Artificial intelligence assisted pulmonary blood vessel image omics acquisition method and device, medium and product

Abstract

The invention discloses an artificial intelligence-assisted pulmonary blood vessel image omics acquisition method and device, a medium and a product, and relates to the technical field of medical image processing, and the method comprises the steps: firstly, based on an input original file, in response to the operation of a user, reconstructing a lung model and a pulmonary blood vessel model; based on the pulmonary blood vessel model, generating a first simplified model of each pulmonary artery model blood vessel branch and a second simplified model of each pulmonary vein model blood vessel branch in response to operation of a user; and finally, based on the lung model, the pulmonary vessel model, the first simplified model and the second simplified model, generating peripheral vessel distribution parameters of pulmonary artery and pulmonary vein and radiomics characteristic parameters of the peripheral vessel distribution parameters, generating radiomics characteristic parameters of the whole vessel, calculating a vessel center line, and grading the vessel. The technical problem of how to improve the yield of information acquired by pulmonary blood vessel imaging omics is solved.

Description

Technical Field

[0001] This application relates to the technical field of medical image processing, and particularly to an artificial intelligence-assisted pulmonary vascular radiomics acquisition method, device, medium, and product. Background Art

[0002] With the rapid development of medical imaging technology, computed tomography angiography (CTA) has become an important means for diagnosing pulmonary vascular diseases. The pulmonary vascular image data obtained through CTA examinations contains rich anatomical and pathological information, which can visually display the morphological characteristics of the pulmonary artery, pulmonary vein, and their branch vessels, and has important guiding significance for the diagnosis, classification, and treatment plan formulation of pulmonary vascular diseases such as pulmonary hypertension and pulmonary embolism. The quantitative analysis of pulmonary vascular images also plays an increasingly important role in the early screening, condition assessment, and prognosis prediction of diseases.

[0003] Currently, the imaging diagnosis of pulmonary vessels mainly relies on simple reconstruction after the hospital radiology department obtains the original data through CT examinations. The specific implementation process is as follows: First, the original DICOM data obtained from the patient's CTA examination is imported into the 3DSlicer software. The target area is outlined on the coronal plane, sagittal plane, and transverse plane of the software interface, and the Grow From Seed module is called to perform 3D model reconstruction on the outlined target area using the improved GrowCut algorithm, thereby dividing the lung model and the pulmonary vascular model. After completing the model reconstruction, switch to the Radiomics module to extract various radiomics parameters in units of segments, including basic radiomics features such as gray values and texture features, for subsequent quantitative analysis.

[0004] The above technical solution has the problem of low information productivity. Since this solution can only obtain the basic parameters of the overall pulmonary vascular model through the Radiomics module and cannot distinguish between the pulmonary artery, pulmonary vein, and their corresponding branch vessels, the obtained radiomics parameters are relatively limited. This results in a small amount of data obtained, making it difficult to meet the needs of clinical diagnosis for refined and personalized analysis. At the same time, due to the lack of independent analysis of the pulmonary artery, pulmonary vein, and their branches, the topological structure characteristics of the vascular network cannot be obtained, making the analysis of the pulmonary vascular system stay at a relatively superficial level, which is not conducive to clinical transformation and application, and it is also difficult to provide sufficient data support for the early diagnosis and precise treatment of diseases. Summary of the Invention

[0005] The main purpose of this application is to provide an artificial intelligence-assisted pulmonary vascular radiomics acquisition method, device, medium, and product, aiming to solve the technical problem of how to improve the information productivity of pulmonary vascular radiomics acquisition.

[0006] To achieve the above object, the present application proposes an artificial intelligence-assisted pulmonary vascular radiomics acquisition method, and the artificial intelligence-assisted pulmonary vascular radiomics acquisition method includes:

[0007] Based on the input original file, in response to the user's operation, reconstruct the lung model and the pulmonary vascular model;

[0008] Based on the pulmonary vascular model, in response to the user's operation, generate a first simplified model of each pulmonary artery model vascular branch and a second simplified model of each pulmonary vein model vascular branch;

[0009] Based on the lung model, the pulmonary vascular model, the first simplified model and the second simplified model, generate the peripheral vascular distribution parameters and their radiomics characteristic parameters of the pulmonary artery and the pulmonary vein, generate the radiomics characteristic parameters of the overall blood vessels, and calculate the blood vessel centerline and grade the blood vessels.

[0010] In one embodiment, the step of reconstructing the lung model and the pulmonary vascular model based on the input original file in response to the user's operation specifically includes:

[0011] Based on the user's operation of selecting 1 layer in the coronal plane, sagittal plane and transverse plane in the original file to draw a target area for the lung and the background, reconstruct the lung model;

[0012] Based on the user's operation of using a painting tool to draw on the main pulmonary artery and the left atrial section in the transverse plane of the original file and defining the pulmonary region, complete the reconstruction of the pulmonary vascular model.

[0013] In one embodiment, the step of generating the peripheral vascular distribution parameters and their radiomics characteristic parameters of the pulmonary artery and the pulmonary vein, generating the radiomics characteristic parameters of the overall blood vessels, and calculating the blood vessel centerline and grading the blood vessels based on the lung model, the pulmonary vascular model, the first simplified model and the second simplified model specifically includes:

[0014] Based on the first simplified model and the second simplified model, calculate the end points of the pulmonary artery model vascular branches and the end points of the pulmonary vein model vascular branches;

[0015] Based on the end points of the branches of the pulmonary artery model and the branches of the pulmonary vein model, generate the peripheral vascular distribution parameters of the pulmonary artery and the pulmonary vein;

[0016] Based on each of the first simplified models, each of the second simplifications and the pulmonary vascular model, generate the radiomics characteristic parameters of the pulmonary artery, the pulmonary vein and the pulmonary blood vessels;

[0017] Based on the end points of the pulmonary artery model vascular branches and the end points of the pulmonary vein model vascular branches, calculate the blood vessel centerline and blood vessel grading.

[0018] In one embodiment, the step of calculating the end points of the blood vessel branches of the pulmonary artery model and the end points of the blood vessel branches of the pulmonary vein model based on the first simplified model and the second simplified model specifically includes:

[0019] Obtain each of the first simplified models and each of the second simplified models;

[0020] Call to detect each of the first simplified models and each of the second simplified models to obtain the first end points of the blood vessel branches of the pulmonary artery model and the second end points of the blood vessel branches of the pulmonary vein model;

[0021] When detecting one of the first end points or the second end points, call again to detect the next first end point or the second end point.

[0022] In one embodiment, the step of generating the peripheral blood vessel distribution parameters of the pulmonary artery and the pulmonary vein based on the end points of the branches of the pulmonary artery model and the branches of the pulmonary vein model specifically includes:

[0023] Extract the background region from the segmentation nodes of the pulmonary artery model and the pulmonary vein model and convert it into a three-dimensional background region model, and remove the background region data in the segmentation nodes of the pulmonary artery model and the pulmonary vein model;

[0024] Calculate the shortest distances from the end points of each blood vessel branch of the pulmonary artery model and the end points of each blood vessel branch of the pulmonary vein model to the three-dimensional background region model;

[0025] Merge the shortest distances to generate the peripheral blood vessel distribution parameters of the pulmonary artery and the pulmonary vein.

