Method for extracting blood vessel centerline, electronic device and storage medium

Abstract

The application discloses a method for extracting a blood vessel center line, an electronic device and a storage medium. The method comprises the following steps: based on blood vessel image data, an initial center line of a blood vessel is extracted by performing blood vessel segmentation on the blood vessel image data using an initial segmentation model; nodes in the initial center line are taken as graph nodes, and connection edges between the graph nodes are constructed to form a graph structure of the blood vessel; the graph nodes and the connection edges in the graph structure are optimized using an optimization model; and during the optimization process, the optimization model is adjusted based on a multi-objective loss function set according to the graph structure to output a final blood vessel center line. By using the scheme, the accuracy of the center line extraction under a complex blood vessel structure can be improved, and reliable analysis can be provided for medical image blood vessels.

Description

Technical Field

[0001] This application generally relates to the field of medical image processing technology. More specifically, this application relates to a method for extracting the centerline of a blood vessel, an electronic device, and a computer-readable storage medium. Background Technology

[0002] Centerline extraction is a core component of vascular analysis in medical imaging, and its accuracy directly impacts subsequent clinical applications such as vascular anatomy reconstruction, vessel diameter measurement, and lesion localization. In multimodal medical imaging, including computed tomography (CT), magnetic resonance imaging (MRI), and digital subtraction angiography (DSA), vascular structures often exhibit complex branching morphologies, tortuous courses, and localized deformations, and may also be accompanied by image noise and tissue overlap. Therefore, efficiently and accurately extracting the centerline of blood vessels has become a critical technical challenge.

[0003] Existing methods for extracting the centerline of three-dimensional blood vessels mainly rely on two technical approaches: one is an end-to-end segmentation network based on traditional image processing algorithms or convolutional neural networks, which first segments the blood vessel region and then extracts the centerline through skeleton extraction; the other is the deformable modeling and graph neural network technology that has emerged in recent years, attempting to adapt the dynamic model to the morphology of blood vessels. However, the former is prone to problems such as blurred segmentation boundaries, broken or offset centerlines when dealing with complex blood vessel branches and small blood vessels, and is not sensitive enough to small blood vessels; the latter, although it introduces the idea of ​​dynamic modeling, mostly fails to effectively integrate the geometric priors of blood vessels (such as connectivity and smoothness), and has not established a collaborative optimization mechanism for segmentation and centerline extraction, resulting in insufficient geometric rationality of the centerline, making it difficult to meet the stringent accuracy requirements of clinical practice.

[0004] Therefore, there is an urgent need to provide a scheme for extracting the centerline of blood vessels in order to improve the accuracy of centerline extraction in complex vascular structures and provide reliable analysis for blood vessels in medical imaging. Summary of the Invention

[0005] In order to at least solve one or more of the technical problems mentioned above, this application proposes a scheme for extracting the centerline of blood vessels in several aspects.

[0006] In a first aspect, this application provides a method for extracting the centerline of a blood vessel, comprising: extracting the initial centerline of the blood vessel by segmenting the blood vessel using an initial segmentation model based on blood vessel image data; using nodal points in the initial centerline as graph nodes and constructing connecting edges between the graph nodes to form a graph structure of the blood vessel; optimizing the graph nodes and connecting edges in the graph structure using an optimization model; and adjusting the optimization model in conjunction with a multi-objective loss function set based on the graph structure during the optimization process to output the final centerline of the blood vessel.

[0007] In some embodiments, the method further includes: performing image normalization and / or data augmentation operations on the vascular imaging data to preprocess the vascular imaging data.

[0008] In some embodiments, optimizing the graph nodes and connecting edges in the graph structure using an optimization model includes: extracting target features in the graph structure using the optimization model, wherein the target features include at least the position information, connection relationship, and diameter information of the graph nodes; adding new graph nodes and connecting edges or deleting graph nodes and connecting edges according to the target features, so as to optimize the graph nodes and connecting edges in the graph structure.

[0009] In some embodiments, the multi-objective loss function includes a first loss function, a second loss function, and a third loss function, and is set by the following operations: calculating the mask information of blood vessels based on the target features in the graph structure to form the first loss function; calculating the edge consistency loss based on the target features in the graph structure to form the second loss function; and calculating the geometric smoothing regularization term based on the target features in the graph structure to form the third loss function.

