Cerebral arteriovenous malformation vascular structure lumped model simulation method and system

Abstract

The invention discloses a brain arteriovenous malformation vascular structure lumped model simulation method and system, and the method comprises the steps: obtaining the three-dimensional image data of a preset region of the head of a human body, and recognizing a brain arteriovenous malformation vascular mass region in the three-dimensional image data; constructing a blood vessel three-dimensional structure model, and acquiring three-dimensional space coordinate data of each voxel in the cerebral arteriovenous malformation vascular cluster region in the model; calibrating an inlet and an outlet of the blood vessel three-dimensional structure model and corresponding pressure set values and / or flow velocity set values; extracting a plurality of skeleton points of the blood vessel based on the blood vessel three-dimensional structure model to obtain a microscopic blood vessel network; based on the three-dimensional space coordinate data, the length and the average diameter of each blood vessel in the microscopic blood vessel network are obtained, and the flow resistance of the blood vessels is calculated; based on the pressure set value and the flow velocity set value of the blood vessel three-dimensional structure model and the flow resistance of each blood vessel, the blood flow velocity and pressure distribution data of each blood vessel are calculated through a lumped parameter model, and the hemodynamic characteristics of the cerebral arteriovenous malformation blood vessel are obtained.

Description

Technical Field

[0001] The present invention relates to the field of vascular detection imaging, and particularly relates to a method and system for simulating a lumped model of a cerebral arteriovenous malformation vascular structure. Background Art

[0002] Existing hemodynamic quantitative analysis methods mainly include computational fluid dynamics analysis (hereinafter referred to as "CFD"), intravascular ultrasound, and quantitative DSA, etc. Currently, the quantitative hemodynamic calculation and simulation methods for AVM vascular malformations are: traditional CFD calculation methods, quantitative DSA technology, intravascular pressure measurement, and intravascular ultrasound imaging.

[0003] First of all, the technical disadvantages of the CFD method mainly include: 1) Complex analysis process: The computational fluid dynamics (CFD) analysis method involves a series of complex operation steps, and each step has high technical requirements. Specifically, it includes: using 3Dslicer software or Mimic software for three-dimensional vascular modeling, using STARCCM+ for hemodynamic analysis, constructing grids in ANSYS ICEM, and performing fitting calculations in ANSYS CFX-pre, etc. These steps not only require operators to have high professional skills, but also need to master a variety of different advanced software tools, and each link involves complex physical and mathematical principles. For clinicians or other medical personnel without an engineering background, it is difficult to master these complex operation processes and their corresponding physical models. Therefore, the complexity of this method limits its popularization and application in clinical practice and the convenience of actual operation, and may affect the timeliness of medical decisions. 2) High time cost: Another major disadvantage of the CFD analysis process is the high time cost. Due to its involvement in a large amount of calculation and data processing, especially in the process of vascular structure meshing and three-dimensional modeling, the consumption of computing resources is particularly obvious. Taking a common computer configuration as an example, when analyzing a complex vascular structure, the calculation process may take up to several hours, or even more than a day to complete. Specifically, only in the three-dimensional modeling and meshing steps, an ordinary computer needs to spend about half a day to complete this operation, and the time required in the subsequent fluid dynamics simulation and result fitting processes will further increase. Therefore, the traditional CFD method not only requires powerful computing capabilities, but also has a significant time delay, which may cause delays to patients in urgent need of treatment in clinical work, and makes it difficult to meet the requirements of rapid response and real-time decision-making in practical applications.

[0004] Secondly, the technical deficiencies of traditional quantitative DSA methods also include: 1) Limited image dimensions: The images obtained by traditional digital subtraction angiography (DSA) technology usually only provide anteroposterior and lateral projections of the cerebral blood vessels, and the obtained image data is the projection of the vascular structure on a two-dimensional plane. Since the vascular structure naturally exists in three-dimensional space, this projection method is difficult to comprehensively reflect the actual spatial morphology of the blood vessels. Therefore, the calculation results of hemodynamics may be significantly affected by factors such as the angle and overlap of blood vessels, thereby reducing the accuracy of the analysis results. Especially when there are complex vascular intersections or distortions in the anatomical structure, the limitations of two-dimensional images will be further amplified, bringing difficulties to the accurate assessment of blood flow conditions and lesion sites. 2) Indistinct boundaries between blood vessels and brain tissue: In DSA images, the contrast agent imaging usually only reflects the fluid distribution within the blood vessels. However, when the vascular structure is complex or abnormal (such as the fine vascular network within a vascular malformation mass), the contrast agent may be simultaneously imaged in the blood vessels within the malformation mass and the surrounding brain tissue. Due to the overlapping effect of the images, the fine vascular structure within the malformation mass and the surrounding brain tissue have blurred boundaries, resulting in the interference of the imaging information of the surrounding tissues in the calculation of its hemodynamics. This phenomenon significantly limits the ability of DSA to accurately locate and quantify the hemodynamic characteristics of the lesion area, posing challenges to the diagnosis and treatment plan design of complex cases. 3) Invasive operation: The DSA angiography technology requires inserting a catheter into the blood vessel through a percutaneous puncture method for operation. This process is an invasive operation and may cause complications such as pain at the puncture site, vasospasm, thrombosis, or local infection in the patient. In addition, this operation has high technical requirements for the operator, increasing the complexity and risk of medical operations, and also posing higher tolerance requirements for the patient's physical condition. Therefore, the invasiveness of traditional DSA limits its applicability in certain patient groups and increases the medical risk of the diagnostic process.

[0005] Thirdly, the technical deficiencies of the intravascular pressure measurement method: 1) Difficulty in vascular super-selection: The intravascular pressure measurement method relies on inserting a pressure sensor into the target blood vessel through a pressure measurement catheter to measure the blood flow pressure at a specific location. However, in practical applications, the malformed vascular mass often has a deep distribution and the vascular path shows significant tortuosity and complexity, resulting in technical difficulties in the selective insertion process of the pressure measurement catheter. Due to the limited flexibility and operation precision of the catheter, it is difficult to penetrate to any position within the malformed mass for pressure measurement, thus limiting the ability of this method to evaluate the all-round and fine pressure distribution in the lesion area. 2) Discontinuous measurement: Another significant limitation of the intravascular pressure measurement method lies in the locality of its measurement method. The pressure measurement catheter can only collect the pressure data at its location at any given moment, and cannot achieve continuous pressure distribution measurement from the feeding artery to the draining vein. This discontinuity leads to obvious deficiencies in evaluating the pressure gradient or dynamic changes of the entire blood flow circuit, thus affecting the accuracy of the comprehensive diagnosis and treatment planning of complex vascular diseases. 3) Invasive operation: Performing intravascular pressure measurement requires percutaneous puncture of the blood vessel for catheter operation.