[0026] In one embodiment, the step of generating the radiomics feature parameters of the pulmonary artery, the pulmonary vein and the pulmonary blood vessels based on each of the first simplified models, each of the second simplifications and the pulmonary blood vessel model specifically includes:

[0027] Create a first segmentation node and a second segmentation node;

[0028] Copy each of the first simplified models and each of the second simplified models into the first segmentation node, and copy the lung model and the pulmonary blood vessel model into the second segmentation node;

[0029] Select the parameters to be extracted, and based on the extracted parameters, perform feature extraction on the first segmentation node and the second segmentation node to generate the radiomics feature parameters of the pulmonary artery, the pulmonary vein and the pulmonary blood vessels.

[0030] In one embodiment, the steps of calculating the vessel centerline and vessel grading based on the endpoints of the pulmonary artery model vessel branches and the endpoints of the pulmonary vein model vessel branches specifically include:

[0031] Extract the vessel centerline based on the endpoints of the pulmonary artery simplified model vessel branches and the endpoints of the pulmonary vein simplified model vessel branches;

[0032] Convert each of the vessel centerlines into corresponding curve nodes and store them in the vessel network attribute table;

[0033] Create a grading column in the vessel network attribute table and initialize the grading value;

[0034] Calculate the intersection points of each of the curve nodes;

[0035] Based on the curve node number of the starting point determined by the user input and each of the intersection points, construct a vessel branch hierarchy tree, determine the hierarchy of each curve node, and further determine each vessel grading.

[0036] In addition, to achieve the above object, the present application also proposes an artificial intelligence-assisted pulmonary vascular imaging omics acquisition device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program is configured to implement the steps of the artificial intelligence-assisted pulmonary vascular imaging omics acquisition method as described above.

[0037] In addition, to achieve the above object, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium, and when the computer program is executed by a processor, it implements the steps of the artificial intelligence-assisted pulmonary vascular imaging omics acquisition method as described above.

[0038] In addition, to achieve the above object, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the artificial intelligence-assisted pulmonary vascular imaging omics acquisition method as described above.

[0039] One or more technical solutions proposed by the present application have at least the following technical effects:

[0040] The pulmonary vascular imaging genomics acquisition method provided by the present invention first imports the original file of the patient's pulmonary artery CTA by the user in the 3D Slicer software. By delineating the target area in the coronal plane, sagittal plane and transverse plane, the Grow FromSeed module is called to complete the 3D reconstruction of the lungs and pulmonary vessels. Subsequently, the system marks according to the main pulmonary artery section and the left atrium section in the transverse plane of the user, and reconstructs the pulmonary artery and pulmonary vein models. Then, the system simplifies these models to generate simplified models of eight main branches. Based on these models, the system automatically generates the vascular end points and calculates the distance between them and the chest wall, extracts the imaging genomics features of each vascular branch, and simultaneously extracts the imaging genomics features of the overall blood vessels, and classifies the blood vessels by analyzing the center line features.

[0041] In this application, since a special plug-in capable of automatically identifying and analyzing the pulmonary artery, pulmonary vein and their branches has been developed, and the automatic extraction of the vascular end distribution, center line features and grading information has been realized, the dimension of the available vascular features has been significantly expanded, effectively solving the technical problem that only the basic parameters of the overall vascular model can be obtained in the prior art and the information productivity is low. Furthermore, a more comprehensive and in-depth quantitative analysis of the pulmonary vascular system is realized. Not only rich feature parameters including vascular morphological features and network topological structure features are obtained, but also multi-level feature extraction from the whole to the local and from the main trunk to the branches is realized, greatly improving the information productivity of pulmonary vascular imaging genomics acquisition and providing richer and more reliable data support for clinical diagnosis and treatment decision-making. Brief Description of the Drawings

[0042] The drawings here are incorporated into the specification and form a part of this specification, showing the embodiments consistent with this application, and are used together with the specification to explain the principles of this application.

[0043] In order to more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

[0044] Figure 1 It is a schematic flowchart provided for the first embodiment of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method of this application;

[0045] Figure 2 It is a schematic flowchart provided for the second embodiment of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method of this application;

[0046] Figure 3 It is a schematic flowchart provided for the third embodiment of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method of this application;

[0047] Figure 4 Another schematic flowchart provided for the third embodiment of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method of this application;

[0048] Figure 5 Another schematic flowchart provided for the third embodiment of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method of this application;

[0049] Figure 6 Another schematic flowchart provided for the third embodiment of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method of this application;

[0050] Figure 7 Schematic diagram of the device structure of the hardware operating environment involved in the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method in the embodiments of this application.

[0051] The implementation, functional features, and advantages of the purpose of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. Specific implementation manners

[0052] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of this application and are not used to limit this application.

[0053] For a better understanding of the technical solutions of this application, the following will be described in detail in conjunction with the accompanying drawings of the specification and specific implementation manners.

[0054] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of implementing the above functions. The following takes the system carrying the improved 3DSlicer software as an example to illustrate this embodiment and the following embodiments.

[0055] Based on this, the embodiments of this application provide an artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method, with reference to Figure 1 , Figure 1 Schematic flowchart of the first embodiment of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method of this application.

[0056] Step S10, based on the input original file, reconstruct the lung model and the pulmonary vascular model in response to the user's operation;

[0057] Among them, the original file represents a DICOM (Digital Imaging and Communications in Medicine) format medical image data file obtained by the patient through CTA (Computed Tomography Angiography) examination, which is used to provide the original scan information of the pulmonary blood vessels. The lung model refers to a digital model of the complete lung organ obtained through three-dimensional reconstruction, which is used to represent the overall shape and boundary of the lungs. The pulmonary vascular model represents a digital model of the reconstructed complete pulmonary vascular system, including the main trunk and branch structures of the pulmonary artery and pulmonary vein. The user's operation refers to interactive behaviors such as target delineation and module invocation on the software interface.

[0058] This step is executed after starting the 3D Slicer software and completing the DICOM data import, and is used to obtain a basic three-dimensional reconstruction model. Specifically, the user first imports the patient's original CTA scan data through the Add DICOM Data function and selects a scanning sequence with better imaging quality. Then, target delineation is performed in the three sectional views of the software interface, and appropriate levels are selected in the coronal plane, sagittal plane, and transverse plane, and the paint tool is used to delineate the lung and background regions. Finally, the Grow From Seed module is invoked, and the three-dimensional reconstruction of the lungs and pulmonary blood vessels is completed based on the marked target area using an improved GrowCut algorithm.

[0059] Step S20: Based on the pulmonary vascular model, in response to the user's operation, generate a first simplified model of each pulmonary artery model vascular branch and a second simplified model of each pulmonary vein model vascular branch;

[0060] Among them, the pulmonary artery model vascular branch represents the arterial branch structure from the main pulmonary artery to each lung lobe, including the right upper, right lower, left upper, and left lower pulmonary artery branches. The pulmonary vein model vascular branch represents the venous branch structure that returns from each lung lobe to the left atrium, also including the right upper, right lower, left upper, and left lower pulmonary vein branches. The first simplified model and the second simplified model refer to the refined vascular models that are more suitable for subsequent analysis obtained through model simplification processing.