[0010] In some embodiments, adjusting the optimization model in conjunction with a multi-objective loss function set based on the graph structure during the optimization process to output the final vascular centerline includes: calculating a weighted total loss function of the first loss function, the second loss function, and the third loss function set based on the graph structure; and adjusting the optimization model in conjunction with the weighted total loss function during the optimization process to output the final vascular centerline.

[0011] In some embodiments, the first loss function is represented by the following formula: ,in, This represents the vascular mask generated based on the graph structure. This represents the actual blood vessel segmentation results. This indicates the smoothing term.

[0012] In some embodiments, the second loss function is represented by the following formula: ,in, and Represents the coordinates of adjacent graph nodes i and j. This represents the Euclidean distance between the i-th and j-th adjacent nodes. Indicates the preset connection side length. Represents the set of all connecting edges in a graph structure. Let represent an edge connecting node i and node j. M This represents the total number of edges in the graph structure.

[0013] In some embodiments, the third loss function is represented by the following formula: ,in, , , Let i represent the coordinates of adjacent graph nodes i-1, i, and i+1. This represents taking the square of the Euclidean norm of the second-order difference vector.

[0014] In a second aspect, this application provides an electronic device, including: a processor; and a memory storing computer instructions for extracting the centerline of a blood vessel, wherein when the computer instructions are executed by the processor, one or more embodiments of the first aspect described above are implemented.

[0015] In a third aspect, this application provides a computer-readable storage medium having stored thereon computer program instructions for extracting the centerline of a blood vessel, which, when executed by one or more processors, cause one or more embodiments of the first aspect to be implemented.

[0016] The scheme for extracting the vascular centerline provided above effectively solves the problems of insufficient segmentation accuracy, inaccurate centerline extraction, and poor sensitivity to small blood vessels in existing technologies by using an initial segmentation model based on vascular image data, constructing a graph structure to represent the vascular topology with the nodes of the centerline, optimizing the model to dynamically adjust the graph nodes and connecting edges, and jointly optimizing the model parameters with a multi-objective loss function. This significantly improves the geometric rationality and clinical usability of the centerline and provides accurate support for the vascular anatomy analysis of multimodal medical images such as CT, MRI, and DSA. Attached Figure Description

[0017] The above and other objects, features, and advantages of exemplary embodiments of this application will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of this application are illustrated by way of example and not limitation, and the same or corresponding reference numerals denote the same or corresponding parts, wherein:

[0018] Figure 1This is an exemplary flowchart illustrating a method 100 for extracting the centerline of a blood vessel according to an embodiment of this application;

[0019] Figure 2 This is an exemplary schematic diagram illustrating blood vessel segmentation according to an embodiment of this application;

[0020] Figure 3 This is an exemplary schematic diagram illustrating the construction of a vascular map structure according to an embodiment of this application;

[0021] Figure 4 This is an exemplary schematic diagram illustrating the acquisition of the final vascular centerline according to an embodiment of this application;

[0022] Figure 5 An exemplary structural block diagram of an electronic device 500 according to an embodiment of this application is shown. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] It should be understood that the terms "comprising" and "including" used in the specification and claims of this application indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0025] It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this specification and claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this specification and claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations.

[0026] As used in this specification and claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0027] The specific embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0028] Figure 1 This is an exemplary flowchart illustrating a method 100 for extracting the centerline of a blood vessel according to an embodiment of this application. Figure 1 As shown, the method 100 may include: step S101: extracting the initial centerline of the blood vessels by segmenting the blood vessels using an initial segmentation model based on the blood vessel image data; step S102: using the nodes in the initial centerline as graph nodes and constructing connecting edges between the graph nodes to form a graph structure of the blood vessels; step S103: optimizing the graph nodes and connecting edges in the graph structure using an optimization model; and step S104: adjusting the optimization model in conjunction with a multi-objective loss function set based on the graph structure during the optimization process to output the final blood vessel centerline.

[0029] First, in step S101, based on the vascular imaging data, an initial segmentation model is used to segment the blood vessels to extract the initial centerline of the vessels. In some implementation scenarios, the vascular imaging data format may include Digital Imaging and Communications in Medicine (DICOM) and the Neuroimaging Informatics Technology Initiative (NIFTI), which may include CT images, MRI images, and DSA images, etc. In some embodiments, the initial segmentation model may be, for example, a U-Net model or a three-dimensional U-Net model.