[0006] Finally, the technical deficiencies of the intravascular ultrasound method include: 1) Shallow detection depth: The intravascular ultrasound technology uses an ultrasonic probe to image and analyze parameters such as blood flow velocity and wall thickness within the blood vessel. However, due to its working principle relying on signal transmission between the probe and the blood vessel wall, the detection depth of intravascular ultrasound is limited, and it can usually only effectively detect relatively shallow blood vessel areas. For blood vessels located in deeper positions, especially those in complex anatomical regions, the signal attenuation is significant, resulting in a significant decline in the ultrasound detection ability. Therefore, this method is difficult to comprehensively and deeply evaluate the hemodynamic characteristics of deep blood vessels, limiting its application within the whole brain. 2) Poor detection accuracy: For intracranial blood vessels, the detection depth of intravascular ultrasound can usually only reach larger blood vessel segments, such as the large blood vessels in the A2, M2, and P2 segments, while its detection ability for smaller blood vessel branches is relatively limited. In addition, in the face of complex vascular structures, the resolution and imaging accuracy of intravascular ultrasound are insufficient to accurately evaluate the minor changes or lesion sites of blood vessels. Especially in the presence of complex lesions such as vascular malformations, aneurysms, or arteriovenous malformations, traditional intravascular ultrasound cannot provide sufficiently accurate hemodynamic information, affecting the precise localization of the lesion and the formulation of treatment plans. 3) Large human differences: The intravascular ultrasound technology relies on the technical level and experience of the operator during the operation. Since precise positioning and adjustment of the ultrasound probe are required during the operation, technical differences between different operators may lead to deviations in the results. Especially in cases where the vascular morphology is complex or the individual differences of patients are large, the interpretation of ultrasound images may be affected by the subjective judgment of the operator, resulting in poor repeatability and consistency of the measurement results. Therefore, the results of the intravascular ultrasound method are easily interfered by human operation differences and it is difficult to ensure consistent diagnostic conclusions among different medical staff. Summary of the Invention

[0007] The objective of the embodiments of the present invention is to provide a lumped model simulation method and system for the cerebrovascular malformation vascular structure, which obtains data of a specific area of the head by using CTA or MRA images, accurately locates the cerebrovascular malformation vascular mass, and then constructs a three-dimensional vascular model; by extracting skeleton points to form a microscopic network, calculating the flow resistance, and combining with the lumped parameter model to calculate the blood flow velocity and pressure distribution, it provides a key basis for accurate diagnosis and effective treatment of this disease.

[0008] To solve the above technical problems, the first aspect of the embodiments of the present invention provides a lumped model simulation method for the cerebrovascular malformation vascular structure, including the following steps:

[0009] Obtain three-dimensional image data of a preset area of the human head, and identify the cerebrovascular malformation vascular mass area in the three-dimensional image data, where the three-dimensional image data is CTA image data or MRA image data;

[0010] Based on the cerebrovascular malformation vascular mass area, construct a three-dimensional vascular structure model, and obtain the three-dimensional spatial coordinate data of each voxel in the cerebrovascular malformation vascular mass area in the model;

[0011] Calibrate the inlet and outlet of the three-dimensional vascular structure model and the corresponding pressure setting value and / or flow rate setting value;

[0012] Based on the three-dimensional vascular structure model, extract several skeleton points of the blood vessels to obtain a microscopic blood vessel network;

[0013] Based on the three-dimensional spatial coordinate data, obtain the length and average diameter of each blood vessel in the microscopic blood vessel network, and calculate the flow resistance of the blood vessels;

[0014] Based on the pressure setting value, flow rate setting value of the three-dimensional vascular structure model and the flow resistance of each blood vessel, calculate the blood flow velocity and pressure distribution data of each blood vessel through the lumped parameter model, and obtain the hemodynamic characteristics of the cerebrovascular malformation blood vessels.

[0015] Further, after obtaining the three-dimensional image data of the preset area of the human head, it further includes:

[0016] Save the three-dimensional image data as a NifTI format file.

[0017] Further, the constructing the three-dimensional vascular structure model and obtaining its three-dimensional spatial coordinate data includes:

[0018] Based on Python software, extract the image volume data and its physical size of the three-dimensional vascular structure model in NifTI format.

[0019] Further, extracting the image volume data and its physical dimensions of the three-dimensional vascular structure model in NifTI format based on Python software includes:

[0020] Use the nib.load() method to load the NIfTI file at the preset path and read it as a NIfTI image object;

[0021] Obtain the volume data of the image by calling the get_fdata() method of the NIfTI image object, and return a NumPy array containing pixel values;

[0022] Use the header.get_zooms() method to extract the actual physical dimensions of the voxels from the image header information, and return a tuple, where the tuple represents the intervals of each voxel in the X-axis, Y-axis, and Z-axis directions.

[0023] Further, extracting several skeleton points of the blood vessels to obtain a microscopic blood vessel network includes:

[0024] Use a three-dimensional skeletonization algorithm to extract the skeleton points of each blood vessel in the three-dimensional vascular structure model, and convert the coordinates of the skeleton points into spatial coordinates in combination with the voxel interval.

[0025] Further, using a three-dimensional skeletonization algorithm to extract the skeleton points of each blood vessel in the three-dimensional vascular structure model includes:

[0026] Remove the boundaries of the blood vessels through erosion operations and retain the central axes of the blood vessels;

[0027] Based on the KD tree, obtain the adjacent skeleton points of the skeleton points and construct skeleton line segments.

[0028] Further, based on the KD tree, obtaining the adjacent skeleton points of the skeleton points and constructing skeleton line segments includes:

[0029] Obtain the coordinate data set of the skeleton points after the erosion operation, select a coordinate dimension and repeatedly perform median splitting on the coordinate data set until each subset contains only one data point or less than the preset data point threshold to obtain the KD tree;

[0030] Based on each node of the KD tree, compare with the coordinate values of the current skeleton point to obtain several candidate adjacent skeleton points closest to the current skeleton point;

[0031] Sort based on the distance values between the candidate adjacent skeleton points and the current skeleton point, determine several adjacent skeleton points of the current skeleton point, and record the corresponding association relationships;

[0032] Traverse all the skeleton points, and based on the association relationship between each current skeleton point and its adjacent skeleton points, connect the two adjacent skeleton points to obtain the skeleton line segments of the skeleton points.

[0033] Further, after obtaining the skeleton line segments of the skeleton points, it further includes:

[0034] Store the coordinate value data of all the skeleton points and the coordinate data of the skeleton line segments connected by the two endpoints of the skeleton points into the data matrix.

[0035] Further, the calculating the blood flow velocity and pressure distribution data of each blood vessel based on the lumped parameter model includes:

[0036] Construct a blood vessel network with the starting point, ending point coordinates and blood flow resistance of each blood vessel, and at the same time initialize the connection relationship and pressure value of each node;

[0037] Based on the pipe network theory, construct a linear equation set of the blood vessel network including blood vessel pressure to obtain the pressure values of all nodes in the blood vessel network;

[0038] Based on the pressure values of all the nodes, calculate the pressure difference between two adjacent nodes, calculate the flow rate and flow velocity of each blood vessel segment according to the blood flow resistance and the pressure difference, and determine the blood flow direction in each blood vessel according to the positive and negative of the flow velocity.

[0039] Correspondingly, a second aspect of the embodiments of the present invention provides a lumped model simulation system for cerebral arteriovenous malformation vascular structures, including:

[0040] An image acquisition module, which is used to acquire three-dimensional image data of a preset area of the human head, and identify the cerebral arteriovenous malformation vascular mass area in the three-dimensional image data, and the three-dimensional image data is CTA image data or MRA image data;

[0041] A model construction module, which is used to construct a three-dimensional vascular structure model based on the cerebral arteriovenous malformation vascular mass area, and obtain the three-dimensional spatial coordinate data of each voxel in the cerebral arteriovenous malformation vascular mass area in the model;

[0042] A parameter calibration module, which is used to calibrate the inlet and outlet of the three-dimensional vascular structure model and the corresponding pressure setting value and / or flow velocity setting value;

[0043] A skeleton extraction module, which is used to extract a number of skeleton points of blood vessels based on the three-dimensional vascular structure model to obtain a microscopic blood vessel network;

[0044] A resistance calculation module, which is used to obtain the length and average diameter of each blood vessel in the microscopic blood vessel network based on the three-dimensional spatial coordinate data, and calculate the flow resistance of the blood vessel;

[0045] A feature extraction module, which is configured to calculate the blood flow velocity and pressure distribution data of each blood vessel based on the lumped parameter model based on the pressure set value, flow velocity set value of the three-dimensional vascular structure model, and the flow resistance of each blood vessel, so as to obtain the hemodynamic characteristics of the cerebral arteriovenous malformation blood vessels.