[0061] This step is executed after completing the reconstruction of the overall pulmonary vascular model and is used to obtain a more detailed vascular branch model. Specifically, the user first generates eight branch models through the clone function in the data module, corresponding to the four main branches of the pulmonary artery and pulmonary vein respectively. Then, the scissors tool in the Segment Editor module is used to refine each branch model, and the model is exported as a three-dimensional surface model. Finally, the Decimate function in the SurfaceToolbox is used to simplify each branch model to generate a simplified model that is convenient for analysis.

[0062] Step S30: Based on the lung model, the pulmonary vascular model, the first simplified model, and the second simplified model, generate the peripheral vascular distribution parameters and their radiomics feature parameters of the pulmonary artery and pulmonary vein, generate the radiomics feature parameters of the overall blood vessels, and calculate the blood vessel centerlines and classify the blood vessels.

[0063] Among them, the peripheral vascular distribution parameters represent the spatial position information of the blood vessel end points and their characteristics such as the distance from the chest wall. The radiomics feature parameters refer to the quantitative features extracted from medical images, including parameters in multiple dimensions such as shape, size, grayscale, and texture. The blood vessel centerline represents the skeleton line of the blood vessel's trajectory and is used to analyze the geometric features of the blood vessel. Blood vessel classification refers to the hierarchical division of the blood vessel network according to the branching levels of the blood vessels.

[0064] This step is executed after obtaining all the necessary models for comprehensive feature extraction and analysis. Specifically, the system first calls the Extract Centerline module to detect the end points of each blood vessel branch through the auto-detect function and calculate the distances between these points and the background model. Then, the Radiomics module is used to extract the radiomics features of each blood vessel branch and the overall blood vessels, including multi-dimensional features such as morphology and texture. Next, the centerline extraction operation is performed on each blood vessel branch model: by setting the source points and target points through the Extract Centerline module and using the VMTK (Vascular Modeling Toolkit) algorithm to extract the blood vessel skeleton, the centerline representing the main trunk direction of the blood vessel is obtained. Finally, based on the extracted centerlines, blood vessel network analysis is carried out. By calculating parameters such as the branch point positions, branch angles, and diameter changes of the centerlines, the hierarchical relationship of the blood vessels is determined to complete the blood vessel classification.

[0065] Generally speaking, in this application, since a dedicated plug-in capable of automatically identifying and analyzing the pulmonary artery, pulmonary vein, and their branches has been developed, and the automatic extraction of blood vessel peripheral distribution, centerline features, and classification information has been realized, the dimension of the available blood vessel features has been significantly expanded, effectively solving the technical problem in the prior art that only the basic parameters of the overall blood vessel model can be obtained and the information productivity is low. Furthermore, a more comprehensive and in-depth quantitative analysis of the pulmonary vascular system has been realized. Not only rich feature parameters including blood vessel morphological features and network topology structure features have been obtained, but also multi-level feature extraction from the whole to the local and from the main trunk to the branches has been realized, greatly improving the information productivity of pulmonary vascular radiomics acquisition and providing more abundant and reliable data support for clinical diagnosis and treatment decision-making.

[0066] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar content as in the above-mentioned first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 3 , based on the first embodiment, this embodiment gives a specific artificial intelligence-assisted pulmonary vascular radiomics acquisition method as follows:

[0067] Step S201, based on the operation of the user selecting 1 layer in the coronal plane, sagittal plane and transverse plane in the original file to draw target areas for the lungs and the background, reconstruct the lung model;

[0068] Among them, the Grow From Seed module represents a functional module for segmentation and reconstruction based on an improved GrowCut algorithm. The seed region refers to the target area manually outlined by the user and serves as the initial region for segmentation. The GrowCut algorithm represents an algorithm that completes segmentation by iteratively expanding the seed region and realizes the growth of the segmentation region by checking and merging adjacent pixels similar to the seed region. The 3D model refers to a three-dimensional visualization model automatically generated based on the segmentation result.

[0069] This step is executed after the DICOM data is imported to obtain a basic three-dimensional lung model. Specifically, the user first selects one layer in each of the three sectional views that can best reflect the lung contour, and uses the paint tool to outline the seed region, including the lung region and the background region. Then, the Grow From Seed module is called. This module uses the outlined target area as the initial seed region and continuously checks adjacent pixels through the improved GrowCut algorithm. When a pixel similar to the seed region is found, it is added to the seed region, and the neighborhood of the newly added pixel is continued to be checked. This process is repeated continuously until no adjacent pixels that meet the similarity condition can be found, and finally the complete lung segmentation result is obtained. The software then automatically generates a three-dimensional lung model based on this segmentation result.

[0070] Step S202, based on the operation of the user using the painting tool to draw and defining the intra-pulmonary region on the main pulmonary artery and left atrium sections in the transverse plane of the original file, complete the reconstruction of the pulmonary vascular model;

[0071] Among them, the similarity calculation of the GrowCut algorithm represents the standard for judging whether adjacent pixels should be added to the seed region, including the comparison of features such as pixel gray value and gradient. The pixel neighborhood refers to the set of pixels to be checked around the current pixel. The iterative process represents the loop process in which the algorithm repeatedly executes operations such as checking the neighborhood, judging similarity, and expanding the seed region. Vascular enhancement refers to a preprocessing operation that highlights the vascular structure features through image processing.

[0072] This step is executed after the reconstruction of the lung model to obtain a complete pulmonary vascular tree model. Specifically, the user first locates the main pulmonary artery and left atrium levels in the cross-sectional view and uses the paint tool to outline the vascular structures at these positions as seed regions. The user then starts the Grow From Seed module, which gradually expands the segmentation region through an iterative process of the GrowCut algorithm based on these initial seed regions. During the expansion process, the algorithm calculates the similarity between each neighboring pixel and the seed region and decides whether to include it in the segmentation region according to a set threshold. To ensure the accuracy of vascular segmentation, the algorithm limits the search range within the reconstructed lung model. This process continues until no eligible new pixels can be found, and finally, a complete pulmonary vascular segmentation result is obtained, and a three-dimensional vascular model is automatically generated.

[0073] Step S203: Based on the pulmonary vascular model, in response to the user's operation, generate a first simplified model of the vascular branches of each pulmonary artery model and a second simplified model of the vascular branches of each pulmonary vein model;

[0074] This step can refer to Step S20 and will not be elaborated here;

[0075] Step S204: Based on the first simplified model and the second simplified model, calculate the endpoints of the vascular branches of the pulmonary artery model and the endpoints of the vascular branches of the pulmonary vein model;

[0076] Among them, the Extract Centerline module represents a functional module in the 3D Slicer software for extracting the centerline and endpoints of blood vessels. The Auto-detect function refers to the function of automatically identifying the endpoints in the vascular network structure. An endpoint represents a position in the vascular network that is connected to only one network unit. A network unit refers to a basic component in the vascular centerline network. A simplified model refers to a vascular branch model after surface extraction processing.