[0030] In the implementation scenario, vascular image data is first input into a trained initial segmentation model to segment the blood vessels, obtaining a coarse segmentation mask (such as a segmentation mask for coronary arteries in CT images or cerebral blood vessels in MRI images). Next, a skeleton extraction algorithm iteratively removes redundant pixels from the edges of the vascular mask, retaining the central skeleton to extract the initial centerline of the blood vessels. Specifically, the coarse segmentation mask (binary image) is denoted as M, and the skeleton image S is initialized as a matrix of all zeros. Each pixel in M ​​is iteratively traversed. If a pixel has a value of 1 (representing a vascular region), its neighborhood is not an isolated point or endpoint, its "intersection number" (reflecting local connectivity) in the three axes is 1 (ensuring that deletion does not disrupt vascular connectivity), and it is not an endpoint, then the corresponding pixel is deleted. The pixels in M ​​that are not deleted form the central skeleton of the blood vessels. The coordinates (i,j,k) of these pixels are arranged according to the order of the blood vessel course to form the initial centerline.

[0031] In some embodiments, before inputting the vascular image data into the initial segmentation model, image normalization and / or data augmentation operations can be performed on the vascular image data to preprocess the data, and then the preprocessed vascular image data can be input into the initial segmentation model. In other embodiments, data augmentation operations may include, but are not limited to, one or more of cropping, scaling, rotation, grayscale transformation, or flipping.

[0032] Next, in step S102: using the nodes in the initial centerline as graph nodes, and constructing connecting edges between the graph nodes to form the graph structure of the blood vessels. Specifically, in some implementation scenarios, each node (pixel) in the initial centerline is used as a graph node, with each node uniquely corresponding to a spatial position on the initial centerline. Each node contains position information, connectivity, and diameter information. Then, potential neighboring nodes are filtered based on spatial proximity (e.g., Euclidean distance filtering, K-nearest neighbor filtering, radius connection filtering), selecting nodes whose Euclidean distance is less than a distance threshold; or the K nodes with the smallest Euclidean distance; or all nodes within the target radius of the sphere as potential neighboring nodes.

[0033] Furthermore, by fitting the local course of blood vessels, the angle between the node connection line and the course direction is calculated. Adjacent nodes that satisfy the preset angle are connected to form connecting edges, and the edge length and direction are recorded. Based on this, the pixel-level information of the initial centerline is transformed into the topological information of the graph structure, realizing the structured representation of the spatial relationship of blood vessels. For example, the branch nodes of cerebral blood vessels can be accurately represented by a graph structure of "one central node connecting multiple child nodes".

[0034] Based on the graph structure that forms blood vessels, in step S103, the graph nodes and connecting edges in the graph structure are optimized using an optimization model. In step S104, the optimization model is adjusted in conjunction with a multi-objective loss function set based on the graph structure during the optimization process to output the final blood vessel centerline.

[0035] In some embodiments, an optimization model can be used to extract target features from the graph structure, wherein the target features include at least the position information, connectivity, and diameter information of the graph nodes; new graph nodes and connecting edges are added or deleted based on the target features to optimize the graph nodes and connecting edges in the graph structure. In some implementation scenarios, the aforementioned optimization model can be, for example, a graph convolutional network (“GCN”) or a label propagation graph convolutional network (“TAGConv”).

[0036] In the implementation scenario, the graph structure is input into the optimization model. The optimization model aggregates information from neighboring nodes to update the features of each node, such as its position, connectivity, and diameter. This feature update can capture local geometric and topological information, thereby better representing the shape and structure of the centerline.

[0037] During the optimization process, the model can insert new nodes or delete redundant nodes as needed to improve the accuracy and simplicity of the centerline. Inserting new nodes refines the centerline, while deleting redundant nodes simplifies the structure. In each optimization cycle, the configuration of nodes and edges can be adaptively adjusted using feedback from a multi-objective loss function to adapt to different blood vessel morphologies and image modalities.

[0038] In some embodiments, the multi-objective loss function may include a first loss function, a second loss function, and a third loss function, and is set by the following operations: calculating the mask information of blood vessels based on the target features in the graph structure to form the first loss function; calculating the edge consistency loss based on the target features in the graph structure to form the second loss function; and calculating the geometric smoothing regularization term based on the target features in the graph structure to form the third loss function. The first loss function ensures that the optimized graph structure accurately reflects the vascular region by generating an approximate vascular mask from the graph nodes and comparing it with the actual vascular segmentation results to constrain the segmentation accuracy. The second loss function ensures the correct connection relationship between adjacent nodes in the centerline graph, avoiding breaks (such as excessively large node spacing in small vascular regions) or excessive density (such as redundant nodes). The third loss function ensures that the centerline maintains smooth curvature, avoiding jagged deformations (such as the broken-line centerline caused by MRI artifacts).

