Rapid pulmonary vein modeling method and device

Abstract

The invention discloses a rapid pulmonary vein modeling method and device in the technical field of pulmonary vein modeling. The method comprises the following steps: acquiring an ultrasonic image sequence; screening and selecting a long-axis image and a short-axis image; segmenting an image boundary, and extracting boundary information; curve fitting and resampling are carried out on the boundary of the long axis, a center coordinate point is obtained through calculation, and the corresponding radius and length of the long axis are calculated; performing curve fitting and resampling on the boundary of the short axis, and obtaining the axial ratio of the pulmonary vein section according to the corresponding radius of the long axis; taking each center coordinate point as the geometric center of the ellipse, and obtaining a plurality of ellipse graphs according to the length of the long axis and the axial ratio; and mapping the plurality of elliptical graphs to a three-dimensional space, and performing triangulation processing to obtain a pulmonary vein three-dimensional graph. The rapid pulmonary vein modeling method is easy and convenient to operate, capable of achieving rapid modeling, low in requirement for fineness of an operator and convenient to apply and popularize, traditional contact type pulmonary vein modeling is replaced, and injuries in an operation are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of pulmonary vein modeling, and in particular to a method and device for rapid pulmonary vein modeling. Background Art

[0002] The left atrium is connected to four pulmonary veins, namely the left superior pulmonary vein (LSPV), the left inferior pulmonary vein (LIPV), the right superior pulmonary vein (RSPV) and the right inferior pulmonary vein (RIPV). Ectopic excitation foci of the pulmonary veins often cause paroxysmal atrial fibrillation. Radiofrequency catheter ablation can cure focal atrial fibrillation. Before radiofrequency ablation of atrial fibrillation, it is very important to understand and evaluate the anatomical structure of the pulmonary veins and the left atrium.

[0003] The purpose of pulmonary vein modeling is to accurately depict the three-dimensional position of these pulmonary veins and the angle at which they enter the left atrium to help doctors perform corresponding treatments. There are usually two ways to model pulmonary veins: contact and non-contact.

[0004] Pulmonary vein modeling is performed through contact method, where a modeling catheter is inserted into the pulmonary vein to perform "blind men and elephant" style modeling, such as Johnson & Johnson's PentaRay catheter. However, contact pulmonary vein modeling may cause damage to the patient.

[0005] In order to avoid the damage that contact pulmonary vein modeling may cause to patients, non-contact pulmonary vein modeling methods can be used at present. By scanning the pulmonary vein sections at different levels, all the sections containing the pulmonary veins are reconstructed in order to establish a pulmonary vein model; however, the scanned pulmonary vein inter-layer data requires the operator to be extremely precise, and the inter-layer thickness cannot be controlled, resulting in modeling difficulties. Therefore, the promotion and application of non-contact pulmonary vein modeling is relatively difficult. Summary of the invention

[0006] The present application solves the problem of difficulty in modeling using non-contact modeling methods in the prior art by providing a method and device for rapid modeling of pulmonary veins, thereby achieving rapid modeling of pulmonary veins.

[0007] The present application embodiment provides a method for rapid pulmonary vein modeling, comprising the following steps:

[0008] S1: Acquire an ultrasound imaging sequence that continuously collects images of the heart's interior;

[0009] S2: screening all frame images containing pulmonary veins from the ultrasound image sequence, and selecting a long-axis image and a short-axis image from all the frame images containing pulmonary veins;

[0010] S3: segmenting the boundaries of the long-axis image and the short-axis image, extracting boundary information, and obtaining a long-axis boundary image and a short-axis boundary image;

[0011] S4: performing curve fitting and uniform interpolation point sampling on all points on the long-axis boundary image to obtain two groups of long-axis sampling points; according to the interpolation order, averaging the coordinates of each pair of points corresponding to the two groups of long-axis sampling points to obtain a group of central coordinate points, and calculating the long-axis radius and long-axis length corresponding to each of the central coordinate points;

[0012] S5: performing curve fitting and uniform interpolation point sampling on all points on the short-axis boundary image to obtain a set of short-axis sampling points, and calculating the centroid point o; obtaining the corresponding short-axis radius according to the long-axis radius corresponding to the centroid point o, thereby obtaining the axial ratio of the pulmonary vein section;