[0046] Correspondingly, a third aspect of the embodiments of the present invention provides an electronic device, including: at least one processor; and a memory 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 above-mentioned cerebral arteriovenous malformation blood vessel structure lumped model simulation method.

[0047] Correspondingly, a fourth aspect of the embodiments of the present invention provides a computer-readable storage medium, on which computer instructions are stored, and when the instructions are executed by a processor, the above-mentioned cerebral arteriovenous malformation blood vessel structure lumped model simulation method is implemented.

[0048] The above technical solutions of the embodiments of the present invention have the following beneficial technical effects:

[0049] (1) Compared with the traditional CFD method, the method of the present invention has significant advantages in computing efficiency. By optimizing the computing process and algorithm structure, the computing power requirement is significantly reduced, the simulation time is reduced, and thus the computing resources are effectively saved; in addition, the present invention has an obvious improvement in reducing the cost and difficulty of simulation calculation, making the method more efficient and feasible in practical applications;

[0050] (2) Compared with the quantitative DSA method, the present invention breaks through the limitation of traditional two-dimensional imaging, realizes precise vascular structure and hemodynamic simulation in three-dimensional space, and avoids the problem of structure overlap that may occur in two-dimensional imaging; through the three-dimensional reconstruction of the vascular structure, the present invention can achieve precise segmentation of blood vessels and surrounding brain tissues, significantly improving the accuracy and resolution of vascular imaging; in addition, using the method of the present invention can effectively reduce the dependence on invasive operations, reduce the surgical risk of patients, and improve the safety and comfort of treatment;

[0051] (3) By performing detailed calculations on the internal conditions of the malformation mass, the present invention can perform fine calculations on the blood flow dynamics inside the vascular malformation mass, rather than simply calculating the average parameters of the malformation mass as a whole; through independent analysis of each region inside the malformation mass, the hemodynamic characteristics can be captured more accurately, providing more reliable data support for subsequent diagnosis and treatment;

[0052] (4) Independently calculate each subdivided point inside the malformation mass. The present invention further independently calculates the pressure and flow velocity for each subdivided point inside the malformation mass, and can further deduce more key hemodynamic information based on the differences in hemodynamic parameters (such as pressure and flow velocity) at different locations; this high-resolution analysis method helps to accurately evaluate the specific conditions of the lesion, thereby providing a more accurate basis for the diagnosis of vascular malformations, the formulation of treatment plans, and the prediction of surgical risks. Description of the Drawings

[0053] Figure 1 is a flowchart of the lumped model simulation method for the cerebrovascular malformation vascular structure provided by the embodiment of the present invention;

[0054] Figure 2a is a schematic diagram of the MRA scan provided by the embodiment of the present invention;

[0055] Figure 2b is a schematic diagram of threshold segmentation of the MRA image provided by the embodiment of the present invention;

[0056] Figure 2c is a schematic diagram of the cross-sectional view after segmentation provided by the embodiment of the present invention;

[0057] Figure 2d is a schematic diagram of the coronal view after segmentation provided by the embodiment of the present invention;

[0058] Figure 2e is a schematic diagram of the sagittal view after segmentation provided by the embodiment of the present invention;

[0059] Figure 2f is a schematic diagram of the three-dimensional reconstruction of the segmented three-dimensional lesion provided by the embodiment of the present invention;

[0060] Figure 3a is a schematic diagram of the flow velocity distribution provided by the embodiment of the present invention;

[0061] Figure 3b is a schematic diagram of the pressure distribution provided by the embodiment of the present invention;

[0062] Figure 4 is a block diagram of the lumped model simulation system for the cerebrovascular malformation vascular structure provided by the embodiment of the present invention.

[0063] Reference Signs:

[0064] 1. Image acquisition module, 2. Model construction module, 3. Parameter calibration module, 4. Skeleton extraction module, 5. Skeleton extraction module, 6. Feature extraction module. Detailed Embodiments

[0065] To make the objectives, technical solutions, and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. In addition, in the following descriptions, the descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present invention.

[0066] An AVM is a vascular malformation with abnormal development of intracerebral arteriovenous, where blood vessels are intertwined and coiled with each other, and rupture and bleeding may occur, seriously endangering the lives of patients. The main treatment methods include: Observation: For AVMs without symptoms or with low risks, regular monitoring can be selected. Surgical resection: Removing the AVM through surgical operation is one of the methods for radical treatment. Endovascular interventional treatment: Injecting embolization substances through a catheter to block abnormal blood vessels. Radiosurgery: Using focused radiation (such as gamma knife) to shrink or close the AVM. The annual bleeding risk of untreated AVM is about 1-3%. Through effective treatment, these risks can be greatly reduced. Since AVM is a congenital disease, there is currently no known preventive method. However, through early diagnosis and appropriate treatment, the risk of complications can be reduced and the prognosis can be improved. It is very necessary for accurate identification of high bleeding risk, high surgical risk, and for making auxiliary decisions on endovascular treatment.

[0067] The diagnosis of AVM usually requires imaging examinations, including: Magnetic Resonance Imaging (MRI): Detailed display of brain structures and blood vessels. Computed Tomography (CT): Especially used when bleeding is suspected. Magnetic Resonance Angiography (MRA) and Computed Tomography Angiography (CTA): Using non-invasive methods to scan cerebral blood vessels, and contrast agents can be injected intravenously to enhance the imaging of cerebral blood vessels. Angiography: Detailed display of blood vessel structures and abnormal connections.

[0068] Please refer to Figure 1 , the first aspect of the embodiment of the present invention provides a method for simulating a lumped model of a cerebral arteriovenous malformation vascular structure, including the following steps:

[0069] Step S100, obtaining three-dimensional image data of a preset area of the human head, and identifying the cerebral arteriovenous malformation vascular mass area in the three-dimensional image data, where the three-dimensional image data is CTA image data or MRA image data.

[0070] Please refer to Figures 2a - 2f, Using medical imaging techniques such as computed tomography angiography (CTA) or magnetic resonance angiography (MRA), specific preset regions of the human head are scanned and imaged. These regions are usually determined based on clinical symptoms, doctors' experience, and preliminary diagnoses that suspect the possible presence of cerebral arteriovenous malformations (AVM). CTA can clearly show the morphology and direction of blood vessels with the help of X-rays and contrast agents, and shows good results for lesions such as vascular calcification; MRA uses the principle of magnetic resonance to image blood vessels without contrast agents, and is especially good at soft tissue contrast, clearly showing the relationship between blood vessels and surrounding brain tissue.

[0071] After the scanning is completed, the obtained three-dimensional image data carries rich blood vessel information. To facilitate subsequent processing, storage, and communication, it is necessary to convert and save it as a DICOM format file. As the standard format for medical digital imaging and communication, DICOM not only contains the pixel data of the image, but also includes the patient's basic information, examination details, and image technical parameters, etc., ensuring compatibility and interoperability between different medical devices and software systems, enabling the smooth flow of image data among various departments in the hospital, and providing a solid foundation for subsequent diagnosis and research.

[0072] Through professional image recognition software or algorithms, the cerebral arteriovenous malformation vascular mass region is accurately locked in the vast amount of three-dimensional image data. Based on the unique morphological characteristics of the AVM vascular mass, such as the degree of blood vessel tortuosity, abnormal changes in blood vessel diameter, and disordered blood flow signals, etc., it is accurately distinguished from normal brain tissue and other vascular structures, preparing for the subsequent construction of a targeted blood vessel model.

[0073] Step S200, based on the cerebral arteriovenous malformation vascular mass region, construct a three-dimensional vascular structure model, and obtain the three-dimensional spatial coordinate data of each voxel in the cerebral arteriovenous malformation vascular mass region in the model.

