A method and system for calculating cross-sectional area and included angle of three-dimensional blood vessel branch

Abstract

The present application relates to a kind of three-dimensional blood vessel branch cross-sectional area and included angle calculation method and system, belong to medical image technology field, including obtaining the outline of main blood vessel and branch blood vessel, and determine the connecting point of main blood vessel and branch blood vessel, three-dimensional branch outer contour and the initial normal vector of branch section, determine its position at branch connection and the intersection section of branch section and branch outer contour;Equivalent two-dimensional branch section contour is obtained by rotating three-dimensional branch section contour, the area of two-dimensional branch section contour is calculated;A series of branch section area and the normal vector corresponding thereto are obtained;The minimum branch section area is taken as the cross-sectional area of this section branch blood vessel, and the included angle of branch blood vessel and main blood vessel is determined, the present application not only can confirm the outline of main blood vessel and branch blood vessel in two-dimensional space, can also carry out three-dimensional branch modeling and its accurate quantification, further analysis and calculation obtain the cross-sectional area of branch blood vessel in three-dimensional space and its included angle with main blood vessel.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical imaging technology, and a three-dimensional blood vessel branch cross-sectional area and included angle calculation method and system. BACKGROUND

[0002] Coronary heart disease is one of the main causes of disease death cases worldwide, and with the acceleration of the trend of social aging, it is more important to strengthen the monitoring and treatment of coronary heart disease. The main cause of coronary heart disease is heart blood supply caused by coronary artery stenosis or obstruction, so it is necessary to quantify the blood supply capacity of the coronary artery. The branch in the stenosis region of the blood vessel is an important factor affecting the coronary blood flow, and the current clinically used coronary angiography imaging can show the degree of vascular stenosis, but due to its low imaging resolution, it cannot realize the accurate identification of the three-dimensional shape of the branch.

[0003] Intravenous ultrasound (IVUS) is a technology that uses a catheter to introduce a high-frequency micro-ultrasound probe into the blood vessel lumen for detection. After full-range scanning in the blood vessel lumen, the electronic imaging system displays the micro-anatomical information of the blood vessel tissue structure and geometric shape. Intravascular optical coherence tomography (OCT) is a rapidly developing intravascular tomography method in recent years, which uses the light transmission of biological structures to convert the reflection and scattering signals of the detected biological tissues into electrical signals, and reconstructs the biological tissue structure image. IVUS and OCT imaging both have high spatial resolution and can accurately identify blood vessel information and branch blood vessels extending from the main blood vessels. Among them, IVUS has strong penetration and OCT imaging has clearer structure, so the integrated image of the two can obtain more comprehensive branch geometric information.

[0004] Currently, the research on branch geometric information based on intravascular imaging mainly focuses on the automatic identification of branch shape in the two-dimensional image sequence, but the quantitative measurement of branch geometric information in the evaluation of vascular structure and function depends on the accurate quantification and reconstruction of the branch in three-dimensional space. For example, the gold standard for evaluating coronary stenosis, Fraction Flow Reservation (FFR), is affected by the cross-sectional area of the lumen branch in three-dimensional space. Clinically, a vascular stent is often implanted to provide radial support for the blocked blood vessels, thereby relieving stenosis and restoring coronary blood flow. The stent needs to be implanted in the connection part of the main blood vessel and the branch to achieve accurate implantation, prevent poor stent adhesion, blood vessel reocclusion, and thrombosis after the stent. In particular, accurate identification and quantification of branches in two-dimensional images do not mean that the real size of the vascular branch in three-dimensional space can be obtained, because three-dimensional space requires additional information such as the angle of the branch relative to the main blood vessel in three-dimensional space; therefore, it is difficult to automatically, quickly and accurately measure the real opening area of the three-dimensional branch.

[0005] It should be noted that the information disclosed in the above background section is only used to strengthen the understanding of the background of the present disclosure, and therefore can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] The purpose of the present application is to overcome the shortcomings of the prior art and provide a three-dimensional vascular branch cross-sectional area and angle calculation method and system, which solves the problem that the prior art does not support geometric quantitative measurement in three-dimensional space and cannot automatically and quickly and accurately measure the real opening area of the three-dimensional vascular branch and the angle.