[0013] S6: taking each of the central coordinate points as the geometric center of the ellipse, calculating and acquiring the short axis length corresponding to each of the central coordinate points according to the long axis length corresponding to each of the central coordinate points and the axial ratio of the pulmonary vein cross section, thereby obtaining a plurality of ellipse figures;

[0014] S7: Mapping the plurality of elliptical figures into a three-dimensional space in combination with corresponding magnetic positioning information, and performing triangulation processing to obtain a three-dimensional map of the pulmonary veins.

[0015] The beneficial effects of the above embodiments are: the rapid pulmonary vein modeling method is easy to operate, can be modeled quickly, has low operator precision requirements, and is easy to promote and apply, thereby replacing traditional contact pulmonary vein modeling and reducing injuries during surgery.

[0016] Based on the above embodiments, the present application can be further improved as follows:

[0017] In one embodiment of the present application, in step S1, the ultrasound image sequence is collected by rotating the ultrasound catheter after it is inserted into the heart.

[0018] In one of the embodiments of the present application, in the step S1, each ultrasound image in the ultrasound image sequence collects the corresponding magnetic positioning information.

[0019] In one embodiment of the present application, in step S2, all frame images containing pulmonary veins are screened by a deep neural network, and a long-axis image and a short-axis image are manually selected from all the frame images containing pulmonary veins. Combined with AI, the screening processing speed is improved, and combined with manual judgment, the accuracy is improved.

[0020] In one embodiment of the present application, in step S3, a model based on a convolutional neural network is used to segment the boundaries of the long-axis image and the short-axis image respectively to extract boundary information. Combined with AI, the processing speed is improved.

[0021] In one embodiment of the present application, cubic spline is used for curve fitting in step S4.

[0022] In one embodiment of the present application, the method for obtaining the minor axis radius in step S5 is as follows:

[0023] Calculate the distance between each of the short axis sampling points and the centroid point o, and find the point p closest to the long axis radius;

[0024] Connect the point p and the centroid point o to obtain a straight line l, calculate the normal line v of the straight line l, the normal line v passes through the centroid point o and intersects the short-axis boundary curve at point c, and the length between the point c and the centroid point o is calculated to be the short-axis radius of the pulmonary vein orifice.

[0025] The embodiment of the present application further provides a pulmonary vein rapid modeling device, according to the above pulmonary vein rapid modeling method, comprising:

[0026] A data acquisition module, used to obtain an ultrasound image sequence for continuously collecting images of the inside of the heart;

[0027] A screening module, used for screening all frame images containing pulmonary veins from the ultrasound image sequence, and selecting a long-axis image and a short-axis image from all the frame images containing pulmonary veins;

[0028] A segmentation module, used to segment the boundaries of the long-axis image and the short-axis image, extract boundary information, and obtain a long-axis boundary image and a short-axis boundary image;

[0029] A long axis calculation module is used to perform curve fitting and uniform interpolation point sampling on all points on the long axis boundary image to obtain two groups of long axis sampling points; according to the interpolation order, according to the coordinates of each pair of points corresponding to the two groups of long axis sampling points, average calculation is performed to obtain a group of central coordinate points, and the long axis radius and long axis length corresponding to each of the central coordinate points are calculated;

[0030] A short axis calculation module is used to perform curve fitting and uniform interpolation point sampling on all points on the short axis boundary image to obtain a set of short axis sampling points and calculate the centroid point o; according to the long axis radius corresponding to the centroid point o, the corresponding short axis radius is obtained, thereby obtaining the axial ratio of the pulmonary vein section;

[0031] A graphics generation module, used to take each of the central coordinate points as the geometric center of the ellipse, calculate and obtain the short axis length corresponding to each of the central coordinate points according to the long axis length corresponding to each of the central coordinate points and the axial ratio of the pulmonary vein cross section, thereby obtaining a plurality of elliptical graphics;

[0032] The three-dimensional modeling module is used to map the plurality of elliptical figures combined with the corresponding magnetic positioning information into a three-dimensional space, and perform triangulation processing to obtain a three-dimensional map of the pulmonary veins.