[0074] Based on the identified cerebral arteriovenous malformation vascular mass region, use computer-aided design (CAD) technology or professional medical modeling software to carefully construct a three-dimensional vascular structure model, accurately constructing the spatial morphology, branch direction, and intricate connection relationships of the blood vessels. During the construction process, at the same time, obtain the three-dimensional spatial coordinate data of each blood vessel in the cerebral arteriovenous malformation vascular mass region in the model. The coordinate data provides an indispensable geometric framework for subsequent accurate calculation of blood vessel parameters and analysis of hemodynamic characteristics.

[0075] Step S300, calibrate the inlet and outlet of the three-dimensional vascular structure model and the corresponding pressure setting value and / or flow rate setting value.

[0076] After the model is constructed, it is further refined by calibrating the inlet and outlet of the three-dimensional vascular structure model. The inlet and outlet determine the flow path of the three-dimensional vascular structure model and directly affect the accuracy of hemodynamic simulation. At the same time, based on clinical experience, preliminary analysis of previous imaging data, and reference to similar case studies, corresponding pressure set values and / or flow rate set values are set for these inlets and outlets, which provides the initial boundary conditions for subsequent blood flow simulation and enables the simulation process to better fit the actual physiological situation and truly reflect the blood flow state in the AVM vessels.

[0077] Step S400: Based on the three-dimensional vascular structure model, several skeleton points of the blood vessels are extracted to obtain a microscopic vascular network.

[0078] Based on the constructed three-dimensional vascular structure model, a skeletonization algorithm is used to deeply explore the internal structure of the blood vessels. By performing erosion operations to remove the boundaries of the blood vessels and gradually retaining the central axis of the blood vessels, several skeleton points of the blood vessels are then extracted. After connecting adjacent skeleton points, a microscopic vascular network is formed. Compared with the complete vascular model, the microscopic vascular network concisely and intuitively shows the topological structure of the blood vessels, not only reducing the complexity of subsequent calculations but also providing key clues for in-depth analysis of the microscopic blood flow characteristics of the blood vessels, which helps to reveal the blood flow laws of AVM vascular lesions at the microscopic level.

[0079] Step S500: Based on the three-dimensional space coordinate data, the length and average diameter of each blood vessel in the microscopic vascular network are obtained, and the flow resistance of the blood vessels is calculated.

[0080] Relying on the three-dimensional space coordinate data obtained in the previous steps, precise geometric measurements are performed on each blood vessel in the three-dimensional vascular structure model. The length and average diameter of each blood vessel are calculated using mathematical methods. According to the principles of fluid mechanics, the flow resistance of a blood vessel is directly proportional to the length of the blood vessel and inversely proportional to the fourth power of the blood vessel radius (as described by Poiseuille's law). Combining the physical properties of blood (such as viscosity), these geometric parameters are substituted into the corresponding calculation formulas to accurately calculate the flow resistance of the blood vessels. Vascular resistance is one of the key factors affecting blood flow velocity and pressure distribution.

[0081] Step S600: Based on the pressure set value, flow rate set value of the three-dimensional vascular structure model, and the flow resistance of each blood vessel, the blood flow velocity and pressure distribution data of each blood vessel are calculated through a lumped parameter model to obtain the hemodynamic characteristics of the cerebral arteriovenous malformation vessels.

[0082] With these key elements, namely, the pressure set value, flow rate set value, and the just-calculated flow resistance for each blood vessel, the hemodynamic simulation process can be carried out. Based on the lumped parameter model, the complex vascular system is simplified into a set of concentrated parameters for description, closely combining the characteristics of blood vessel resistance, elasticity, etc. with variables such as blood flow velocity and pressure. Substitute these parameters into the lumped parameter equations, and through mathematical operations, simulate and calculate the blood flow velocity and pressure distribution data for each blood vessel. Finally, through the collation and analysis of these data, the hemodynamic characteristics of the cerebral arteriovenous malformation blood vessels are obtained. These characteristics cover key information such as the speed and direction of blood flow in the blood vessels, the high and low pressure, the direction of blood flow, and the mutual influence between different blood vessels, providing extremely valuable scientific basis for clinicians to diagnose the condition of AVM and formulate treatment plans.

[0083] In a specific implementation manner of an embodiment of the present invention, the process of identifying the cerebral arteriovenous malformation vascular mass region in the three-dimensional image data in step S100 may be as follows: Start the 3D Slicer software, click the "Add Data" button on the main interface, and select the "Add DICOM Data" option; Browse and select the corresponding Dicom image folder in the pop-up window to load the patient's image data. Ensure that the Dicom image data is correctly loaded, and check whether the slices of each perspective of the patient's whole-body image (including axial, coronal, and sagittal planes) are complete and error-free. After the image data is loaded, use the "SliceView" window of 3D Slicer to check the location of the patient's lesion by scrolling the image slices. Use the zoom in, zoom out, and move tools in the software to adjust the image, which can more clearly locate the lesion area and accurately confirm the serial number of the image slice where the target lesion is located and the corresponding three-dimensional position range. After confirming the lesion area, use the "Threshold Segmentation" module in 3D Slicer to segment the vascular structure. The specific operations include: Enter the "Segment Editor" module, click the "Add" button to create a new segmentation area and name it "vascular structure". Select the "Threshold" tool and set the upper and lower threshold ranges (adjust according to the image gray value to ensure that the vascular structure is completely covered and the background tissue is not mis-segmented). Click the "Apply" button to generate the threshold-segmented vascular area, and use the three-dimensional view to check the segmentation effect. If an error is found, it can be corrected by adjusting the threshold or using the "Paint" and "Erase" tools. After the overall segmentation of the vascular structure is completed, switch to the "Scissors" tool to cut and segment the malformed vascular mass (such as the AVM area). Perform a geometric boundary extraction operation on the segmented vascular structure, switch to the "Segmentations" module, and select the malformed vascular mass segmentation area just generated. Click the "Export to Files" option to export the segmented vascular geometric structure as a standard three-dimensional model (such as STL or OBJ format). Enable the "Smooth Surface" option in the export settings to ensure that the exported geometric boundary model has good smoothness and accuracy.

[0084] Import the mesh file into CAD software to model and reconstruct the segmented blood vessels, forming a three-dimensional structural model of the blood vessels. View the three-dimensional view based on the blood vessel mesh file to check whether the model is intact and confirm whether the segmented blood vessel structure is clear and without missing parts. Perform smoothing processing, and use the smoothing tool in CAD software to optimize the imported mesh to remove possible noise or irregular regions. Repair incomplete parts. When observing missing or discontinuous parts in the mesh, use the repair tool to supplement these regions. Optimize the details as needed. Improve the accuracy and usability of the model by modifying the details of the mesh (such as densifying the mesh, optimizing the edge smoothness, etc.). Finally, save it in the medical image NifTI format.

[0085] Further, constructing a three-dimensional structural model of the blood vessels in step S200 and obtaining its three-dimensional spatial coordinate data includes:

[0086] Step S210, extract the image volume data and its physical dimensions of the three-dimensional structural model of the blood vessels in NifTI format based on Python software.

[0087] Write the model into a preset Python program, establish a three-dimensional space coordinate system according to the actual scale, use the nibabel library to load the medical image file in NIfTI format, and extract the volume data and its physical dimensions of the image.

[0088] Use the nibabel library in Python to load the medical image file in NIfTI format, which has powerful file reading and parsing capabilities. When using this library to load a NIfTI file, the volume data of the image can be extracted from it. The volume data refers to the set of values of all pixel points in three-dimensional space, and these values represent the imaging characteristics of the blood vessels at different positions, such as gray values, etc. They are the basic raw materials for constructing the three-dimensional model; the physical dimension information of the image can also be extracted, including the size of each pixel in the actual space, such as the length unit in the x, y, and z-axis directions, usually measured in millimeters, etc.