[0007] The purpose of the present application is achieved by the following technical solution: a three-dimensional vascular branch cross-sectional area and angle calculation method, the calculation method comprising:

[0008] S1, automatically detecting the profiles of the main blood vessel and the branch blood vessel according to intravascular imaging data, and determining the connection point of the main blood vessel and the branch blood vessel;

[0009] S2, determining the three-dimensional branch outer profile and determining the initial normal vector of the branch cross section;

[0010] S3, translating the branch cross section to the branch connection point, determining the position of the branch cross section at the branch connection point, and determining the intersection cross section of the branch cross section and the branch outer profile;

[0011] S4, obtaining an equivalent two-dimensional branch cross-sectional profile by rotating the three-dimensional branch cross-sectional profile, and calculating the area of the two-dimensional branch cross-sectional profile;

[0012] S5. Rotate the normal vector of the branch section and repeat steps S3 and S4 to obtain a series of branch section areas and their corresponding normal vectors.

[0013] S6. Select the smallest branch section area s among all branch sections. Min The cross-sectional area of ​​the branch vessel is used as the reference value, and the angle β between the branch vessel and the main vessel is determined by the angle between the normal vector corresponding to the minimum branch cross-sectional area and the z-axis.

[0014] In step S1, the contours of the main blood vessel and branch vessels are automatically detected based on intravascular imaging data, and the connection points of the main blood vessel and branch vessels are determined, including:

[0015] High-level semantic features within blood vessel images are obtained through convolutional neural networks. Contextual information is extracted from these high-level semantic features. Foreground information is erased through reverse attention to identify specific tissue regions, resulting in image frames that characterize the structural details of blood vessel segmentation boundaries. It is then determined whether branches exist in the current frame image. If they do, the segmentation results of the main blood vessel and branch blood vessels are output separately, and the contours are extracted to obtain the main blood vessel contour L1 and the branch blood vessel contour L2.

[0016] Obtain the intersection segment of the main vessel contour L1 and the branch vessel contour L2 as the connection contour L3, i.e., L3 = L1 ∩ L2, where contour L3 is composed of a continuous point set P = {p1, p2, ..., p...} i} is composed of the contour points p1 and p2 at both ends of the point set P. i Let Q1 and Q2 be the connection points between the main blood vessel and the branch blood vessel in this frame image, where i is the number of points in the point set P.

[0017] In step S2, determining the three-dimensional branch outer contour includes:

[0018] The true two-dimensional outer contour L of the blood vessel in each frame is obtained from the r-frame sequence of images with branches. Out =L2-L3, by stacking the frames in sequence, the three-dimensional contour L is obtained. V =∑L Out ;

[0019] Insert m virtual frames in the middle of each frame, and use the real branch contours L of the preceding and following frames. Out The branch contours L of each virtual frame are synthesized by combining different influence factors. Vir After inserting the virtual frame branch outline, the total branch outer outline is obtained as L. New =L V +∑L Vir .

[0020] In step S2, determining the initial normal vector of the branch tangent includes:

[0021] Based on the sequence segmentation results, a three-dimensional model is performed, and the centroid coordinates of the three-dimensional region V1 of the main blood vessel and the three-dimensional region V2 of the branch blood vessel are calculated and denoted as M1 and M2.

[0022] The vector is obtained from M1 and M2. The initial normal vector N of the branch tangent Ori The initial sectional plane S of the branch is obtained through the centroid M1 of the main blood vessel. Ori .

[0023] In step S3, translating the branch cut surface to the branch connection point and determining its position at the branch connection point, as well as the intersection section of the branch cut surface and the outer contour of the branch, includes:

[0024] The initial cut surface S Ori Translate along the normal vector direction to the profile L that connects to the branch. C The intersection and the translated tangent plane are denoted as S. F ;

[0025] Section S F In three-dimensional space, the outer contour L of the branch New Intersecting frames, and arranging all intersection points in counter-clockwise order, yields the three-dimensional branch section profile point set C for all branch section profiles. 3d ={c1, c2, ... c h}, where h is the set of three-dimensional branch section contour points C 3d The number of.

[0026] After obtaining multiple branch section profiles, the multiple branch section profiles are filtered according to branch section filtering conditions to obtain valid branch section profiles; the branch section filtering conditions include:

[0027] Section S F With branch outer contour L New There are two intersection points at the most recent frame position, and the subsequent consecutive a frames intersect with the branch outer contour L. New There are two intersection points, a≥2, and the last frame of consecutive a frames has only one intersection point;

[0028] If frame a is not the furthest frame in the branch segment, then the branch outer contour L New All frames after frame a+1 are not related to the tangent S. F intersect.