[0033] The method and device for rapid pulmonary vein modeling in the above-mentioned embodiments of the present application have at least the following technical effects or advantages:

[0034] 1. This rapid pulmonary vein modeling method is easy to operate, has low requirements on the operator's precision, is easy to promote and apply, can replace the traditional contact pulmonary vein modeling, and reduce damage during surgery;

[0035] 2. This rapid pulmonary vein modeling method combines the guidance of the AI ​​model on steps such as image acquisition and boundary segmentation, and can perform rapid pulmonary vein modeling with high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0037] Figure 1 This is a schematic diagram of the connection structure of the ultrasonic device in the embodiment of the present application;

[0038] Figure 2 A flowchart of the steps of a method for rapid pulmonary vein modeling in an embodiment of the present application;

[0039] Figure 3 (A) is a schematic diagram of rotating ultrasound image acquisition in an embodiment of the present application. Figure 3 (B) is a schematic diagram of an ultrasound image collected in an embodiment of the present application;

[0040] Figure 4 (A) is a schematic diagram of the segmentation result of the pulmonary vein long axis in the embodiment of the present application, Figure 4 (B) is a schematic diagram of the segmentation result of the short axis of the pulmonary vein in an embodiment of the present application;

[0041] Figure 5 This is a schematic diagram of the resampling results of the curve fitting in the embodiment of the present application;

[0042] Figure 6 This is a schematic diagram of the centroid calculation result in the embodiment of the present application;

[0043] Figure 7 This is a schematic diagram of matching and calculating the short axis in an embodiment of the present application;

[0044] Figure 8 This is a schematic diagram of fusion calculation in an embodiment of the present application;

[0045] Fig. 9 This is a schematic diagram of the mapping process in the embodiment of the present application;

[0046] Fig.10 This is a schematic diagram of the three-dimensional model of the pulmonary vein in an embodiment of the present application. DETAILED DESCRIPTION

[0047] The present invention is further explained below in conjunction with specific implementation methods. It should be understood that these implementation methods are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0048] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0049] The embodiment of the present application solves the problem of difficulty in modeling using non-contact modeling methods in the prior art by providing a method for rapid modeling of pulmonary veins, thereby achieving rapid modeling of pulmonary veins.

[0050] In the context of ICE (intracavitary ultrasound) technology, the core of pulmonary vein modeling relies on the combination of real-time imaging and three-dimensional reconstruction. ICE catheter technology can accurately observe the heart and its related vascular structures, such as the connection between the left atrium and the pulmonary veins, by performing ultrasound imaging in the heart cavity. ICE not only provides high-resolution two-dimensional images, but also can obtain data from multiple angles through the rotation and movement of the catheter. These multi-angle images are used to reconstruct the three-dimensional geometric model of the pulmonary veins to help doctors better plan and perform surgery.

[0051] The technical solution in the embodiment of the present application is to solve the above problems, and the device used is described as follows:

[0052] Ultrasound diagnosis is an imaging technology that uses the principle of echo. This technology uses a multi-element array transducer to transmit ultrasound into the human body, and by adjusting the relative delay and amplitude between the elements, a focused sound beam pointing in a specific direction is formed. When this sound beam encounters the interface of different organs and tissues in the body, it will produce reflected echoes, which are then received by the same transducer array. By receiving echo signals of different times and intensities, the device can construct an image of the structure in the body. The entire device consists of multiple interconnected components, and their connection methods are usually shown in detail in the schematic diagram in the device manual, such as Figure 1 This technology is widely used in the medical field and can provide real-time images of organs and tissues, which is very helpful for diagnosis and treatment planning. Figure 1 The meaning and function of each symbol are shown in Table 1 below:

[0053]

[0054] In order to compensate for the limitation of ICE imaging field of view, magnetic positioning technology is used in combination with ICE. Magnetic sensors are embedded in the catheter to track the position of the catheter in real time, and the acquired images are associated with spatial coordinates to generate a more accurate three-dimensional model. This can ensure the accuracy of the model and improve the precise navigation capability of the surgery.

