An embodiment of the present application provides a computer program product. When the computer program product runs on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0113] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0114] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0115] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0116] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0117] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0118] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A coronary artery scanning control method based on deep learning, characterized in that: The method comprises: Based on the patient's coronal film imaging through the diaphragm top and myocardium, the patient's diaphragm motion image, left ventricular myocardial motion image, right ventricular myocardial motion image and apex motion image are determined respectively, and the patient's arrhythmia monitoring index is obtained through ECG calibration; The patient's diaphragm motion characteristics are determined through diaphragm motion images, the patient's left ventricular myocardial motion characteristics are determined through left ventricular myocardial motion images, the patient's right ventricular myocardial motion characteristics are determined through right ventricular myocardial motion images, and the patient's apical motion characteristics are determined through apical motion images; the patient's coronary artery navigation mode is determined through a navigation prediction model according to arrhythmia monitoring indicators, diaphragm motion characteristics, left ventricular myocardial motion characteristics, right ventricular myocardial motion characteristics and apical motion characteristics; wherein the navigation modes include diaphragm navigation, left ventricular myocardial navigation, right ventricular myocardial navigation and apical navigation.

2. The method according to claim 1, characterized in that The method further comprises: The diaphragm area is determined by the patient's diaphragm motion image, the left ventricular myocardium area is determined by the left ventricular myocardium motion image, the left ventricular myocardium position is determined by the left ventricular myocardium motion image, the right ventricular myocardium position is determined by the right ventricular myocardium motion image, and the apex position is determined by the apex motion image; wherein the diaphragm motion image, the left ventricular myocardium motion image, the right ventricular myocardium motion image and the apex motion image respectively include multiple frames of static images in a time sequence within 2 to 3 respiratory motion cycles and multiple cardiac cycles; The patient's diaphragm motion characteristics are determined by the diaphragm motion image, the patient's left ventricular myocardial motion characteristics are determined by the left ventricular myocardial motion image, the patient's right ventricular myocardial motion characteristics are determined by the right ventricular myocardial motion image, and the patient's apical motion characteristics are determined by the apical motion image, including: The patient's diaphragm motion characteristics are determined by the diaphragm area in the diaphragm motion image, the patient's left ventricular myocardial motion characteristics are determined by the left ventricular myocardial area in the left ventricular myocardial motion image, the patient's right ventricular myocardial motion characteristics are determined by the right ventricular myocardial position in the right ventricular myocardial motion image, and the patient's apex motion characteristics are determined by the apex position in the apex motion image.

3. The method according to claim 2, characterized in that The method further comprises: Acquire a first arrhythmia monitoring index of the patient through an ECG signal in a first time period, and when the first arrhythmia monitoring index is greater than or equal to a preset arrhythmia monitoring index, acquire a coronary myocardial transmural cine imaging of the patient in a second time period after the first time period, and respectively determine a left ventricular myocardial motion image, a right ventricular myocardial motion image, and an apex motion image of the patient; In a third time period after the second time period, the patient's left ventricular myocardial motion characteristics are determined through the left ventricular myocardial motion image, the patient's right ventricular myocardial motion characteristics are determined through the right ventricular myocardial motion image, and the patient's apical motion characteristics are determined through the apical motion image; the patient's coronary artery navigation mode is determined according to the left ventricular myocardial motion characteristics, the right ventricular myocardial motion characteristics and the apical motion characteristics through the navigation prediction model; wherein the navigation modes include left ventricular myocardial navigation, right ventricular myocardial navigation and apical navigation.

4. The method according to claim 3, characterized in that The method further comprises: During the second time period and the third time period, the diastolic time period of the cardiac cycle is determined by the ECG signal; and during the diastolic time period, the navigation window prediction model is used to determine the area and monitoring time window of the left ventricular myocardial navigation, right ventricular myocardial navigation and apex navigation based on the ECG signal, the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image.

5. The method according to claim 4, characterized in that The navigation window prediction model includes an input layer, a CNN layer, an LSTM layer, a merging layer, a fully connected layer and an output layer. The input layer is used to receive the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image. The CNN layer is used to receive the input of the input layer and extract the spatial features and local features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image. The LSTM layer is used to receive the input of the input layer and extract the time series features of the left ventricular myocardial motion image, the right ventricular myocardial motion image and the apex motion image. The merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging. The fully connected layer is used to receive the output of the merging layer and input it to the output layer. The output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, the right ventricular myocardial navigation and the apex navigation.

6. The method according to claim 5, characterized in that The input layer is used to receive ECG signals, left ventricular myocardial motion images, right ventricular myocardial motion images and apical motion images. The LSTM layer is also used to receive input from the input layer and extract time series features of the ECG signals, left ventricular myocardial motion images, right ventricular myocardial motion images and apical motion images. When the first arrhythmia monitoring index is greater than or equal to the preset arrhythmia monitoring index, the LSTM layer is used to determine the arrhythmia monitoring features according to the ECG signal. The LSTM layer is also used to extract time series features of the left ventricular myocardial motion images, right ventricular myocardial motion images and apical motion images. The merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging. The fully connected layer is used to receive the output of the merging layer and input it to the output layer. The output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, right ventricular myocardial navigation and apical navigation.

7. The method according to claim 5, characterized in that It also includes a front LSTM unit, which is used to receive input from the input layer and determine arrhythmia monitoring features based on the ECG signal. The front LSTM unit inputs the arrhythmia monitoring features into the LSTM layer. The LSTM layer is used to extract time series features of arrhythmia monitoring features, left ventricular myocardial motion images, right ventricular myocardial motion images, and apex motion images. The merging layer is used to merge the outputs of the CNN layer and the LSTM layer and perform feature merging. The fully connected layer is used to receive the output of the merging layer and input it into the output layer. The output layer is used to output the area and monitoring time window of the monitoring window in the left ventricular myocardial navigation, right ventricular myocardial navigation, and apex navigation.

8. The method according to claim 7, characterized in that The method further comprises: By using the navigation window prediction model, according to the ECG signal and the left ventricular myocardial motion image, the region and monitoring time window of the left ventricular myocardial navigation monitoring window are determined, and the window area of ​​the region of the monitoring window is determined; When the window area is smaller than the preset window area, or the duration of the monitoring time window is smaller than the preset duration, the navigation window prediction model is used to determine the area and monitoring time window of the right ventricular myocardial navigation and apex navigation monitoring windows based on ECG signals, right ventricular myocardial motion images and apex motion images.

9. The method according to claim 8, characterized in that The method further comprises: When the window area of ​​the monitoring window determined based on the ECG signal, left ventricular myocardial motion image, right ventricular myocardial motion image and apical motion image is smaller than the preset window area, or the duration of the monitoring time window is less than the preset duration, the ECG signal of the patient is reacquired after adjusting the position of the ECG patch, and the area of ​​the monitoring window and the monitoring time window in the left ventricular myocardial navigation are determined based on the reacquired ECG signal and left ventricular myocardial motion image through the navigation window prediction model.

10. A coronary artery scanning control system based on deep learning, characterized in that: The invention comprises a unit for executing the coronary artery scanning control method based on deep learning according to any one of claims 1 to 9.

Images