[0029] In step S4, obtaining an equivalent two-dimensional branch cross-section profile by rotating the three-dimensional branch cross-section profile includes:

[0030] In the three-dimensional branch section profile C 3d For any point c on K, calculate its perpendicular point c on K. K and its relationship with S Kthe angle α between the line and the plane S, and the point c K is the center of the plane S K is rotated by an angle α in the direction, and the two-dimensional equivalent point c′ of the point is obtained;

[0031] The three-dimensional branch cross-sectional profile C 3d is rotated to the plane S K , and the equivalent two-dimensional profile of the three-dimensional branch cross-sectional profile is obtained, denoted as C 2d .

[0032] In the S4 step, the area of the two-dimensional branch cross-sectional profile is calculated, including:

[0033] The straight line parts at both ends of the starting frame and the ending frame are replaced by a circular arc profile, and the corrected two-dimensional profile C Arc is obtained, and the area of the simply connected domain formed by C Arc is calculated by using the Green formula, that is, the area of the branch cross-sectional profile s Now is obtained, wherein the height h of the circular arc profile is linearly or nonlinearly related to the actual physical height H between the frames and the current cross-sectional angle.

[0034] In the S5 step, the rotation mode of the initial normal vector of the branch cross section includes proportionally changing the angle for rotation or changing the angle according to the mode of sparse first and dense later; and the intravascular imaging data includes intravascular optical coherence tomography data and intravascular ultrasound data.

[0035] A three-dimensional blood vessel branch cross-sectional area and angle calculation system includes an image acquisition module, a blood vessel branch calculation module, a subsequent processing module, and a display module;

[0036] The image acquisition module is used to acquire an OCT image or an IVUS image of a target blood vessel, or simultaneously acquire an OCT image and an IVUS image;

[0037] The blood vessel branch calculation module is used to automatically detect the profiles of a main blood vessel and a branch blood vessel according to intravascular imaging data, determine the connection point of the main blood vessel and the branch blood vessel, determine the three-dimensional branch outer profile, and determine the initial normal vector of the branch cross section; the branch cross section is translated to the branch connection, and the position of the branch cross section at the branch connection and the intersection of the branch cross section and the branch outer profile are determined; the three-dimensional branch cross-sectional profile is rotated to obtain an equivalent two-dimensional branch cross-sectional profile, and the area of the two-dimensional branch cross-sectional profile is calculated; the normal vector of the branch cross section is rotated to obtain a series of branch cross-sectional areas and the corresponding normal vectors; the smallest branch cross-sectional area s Min is selected as the cross-sectional area of the branch blood vessel, and the angle β between the normal vector corresponding to the smallest branch cross-sectional area and the z-axis is used to determine the angle between the branch blood vessel and the main blood vessel;

[0038] The subsequent processing module is used for carrying out corresponding subsequent processing on the calculated branch area and the included angle information.

[0039] The display module is used for displaying the target blood vessel branch detection and calculation result through various dimensional display modes.

[0040] The present application has the following advantages: a three-dimensional blood vessel branch cross-sectional area and included angle calculation method and system can not only confirm the profile of the main blood vessel and the branch blood vessel in the two-dimensional space, but also can perform three-dimensional branch modeling and accurate quantification, further analyze and calculate the cross-sectional area of the branch blood vessel in the three-dimensional space and the included angle with the main blood vessel; can guide the subsequent quantitative evaluation of blood vessel morphology and function, etc. For example: FFR calculation, accurate branch blood vessel cross-sectional area in three-dimensional space is needed to calculate the shunt; when a stent is implanted at the connection between the main blood vessel and the branch in clinic, the accurate included angle of the branch blood vessel relative to the main blood vessel in the three-dimensional space is needed for directional positioning; and promotes the development of three-dimensional quantification technology of blood vessel branches. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 It is a schematic diagram of the connection between the main blood vessel and the branch blood vessel based on the blood vessel morphology and fluid mechanics theory;

[0042] Figure 2 It is a flowchart for automatically identifying the branch blood vessel and calculating the three-dimensional cross-sectional area thereof and the included angle with the main blood vessel;