[0055] like Figure 2 As shown, a method for rapid modeling of pulmonary veins includes the following steps:

[0056] S1: Data collection:

[0057] Using intracardiac ultrasound catheter technology, the catheter is inserted into the left atrium of the heart and rotated around the ultrasound probe within a certain angle range to continuously collect an ultrasound image sequence of the internal heart; each ultrasound image in the ultrasound image sequence collects corresponding magnetic positioning information.

[0058] Specifically, when imaging the left atrium and pulmonary veins, intracardiac ultrasound catheter technology is used. The catheter is inserted into the left atrium and rotated around the ultrasound probe within a certain angle range, such as Figure 3As shown in (A), a sequence of ultrasound images of the left atrium and pulmonary veins can be continuously collected. During this process, the rotation of the catheter is controlled by the motor in a stepping manner, while waiting for the sampling signal provided by the electrocardiogram to perform image sampling and record the rotation angle of the catheter at each specific moment. This rotation method supports high-speed 360° real-time internal rotation. In order to ensure that the collected images can cover different perspectives and positions of the left atrium and each pulmonary vein, and to ensure image quality, the sampling frequency needs to be adjusted according to the diastolic phase of the patient's ECG, which helps to ensure the accuracy of the subsequently generated three-dimensional model. The entire acquisition and analysis process can be updated in real time, and the model can be dynamically reconstructed to display the detailed structure inside the left atrium and pulmonary veins. The specific acquired images are shown in the figure. Figure 3 (B) shown.

[0059] S2: Similarity detection:

[0060] Through a deep neural network, all frame images containing the pulmonary veins are screened, and one long-axis image and one short-axis image are manually selected from all frame images containing the pulmonary veins.

[0061] Similarity detection is a deep neural network based on images trained with a large amount of clinical data. These data are annotated by professional doctors. It is mainly used here to identify whether the currently acquired pulmonary vein frame image meets the standards required for modeling and give a confidence score.

[0062] When the detection starts, the system will prompt the doctor to put the catheter into the Homeview surface. During this process, the system will prompt the doctor whether the target area has been reached. When the correct position is detected, the doctor clicks the start button, the motor automatically rotates and bends, and each frame of the image is detected using the similarity detection network. When an image frame that meets the conditions is found, a confidence score will be given to the frame, and the doctor will manually confirm whether to use the image.

[0063] S3: Boundary segmentation:

[0064] A convolutional neural network-based model is used to segment the boundaries of the long-axis image and the short-axis image, extract boundary information, and obtain the long-axis boundary image and the short-axis boundary image.

[0065] The boundary is segmented and the boundary information is extracted using a model framework based on a convolutional neural network. The boundary model is trained using ultrasound images of pulmonary vein boundaries annotated by professional physicians. The model can segment the boundaries of pulmonary veins in the long axis and short axis. The segmentation results are shown in Figure 4 , Figure 4 (A) is the segmentation result of the pulmonary vein long axis. Figure 4 (B) is the segmentation result of the pulmonary vein in the short axis.

[0066] S4: Calculation of the long axis of the pulmonary veins:

[0067] Obtain all points on the long axis boundary image (on two long line segments), perform curve fitting on these points to obtain two long axis boundary curves, perform point sampling of uniform interpolation on the two fitted long axis boundary curves respectively, and obtain two groups of long axis sampling points;

[0068] In the interpolation order, according to the coordinates of each pair of points corresponding to the two sets of major axis sampling points, a set of center coordinate points is obtained by averaging, and the radius corresponding to the set of center coordinate points is calculated. The major axis length is twice the radius.

[0069] The specific steps include:

[0070] S4.1: Line fitting:

[0071] All points on the two long line segments are obtained, and curve fitting is performed on these points. The fitting method is cubic spline, that is, a cubic polynomial is used to approximate the data points in each interval, and it is ensured that the fitting curve has continuous function values ​​and continuous derivatives at the connection of the intervals.