[0089] Further, extracting the image volume data and its physical dimensions of the three-dimensional structural model of the blood vessels in NifTI format in step S210 includes:

[0090] Step S211, use the nib.load() method to load the NIfTI file at the preset path and read it as a NIfTI image object.

[0091] The 3D vascular structure model file in NIfTI (Neuroimaging Informatics Technology Initiative) format is a commonly used format in the field of medical imaging. It is often used to store neuroimaging data such as brain data. Similar to the DICOM format, it also contains rich image information. It not only has pixel data reflecting the image content, but also covers key physical parameters such as spatial coordinates and image resolution. These information are crucial for constructing an accurate 3D model.

[0092] After performing the nib.load() operation, the binary data file is read and converted into a NIfTI image object, which contains various information of the image.

[0093] Step S212, obtain the volume data of the image by calling the get_fdata() method of the NIfTI image object, and return a NumPy array containing pixel values.

[0094] By calling the get_fdata() method of this object, the key volume data of the 3D model is obtained and returned in the form of a NumPy array. Each element in the returned NumPy array corresponds to a pixel value in the image. The pixel value contains the imaging feature information of the blood vessel at different positions in the 3D space, such as grayscale value, etc. Based on the distribution of the pixel values and the combination with the physical size, the 3D shape of the blood vessel is obtained.

[0095] Step S213, use the header.get_zooms() method to extract the actual physical size of the voxel from the image header information, and return a tuple. The tuple represents the intervals of each voxel in the X-axis, Y-axis, and Z-axis directions.

[0096] Through the header.get_zooms() method, the NIfTI image header information is mined to extract the actual physical size of the voxel, and a tuple is returned. The three elements in the tuple respectively represent the intervals of each voxel in the X-axis, Y-axis, and Z-axis directions, usually measured in actual length units such as millimeters. These size information are the bridge to accurately convert 2D pixel information into 3D spatial coordinates. Knowing the actual size of the voxel, the true coordinates of each point in the 3D space coordinate system can be accurately calculated based on the pixel positions in the image volume data, thus constructing a 3D model that accurately reflects the actual blood vessel structure.

[0097] Furthermore, extracting several skeleton points of the blood vessel in step S400 to obtain the microscopic blood vessel network includes:

[0098] Step S410: Use a three-dimensional skeletonization algorithm to extract the skeleton points of each blood vessel in the three-dimensional blood vessel structure model, and convert the coordinates of the skeleton points into spatial coordinates in combination with the voxel spacing.

[0099] The three-dimensional skeletonization algorithm is used to analyze complex three-dimensional blood vessel structures. Facing a three-dimensional blood vessel model composed of numerous voxels, through a series of complex operations, the redundant boundary information of the blood vessels is gradually removed, and only the most core central axis part is retained. The key nodes on the central axis are the skeleton points to be extracted. Taking a common skeletonization algorithm based on morphological erosion operation as an example, it starts from the edge of the blood vessel and erodes inward layer by layer according to the set rules until only a continuous central line representing the blood vessel's trend remains. The discrete points on this line are the skeleton points.

[0100] Furthermore, the use of the three-dimensional skeletonization algorithm in step S410 to extract the skeleton points of each blood vessel in the three-dimensional blood vessel structure model includes:

[0101] Step S411: Remove the boundary of the blood vessel through erosion operation and retain the central axis of the blood vessel.

[0102] Erosion is a basic morphological image processing operation. In the context of extracting blood vessel skeleton points, it processes the three-dimensional blood vessel structure model voxel by voxel according to a specific structuring element (usually a small pixel set, such as a 3x3 or 5x5 square). During this process, the structuring element slides on the image. When all voxels within the area it covers belong to the blood vessel, the central voxel is retained; otherwise, it is removed.

[0103] As the erosion operation continues, the boundary of the blood vessel is continuously thinned until finally only a continuous central axis representing the blood vessel's trend remains. This central axis removes the interference caused by complex factors such as blood vessel wall thickness and reflects the basic shape and trend of the blood vessel. For example, in the study of cerebral arteriovenous malformation blood vessels, the malformed blood vessels often have complex and tortuous shapes. The central axis obtained through erosion enables researchers to quickly grasp the main context of the blood vessels, providing a clear framework for subsequent in-depth analysis of blood flow characteristics inside the blood vessels and construction of a precise microscopic blood vessel network.

[0104] Step S412: Based on the KD tree, obtain the adjacent skeleton points of the skeleton points and construct skeleton line segments.

[0105] The KD tree (K-Dimensional Tree) is a data structure used for efficiently storing and retrieving multi-dimensional data. In this scenario, first, the coordinate data of all the skeleton points obtained after the erosion operation is used as input to construct a KD tree. The construction process is like carefully building a multi-layer bookshelf. Starting from the initial skeleton point data set, the data set is recursively divided into two parts according to different dimensions (usually according to the coordinate axes, such as first by the X-axis, then by the Y-axis, and then by the Z-axis, and so on in a cycle). Each division divides the data set into left and right parts until each node contains only one data point or meets a pre-set threshold (for example, the number of skeleton points contained in a node does not exceed 3, etc.). The KD tree constructed in this way can organize the skeleton point data in an efficient manner, facilitating the subsequent rapid search for adjacent skeleton points.

[0106] Furthermore, obtaining adjacent skeleton points of the skeleton points based on the KD tree in step S412 and constructing skeleton line segments includes:

[0107] Step S412a, obtain the coordinate data set of the skeleton points after the erosion operation, and repeatedly perform median splitting on the coordinate data set by selecting a coordinate dimension until each subset contains only one data point or is less than the preset data point threshold, obtaining a KD tree.

[0108] First, the coordinate information of the skeleton points remaining after the erosion operation needs to be collected, and these coordinates form the original data set for constructing the subsequent KD tree. Then, a coordinate dimension is selected to start the division. For example, start from the X-axis dimension. Calculate the median of the X-axis coordinates of all the skeleton points. This median is like a dividing line that divides the entire coordinate data set into two parts, such that the X-axis coordinate values of the skeleton points in the left subset are all less than the median, and the X-axis coordinate values of the skeleton points in the right subset are greater than or equal to the median.

[0109] After the first division is completed, the above operations need to be repeated for the left and right subsets after division. However, the dimension selection will switch according to a certain rule during the next division. For example, for the second division, the Y-axis is selected for the left subset and the Z-axis is selected for the right subset for median splitting, and so on in a cycle, continuously subdividing the data set. At this time, a multi-layer nested KD tree is constructed, storing the skeleton point coordinate data in an efficient and orderly manner, facilitating the subsequent rapid search for adjacent skeleton points.

[0110] Step S412b, based on each node of the KD tree, compare with the coordinate value of the current skeleton point to obtain several candidate adjacent skeleton points that are closest to the current skeleton point.

[0111] When it is necessary to find the adjacent skeleton points of a certain current skeleton point, the exploration journey starts from the top of the KD tree - the root node. Compare the coordinate values of the current skeleton point with the splitting point on the splitting dimension represented by the root node. For example, if the root node is split according to the X-axis, then check whether the X-axis coordinate of the current skeleton point is less than or greater than or equal to the value of this splitting point. Based on the comparison result, it can be determined whether the current skeleton point should search for closer skeleton points in the left subtree or the right subtree of the KD tree, and decide which side to go according to the size of the coordinates.

[0112] Continue to go deeper along the selected subtree path. Every time a new node is reached, repeat the above comparison operation to continuously narrow the search range and gradually approach the several nodes closest to the current skeleton point. Here, it is set to find several candidate adjacent skeleton points that are the closest. This quantity can be preset according to the actual algorithm requirements and computing resources. For example, find the 3 closest or 5 closest nodes, etc.

[0113] Step S412c, sort based on the distance values between the candidate adjacent skeleton points and the current skeleton point, determine several adjacent skeleton points of the current skeleton point, and record the corresponding association relationships.