[0077] This step is executed after obtaining the simplified vascular model to automatically detect the endpoint positions of vascular branches. Specifically, the system first switches to the Extract Centerline module and accesses the functional interface through the logical object of the module. Then, each simplified vascular branch model is processed in turn and set as the input of the module. Next, a new Markup node is created for each vascular model to store endpoint information. By calling the auto-detect function, the Extract Centerline module punches holes in the vascular model and starts growing to extract the vascular network structure. Finally, each unit in the network is traversed to check whether its endpoint is connected to only one unit, and the points that meet the conditions are recorded as endpoints.

[0078] Step S205: Generate the distribution parameters of the terminal blood vessels of the pulmonary artery and the pulmonary vein based on the endpoints of the branches of the pulmonary artery model and the branches of the pulmonary vein model.

[0079] Among them, the terminal blood vessel distribution parameter is a quantitative index that describes the distribution characteristics of the endpoints of the pulmonary artery and the pulmonary vein in three-dimensional space. The endpoint refers to the end position of the blood vessel branch. The spatial distribution characteristics include information such as the three-dimensional coordinates, distribution density, and relative position of the endpoints. The shortest distance refers to the Euclidean distance from the endpoint to a specific reference position. The statistical index refers to the numerical characteristics that describe the overall distribution law of the endpoints, such as the average distance, standard deviation, etc.

[0080] This step is executed after calculating the endpoints of the blood vessel branches and is used to analyze the spatial distribution characteristics of the blood vessel endpoints. Specifically, first collect the position information of all the endpoints of the pulmonary artery and pulmonary vein branches and represent them in a unified three-dimensional coordinate system. Then use the FiducialsToModelDistance module to calculate the shortest distance from each endpoint to the reference position and generate a Distance Table containing distance information. Next, based on these raw data, calculate statistical features such as endpoint density distribution, average distance, distribution range, etc., and generate a Metrics Table containing statistical information. Finally, integrate all the parameters into a unified data table to form a complete set of terminal blood vessel distribution parameters.

[0081] Step S206: Generate the radiomics feature parameters of the pulmonary artery, the pulmonary vein, and the pulmonary vasculature based on each of the first simplified models, each of the second simplifications, and the pulmonary vasculature model.

[0082] Among them, the Radiomics module represents a functional module for extracting quantitative features of medical images. Feature Classes refer to different types of radiomics features, including First Order features, Gray Level Co-occurrence Matrix features, etc. The radiomics feature parameter is a quantitative index extracted from medical images and is used to describe the gray scale distribution, texture features, and shape features of the images. The segmentation node is a data container that stores the segmentation results of different anatomical structures.

[0083] This step is executed after the vascular model reconstruction is completed and is used to extract quantitative radiomics features. Specifically, the system first creates two new segmentation nodes, which are used to store the segmentation results of eight blood vessels and the overall pulmonary blood vessels respectively. Then the Radiomics module is enabled, and the feature categories to be extracted are set, usually including Firstorder, GLCM, GLDM, GLRLM, GLSZM, NGTDM, Shape, and Shape2D, etc. Next, feature extraction is performed on each segmentation node, and the extraction process is started through the CLI interface, and the extracted feature results are saved to the output table. Finally, the completion status of the feature extraction is monitored through a timer to ensure that all features are successfully extracted.

[0084] Step S207: Calculate the vessel centerline and vessel grading based on the endpoints of the blood vessel branches of the pulmonary artery model and the endpoints of the blood vessel branches of the pulmonary vein model;

[0085] Among them, the centerline represents the skeleton structure path of the blood vessel. The vtkvmtkPolyDataCenterlines algorithm is a centerline extraction algorithm based on the Voronoi diagram. This algorithm first calculates the Voronoi diagram of the blood vessel surface mesh, and then finds the shortest path on the Voronoi diagram to obtain the centerline. The Network model refers to the model that describes the connection relationship of blood vessel branches. The Network curve represents the geometric shape of the blood vessel centerline. The Network properties refer to the quantitative parameters that describe the blood vessel characteristics. The vessel grading represents the hierarchical relationship of the blood vessels from the main trunk to the branches. The Level represents the generation of the blood vessel branches relative to the main trunk. The intersection point refers to the connection position of different blood vessel branches.

[0086] This step is executed after the vessel endpoint positions are obtained and is used to extract the vessel centerline and determine the branch hierarchy. Specifically, first use the Extract Centerline module to extract the vessel centerline based on the previously detected endpoint positions through the vtkvmtkPolyDataCenterlines algorithm. The module will generate a Network model that describes the blood vessel network structure, a Network curve that represents the centerline shape, and Networkproperties that record the blood vessel properties. Then, by analyzing the intersection point positions of the centerlines, a tree structure of the blood vessel branches is established, and the hierarchical relationship of each branch is determined using breadth-first search. Finally, the grading results are updated to the network property table to complete the vessel grading.

[0087] In general, this solution first delineates the target area based on the layers selected by the user on the coronal, sagittal and transverse planes of the original CTA file, and reconstructs the lung model using the improved GrowCut algorithm. Then, the reconstruction of the pulmonary vascular model is completed through the user's marking operations on the main pulmonary artery and left atrial sections in the cross section. Next, based on the reconstructed pulmonary vascular model, the system generates simplified models of the branches of the pulmonary artery and pulmonary vein. Subsequently, the vascular endpoints are calculated using the Extract Centerline module, and the endpoint distribution characteristics are analyzed using the FiducialsToModelDistance module. The system also extracts features of vascular structures at different levels through the Radiomics module, and finally performs vascular grading analysis based on the extracted centerlines.

[0088] In the embodiments of the present application, due to the development of a special plug-in that can automatically call up modules such as Radiomics, Extract Centerline, and Fiducials To Model Distance, the manual operation cost is significantly reduced, and the problem of inconvenient use of the prior art is effectively solved. At the same time, due to the introduction of additional modules to assist in the segmentation of the pulmonary artery, pulmonary vein and its branches, step-by-step feature collection from the whole to the branch is achieved, which effectively solves the problem of poor extraction effect of relying solely on the Radiomics module in the prior art. In addition, by modifying the underlying code to add new functions, the system can automatically manage and integrate various types of feature data, effectively solving the problem of low information yield in the prior art. This realizes the automation, refinement and systematization of feature extraction, greatly improving the efficiency and completeness of pulmonary vascular imaging genomics acquisition, and providing more comprehensive and reliable data support for clinical diagnosis.

[0089] Based on the first embodiment and / or the second embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those of the first and second embodiments can be referred to the above description, and will not be described in detail later. Figure 3 , Figure 4 , Figure 5 and Figure 6 , this embodiment provides a more detailed solution for each step from step S204 to step S207;

[0090] For step S204, refer to Figure 3 ,as follows:

[0091] Step S301, obtaining each of the first simplified models and each of the second simplified models;

[0092] The model data refers to the digital information describing the geometric shape of the blood vessel. The reading process refers to the operation of loading data from the storage space to the memory.