[0072] S4.2: Uniform sampling:

[0073] The above steps have obtained the fitted curve equation. Then, the line segment is uniformly sampled. The generation process of sampling points can be concisely described as:

[0074] Line representation: Assume that the two endpoints of the line segment are p0 = (x0, y0) and p1 = (x1, y1);

[0075] Number of sampling points: We hope to generate n evenly distributed sampling points on this line segment;

[0076] Uniform interpolation: Generate n points at equal intervals on the line segment between the starting point and the end point. The coordinates of these points can be calculated using the following formula (1):

[0077]

[0078] Among them, p k are the coordinates of the kth sampling point.

[0079] Through the above method, n points will be evenly generated on the line segment, and the interval between each point is equal.

[0080] S4.3: Centerline and Radius Calculations:

[0081] After obtaining the sampling points, according to the interpolation order, the coordinates of each two groups of points are averaged to calculate their center coordinate points. Assuming that the coordinates of one group of points are: A = (x a ,y a ) and B=(xb ,y b ), then their center coordinates are AB for:

[0082]

[0083] The major axis radius corresponding to the center coordinate point is calculated as:

[0084]

[0085] The curve formed by all the center coordinate points is the center line; the length of the major axis corresponding to each center coordinate point is twice the corresponding radius.

[0086] S5: Calculation of the short axis of the pulmonary veins:

[0087] S5.1: Get all the points on the short-axis boundary image (quasi-circular), perform curve fitting on these points to obtain a short-axis boundary curve, perform point sampling of uniform interpolation on the fitted short-axis boundary curve to obtain a set of short-axis sampling points; the short-axis curve fitting and uniform sampling methods are consistent with the corresponding long-axis processing methods in S4. The results of the curve fitting and uniform sampling are shown in the figure. Figure 5 shown.

[0088] S5.2: Calculate the centroid point o of the group of short axis sampling points;

[0089] Calculate the centroid o of the uniformly sampled point set, assuming that the point set is (x1, y1), (x2, y2), …, (x n ,y n ), center of mass (o x ,o y ) is calculated as follows:

[0090]

[0091] The calculated centroid point is Figure 6 shown.

[0092] S5.3: According to the long-axis radius of the centroid point o corresponding to the short-axis boundary image, the corresponding short-axis radius is obtained, thereby obtaining the axial ratio of the pulmonary vein cross section.

[0093] The pulmonary vein orifice is located at the position where the pulmonary vein is directly connected to the left atrium. The default position is the short-axis boundary image position and the long-axis boundary image endpoint position. During the long-axis calculation process, the long-axis length of the pulmonary vein orifice has been obtained, such as Figure 7 As shown, the long axis is matched with the short axis to obtain the axial ratio of the pulmonary vein section.

[0094] S5.3.1: Obtain the long axis radius corresponding to the central coordinate point corresponding to the pulmonary vein orifice, where the central coordinate point corresponding to the pulmonary vein orifice is the centroid point o;

[0095] S5.3.2: Calculate the distance between each minor axis sampling point and the centroid point o, and find the point p that is closest to the major axis radius;

[0096] S5.3.3: Connect point p and centroid o to obtain line l, calculate the normal v of line l, and the normal v will intersect the short axis boundary curve through the centroid o at point c. At this time, the length between point c and centroid o is calculated as the short axis radius of the pulmonary vein orifice. The short axis length is twice the short axis radius.

[0097] S5.3.4: Obtain the axial ratio of the pulmonary vein cross-section based on the major and minor axis lengths corresponding to the pulmonary vein orifices.

[0098] S6: Fusion computing:

[0099] like Figure 8 As shown, each central coordinate point is taken as the geometric center of the ellipse, and the short axis length corresponding to each central coordinate point is calculated and obtained according to the long axis length corresponding to each central coordinate point and the axial ratio of the pulmonary vein section, thereby obtaining multiple elliptical figures.

[0100] S7: 3D imaging:

[0101] Multiple elliptical figures combined with corresponding magnetic positioning information are mapped into three-dimensional space and triangulated to obtain a three-dimensional map of the pulmonary veins.