[0043] Figure 3 It is a schematic diagram of the profile of the main blood vessel and the branch blood vessel and the connection therebetween in the two-dimensional space;

[0044] Figure 4 It is a schematic diagram of the outer profile after three-dimensional modeling of the branch;

[0045] Figure 5 It is a schematic diagram of determining the initial tangent plane in the three-dimensional space and translating the tangent plane to the branch connection;

[0046] Figure 6 It is a schematic diagram of the cross-sectional profile obtained by intersecting the three-dimensional tangent plane with the branch;

[0047] Figure 7 It is a schematic diagram of rotating the three-dimensional cross-sectional profile to the two-dimensional plane and adjusting the profile; DETAILED DESCRIPTION

[0048] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in combination with the drawings is not intended to limit the protection scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the present application. The present application will be further described below in combination with the drawings.

[0049] One of the embodiments of the present application relates to a method for automatically identifying a branch blood vessel and calculating its three-dimensional cross-sectional area and the included angle with the main blood vessel, which mainly solves the technical problem of detecting the branch profile by using intravascular image data, calculating the cross-sectional area of the blood vessel branch in three-dimensional space and the included angle of the branch blood vessel with the main blood vessel, so as to guide the subsequent quantitative evaluation of blood vessel morphology and function, such as FFR calculation and direction positioning of branch stent implantation. The present application can not only extract the two-dimensional branch profile and its position, but also perform three-dimensional branch modeling on this basis, and further analyze and calculate the cross-sectional area of the branch blood vessel relative to the main blood vessel in three-dimensional space and its related three-dimensional profile and included angle data; the specific contents include the following:

[0050] As shown in Figure 1 , based on the blood vessel morphology and fluid mechanics theory, the direction of the branch blood vessel represents the flow direction v when the blood flow is divided into the branch, and the cross section of the branch is the plane S perpendicular to the blood flow direction and located at the connection between the branch and the main blood vessel. Any direction plane at the branch connection obtains a cross section S` intersecting with the branch blood vessel, and the plane S is perpendicular to the extension direction of the branch, so the area of S is smaller than that of S`. Therefore, the cross section with the smallest area at the branch connection is obtained, which is the three-dimensional cross section of the blood vessel branch, and the normal direction is the included angle direction of the branch.

[0051] The flow chart for calculating the above data is shown in Figure 2 , and the specific steps are as follows.

[0052] Step one, according to the intravascular imaging data, the profiles of the main blood vessel and the branch blood vessel are automatically detected.

[0053] A sequence of two-dimensional intravascular images of a patient's blood vessels is obtained using an intravascular imaging system. The original vascular information in three-dimensional space is reflected by the sequential stacking of these frames. In this three-dimensional space, the positive directions of the width and height of each frame are used as the positive x-axis and y-axis, respectively, and the stacking direction of the frames is used as the positive z-axis. Since intravascular imaging is based on a pull-back system, the positive direction of blood flow in the main vessel is the negative z-axis. Based on the large amount of data obtained from the intravascular imaging system, artificial intelligence algorithms are used for automated detection of vessel and branch contours.

[0054] Specifically, a convolutional neural network is used for the algorithm implementation, comprising an encoder and a parallel partial decoder to acquire high-level semantic global information within the blood vessel image. It includes a channel feature pyramid module to extract contextual information from high-level features. It also includes an axial inverse attention module to erase foreground information through inverse attention to identify specific tissue regions and obtain structural details of the blood vessel segmentation boundaries.

[0055] Determine if a branch exists in the current frame. If no branch exists, directly output the main vessel segmentation result without branch analysis. If a branch exists, output the segmentation results for the main vessel and branch vessels separately, such as... Figure 3 As shown, the output maintains the same dimensions as the original image. In the segmentation result, S1 represents the main vessel region, and S2 represents the branch vessel region. S1 and S2 are mutually exclusive but adjacent. Further contour extraction yields L1 representing the main vessel contour and L2 representing the branch vessel contour. L1 and L2 intersect at their connection points.

[0056] Step 2: Determine the connection point between the main blood vessel and the branch blood vessels.