[0093] This step is executed after the simplification process of the blood vessel model to prepare data for subsequent endpoint detection. Specifically, the system reads the simplified models of eight blood vessels from the system storage space, including each first simplified model and each second simplified model. These models retain the main blood vessel structure features and remove the fine branches, providing ideal input data for subsequent endpoint detection.

[0094] In some embodiments, the acquisition of the simplified model can be achieved in multiple ways: Optionally, first determine the storage path, then read the data of the simplified first simplified model, then read the data of the simplified second simplified model, then verify the data integrity, and finally load the model into the memory; Optionally, first initialize the data reading environment, then load the simplified models one by one according to the identifiers of the first simplified model and the second simplified model, then perform format conversion on the loaded data, then verify the data validity, and finally complete the acquisition of all the first simplified models and the second simplified models. It can be understood that other data acquisition methods can also be used to implement the reading and loading of the model, which is not limited here.

[0095] Step S302, call to detect each of the first simplified models and each of the second simplified models to obtain the first endpoint of the blood vessel branches of the pulmonary artery model and the second endpoint of the blood vessel branches of the pulmonary vein model;

[0096] Among them, the call represents the operation of starting the Extract Centerline module. The detection process refers to the process of using the auto-detect function to identify the endpoints. The first endpoint represents the endpoint position of the blood vessel branches of the pulmonary artery. The second endpoint refers to the endpoint position of the blood vessel branches of the pulmonary vein. The punching position represents the starting point of the detection algorithm. The network structure refers to the data describing the blood vessel connection relationship. Cell represents the basic unit in the network.

[0097] This step is executed after the simplified model is obtained to detect the blood vessel endpoints. Specifically, the system calls the ExtractCenterline module and performs endpoint detection through the auto-detect function of the module. During the detection, first select a punching position on the surface of the first simplified model or the second simplified model, start growing from this position to extract the network structure, and then analyze the connection relationship of each Cell in the network, and identify the endpoint position as the endpoint that is only connected to one unit. Create Markup nodes for each first simplified model and second simplified model to save the detected endpoint information.

[0098] In some embodiments, the detection of the end points can be achieved in various ways: Optionally, first initialize the Extract Centerline module, then create an end point storage node, then execute an automatic detection algorithm, then analyze the network connection relationship, and finally determine the end point position; Optionally, first load the module parameters, then select the starting detection position, then generate the network structure, then analyze the end point features, and finally save the detection results. It can be understood that other end point detection methods can also be used to identify the ends of blood vessels, which are not limited herein.

[0099] Step S303, when detecting one of the first end points or the second end points, re - call to perform the detection of the next first end point or the second end point;

[0100] Among them, re - calling means resetting the module state and starting a new detection. Completion of the detection means that the identification of the end points of the current blood vessel has ended. Detection of the next end point means processing a new blood vessel model. The module state refers to the running environment of the ExtractCenterline module. The reset operation means clearing the temporary data and initializing the parameters.

[0101] This step is executed after the detection of each blood vessel end point to ensure continuous and stable detection of all blood vessel end points. Specifically, after the detection of the end points of a first simplified model or a second simplified model is completed, the system resets the module state, clears the temporary data, and prepares for the detection of the next first simplified model or second simplified model. This process will continue until the detection of the end points of all eight first simplified models or second simplified models is completed, and finally, the user is notified of the completion of the detection through a prompt message.

[0102] In some embodiments, the continuous execution of the detection can be achieved in various ways: Optionally, first save the current detection results, then clear the module environment, then update the processing progress, then initialize new detection parameters, and finally start the next detection; Optionally, first verify the current detection state, then reset the module configuration, then prepare a new detection object, then update the detection queue, and finally start a new detection process. It can be understood that other ways can also be used to achieve the continuous processing of end point detection, which are not limited herein.

[0103] Regarding step S205, refer to Figure 4 , as follows:

[0104] Step S304, extract the background region from the segmentation nodes of the pulmonary artery model and the pulmonary vein model and convert it into a three - dimensional background region model, and remove the background region data in the segmentation nodes of the pulmonary artery model and the pulmonary vein model;

[0105] Among them, the segmentation node represents a data structure storing the segmentation results of the blood vessel model. The background region refers to the tissue region outside the blood vessel model. The three-dimensional model of the background region represents the result after converting the background region into a three-dimensional grid. The removal operation refers to deleting the data in the segmentation node. Data conversion represents the process of converting the segmentation data into a three-dimensional model.

[0106] This step is executed after the end point detection and is used to extract and process the background region data. Specifically, the system first identifies the segment named "background" from the segmentation nodes of the pulmonary artery model and the pulmonary vein model and exports it as an independent three-dimensional model. Then, the data of the background region is deleted from the original segmentation node to avoid interference during subsequent processing. The three-dimensional model of the background region obtained in this way will be used for subsequent distance calculation.

[0107] In some embodiments, the extraction and processing of the background region can be implemented in various ways: Optionally, first identify the background region in the segmentation node, then convert the background data into a three-dimensional model, then verify the integrity of the model, then clean the background data in the original node, and finally save the background model; Optionally, first load the segmentation data, then separate the background region information, then generate the background three-dimensional grid, then remove the original background data, and finally verify the processing result. It can be understood that other methods can also be used to implement the extraction and conversion of the background region, which is not limited here.

[0108] Step S305, calculate the shortest distance from the end points of each blood vessel branch of the pulmonary artery model and the end points of each blood vessel branch of the pulmonary vein model to the three-dimensional model of the background region;

[0109] Among them, the shortest distance represents the closest distance from the end point to the surface of the three-dimensional model of the background region. The calculation process refers to using the FiducialsToModelDistance module for distance measurement. The end point represents the end position of the blood vessel branch. Distance measurement represents the process of calculating the shortest path between a point and a surface in three-dimensional space. The calculation result represents the distance value from each end point to the background model.

[0110] This step is executed after obtaining the three-dimensional model of the background region and is used to measure the distance from the blood vessel end point to the background. Specifically, the system uses the FiducialsToModelDistance module, takes the previously detected blood vessel end points as reference points, and the three-dimensional model of the background region as the target surface, and calculates the shortest distance from each end point to the surface of the background model. The distance calculation is performed for all end points of each blood vessel, and the results are saved to the corresponding Metrics Table and Distance Table.

[0111] In some embodiments, distance calculation can be achieved in various ways: Optionally, first load the end point data and the background model, then initialize the distance calculation module, then perform the point-to-plane distance calculation, then verify the calculation result, and finally save the distance data; Optionally, first prepare the calculation parameters, then process each end point one by one, then calculate the distance to the background, then generate a result table, and finally organize the calculation data. It can be understood that other distance calculation methods can also be used to measure the distance from the end point to the background, which is not limited here.

[0112] Step S306: Merge all the shortest distances to generate the peripheral vascular distribution parameters of the pulmonary artery and pulmonary vein.