[0039] Specifically, in some embodiments, a weighted total loss function based on a first loss function, a second loss function, and a third loss function set according to the graph structure can be calculated. During the optimization process, the optimization model is adjusted using the weighted total loss function to output the final blood vessel centerline. Specifically, in one implementation scenario, the aforementioned weighted total loss function... It can be expressed by the following formula:

[0040]

[0041] in, Denotes the first loss function. This represents the second loss function. This represents the third loss function. , and These are weighted parameters.

[0042] In some embodiments, the first loss function described above is expressed by the following formula:

[0043]

[0044] in, This represents the vascular mask generated based on the graph structure. This represents the actual blood vessel segmentation results. This represents a smoothing term, used to prevent the denominator from being zero and to improve the stability of numerical calculations.

[0045] In some embodiments, the second loss function described above is expressed by the following formula:

[0046]

[0047] in, and Represents the coordinates of adjacent graph nodes i and j. This represents the Euclidean distance between the i-th and j-th adjacent nodes. Indicates the preset connection side length. Represents the set of all connecting edges in a graph structure. Let represent an edge connecting node i and node j. M This represents the total number of edges in the graph structure.

[0048] In some embodiments, the third loss function described above is expressed by the following formula:

[0049]

[0050] in, , , Let i represent the coordinates of adjacent graph nodes i-1, i, and i+1. This represents the summation of the squared Euclidean norm of the second-order difference vector across all nodes, used to constrain the overall geometric smoothness of the centerline.

[0051] As described above, the embodiments of this application construct a graph structure from the initial centerline extracted by the initial segmentation model. An optimization model combined with a multi-objective loss function is used to iteratively optimize the graph nodes and connecting edges. This significantly improves the accuracy of vascular centerline extraction in complex vascular scenarios (such as dense branching, vascular deformation, and multimodal image differences), effectively solving problems such as insufficient segmentation accuracy, centerline offset or breakage, and missed detection of small vessels in existing technologies. Through the synergistic effect of segmentation accuracy loss, connectivity consistency loss, and geometric smoothness loss in the multi-objective loss function, a high degree of matching between the centerline and the real vascular region is ensured, while also giving the centerline continuous, unbroken, and geometrically smooth characteristics that conform to physiological structures, greatly improving the ability to identify small vessels and branch nodes.

[0052] Furthermore, this application embodiment also enhances the model's adaptability to multimodal medical images such as CT, MRI, and DSA, as well as blood vessels in different locations, through image normalization and enhancement operations during data preprocessing. This eliminates the need for retraining for a single scenario, significantly improving generalization ability. The final output of a precise centerline can directly provide reliable support for clinical diagnostic applications such as blood vessel diameter measurement, stenosis assessment, and interventional surgery planning, fully meeting the high-precision requirements of clinical vascular anatomy analysis.

[0053] Figure 2 This is an exemplary schematic diagram illustrating blood vessel segmentation according to an embodiment of this application. Figure 2 As shown, vascular image data 201 is input into the initial segmentation model 202 for vascular segmentation to obtain a coarse vascular segmentation mask 203. In some embodiments, vascular image data 201 may be, for example, CT images, MRI images, and DSA images, and the initial segmentation model 202 may be, for example, a three-dimensional U-Net model. In other embodiments, before inputting the vascular image data into the initial segmentation model, preprocessing operations such as image normalization and / or data augmentation may be performed on the vascular image data.

[0054] Figure 3 This is an exemplary schematic diagram illustrating the construction of a vascular map structure according to an embodiment of this application. Figure 3 As shown, based on the blood vessel coarse segmentation mask 203, the initial centerline 301 of the blood vessel is first extracted. As mentioned earlier, the initial centerline 301 of the blood vessel can be extracted by iteratively deleting redundant pixels at the edges of the blood vessel mask and retaining the central skeleton through a skeleton extraction algorithm. Next, the nodes in the initial centerline 301 are used as graph nodes, and connecting edges are constructed between the graph nodes to form the graph structure 302 of the blood vessel. For more details on forming the graph structure of the blood vessel, please refer to [reference needed]. Figure 1 The descriptions in the document are not repeated here.