[0057] like Figure 3 As shown, the intersection of the main vessel contour L1 and the branch vessel contour L2 is taken as the connection contour L3, i.e., L3 = L1 ∩ L2. The image scale is n*n, and the coordinates of the image center point C are (n / 2, n / 2). The central region of the intravascular imaging represents the imaging duct. The main vessel contour is a closed curve surrounding the outside of the duct, and the branch contours are outside the main vessel contour. Taking the image center point C as the reference point, and rotating counterclockwise as the positive direction, contour L3 is formed by a continuous point set P = {p1, p2, ..., p...}. i It consists of the contour points p1 and p2 located at the two ends of the point set P. i We take these as the connection points Q1 and Q2 between the main blood vessel and the branch blood vessel in this frame of imaging.

[0058] Step 3: Determine the outer contour of the three-dimensional branch.

[0059] like Figure 3 As shown, the branch connection points Q1 and Q2 obtained above can divide the branch profile L2 into two parts, L In L is the virtual contour obtained during imaging segmentation.Out The actual outer contour of the blood vessel branch shows that L Out =L2-L3. Stack the r-frame sequence images with branches sequentially to obtain the result from each frame L. Out The true outer contour L of the branches in the three-dimensional space V .like Figure 4 The example shown is for r=3, L V =∑L Out The connection points Q1 and Q2 of each frame branch in this segment are represented in three-dimensional space, and the points are rearranged in a counterclockwise direction to obtain the point set Open = {o1, o2, ... o...}. k}, which is the connection contour L of the main blood vessel and branch blood vessels in three-dimensional space. C .

[0060] Due to limitations in the pullback parameters of intravascular imaging systems, the total number of frames in the branch section is usually small, and abrupt changes in the contour between consecutive frames can occur. Interpolation can be performed between consecutive frames to simulate the contour and buffer the magnitude of these abrupt changes. m virtual frames are inserted in the middle of each frame, and the branch contour L in each virtual frame is... Vir Based on the true branch contours L of the preceding and following frames Out The influencing factors are synthesized by combining different factors, and these factors vary with the position of the virtual frame, and can be linear or nonlinear. The total branch outer contour after inserting the simulated contour is denoted as L. New ,like Figure 4 The image shows an example where m = 2, L New =L V +∑L Vir .

[0061] Step 4: Determine the initial normal vector of the branch tangent.

[0062] like Figure 5 As shown, 3D modeling is performed based on the sequence segmentation results, yielding the 3D region V1 of the main vessel and the 3D region V2 of the branch vessels. The centroid coordinates of V1 and V2 are calculated and denoted as M1 and M2, respectively. Vectors are then used. The initial normal vector N of the branch tangent Ori The initial sectional plane S of the branch is obtained through the centroid M1 of the main blood vessel. Ori .

[0063] Step 5: Determine the position of the above-mentioned cut surface at the branch connection point.

[0064] like Figure 5 As shown, the initial cut surface S Ori Translate along the normal vector direction until it connects with the profile L of the branch. C Intersection. First, calculate the branch connection profile L. C From each point on the surface to the initial tangent S OriCalculate the Cartesian distance to obtain the farthest distance d and the current contour point F. Then, calculate S... Ori Along N Ori Translate it by a distance d in the direction so that it is aligned with L C The plane intersects at contour point F, and the translated plane is the plane with N as its boundary. Ori Let S be the three-dimensional branch section along the direction of the normal vector. F .

[0065] Step 6: Determine the intersection section between the above-mentioned cut surface and the outer contour of the branch.

[0066] Branch Outer Contour L New It consists of r real frames and m virtual frames inserted in the middle of each frame, with a cross-section S. F In three-dimensional space, the outer contour L of the branch New Each frame intersects at two points. Arranging all these intersection points in counter-clockwise order yields the point set C. 3d ={c1, c2, ... c h This is the profile of the branch section. For example... Figure 6 The example shown is where r = 3 and m = 2.

[0067] But the cross-section S F The direction and position determine the relationship with L New The intersection of each frame is not limited to all slices S. F All can be with L New Each frame has two intersection points. If the frames are arranged along the positive z-axis and the blood flow direction is along the negative z-axis, then the effective branch section profile must satisfy the condition of section S. F With L New There are two intersections at the most recent frame position, and the next consecutive a frames intersect with L. New There are two intersection points, a≥2, and the last frame of consecutive 'a' frames can have only one intersection point. Furthermore, if frame 'a' is not the farthest frame in this branch segment, then L... New All frames after frame a+1 are not related to the tangent S. F Intersecting. There are various cases where the branch section profile does not meet the above valid conditions, and all of them are considered invalid profiles and need to be screened out.