[0113] Among them, the merging operation means integrating multiple distance data together. The peripheral vascular distribution parameter refers to a set of data describing the distribution characteristics of the vascular end points. The parameter table refers to a structured table storing the merged data. Data integration means the process of combining the measurement results of different blood vessels. The merged result refers to the final generated comprehensive data table.

[0114] This step is executed after the distance calculation is completed and is used to integrate all the measurement results. Specifically, the system traverses all the generated metric tables and distance tables, merges the data containing "Metrics Table" and "Distances Table", and generates a unified Merged Metrics Table and Merged Distance Table. These tables contain the location distribution characteristics of all the end points of the pulmonary artery and pulmonary vein.

[0115] In some embodiments, data merging can be achieved in various ways: Optionally, first collect the distance data of all blood vessels, then unify the table format, then merge the data of the same type, then organize the merged result, and finally generate the final table; Optionally, first prepare the merging template, then import the distance data one by one, then check the data consistency, then organize the data structure, and finally output the merged table. It can be understood that other data merging methods can also be used to integrate the parameters, which is not limited here.

[0116] Regarding step S206, refer to Figure 5 , as follows:

[0117] Step S307: Create the first segmentation node and the second segmentation node.

[0118] Among them, the first segmentation node represents a data structure for storing all the first simplified models and the second simplified models. The second segmentation node refers to a data structure for storing the lung model and the pulmonary vascular model. The creation operation represents generating a new data node in the system. The segmentation node represents a container for organizing and managing segmented data. Node creation represents initializing a new data storage space.

[0119] This step is executed after parameter merging and is used to prepare the data structure for feature extraction. Specifically, the system creates two new segmentation nodes, named Segmentation1 and Segmentation2 respectively, for storing different types of segmented data subsequently. These nodes provide independent data management spaces, facilitating subsequent feature extraction operations.

[0120] In some embodiments, the creation of the segmentation node can be implemented in various ways: Optionally, first initialize the node creation environment, then create the first segmentation node, then set the node attributes, subsequently create the second segmentation node, and finally verify the node status; Optionally, first prepare the node template, then generate two segmentation nodes, then configure the node parameters, then check the node validity, and finally complete the node initialization. It can be understood that other methods can also be used to implement the creation of the segmentation node.

[0121] Step S308, copy each of the first simplified models and each of the second simplified models into the first segmentation node, and copy the lung model and the pulmonary vascular model into the second segmentation node;

[0122] Among them, the copy operation represents copying the model data to the target node. Data migration represents the process of transferring the model data to a new node. Model organization represents organizing the model data according to a specific structure.

[0123] This step is executed after creating the segmentation node and is used to organize the data structure. Specifically, the system first copies all the first simplified models and the second simplified models into the first segmentation node, and then copies the lung model, the pulmonary vascular model, the overall pulmonary vein model, and the overall pulmonary artery model into the second segmentation node.

[0124] In some embodiments, the copying of the model data can be implemented in various ways: Optionally, first load the source model data, then verify the data integrity, then execute the data copy operation, subsequently confirm the copy result, and finally organize the node data; Optionally, first prepare the data transfer environment, then copy the models by category, then check the data consistency, then organize the model structure, and finally complete the data migration. It can be understood that other methods can also be used to implement the copying and organization of the model data, which is not limited here.

[0125] Step S309: Select extraction parameters, and based on the extraction parameters, extract features from the first segmentation node and the second segmentation node to generate the radiomics feature parameters of the pulmonary artery, the pulmonary vein, and the pulmonary blood vessels;

[0126] Among them, the extraction parameters represent the configuration options for radiomics feature extraction. Feature extraction represents the process of calculating the imaging features of the model. The radiomics feature parameters refer to the quantitative indicators describing the model features. The feature categories include Firstorder, glcm, gldm, glrlm, glszm, ngtdm, shape, shape2D, etc. Parameter generation represents the process of calculating and outputting the feature values.

[0127] This step is executed after the data organization is completed and is used to extract the model features. Specifically, first, the user selects the feature extraction parameters in the Radiomics module, including feature categories such as Firstorder, glcm, gldm, glrlm, glszm, ngtdm, shape, shape2D, etc. Then the system extracts the features of the models in the first segmentation node and the second segmentation node, starts the extraction process through the CLI interface, and finally organizes the extracted feature data into a radiomics feature parameter table.

[0128] In some embodiments, feature extraction can be implemented in multiple ways: Optionally, first prompt the user to configure the extraction parameters, then initialize the Radiomics module, then perform the feature extraction calculation, then monitor the extraction progress, and finally generate the feature parameters; Optionally, first prompt Tong Hu to set the feature categories, then prepare the extraction environment, then start the feature calculation, then verify the extraction results, and finally output the feature data. It can be understood that other methods can also be used to implement the extraction of radiomics features, which are not limited here.

[0129] Regarding step S206, refer to Figure 6 , as follows:

[0130] Step S310: Based on the endpoints of the blood vessel branches of the simplified model of the pulmonary artery and the endpoints of the blood vessel branches of the simplified model of the pulmonary vein, extract the blood vessel centerline;

[0131] Among them, the centerline represents the central path inside the blood vessel. The endpoint refers to the end position of the blood vessel branch. The extraction process represents the operation of generating the centerline using the vtkvmtkPolyDataCenterlines algorithm. The vtkvmtkPolyDataCenterlines algorithm is an algorithm for extracting the centerline from a blood vessel model. This algorithm grows along the blood vessel model and identifies the centerline of the blood vessel. Blood vessel growth represents the process of the algorithm extending along the blood vessel model. The Voronoi diagram represents the geometric structure describing the area around the centerline.

[0132] This step is executed after feature extraction and is used to generate the vascular centerline. Specifically, the ExtractCenterline module is used to extract the centerline of the blood vessels through the vtkvmtkPolyDataCenterlines algorithm based on the endpoints of the vascular branches of the simplified pulmonary artery model and the endpoints of the vascular branches of the simplified pulmonary vein model detected previously. The algorithm identifies the central path by growing inside the vascular model, and simultaneously generates centerline data and Voronoi diagram data.

[0133] In some embodiments, the centerline extraction can be implemented in various ways: Optionally, first load the endpoint data, then initialize the extraction algorithm, then perform centerline calculation, then generate the geometric structure, and finally verify the extraction result; Optionally, first set the algorithm parameters, then process the vascular model, then calculate the central path, then generate auxiliary data, and finally complete the centerline extraction. It can be understood that other methods can also be used to implement the extraction of the vascular centerline, which is not limited here.

[0134] Step S311: Convert each of the vascular centerlines into corresponding curve nodes and store them in the vascular network attribute table;

[0135] Among them, the table node represents the data structure for storing the centerline attributes. The data conversion represents the process of converting the centerline geometric information into a table form. The attribute calculation represents the operation of quantitatively analyzing the centerline. The quantization result represents the numerical characteristics of the centerline. The table storage represents storing the data in a structured table.

[0136] This step is executed after extracting the centerline and is used to process and store the centerline data. Specifically, each extracted vascular centerline is converted into the corresponding curve node (centerlineCurveNode) form, which makes the centerline data more convenient for subsequent processing and analysis. The converted curve node information will be stored in the vascular network attribute table. This attribute table is a structured data storage form for recording and managing the characteristic data of all vascular centerlines.