[0102] Since the center coordinate points corresponding to each ellipse image are located on a plane (long axis image), the spatial coordinates of the geometric center of each ellipse can be calculated by simply obtaining the quaternion and position information of the ultrasound plane (long axis image) through magnetic positioning information. Fig. 9 As shown, after the points are initialized, they are combined with magnetic positioning information to convert all points into points in three-dimensional space.

[0103] After obtaining the three-dimensional space points, these points are triangulated to obtain the final three-dimensional map of the pulmonary veins, such as Fig.10 shown.

[0104] A pulmonary vein rapid modeling device, based on the aforementioned pulmonary vein rapid modeling method, comprises:

[0105] A data acquisition module, used to obtain an ultrasound image sequence for continuously collecting images of the inside of the heart;

[0106] A screening module, used for screening all frame images containing pulmonary veins from the ultrasound image sequence, and selecting a long-axis image and a short-axis image from all frame images containing pulmonary veins;

[0107] A segmentation module is used to segment the boundaries of the long-axis image and the short-axis image, extract boundary information, and obtain a long-axis boundary image and a short-axis boundary image;

[0108] The long axis calculation module is used to perform curve fitting and uniform interpolation point sampling on all points on the long axis boundary image to obtain two groups of long axis sampling points; according to the interpolation order, according to the coordinates of each pair of points corresponding to the two groups of long axis sampling points, a group of center coordinate points is obtained by averaging and calculating, and the long axis radius and long axis length corresponding to each center coordinate point are calculated;

[0109] The short axis calculation module is used to perform curve fitting and uniform interpolation point sampling on all points on the short axis boundary image to obtain a set of short axis sampling points and calculate the centroid point o; according to the long axis radius corresponding to the centroid point o, the corresponding short axis radius is obtained, thereby obtaining the axial ratio of the pulmonary vein section;

[0110] A graphics generation module is used to take each central coordinate point as the geometric center of the ellipse, calculate and obtain the short axis length corresponding to each central coordinate point according to the long axis length corresponding to each central coordinate point and the axis ratio of the pulmonary vein section, so as to obtain multiple elliptical graphics;

[0111] The three-dimensional modeling module is used to map multiple elliptical figures combined with corresponding magnetic positioning information into three-dimensional space, and perform triangulation processing to obtain a three-dimensional map of the pulmonary veins.

[0112] The functions of each module in the embodiment correspond to the contents in the corresponding method embodiment and will not be repeated here.

[0113] The method and device for rapid modeling of pulmonary veins in the above-mentioned embodiment of the present application is a reconstruction method based entirely on ultrasound images. First, the image of the heart cavity acquired by the ultrasound catheter is automatically identified through a similarity network, and then the cavity boundary is determined through an automatic segmentation method. Then, the pulmonary vein centerline, the corresponding long axis and short axis length of the pulmonary vein are calculated to simulate the pulmonary vein, and finally the surface reconstruction is performed to display the pulmonary vein. It has at least the following technical effects or advantages:

[0114] 1. This pulmonary vein rapid modeling method combines the guidance of the AI ​​model on image acquisition and boundary segmentation, and can perform rapid pulmonary vein modeling with high accuracy;

[0115] 2. This rapid pulmonary vein modeling method is easy to operate and requires low precision from the operator. It can replace the traditional contact pulmonary vein modeling and reduce damage during surgery.

[0116] 3. This rapid pulmonary vein modeling method is real-time and fast, can achieve modeling while collecting data, and supports real-time updates.