[0068] Step 7: Rotate the three-dimensional branch section profile to obtain an equivalent two-dimensional profile.

[0069] After satisfying the valid conditions of the profile, the area is further calculated. The area calculation for a three-dimensional cross-sectional profile is extensive; therefore, it can be rotated to a two-dimensional plane to obtain its equivalent two-dimensional profile before performing the area calculation. For example... Figure 7 As shown, taking the intersecting contour line K of the most recent frame as the reference, a plane S parallel to the coordinate plane XOY is constructed. K This is taken as the plane containing the equivalent two-dimensional contour. Three-dimensional contour C 3d For any point c on K, calculate its perpendicular point c on K.K and its relationship with S K The angle α between point c and point c. K Centered on plane S K Rotate the direction by an angle α to obtain the two-dimensional equivalent point c'. Then, transform the three-dimensional contour C... 3d Rotate all points to plane S K The equivalent two-dimensional profile of the three-dimensional branch section profile is obtained, denoted as C. 2d .

[0070] Step 8: Calculate the area of ​​the two-dimensional branch cross-section outline.

[0071] like Figure 7 As shown, the branch section contour obtained by the above method appears as a straight line at the nearest and farthest frames, but the actual branch section contour as a whole is a uniform curve. This is because the pullback parameter of the intravascular imaging system is a given value and cannot guarantee that it will image exactly at the start and end positions of the branch contour. Typically, the start position is between the frame where the branch can be automatically detected and the frame preceding it, and the end position is similarly determined. Therefore, to further restore the actual branch section contour, circular arc contours need to be added at both ends to replace the straight line, resulting in the modified two-dimensional contour C. Arc .

[0072] The arc height h is related to the actual physical height H between frames and the current cross-section angle, and this relationship can be linear or non-linear. Simply put, considering only the distance between frames, we can take the probability average, i.e., H = 2*h. C is then calculated using Green's theorem. Arc The area of ​​the formed simply connected region is equal to the required branch cross-sectional area s. Now .

[0073] Step nine: Rotate the normal vector of the branch cross-section. Repeat steps five through eight to obtain a series of branch cross-sectional areas and their corresponding normal vectors. When rotating the branch cross-section, the angle can be changed proportionally in each direction in three-dimensional space, or the angle can be changed from sparse to dense to save computational resources.

[0074] Step 10: Select the smallest cross-sectional area s. Min The cross-sectional area of ​​this branch vessel is represented by the angle β between the normal vector of the tangent plane and the z-axis, which is the angle between the branch vessel and the main vessel.

[0075] Another embodiment of the present invention relates to a system for calculating the cross-sectional area of ​​vascular branches and the angle between branch vessels and main vessels in three-dimensional space based on intravascular imaging information. It includes an image acquisition module, a vascular branch calculation module, a post-processing module and a display module.

[0076] The image acquisition module is configured to acquire an OCT image or an IVUS image of the target blood vessel, or simultaneously acquire an OCT and IVUS image;

[0077] The blood vessel branch calculation module is configured to automatically detect the profiles of the main blood vessel and the branch blood vessel according to the intravascular imaging data, determine the connection points of the main blood vessel and the branch blood vessel, determine the three-dimensional outer profile of the branch and the initial normal vector of the branch cross section, translate the branch cross section to the branch connection point, determine the position of the branch cross section at the branch connection point and the intersection section of the branch cross section and the outer profile of the branch, obtain the equivalent two-dimensional branch cross section profile by rotating the three-dimensional branch cross section profile, and calculate the area of the two-dimensional branch cross section profile; the normal vector of the branch cross section is rotated to obtain a series of branch cross section areas and the corresponding normal vectors; the minimum branch cross section area s Min is selected from all the branch cross section areas, and the included angle β between the normal vector corresponding to the minimum branch cross section area and the z-axis is used to determine the included angle between the branch blood vessel and the main blood vessel;

[0078] The subsequent processing module is configured to perform corresponding subsequent processing on the calculated branch area and the included angle information.

[0079] The subsequent processing includes but is not limited to calculation of FFR, automatic identification and classification of atherosclerotic plaques, simulation and modeling of blood flow dynamics of the target blood vessel (such as calculation of the shear force of the blood vessel wall), virtual stent implantation, surgical planning, treatment of bifurcation divergence, and the like; the FFR is the ratio of the maximum blood flow that can be obtained by the myocardium in the stenosis region to the maximum blood flow that can be obtained by the myocardium in the case of no stenosis after the stenosis of the coronary artery.