[0055] Figure 4 This is an exemplary schematic diagram illustrating the acquisition of the final vascular centerline according to an embodiment of this application. Figure 4 As shown, based on the graph structure 302 forming blood vessels, it is input into the optimization model 401, and simultaneously, a multi-objective loss function is used to adjust the optimization model, optimizing the graph nodes and connecting edges in the graph structure 302 (including, for example, inserting new nodes or deleting redundant nodes), to output the final blood vessel centerline 402. For more details on the aforementioned optimization process and the multi-objective loss function, please refer to the aforementioned... Figure 1 The description of the subject matter will not be repeated here.

[0056] Figure 5An exemplary structural block diagram of an electronic device 500 according to an embodiment of this application is shown. It will be understood that the device implementing the solution of this application may be a single device (e.g., a computing device) or a multifunctional device including various peripheral devices.

[0057] like Figure 5 As shown, the electronic device of this application may include a central processing unit (“CPU”) 511, which may be a general-purpose CPU, a dedicated CPU, or other information processing and program execution unit. Furthermore, the electronic device 500 may also include a mass storage device 512 and a read-only memory (“ROM”) 513. The mass storage device 512 may be configured to store various types of data, including various vascular imaging data, initial centerlines of blood vessels, graphical structures, multi-objective loss functions, final vascular centerlines, algorithm data, intermediate results, and various programs required to run the electronic device 500. The ROM 513 may be configured to store power-on self-tests of the electronic device 500, initialization of various functional modules in the system, drivers for the system's basic input / output, and data and instructions required to boot the operating system.

[0058] Optionally, the electronic device 500 may also include other hardware platforms or components, such as the tensor processing unit (“TPU”) 514, graphics processing unit (“GPU”) 515, field-programmable gate array (“FPGA”) 516, and machine learning unit (“MLU”) 517 shown. It is understood that although various hardware platforms or components are shown in the electronic device 500, they are merely exemplary and not limiting, and those skilled in the art can add or remove appropriate hardware as needed. For example, the electronic device 500 may implement the method for extracting the vascular centerline of this application by including only a CPU, associated storage devices, and interface devices.

[0059] In some embodiments, to facilitate data transmission and interaction with external networks, the electronic device 500 of this application further includes a communication interface 518, through which it can connect to a local area network / wireless local area network (“LAN / WLAN”) 505, and further through the LAN / WLAN to connect to a local server 506 or to the Internet (“Internet”) 507. Alternatively or additionally, the electronic device 500 of this application can also directly connect to the Internet or cellular network via the communication interface 518 based on wireless communication technology, such as wireless communication technology based on 3G (“3G”), 4G (“4G”), or 5G (“5G”). In some application scenarios, the electronic device 500 of this application can also access the server 508 and database 509 of an external network as needed to obtain various known algorithms, data, and modules, and can remotely store various data, such as various data or instructions for presenting vascular imaging data, initial centerlines of blood vessels, graph structures, multi-objective loss functions, final centerlines of blood vessels, etc.

[0060] Peripherals of the electronic device 500 may include a display device 502, an input device 503, and a data transmission interface 504. In one embodiment, the display device 502 may include, for example, one or more speakers and / or one or more visual displays, configured to provide voice prompts and / or display images and videos for the extraction of vascular centerlines according to this application. The input device 503 may include, for example, a keyboard, mouse, microphone, posture capture camera, and other input buttons or controls, configured to receive audio data input and / or user commands. The data transmission interface 504 may include, for example, a serial interface, parallel interface, or Universal Serial Bus interface (“USB”), Small Computer System Interface (“SCSI”), Serial ATA, FireWire (“FireWire”), PCI Express, and High Definition Multimedia Interface (“HDMI”), configured for data transmission and interaction with other devices or systems. According to the scheme of this application, the data transmission interface 504 can receive vascular imaging data and transmit vascular imaging data or various other types of data or results to the electronic device 500.

[0061] The aforementioned CPU 511, mass storage 512, ROM 513, TPU 514, GPU 515, FPGA 516, MLU 517, and communication interface 518 of the electronic device 500 of this application can be interconnected via bus 519, and can interact with peripheral devices through this bus. In one embodiment, the CPU 511 can control other hardware components and peripheral devices in the electronic device 500 through bus 519.