[0117] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for rapid modeling of pulmonary veins, characterized in that: The following steps are involved: S1: Acquire an ultrasound imaging sequence that continuously collects images of the heart's interior; S2: screening all frame images containing pulmonary veins from the ultrasound image sequence, and selecting long-axis images and short-axis images from all frame images containing pulmonary veins; S3: segmenting the boundaries of the long-axis image and the short-axis image, extracting boundary information, and obtaining a long-axis boundary image and a short-axis boundary image; S4: performing curve fitting and uniform interpolation point sampling on all points on the long-axis boundary image to obtain two groups of long-axis sampling points; according to the interpolation order, averaging the coordinates of each pair of points corresponding to the two groups of long-axis sampling points to obtain a group of central coordinate points, and calculating the long-axis radius and long-axis length corresponding to each of the central coordinate points; S5: performing curve fitting and uniform interpolation point sampling on all points on the short-axis boundary image to obtain a set of short-axis sampling points, and calculating the centroid point o; According to the long axis radius corresponding to the centroid point o, the corresponding short axis radius is obtained, thereby obtaining the axial ratio of the pulmonary vein cross section; S6: taking each of the central coordinate points as the geometric center of the ellipse, calculating and acquiring the short axis length corresponding to each of the central coordinate points according to the long axis length corresponding to each of the central coordinate points and the axial ratio of the pulmonary vein cross section, thereby obtaining a plurality of ellipse figures; S7: Mapping the plurality of elliptical figures into a three-dimensional space in combination with corresponding magnetic positioning information, and performing triangulation processing to obtain a three-dimensional map of the pulmonary veins.

2. The method for rapid pulmonary vein modeling according to claim 1, characterized in that: In the step S1, the ultrasound image sequence is collected by rotating the ultrasound catheter after it is inserted into the heart.

3. The rapid pulmonary vein modeling method according to claim 1, characterized in that: In the step S1, each ultrasound image in the ultrasound image sequence collects the corresponding magnetic positioning information.

4. The rapid pulmonary vein modeling method according to claim 1, characterized in that: In the step S2, all frame images containing the pulmonary veins are screened by a deep neural network, and a long-axis image and a short-axis image are manually selected from all the frame images containing the pulmonary veins.

5. The method for rapid pulmonary vein modeling according to claim 1, characterized in that: In step S3, a model based on a convolutional neural network is used to segment the boundaries of the long-axis image and the short-axis image respectively to extract boundary information.

6. The method for rapid pulmonary vein modeling according to claim 1, characterized in that: In step S4, cubic spline is used for curve fitting.

7. The method for rapid pulmonary vein modeling according to claim 1, characterized in that: The method for obtaining the minor axis radius in step S5 is as follows: Calculate the distance between each of the short axis sampling points and the centroid point o, and find the point p closest to the long axis radius; Connect the point p and the centroid point o to obtain a straight line l, calculate the normal line v of the straight line l, the normal line v passes through the centroid point o and intersects the short-axis boundary curve at point c, and the length between the point c and the centroid point o is calculated to be the short-axis radius of the pulmonary vein orifice.

8. A pulmonary vein rapid modeling device, characterized in that: The method for rapid pulmonary vein modeling according to any one of claims 1 to 7 comprises: A data acquisition module, used to obtain an ultrasound image sequence for continuously collecting images of the inside of the heart; A screening module, used for screening all frame images containing pulmonary veins from the ultrasound image sequence, and selecting a long-axis image and a short-axis image from all the frame images containing pulmonary veins; A segmentation module, used to segment the boundaries of the long-axis image and the short-axis image, extract boundary information, and obtain a long-axis boundary image and a short-axis boundary image; A long axis calculation module is used to perform curve fitting and uniform interpolation point sampling on all points on the long axis boundary image to obtain two groups of long axis sampling points; according to the interpolation order, according to the coordinates of each pair of points corresponding to the two groups of long axis sampling points, average calculation is performed to obtain a group of central coordinate points, and the long axis radius and long axis length corresponding to each of the central coordinate points are calculated; A short axis calculation module is used to perform curve fitting and uniform interpolation point sampling on all points on the short axis boundary image to obtain a set of short axis sampling points and calculate the centroid point o; according to the long axis radius corresponding to the centroid point o, the corresponding short axis radius is obtained, thereby obtaining the axial ratio of the pulmonary vein section; A graphics generation module, used to take each of the central coordinate points as the geometric center of the ellipse, calculate and obtain the short axis length corresponding to each of the central coordinate points according to the long axis length corresponding to each of the central coordinate points and the axial ratio of the pulmonary vein cross section, thereby obtaining a plurality of elliptical graphics; The three-dimensional modeling module is used to map the plurality of elliptical figures combined with the corresponding magnetic positioning information into a three-dimensional space, and perform triangulation processing to obtain a three-dimensional map of the pulmonary veins.

Images