[0080] The display module is configured to display the target blood vessel branch detection and calculation results together with the OCT or IVUS or coronary angiography image information in one-dimensional, two-dimensional or three-dimensional manner on a display terminal; a user can observe the branch position in the three-dimensional space through a software interactive interface and simulate the surgical planning in real time.

[0081] The above description is only the preferred embodiments of the present application, and it should be understood that the present application is not limited to the forms disclosed herein, and should not be considered as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concepts described herein by the above teachings or related art or knowledge. Any modification and change made by those skilled in the art without departing from the spirit and scope of the present application shall be within the protection scope of the appended claims of the present application.

Claims

1. A method for calculating cross-sectional area and included angle of three-dimensional blood vessel branch, characterized in that: The calculation method comprises: S1, automatically detecting the profiles of the main blood vessel and the branch blood vessel according to the intravascular imaging data, and determining the connection points of the main blood vessel and the branch blood vessel; S2, determining a three-dimensional branch outer profile and determining an initial normal vector of a branch cross section; S3, translating the branch cross section to the branch connection and determining the position of the branch cross section at the branch connection and the intersection section of the branch cross section and the branch outer profile; S4, obtaining an equivalent two-dimensional branch cross section profile by rotating the three-dimensional branch cross section profile, and calculating the area of the two-dimensional branch cross section profile; S5, rotating the initial normal vector of the branch cross section, and repeating steps S3 and S4 to obtain a series of branch cross section areas and the corresponding normal vectors; S6, selecting the minimum branch cross-sectional area s among all branch cross-sectional areas Min The cross-sectional area of the branch vessel as the segment is determined, and the angle between the branch vessel and the main vessel is determined by the angle β between the normal vector corresponding to the minimum branch cross-sectional area and the z-axis.

2. The method for calculating cross-sectional area and included angle of three-dimensional blood vessel branch according to claim 1, characterized in that: In the S1 step, the profiles of the main blood vessel and the branch blood vessel are automatically detected according to the intravascular imaging data, and the connection points of the main blood vessel and the branch blood vessel are determined, which comprises: obtaining high-level semantic features in the blood vessel image through a convolutional neural network, obtaining context information from the high-level semantic features, identifying specific tissue regions by erasing foreground information through a reverse attention, obtaining an image frame representing the structural details of the blood vessel segmentation boundary, judging whether the current frame image exists a branch, and if so, outputting the segmentation results of the main blood vessel and the branch blood vessel respectively, and extracting the profiles to obtain the main blood vessel profile L1 and the branch blood vessel profile L2; The intersecting section of the main blood vessel contour L1 and the branch blood vessel contour L2 is obtained as the connection contour L3, i.e. L3=L1∩L2, wherein the contour L3 is composed of a section of continuous point set P={p1, p2,..., p i} and the contour points p1 and p i at both ends of the point set P are taken as the connection points Q1 and Q2 of the main blood vessel and the branch blood vessel in the frame image, wherein i is the number of the point set P.

3. The method of claim 2, wherein: In the S2 step, the three-dimensional branch outer profile is determined, which comprises: The real two-dimensional vessel outer contour L of each frame is obtained from the r-frame sequence image with branches respectively Out The three-dimensional contour L is obtained by stacking each frame in sequence V The three-dimensional contour L is obtained by stacking each frame in sequence Out ; Insert m virtual frames in the middle of each frame, and use the real branch contours L of the preceding and following frames. Out The branch contours L of each virtual frame are synthesized by combining different influence factors. Vir After inserting the virtual frame branch outline, the total branch outer outline is obtained as L. New =L V +∑L Vir .