[0114] After obtaining the candidate adjacent skeleton points, they cannot be directly recognized as the final adjacent skeleton points and need to be further screened. At this time, according to a distance metric rule, the most commonly used is the Euclidean distance formula (in three-dimensional space, the Euclidean distance between two points is equal to the square root of the sum of the squares of the coordinate differences on the X, Y, and Z axes), calculate the actual distance between each candidate adjacent skeleton point and the current skeleton point, and then sort the candidate adjacent skeleton points in ascending order of the distance value, with the closest one ranked at the front.

[0115] From the sorted candidate list, select several skeleton points in the front as the finally determined adjacent skeleton points. This quantity also depends on the actual situation. At the same time, the connection relationships between these adjacent skeleton points and the current skeleton point need to be recorded to facilitate knowing which points should be connected together when constructing the skeleton line segments later. These association relationship information will become an important clue for constructing the microvascular network.

[0116] Step S412d, traverse all skeleton points, and connect two adjacent skeleton points according to the association relationship between each current skeleton point and its adjacent skeleton points to obtain the skeleton line segments of the skeleton points.

[0117] To construct a complete skeleton line segment network, all the previously extracted skeleton points need to be processed one by one. Starting from the first skeleton point, according to the adjacent skeleton point association relationship recorded in step S312c, find the adjacent skeleton point. Connect the two adjacent skeleton points with a straight line, and this straight line is the skeleton line segment. After traversing all the skeleton points, numerous skeleton line segments intersect with each other, clearly showing the basic topological structure of the blood vessels, providing an intuitive and crucial geometric model basis for subsequent in-depth research on the hemodynamic characteristics of blood vessels, calculation of blood vessel parameters, etc.

[0118] Further, after obtaining the skeleton line segments of the skeleton points in step S412d, it also includes:

[0119] Step S412e, storing the coordinate value data of all skeleton points and the coordinate data of the skeleton line segments connected by the two endpoints of the skeleton points into a data matrix.

[0120] After the construction of the skeleton line segments, although there is visually a microscopic blood vessel network that can reflect the basic topological structure of the blood vessels, for subsequent more in-depth and efficient analysis, simulation of the blood vessel model, and interaction with other algorithm modules, the relevant data needs to be systematically integrated. The coordinate values of each skeleton point are the cornerstones for constructing the microscopic blood vessel network, which accurately locate the orientation of the blood vessels in three-dimensional space. These coordinate values usually exist in the form of a three-dimensional array, corresponding to the X, Y, and Z axis coordinates in space. Storing them into the data matrix is like registering the "node" information of the microscopic blood vessel network one by one. For example, if the coordinate of a skeleton point is (x1, y1, z1), according to the established storage rules, this triple will be accurately placed in the corresponding row and column positions of the data matrix, ensuring that subsequent queries about the position of a specific skeleton point or further geometric calculations based on these coordinates (such as calculating the length and angle of the blood vessel) can be quickly achieved. The skeleton line segment is a key element connecting the skeleton points and visualizing the blood vessel morphology. Storing the coordinate data of the skeleton line segment is relatively more complex because a line segment usually requires at least two endpoint coordinates to be determined. For each skeleton line segment, the coordinates of its two endpoints (assumed to be (x1, y1, z1) and (x2, y2, z2)) need to be completely recorded into the data matrix. This not only involves the storage of the endpoint coordinates themselves but may also include some additional information describing the line segment attributes, such as the length of the line segment (which can be calculated by the distance formula between two points) and the direction vector of the line segment (determined by the difference in endpoint coordinates).

[0121] The organization form of the data matrix is usually designed according to the subsequent algorithm requirements and computational convenience. Generally speaking, different skeleton points or skeleton line segments may correspond between rows, and columns are used to distinguish different types of data such as coordinate values and attribute values. Such a regular structure makes data retrieval, update, and batch processing efficient. For example, when calculating the total length of all skeleton line segments, only by extracting the data column related to the line segment length according to the established column index rule and using a simple summation function can the result be quickly obtained, greatly improving the efficiency of data analysis and providing a solid data foundation for subsequent in-depth research on the hemodynamic characteristics of the cerebral arteriovenous malformation blood vessels and the construction of blood vessel models.

[0122] Furthermore, calculating the blood flow velocity and pressure distribution data of each blood vessel in step S600 based on the lumped parameter model includes:

[0123] Step S611, construct a blood vessel network with the starting point, ending point coordinates and blood flow resistance of each blood vessel, and at the same time initialize the connection relationship and pressure value of each node.

[0124] Using the starting point, ending point coordinates of each blood vessel obtained previously and the accurately calculated blood flow resistance information, start to build a blood vessel network model. Combine these blood vessels according to their actual spatial layout and connection relationship in the human head to outline a network prototype reflecting the architecture of the cerebral arteriovenous malformation blood vessels. When constructing the blood vessel network, the connection relationship of each node (blood vessel branch point or intersection point) should be clarified. By setting a suitable data structure, such as an adjacency matrix or a linked list, to record this connection information for subsequent quick query and calculation. For example, if node A is connected to blood vessels a, b, and c, then these associations will be recorded accordingly in the data structure, providing a basis for subsequent analysis of the shunt and confluence of blood flow at the node. Assign an initial pressure value to each node, and the initial pressure value is set based on the previous imaging data, clinical experience, and research on similar cases. For example, for nodes close to the arterial end, a relatively high initial pressure value is given based on the known physiological blood pressure range of the human body; for nodes close to the venous end, a relatively low pressure value is set.

[0125] Step S612, based on the pipe network theory, construct a linear equation system of the blood vessel network containing blood vessel pressure to obtain the pressure values of all nodes in the blood vessel network.

[0126] Based on the pipe network theory, the complex vascular network is analogized to an electrical circuit network. In an electrical circuit, Kirchhoff's Current Law (KCL) and Kirchhoff's Voltage Law (KVL) are used to describe the distribution laws of current and voltage. Similarly, in a vascular network, a mathematical model can be constructed based on the law of conservation of mass (equivalent to KCL, that is, the blood flow rate into a node is equal to the blood flow rate out of the node) and the law of conservation of energy (similar to KVL, considering that the pressure difference in a blood vessel segment follows a certain law, and blood always flows from a high-pressure area to a low-pressure area, and the magnitude of the flow rate is proportional to the pressure difference and inversely proportional to the resistance of the blood vessel). Represent the pressure of each node in the vascular network as an unknown quantity. According to the above conservation laws, an equation is listed for each node. The left side of the equation usually contains the ratio of the pressure difference between this node and adjacent nodes (combined with the blood vessel resistance, similar to the resistance voltage division relationship in an electrical circuit), and the right side may contain known node pressures (if any) or other constant terms. In this way, the equations of all nodes form a linear equation system containing vascular pressures, and its form can usually be simply expressed in matrix form as: A×p = b, where A is the coefficient matrix, containing the relationship between each node and its adjacent nodes; p is the column vector containing the pressures of all nodes; b is the constant term on the right side, containing the pressure values of known nodes. By solving this linear equation system, the pressure values of all nodes in the vascular network can be obtained.

[0127] Step S613: Based on the pressure values of all nodes, calculate the pressure difference between two adjacent nodes, calculate the flow rate and velocity of each blood vessel segment according to the blood flow resistance and the pressure difference, and determine the blood flow direction in each blood vessel according to the positive or negative velocity.

[0128] After obtaining the pressure values of all nodes, it becomes natural to calculate the pressure difference between two adjacent nodes. This is similar to measuring the water pressure difference between adjacent nodes in a city water supply pipeline, and it can be obtained by simply subtracting the pressure values of two adjacent nodes. For example, the pressure of node A is P A , and the pressure of adjacent node B is P B , then the pressure difference ΔP = P A - P B . These pressure differences are the driving force for blood flow in blood vessels and provide key inputs for subsequent calculation of blood flow parameters.