[0137] In some embodiments, the data conversion and storage can be implemented in various ways: Optionally, first convert the centerline into curve nodes, and then store the node information in the attribute table; Optionally, first create the attribute table structure, then perform the centerline conversion, and finally complete the data storage. It can be understood that other methods can also be used to implement the conversion and storage from the centerline to the curve nodes, which is not limited here.

[0138] Step S312: Create a hierarchical column in the vascular network attribute table and initialize the hierarchical value;

[0139] Among them, the hierarchical column represents the table column used to record the vascular hierarchy. Initialization represents setting the starting state of the hierarchical value. The attribute table represents the data table storing the characteristics of the vascular network. The hierarchical value represents the hierarchical number of the vascular branches. The creation operation represents adding a new data column to the table.

[0140] This step is executed after generating the vascular network attribute table and is used to prepare the hierarchical data structure. Specifically, the system creates a new data column named "Level" in the existing vascular network attribute table, and this column is specifically used to record the hierarchical information of the vascular branches. After creation, the hierarchical values of all rows in this column are initialized to 0, and this initialization operation provides a basis for subsequent hierarchical calculations.

[0141] In some embodiments, the creation and initialization of the hierarchical column can be implemented in various ways: Optionally, first add the Level column to the attribute table, and then set all values to 0; Optionally, first prepare the table structure, and then create a new column and initialize the data. It can be understood that other methods can also be used to implement the creation and initialization of the hierarchical column, which is not limited here.

[0142] Step S313, calculate the intersection points of each of the curve nodes;

[0143] Among them, the curve node represents the data structure describing the geometric shape of the centerline. The intersection point represents the position where different centerlines intersect. The coordinate point represents the spatial position on the centerline. The node conversion represents the process of converting the centerline data into curve nodes. The intersection point calculation represents the operation of identifying the intersection position of the centerlines.

[0144] This step is executed after creating the hierarchical column and is used to determine the spatial relationship of the vascular branches. Specifically, the system calculates the intersection points between each curve node through the find_intersections function, and this process will identify and record the intersection positions of different curve nodes in space. The obtained intersection point information is crucial for understanding the topological structure of the vascular network and subsequent construction of the branch hierarchy tree.

[0145] In some embodiments, the intersection point calculation can be implemented in various ways: Optionally, first load the curve node data, and then use the find_intersections function to calculate the intersection points; Optionally, first prepare the node data, then perform the intersection point calculation, and finally record the intersection information. It can be understood that other methods can also be used to implement the calculation of the intersection points of the curve nodes, which is not limited here.

[0146] Step S314, based on the user input, determine the curve node number of the starting point and each of the intersection points to construct a vascular branch hierarchy tree, determine the hierarchy of each curve node, and further determine each vascular grade;

[0147] Among them, the hierarchical tree represents a tree structure describing the vascular branching relationship. The starting point number represents the initial node identifier for starting to construct the tree. The hierarchical determination represents the depth of the analysis node in the tree. The branching level represents the level relationship of vascular branches. The tree structure represents the data form used to organize the branching relationship.

[0148] This step is executed after calculating the intersection points and is used to determine the vascular grading. Specifically, the system determines the curve node number of the starting point through user input, uses this starting point and the intersection point information calculated previously, and constructs a vascular branching hierarchical tree using the breadth-first search method. Based on the structure of the tree, the level of each curve node is determined, and then the levels of each vascular branch are determined.

[0149] In some embodiments, the grading determination can be achieved in various ways: Optionally, first determine the starting point position, then construct the hierarchical tree, then analyze the node depth, then update the grading information, and finally verify the grading result; Optionally, first prepare the tree structure environment, then perform the breadth-first search, then determine the node level, then update the grading data, and finally complete the hierarchical determination. It can be understood that other methods can also be used to achieve the grading of vascular branches, which is not limited here.

[0150] Generally speaking, first the system obtains the simplified vascular model, detects the endpoints of each vascular branch through the auto-detect function of the Extract Centerline module, and resets the module state after each detection is completed to continue the next detection. Then extract the background region from the segmentation nodes and convert it into a three-dimensional model, calculate the shortest distance from the vascular endpoints to the background through the FiducialsToModelDistance module, and merge these distance data to generate the distribution parameters of the peripheral blood vessels. Then create two segmentation nodes, store the simplified model and the overall model respectively, select the characteristic parameters and then use the Radiomics module for feature extraction. Finally, based on the endpoint position, extract the vascular centerline through the vtkvmtkPolyDataCenterlines algorithm, convert the centerline into curve nodes and store them in the attribute table, create a grading column and initialize it, calculate the intersection points of each curve node, and finally complete the grading by constructing a vascular branching hierarchical tree based on the starting point specified by the user.

[0151] In the embodiments of the present application, due to the adoption of a modular feature extraction architecture, functions such as endpoint detection, distance calculation, feature extraction, and vascular grading are decoupled into relatively independent solutions, and the coordinated work between modules is ensured through automated data flow, so the maintainability and scalability of the system are significantly improved, effectively solving the problems of high functional coupling degree and difficult maintenance in traditional methods. At the same time, the integrity and accuracy of vascular feature data are ensured, and the reliability and practicality of pulmonary vascular radiomics analysis are greatly improved.

[0152] It should be noted that the above examples are only for understanding the present application and do not constitute a limitation on the artificial intelligence-assisted pulmonary vascular imagingomics acquisition method of the present application. Based on this technical concept, more forms of simple transformations are within the protection scope of the present application.

[0153] The present application provides an artificial intelligence-assisted pulmonary vascular imagingomics acquisition device. The artificial intelligence-assisted pulmonary vascular imagingomics acquisition device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the artificial intelligence-assisted pulmonary vascular imagingomics acquisition method in the first embodiment above.

[0154] Reference is made below to Figure 7 , which shows a schematic structural diagram of an artificial intelligence-assisted pulmonary vascular imagingomics acquisition device suitable for implementing the embodiments of the present application. The artificial intelligence-assisted pulmonary vascular imagingomics acquisition device in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 7 The artificial intelligence-assisted pulmonary vascular imagingomics acquisition device shown is only an example and should not impose any limitation on the functions and usage scope of the embodiments of the present application.

[0155] As Figure 7As shown, the artificial intelligence-assisted pulmonary vascular radiomics acquisition device may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM: Read Only Memory) 1002 or the program loaded from the storage device 1003 into the random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the artificial intelligence-assisted pulmonary vascular radiomics acquisition device are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. The input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the artificial intelligence-assisted pulmonary vascular radiomics acquisition device to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows an artificial intelligence-assisted pulmonary vascular radiomics acquisition device with various systems, it should be understood that it is not required to implement or have all the shown systems. More or fewer systems can be alternatively implemented or had.

[0156] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above functions defined in the methods of the embodiments disclosed in the present application are executed.