4. The method for calculating cross-sectional area and included angle of three-dimensional blood vessel branch according to claim 1, characterized in that: In the S2 step, the initial normal vector of the branch cross section is determined, which comprises: According to the three-dimensional modeling based on the sequence segmentation results, the body center coordinates of the three-dimensional regions V1 and V2 of the main blood vessel and the branch blood vessel are calculated, denoted as M1 and M2; Vector according to M1 and M2 Initial normal vector N to branch cut surface ori Branch initial cut surface S from M1 Ori ​ 5. The method for calculating cross-sectional area and included angle of three-dimensional blood vessel branch according to claim 4, characterized in that: In the S3 step, the branch cross section is translated to the branch connection, and the position of the branch cross section at the branch connection and the intersection section of the branch cross section and the branch outer profile are determined, which comprises: Translate the initial section S Ori along the normal vector direction to intersect the branch connection contour L C , the translated section is denoted as S F ; section S F In three-dimensional space with the branch outer contour L New Each frame intersection, all intersection points are arranged in anticlockwise order to obtain all branch cross-sectional profile of three-dimensional branch cross-sectional profile point set C 3d ={c1, c2,... c h}, wherein h is the number of three-dimensional branch cross-sectional profile point set C 3d .

6. The method for calculating cross-sectional area and included angle of three-dimensional blood vessel branch according to claim 5, characterized in that: After obtaining a plurality of branch cross section profiles, the effective branch cross section profiles are obtained by screening the plurality of branch cross section profiles according to the branch cross section screening conditions; The branch cross section screening conditions comprise: section S F with the branch outer contour L New There are two intersection points at the last frame position, and then a consecutive a frames with the branch outer contour L New There are two intersection points, a≥2, the last frame of the consecutive a frames has only one intersection point; If the a-frame is not the farthest frame of the branch segment, then the outer contour L New All frames after the a+1-frame do not intersect the surface S F .

7. The method of claim 1, wherein: In the S4 step, the three-dimensional branch cross section profile is rotated to obtain an equivalent two-dimensional branch cross section profile, which comprises: At any point c of the three-dimensional branch cross-sectional profile C 3d a perpendicular point c of c on K is calculated K and the angle α between c and S K is calculated, and the point c K is rotated by an angle α in the direction of the plane S K to obtain the two-dimensional equivalent point c′ of the point; The three-dimensional branch cross-sectional profile C 3d All the points above are rotated to the plane S K Above, resulting in an equivalent two-dimensional profile of the three-dimensional branch cross-sectional profile, denoted C 2d .

8. The method for calculating cross-sectional area and included angle of three-dimensional blood vessel branch according to claim 1, characterized in that: In the S4 step, the area of the two-dimensional branch cross section profile is calculated, which comprises: The straight line parts at both ends of the starting frame and the ending frame are replaced by arc profiles to obtain a modified two-dimensional profile C Arc And the area of the simply connected domain formed is the required branch cross-sectional profile area s Arc Now Wherein the arc profile height h is linearly or nonlinearly related to the actual physical height H between the frames and the current cross-sectional angle.​ 9. The method of claim 1-8, wherein: The rotation mode of rotating the initial normal vector of the branch cross section in the S5 step comprises proportionally changing the angle for rotation, or changing the angle according to the mode of sparse first and dense later; the intravascular imaging data comprises intravascular optical coherence tomography data and intravascular ultrasound data.

10. A three-dimensional system for calculating the cross-sectional area and included angle of blood vessel branches, characterized in that: It comprises an image acquisition module, a blood vessel branch calculation module, a subsequent processing module and a display module; The image acquisition module is used to acquire the OCT image or the IVUS image of the target blood vessel, or simultaneously acquire the OCT and IVUS images; The blood vessel branch calculation module is used for automatically detecting the profiles of the main blood vessel and the branch blood vessel according to the intravascular imaging data, determining the connection points of the main blood vessel and the branch blood vessel, determining the three-dimensional branch outer profile, and determining the initial normal vector of the branch cross section; the branch cross section is translated to the branch connection, and the position of the branch cross section at the branch connection and the intersection section of the branch cross section and the branch outer profile are determined; the three-dimensional branch cross section profile is rotated to obtain an equivalent two-dimensional branch cross section profile, and the area of the two-dimensional branch cross section profile is calculated; the normal vector of the branch cross section is rotated to obtain a series of branch cross section areas and the normal vectors corresponding to the branch cross section areas; the minimum branch cross section area s Min is selected from all the branch cross sections as the cross section area of the segment branch blood vessel, and the included angle β between the normal vector corresponding to the minimum branch cross section area and the z axis is used to determine the included angle between the branch blood vessel and the main blood vessel. The subsequent processing module is used to perform corresponding subsequent processing on the calculated branch area and the included angle information; The display module is used to display the target blood vessel branch detection and calculation results through various dimensional display modes.

Images