Coronary artery hemodynamic information prediction method, device, equipment, medium and product

Abstract

The invention discloses a coronary artery hemodynamic information prediction method, device and equipment, a medium and a product, and relates to the field of biomechanics, and the method comprises the steps: obtaining clinical data, and segmenting the clinical data to obtain a coronary artery three-dimensional model and a heart three-dimensional model; constructing a human body blood circulation system lumped parameter model; determining a personalized human body blood circulation system lumped parameter model of the patient after assignment; constructing a 0-3-dimensional myocardial mechanical coupling model; constructing a first database; obtaining personalized systolic function parameters and diastolic function parameters of the patient based on the first mapping network; constructing a blood vessel 0-3-dimensional coupling model; constructing a second database; obtaining personalized coronary flow field information of the patient based on a second mapping network; the personalized coronary flow field information comprises a pressure field, a velocity field and concentration polarization characteristics of low-density lipoprotein on a blood vessel wall surface. Through network architecture innovation, the bottleneck of the model in the aspects of efficiency and personalization is solved, and the diagnosis precision of the coronary heart disease is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of biomechanics, and in particular to a coronary blood flow hemodynamic information prediction method, device, equipment, medium and product. BACKGROUND

[0002] Coronary heart disease is one of the most important cardiovascular diseases threatening human health, and its main pathological mechanism is myocardial perfusion deficiency or myocardial ischemic necrosis caused by coronary stenosis or rupture of atherosclerotic plaques. At present, the risk assessment of coronary stenosis in clinical practice mainly depends on the occurrence site and morphological characteristics of coronary stenosis (such as the assessment of the degree of vascular occlusion by stenosis rate and stenosis length, and the development of corresponding treatment strategies) obtained from imaging information, and further considering family history, hypertension and other risk factors. The above method has good clinical operability, but this method is based on imaging information only, and is highly subjective, and cannot accurately assess the degree of influence of stenosis on the distal blood flow. Therefore, identifying blood flow hemodynamic factors closely related to plaque progression / rupture, reasonably assessing the quantitative relationship between stenosis and myocardial ischemia, and accurately and quickly screening high-risk stenosis lesions in need of treatment are urgent clinical problems. Although the coronary flow reserve fraction (FFR) can assess the functional state of coronary stenosis, the clinical measurement is an invasive operation, and the cost is high. Neither the imaging assessment nor the functional assessment method can provide blood flow hemodynamic information such as wall shear stress and low-density lipoprotein concentration distribution, which is closely related to the development trend and stability of the plaque, which to some extent limits its application value in plaque risk assessment. In particular, considering the differences and physiological variability of the physiological / pathological state of patients, how to comprehensively and accurately assess the individualized coronary blood flow hemodynamic state of the patient without invasive operation is another urgent clinical problem. The coronary evaluation system of the present application adopts a double mapping network architecture, uses more comprehensive patient individualized information as the input features of the machine learning model, realizes the purpose of quickly and comprehensively predicting related blood flow hemodynamic factors through non-invasive imaging examination (cCTA), and thus makes it possible for the "imaging + mechanical diagnosis" mode to further change into the "imaging + mechanical + machine learning diagnosis" mode. SUMMARY

[0003] The purpose of the present application is to provide a coronary blood flow hemodynamic information prediction method, device, equipment, medium and product, which solves the bottleneck of the model in efficiency and individualization through network architecture innovation, and improves the diagnosis accuracy of coronary heart disease.

[0004] To achieve the above-mentioned purpose, the present application provides the following solutions:

[0005] In a first aspect, the present application provides a coronary blood flow hemodynamic information prediction method, comprising:

[0006] Acquire clinical data; the clinical data includes: coronary CTA imaging data and individualized data; the individualized data includes: physiological and pathological data of patient heart rate, blood pressure, pulse wave velocity, blood low-density lipoprotein concentration, and patient personal information;

[0007] The coronary CTA imaging data is segmented to obtain a three-dimensional model of the coronary arteries and a three-dimensional model of the heart.

[0008] Constructing a lumped parameter model of the human blood circulation system;

[0009] The individualized data is converted into model parameters using a parameter optimization algorithm;

[0010] The model parameters are assigned to the human circulatory system lumped parameter model to obtain the patient-personalized human circulatory system lumped parameter model after assignment.

[0011] A 0-3D myocardial mechanical coupling model is constructed based on the aforementioned three-dimensional cardiac model and the assigned value of the patient's personalized human blood circulation system lumped parameter model.

[0012] Based on the aforementioned 0-3D myocardial mechanics coupling model, finite element simulation of the myocardium is performed to generate solid mechanics simulation results for a three-dimensional model of the heart.

[0013] A first database is constructed; the first database includes: three-dimensional cardiac models of a batch of patients, solid mechanical simulation results of each three-dimensional cardiac model, a first mapping network, and individualized patient data; the first mapping network is trained from the solid mechanical simulation results of each three-dimensional cardiac model in the first database, the three-dimensional cardiac models of a batch of patients, and the individualized patient data;

[0014] Personalized cardiac systolic and diastolic function parameters of patients are obtained based on the trained first mapping network.