[0157] The artificial intelligence-assisted pulmonary vascular imaging genomics acquisition device provided by this application adopts the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method in the above-mentioned embodiment, and can solve the technical problem of how to improve the information productivity of pulmonary vascular imaging genomics acquisition. Compared with the prior art, the beneficial effects of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition device provided by this application are the same as those of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method provided by the above-mentioned embodiment, and other technical features in the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition device are the same as those disclosed in the method of the previous embodiment, and will not be elaborated here.

[0158] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0159] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all of them should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0160] This application provides a computer-readable storage medium with computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method in the above-mentioned embodiment.

[0161] The computer-readable storage medium provided by this application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, device, or component. The program code contained on the computer-readable storage medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0162] The above computer-readable storage medium can be included in the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition device; it can also exist separately and not be assembled into the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition device.

[0163] The modules described in the embodiments of this application can be implemented in software or in hardware. In some cases, the name of the module does not constitute a limitation on the unit itself.

[0164] The readable storage medium provided by this application is a computer-readable storage medium, and the computer-readable storage medium stores computer-readable program instructions (i.e., computer programs) for executing the above artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method, which can solve the technical problem of how to improve the information productivity of pulmonary vascular imaging genomics acquisition. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by this application are the same as those of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method provided in the above embodiments, and will not be elaborated here.

[0165] This application also provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the steps of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method as described above.

[0166] The computer program product provided by the present application can solve the technical problem of how to improve the information productivity of pulmonary vascular imaging genomics. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as those of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method provided in the above embodiments, and will not be elaborated herein.

[0167] The foregoing are only partial embodiments of the present application, and thus do not limit the patent scope of the present application. Any equivalent structural transformation made under the technical concept of the present application by using the content of the specification and drawings of the present application, or any direct / indirect application in other related technical fields, is included in the patent protection scope of the present application.

Claims

1. An artificial intelligence-assisted pulmonary vascular imaging omics acquisition method, characterized in that: The artificial intelligence-assisted pulmonary vascular imaging omics acquisition method includes: Reconstructing a lung model and a pulmonary vascular model in response to a user's operation based on the input original file; Based on the pulmonary vascular model, generating a first simplified model of each pulmonary artery model vascular branch and a second simplified model of each pulmonary vein model vascular branch in response to a user's operation; Based on the lung model, the pulmonary vascular model, the first simplified model and the second simplified model, the peripheral vascular distribution parameters of the pulmonary artery and pulmonary vein and their imaging omics feature parameters are generated, the imaging omics feature parameters of the entire blood vessel are generated, the vascular centerline is calculated, and the blood vessels are graded.

2. The method according to claim 1, characterized in that The step of reconstructing the lung model and the pulmonary vascular model based on the input original file and in response to the user's operation specifically includes: Reconstructing the lung model based on the user's operation of selecting one plane each from the coronal plane, sagittal plane, and transverse plane in the original file to face the target area of ​​the lung and the background image; The reconstruction of the pulmonary vascular model is completed based on the user's operation of drawing the cross-sectional main pulmonary artery and left atrium sections in the original file using drawing tools and defining the intrapulmonary area.

3. The method according to claim 1, characterized in that The steps of generating peripheral vascular distribution parameters of pulmonary arteries and pulmonary veins and their radiomics characteristic parameters, generating radiomics characteristic parameters of whole blood vessels based on the lung model, the pulmonary vascular model, the first simplified model and the second simplified model, calculating the vascular centerline and grading the blood vessels specifically include: Based on the first simplified model and the second simplified model, calculating the end point of the pulmonary artery model blood vessel branch and the end point of the pulmonary vein model blood vessel branch; generating peripheral vascular distribution parameters of the pulmonary artery and pulmonary vein based on the branches of the pulmonary artery model and the endpoints of the branches of the pulmonary vein model; generating radiomics feature parameters of the pulmonary artery, the pulmonary vein, and the pulmonary blood vessel based on each of the first simplified models, each of the second simplified models, and the pulmonary blood vessel model; Based on the end point of the pulmonary artery model vascular branch and the end point of the pulmonary vein model vascular branch, a vascular centerline and a vascular grade are calculated.

4. The method according to claim 3, characterized in that The step of calculating the end point of the pulmonary artery model blood vessel branch and the end point of the pulmonary vein model blood vessel branch based on the first simplified model and the second simplified model specifically includes: Acquire each of the first simplified models and each of the second simplified models; Calling and detecting each of the first simplified models and each of the second simplified models to obtain a first end point of the vascular branch of the pulmonary artery model and a second end point of the vascular branch of the pulmonary vein model; After detecting one of the first endpoints or the second endpoint, the detection of the next first endpoint or the second endpoint is called again.

5. The method according to claim 3, characterized in that The step of generating the peripheral blood vessel distribution parameters of the pulmonary artery and the pulmonary vein based on the branches of the pulmonary artery model and the endpoints of the branches of the pulmonary vein model specifically includes: Extracting background regions from the segmentation nodes of the pulmonary artery model and the pulmonary vein model and converting them into background region three-dimensional models, and removing background region data in the segmentation nodes of the pulmonary artery model and the pulmonary vein model; Calculating the shortest distances from the endpoints of each of the pulmonary artery model blood vessel branches and the endpoints of each of the pulmonary vein model blood vessel branches to the three-dimensional model of the background area; The shortest distances are combined to generate the peripheral vascular distribution parameters of the pulmonary artery and pulmonary vein.

6. The method according to claim 3, characterized in that The step of generating radiomics feature parameters of the pulmonary artery, the pulmonary vein and the pulmonary blood vessel based on each of the first simplified models, each of the second simplified models and the pulmonary blood vessel model specifically includes: Create a first split node and a second split node; Copying each of the first simplified models and each of the second simplified models to the first segmentation node, and copying the lung model and the pulmonary vascular model to the second segmentation node; Extraction parameters are selected, and based on the extraction parameters, feature extraction is performed on the first segmentation node and the second segmentation node to generate radiomics feature parameters of the pulmonary artery, the pulmonary vein, and the pulmonary blood vessel.

7. The method according to claim 3, characterized in that The step of calculating the vascular centerline and vascular classification based on the endpoints of the pulmonary artery model vascular branches and the endpoints of the pulmonary vein model vascular branches specifically includes: Extracting the blood vessel centerline based on the endpoint of the blood vessel branch of the simplified pulmonary artery model and the endpoint of the blood vessel branch of the simplified pulmonary vein model; Convert each of the blood vessel center lines into corresponding curve nodes and store them in a blood vessel network attribute table; Create a classification column in the vascular network attribute table and initialize the classification value; Calculating the intersection points of the curve nodes; Based on the curve node number of the starting point determined by the user input and each of the intersection points, a blood vessel branching hierarchical tree is constructed, the level of each curve node is determined, and then the grade of each blood vessel is determined.

8. An artificial intelligence-assisted pulmonary vascular imaging omics acquisition device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method as described in any one of claims 1 to 7.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method according to any one of claims 1 to 7 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, which, when executed by a processor, implements the steps of the artificial intelligence-assisted pulmonary vascular imaging genomics acquisition method as described in any one of claims 1 to 7.

Images