[0062] The above combination Figure 5An electronic device for extracting the centerline of a blood vessel, which can be used to perform the present application, is described. It should be understood that the device structure or architecture described herein is merely exemplary, and the implementation method and implementation entity of the present application are not limited thereto, but can be modified without departing from the spirit of the present application.

[0063] Based on the foregoing description in conjunction with the accompanying drawings, those skilled in the art will understand that the embodiments of this application can also be implemented by software programs. Therefore, this application also provides a computer-readable storage medium storing computer-readable instructions for extracting the centerline of a blood vessel. When executed by one or more processors, these computer-readable instructions can be used to implement the embodiments of this application in conjunction with the accompanying drawings. Figure 1 The method described is for extracting the centerline of blood vessels.

[0064] It should be noted that although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart can be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0065] It should be understood that when the terms "first," "second," "third," and "fourth," etc., are used in the claims, specification, and drawings of this application, they are used only to distinguish different objects and not to describe a specific order. The terms "comprising" and "including" as used in the specification and claims of this application indicate the presence of the described features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof.

[0066] It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this specification and claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this specification and claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations.

[0067] While numerous embodiments of this application have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will arise for those skilled in the art without departing from the spirit and intent of this application. It should be understood that various alternatives to the embodiments of this application described herein may be employed in the practice of this application. The appended claims are intended to define the scope of protection of this application and therefore cover equivalents or alternatives within the scope of these claims.

Claims

1. A method for extracting the centerline of a blood vessel, comprising: Based on vascular imaging data, an initial segmentation model is used to segment blood vessels and extract the initial centerline of the blood vessels. Using the nodes in the initial centerline as graph nodes, and constructing connecting edges between the graph nodes, a graph structure of blood vessels is formed; The graph nodes and connecting edges in the graph structure are optimized using an optimization model; During the optimization process, the optimization model is adjusted in conjunction with a multi-objective loss function set based on the graph structure to output the final blood vessel centerline; The multi-objective loss function includes a first loss function, a second loss function, and a third loss function, and is set by the following operations: The mask information of blood vessels is calculated based on the target features in the graph structure to form the first loss function; The consistency loss of the connection edges is calculated based on the target features in the graph structure to form the second loss function; The geometric smoothing regularization term is calculated based on the target features in the graph structure to form the third loss function; The optimization process, which involves adjusting the optimization model using a multi-objective loss function based on the graph structure to output the final vessel centerline, includes: Calculate the weighted total loss function based on the first loss function, the second loss function, and the third loss function set according to the graph structure; During the optimization process, the optimization model is adjusted in conjunction with the weighted total loss function to output the final blood vessel centerline.

2. The method according to claim 1, further comprising: Image normalization and / or data augmentation operations are performed on the vascular imaging data to preprocess the vascular imaging data.

3. The method according to claim 1 or 2, wherein optimizing the graph nodes and connecting edges in the graph structure using an optimization model includes: The target features in the graph structure are extracted using an optimization model, wherein the target features include at least the location information, connection relationship and diameter information of the graph nodes; Based on the target features, new graph nodes and connecting edges are added or deleted to optimize the graph nodes and connecting edges in the graph structure.

4. The method of claim 1, wherein the first loss function is expressed by the following formula: in, This represents the vascular mask generated based on the graph structure. This represents the actual blood vessel segmentation results. This indicates the smoothing term.

5. The method of claim 1, wherein the second loss function is expressed by the following formula: in, and Represents the coordinates of adjacent graph nodes i and j. This represents the Euclidean distance between the i-th and j-th adjacent nodes. Indicates the preset connection side length. Represents the set of all connecting edges in a graph structure. This represents an edge connecting node i and node j. M This represents the total number of edges in the graph structure.

6. The method of claim 1, wherein the third loss function is expressed by the following formula: in, , , Let i represent the coordinates of adjacent graph nodes i-1, i, and i+1. This represents taking the square of the Euclidean norm of the second-order difference vector.

7. An electronic device, comprising: processor; as well as A memory having stored computer instructions for extracting the centerline of a blood vessel, which, when executed by a processor, cause the method according to any one of claims 1-6 to be implemented.

8. A computer-readable storage medium having stored thereon computer program instructions for extracting the centerline of a blood vessel, the computer program instructions, when executed by one or more processors, causing the method according to any one of claims 1-6 to be implemented.

Images