[0129] According to Ohm's Law in fluid mechanics (analogous in the context of blood vessels), the relationship between the flow rate Q, the pressure difference ΔP, and the resistance R is: Q = ΔP / R. That is, given the pressure difference and the resistance, the flow rate of each blood vessel segment can be calculated. And the velocity v is the flow rate Q divided by the cross-sectional area S (assuming the cross-sectional area of the blood vessel is known and can be calculated from parameters such as the previously measured blood vessel diameter), that is, v = Q / S. Through successive derivations and calculations, the velocity of each blood vessel segment can be obtained.

[0130] Determine the blood flow direction in each blood vessel according to the positive or negative value of the flow velocity. If the calculated flow velocity is positive, it indicates that the blood flow direction is consistent with the pre-set positive direction; if the flow velocity is negative, it means that the flow direction is opposite to the expectation. At this time, the starting point and the ending point need to be exchanged, and the absolute value of the flow velocity is taken. The positive or negative value of the flow velocity provides an intuitive basis for judging the blood flow direction and ensures the accurate simulation of the blood flow situation in the blood vessel.

[0131] Store the pressure and flow velocity values of each section of the blood vessel skeleton in a data matrix. Then, display them in the form of pictures through a visualization method. Use the scatter() method in the matplotlib library to create a 3D scatter plot to show the distributions of pressure, flow velocity, and flow rate respectively. In each graph, set the color of the points according to the physical quantities (pressure, flow velocity, flow rate) to make the distribution of the numerical values clearer and more understandable. The results are as Figure 3a and Figure 3b shown.

[0132] Correspondingly, please refer to Figure 4 , the second aspect of the embodiment of the present invention provides a lumped model simulation system for a cerebral arteriovenous malformation vascular structure, including:

[0133] An image acquisition module 1, which is used to acquire three-dimensional image data of a preset area of the human head and identify the cerebral arteriovenous malformation vascular mass area in the three-dimensional image data. The three-dimensional image data is CTA image data or MRA image data;

[0134] A model construction module 2, which is used to construct a three-dimensional vascular structure model based on the cerebral arteriovenous malformation vascular mass area and obtain the three-dimensional spatial coordinate data of each voxel in the cerebral arteriovenous malformation vascular mass area in the model;

[0135] A parameter calibration module 3, which is used to calibrate the inlet and outlet of the three-dimensional vascular structure model and the corresponding pressure setting value and / or flow velocity setting value;

[0136] A skeleton extraction module 4, which is used to extract a number of skeleton points of the blood vessel based on the three-dimensional vascular structure model to obtain a microscopic blood vessel network;

[0137] A resistance calculation module 5, which is used to obtain the length and average diameter of each blood vessel in the microscopic blood vessel network based on the three-dimensional spatial coordinate data and calculate the flow resistance of the blood vessel;

[0138] A feature extraction module 6, which is used to calculate the blood flow velocity and pressure distribution data of each blood vessel based on the pressure setting value, flow velocity setting value and flow resistance of each blood vessel of the three-dimensional vascular structure model, and obtain the hemodynamic characteristics of the cerebral arteriovenous malformation blood vessels.

[0139] Accordingly, a third aspect of the embodiments of the present invention provides an electronic device, including: at least one processor; and a memory 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 above-mentioned lumped model simulation method for cerebral arteriovenous malformation vascular structures.

[0140] Accordingly, a fourth aspect of the embodiments of the present invention provides a computer-readable storage medium, on which computer instructions are stored, and when the instructions are executed by a processor, the above-mentioned lumped model simulation method for cerebral arteriovenous malformation vascular structures is implemented.

[0141] The embodiments of the present invention aim to protect a lumped model simulation method and system for cerebral arteriovenous malformation vascular structures. The method includes the following steps: obtaining three-dimensional image data of a preset region of a human head, identifying the cerebral arteriovenous malformation vascular mass region in the three-dimensional image data, and the three-dimensional image data being CTA image data or MRA image data; constructing a three-dimensional vascular structure model based on the cerebral arteriovenous malformation vascular mass region, and obtaining the three-dimensional spatial coordinate data of each voxel in the cerebral arteriovenous malformation vascular mass region in the model; calibrating the inlet and outlet of the three-dimensional vascular structure model and the corresponding pressure setting value and / or flow rate setting value; extracting a number of skeleton points of the blood vessels based on the three-dimensional vascular structure model to obtain a microscopic vascular network; obtaining the length and average diameter of each blood vessel in the microscopic vascular network based on the three-dimensional spatial coordinate data, and calculating the flow resistance of the blood vessels; calculating the blood flow velocity and pressure distribution data of each blood vessel through a lumped parameter model based on the pressure setting value, flow rate setting value of the three-dimensional vascular structure model and the flow resistance of each blood vessel, so as to obtain the hemodynamic characteristics of the cerebral arteriovenous malformation blood vessels. The above technical solution has the following effects:

[0142] 1. Simplify the traditional CFD technology, save time and computing power, reduce labor costs, and have a large number of automated calculation processes. The traditional computational fluid dynamics (CFD) simulation method usually requires very complex mesh generation, long-time calculation and a large amount of manual intervention, which is not only time-consuming but also has high requirements for computing resources. However, the present invention simplifies the blood vessel morphology and automates the calculation process by introducing a lumped parameter model and a skeletonization algorithm. Compared with the traditional CFD method, the computational complexity and time are significantly reduced, the demand for high-performance computing resources is reduced, and a large amount of labor costs are saved through an automated segmentation and modeling process, making the application of hemodynamic simulation in clinical practice more efficient and reducing the implementation threshold;

[0143] 2. It can obtain quantitative data, get more refined data inside the malformation mass, and retain the structural form of the malformation mass. The obtained data has a higher dimension. Through precise image segmentation and skeletonization processing, the present invention can accurately extract the three-dimensional geometric shape of the blood vessels of cerebral arteriovenous malformation (AVM), while retaining the detailed structure of the malformation mass. Compared with traditional methods, it can not only provide hemodynamic simulation results, but also provide more refined quantitative data inside the malformation mass, such as parameters such as blood flow velocity, pressure, and flow resistance of each tiny blood vessel segment. The high-dimensional data can more comprehensively reflect the complexity of AVM, help doctors better understand blood flow behavior and pathological characteristics, and provide a more accurate decision-making basis for the design of treatment plans;

[0144] 3. Other hemodynamic parameters can be further obtained from a large amount of quantitative original data. Based on the precise calculation of basic blood flow parameters such as blood flow velocity, pressure, and flow resistance, a large amount of original data has been obtained, which is not only the basis for hemodynamic analysis, but also can be further used to deduce other related hemodynamic parameters, such as shear stress of blood vessels, blood flow volatility, and pressure difference between arteries and veins. The above-mentioned deep-level parameters are crucial for clinical diagnosis and personalized adjustment of treatment plans. Especially in the process of formulating treatment plans for AVM, they can help doctors more accurately evaluate the vulnerability of diseased blood vessels, predict rupture risks, and achieve dynamic monitoring;

[0145] 4. It has strong scalability. By combining clinical data with other technologies such as machine learning, it can help with the diagnosis, rupture prediction, and treatment decision-making of arteriovenous malformations. It can combine clinical data and other advanced technologies (such as machine learning, big data analysis, etc.) to further improve the accuracy of diagnosis and treatment. And through the integration with clinical data, multi-dimensional analysis of AVM can be achieved, not only providing hemodynamic simulation results, but also using machine learning models to classify the malformation mass, predict rupture risks, and evaluate treatment effects, etc. Through this data-driven method, more intelligent support can be provided in the formulation of treatment plans, improve the effect of personalized medicine, and promote the early diagnosis and prevention of cerebrovascular diseases such as AVM;

[0146] 5. Promote the implementation of precision medicine and personalized treatment. By using the three-dimensional vascular imaging data of individual patients and precise hemodynamic simulation, treatment plans can be customized for each patient. Based on the calculation of multiple hemodynamic parameters such as blood flow velocity, pressure, and flow resistance, the severity of vascular lesions and their impact on surrounding tissues can be more accurately evaluated. Based on the above quantitative data, doctors can provide more personalized treatment plans for each patient, including choosing the most suitable surgical method or interventional treatment, and can also provide real-time data support for the evaluation of treatment effects and postoperative monitoring, promoting the application of precision medicine and personalized treatment in cerebrovascular diseases.

[0147] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0148] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0149] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implement the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0150] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0151] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: still can modify the specific implementation manners of the present invention or make equivalent substitutions, and any modification or equivalent substitution that does not depart from the spirit and scope of the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. A method for simulating a lumped model of a cerebral arteriovenous malformation vascular structure, characterized in that: The steps include: Acquire three-dimensional image data of a preset area of ​​a human head, and identify a brain arteriovenous malformation vessel mass area in the three-dimensional image data, wherein the three-dimensional image data is CTA image data or MRA image data; Based on the cerebral arteriovenous malformation vascular mass region, a vascular three-dimensional structure model is constructed, and the three-dimensional spatial coordinate data of each voxel of the cerebral arteriovenous malformation vascular mass region in the model is obtained; Calibrate the inlet and outlet of the three-dimensional structure model of the blood vessel and the corresponding pressure setting value and / or flow rate setting value; Based on the three-dimensional structure model of the blood vessel, a number of skeleton points of the blood vessel are extracted to obtain a microscopic blood vessel network; Based on the three-dimensional spatial coordinate data, obtaining the length and average diameter of each blood vessel in the microscopic blood vessel network, and calculating the flow resistance of the blood vessels; Based on the pressure setting value, flow rate setting value and flow resistance of each blood vessel of the three-dimensional structure model of the blood vessel, the blood flow velocity and pressure distribution data of each blood vessel are calculated through a lumped parameter model to obtain the hemodynamic characteristics of the cerebral arteriovenous malformation blood vessel.

2. The method for simulating the lumped model of cerebral arteriovenous malformation vascular structure according to claim 1, characterized in that: After acquiring the three-dimensional image data of the preset area of ​​the human head, the method further includes: The three-dimensional image data is saved as a NifTI format file.

3. The method for simulating the lumped model of cerebral arteriovenous malformation vascular structure according to claim 1, characterized in that: The step of constructing a three-dimensional structure model of a blood vessel and obtaining three-dimensional spatial coordinate data thereof includes: The image volume data and physical dimensions of the three-dimensional structural model of the blood vessel in NifTI format are extracted based on Python software.

4. The method for simulating the lumped model of cerebral arteriovenous malformation vascular structure according to claim 3, characterized in that: The method of extracting the image volume data and physical dimensions of the three-dimensional structure model of the blood vessel in NifTI format based on Python software includes: Use the nib.load() method to load the NIfTI file in the preset path and read it as a NIfTI image object; Get the volume data of the image by calling the get_fdata() method of the NIfTI image object, and return a NumPy array containing pixel values; The actual physical size of the voxel is extracted from the image header information using the header.get_zooms() method, and a tuple is returned, which indicates the interval of each voxel in the X-axis, Y-axis, and Z-axis directions.

5. The method for simulating the lumped model of cerebral arteriovenous malformation vascular structure according to claim 4, characterized in that: The extraction of several skeleton points of blood vessels to obtain a microscopic blood vessel network includes: A three-dimensional skeletonization algorithm is used to extract skeleton points of each blood vessel in the three-dimensional structure model of the blood vessel, and the coordinates of the skeleton points are converted into spatial coordinates in combination with voxel intervals.

6. The method for simulating the lumped model of cerebral arteriovenous malformation vascular structure according to claim 5, characterized in that: The step of extracting skeleton points of each blood vessel in the three-dimensional structure model of the blood vessel using a three-dimensional skeletonization algorithm includes: removing the boundary of the blood vessel by an erosion operation and retaining the central axis of the blood vessel; Based on the KD tree, adjacent skeleton points of the skeleton point are obtained to construct skeleton line segments.

7. The method for simulating the lumped model of cerebral arteriovenous malformation vascular structure according to claim 6, characterized in that: The step of acquiring adjacent skeleton points of the skeleton point based on the KD tree and constructing skeleton line segments includes: Obtaining a coordinate data set of the skeleton points after the corrosion operation, selecting a coordinate dimension to repeatedly perform median segmentation on the coordinate data set until each subset contains only one data point or is less than a preset data point threshold, thereby obtaining the KD tree; Based on each node of the KD tree, compare the coordinate value of the current skeleton point with that of the current skeleton point to obtain a number of candidate adjacent skeleton points that are closest to the current skeleton point; Sorting the candidate adjacent skeleton points based on the distance values ​​between the candidate adjacent skeleton points and the current skeleton point, determining a number of adjacent skeleton points of the current skeleton point, and recording corresponding association relationships; All skeleton points are traversed, and according to the association relationship between each current skeleton point and its adjacent skeleton point, two adjacent skeleton points are connected to obtain the skeleton line segment of the skeleton point.

8. The method for simulating the lumped model of cerebral arteriovenous malformation vascular structure according to claim 7, characterized in that: After obtaining the skeleton line segments of the skeleton points, the method further includes: The coordinate value data of all the skeleton points and the coordinate data of the skeleton line segment connected by two skeleton point endpoints are stored in a data matrix.

9. The method for simulating the lumped model of the vascular structure of a brain arteriovenous malformation according to any one of claims 1 to 8, characterized in that: The blood flow velocity and pressure distribution data of each blood vessel are calculated based on the lumped parameter model, including: The starting point and end point coordinates and blood flow resistance of each blood vessel are used to construct a vascular network, and the connection relationship and pressure value of each node are initialized; Based on the pipe network theory, a vascular network linear equation group including vascular pressure is constructed to obtain the pressure values ​​of all nodes in the vascular network; Based on the pressure values ​​of all the nodes, the pressure difference between two adjacent nodes is calculated, the flow rate and flow velocity of each blood vessel are calculated according to the blood flow resistance and the pressure difference, and the flow direction of the blood flow in each blood vessel is determined according to the positive and negative flow velocity.

10. A lumped model simulation system for cerebral arteriovenous malformation vascular structure, characterized in that: include: An image acquisition module, which is used to acquire three-dimensional image data of a preset area of ​​a human head and identify a brain arteriovenous malformation vessel mass area in the three-dimensional image data, wherein the three-dimensional image data is CTA image data or MRA image data; A model building module, which is used to build a three-dimensional vascular structure model based on the cerebral arteriovenous malformation vascular mass area, and obtain the three-dimensional spatial coordinate data of each voxel of the cerebral arteriovenous malformation vascular mass area in the model; A parameter calibration module, which is used to calibrate the inlet and outlet of the three-dimensional structure model of the blood vessel and the corresponding pressure setting value and / or flow rate setting value; A skeleton extraction module, which is used to extract several skeleton points of the blood vessels based on the three-dimensional structure model of the blood vessels to obtain a microscopic blood vessel network; a resistance calculation module, which is used to obtain the length and average diameter of each blood vessel in the microscopic blood vessel network based on the three-dimensional spatial coordinate data, and calculate the flow resistance of the blood vessels; A feature extraction module is used to calculate the blood flow velocity and pressure distribution data of each blood vessel based on the pressure setting value, flow rate setting value and flow resistance of each blood vessel of the three-dimensional structure model of the blood vessel and the lumped parameter model to obtain the hemodynamic characteristics of the cerebral arteriovenous malformation blood vessel.

Images