[0015] A 0-3D coupled model of the coronary arteries is constructed based on the aforementioned 3D model of the coronary arteries and the assigned lumped parameter model of the human blood circulation system.

[0016] Based on the 0-3D coupled model and the patient's personalized cardiac systolic and diastolic function parameters, hemodynamic simulation is performed to generate high-precision flow field feature data and low-density lipoprotein vessel wall polarization data.

[0017] A second database is constructed; the second database includes: individualized parameters corresponding to the patient's physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, low-density lipoprotein vessel wall polarization feature data, and a second mapping network; the second mapping network is trained from the individualized parameters corresponding to the patient's physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, and low-density lipoprotein vessel wall polarization feature data in the second database; the individualized parameters corresponding to the patient's physiological and pathological indicators include: cardiac systolic function parameters and diastolic function parameters corresponding to the patient obtained from the first mapping network;

[0018] Personalized coronary flow field information of patients is obtained based on a trained second mapping network; the personalized coronary flow field information includes pressure field, velocity field, and concentration polarization characteristics of low-density lipoprotein on the vessel wall.

[0019] Optionally, segmenting the coronary CTA imaging data to obtain a three-dimensional model of the coronary arteries and a three-dimensional model of the heart specifically includes the following steps:

[0020] The coronary CTA imaging data is imported into open-source image processing software for three-dimensional reconstruction to obtain the reconstructed three-dimensional image.

[0021] The locations of the myocardial tissue are circled in the three views of the reconstructed three-dimensional image;

[0022] The heart region is segmented in the reconstructed 3D image based on the location of myocardial tissue.

[0023] Threshold segmentation was used to obtain the complete coronary artery region;

[0024] The location of the coronary arteries was marked on the complete coronary artery region using a three-dimensional ball brush.

[0025] The coronary artery locations are then refined and cropped to obtain the final segmentation results of the heart and coronary vessels.

[0026] Optionally, the coronary hemodynamics information prediction method further includes, between the steps of "segmenting the coronary CTA imaging data to obtain a three-dimensional model of the coronary vessels and a three-dimensional model of the heart" and "constructing a lumped parameter model of the human circulatory system":

[0027] The coronary artery 3D model and the heart 3D model are preprocessed, including format conversion and time-space discretization.

[0028] Optionally, the specific method for converting the individualized data into model parameters using a parameter optimization algorithm is as follows:

[0029] Based on the patient's individualized data and the numerical calculation results of the lumped parameter model of the human circulatory system, a parameter optimization objective function is established. The global optimization algorithm is used to solve the parameter optimization objective function to obtain cardiovascular function parameters that correspond to the patient's actual physiological condition.

[0030] Optionally, the loss function of the first mapping network is:

[0031] Loss1=L data1 +L physicsU +L physicsT ;

[0032] Where Loss1 is the total loss of the first mapping network; L data1 For data loss; L physicsU For the loss of strain energy; L physicsT This refers to stress loss.

[0033] Optionally, the loss function of the second mapping network is:

[0034] Loss2 = Loss data2 +ω*L physcis

[0035]

[0036] Where Loss2 is the loss of the second mapping network; L data2 The loss of the data is represented by ω; the weighting coefficient is represented by L. physics N represents the residual term of the hemodynamic equation. data U represents the number of data points used in the training process. true The simulated hemodynamic values ​​are used as the gold standard for training; x j Spatial coordinates of the point cloud of the blood vessel wall; t j For time; U pred The values ​​of hemodynamic parameters predicted by the network; i is the governing equation number; e i This is the value obtained by subtracting both sides of the three-dimensional hemodynamic equation.

[0037] Secondly, this application provides a coronary hemodynamic information prediction device, comprising:

[0038] The data acquisition module is used to acquire clinical data, which includes coronary CTA imaging data and individualized data. The individualized data includes physiological and pathological data of patient heart rate, blood pressure, pulse wave velocity, blood low-density lipoprotein concentration, and patient personal information.

[0039] The segmentation module is used to segment the coronary CTA imaging data to obtain a three-dimensional model of the coronary arteries and a three-dimensional model of the heart.

[0040] The module for constructing a lumped parameter model of the human circulatory system is used to construct a lumped parameter model of the human circulatory system.

[0041] The parameter conversion module is used to convert the individualized data into model parameters using a parameter optimization algorithm;

[0042] The parameter assignment module is used to assign the model parameters to the human blood circulation system lumped parameter model to obtain the assigned patient-personalized human blood circulation system lumped parameter model.

[0043] The 0-3D myocardial mechanics model construction module is used to construct a 0-3D myocardial mechanics coupling model based on the aforementioned three-dimensional cardiac model and the assigned patient-personalized human blood circulation system lumped parameter model.

[0044] The first simulation module is used to generate solid mechanics simulation results of a three-dimensional cardiac model based on the 0-3D myocardial mechanics coupling model.

[0045] A first database construction module is used to construct a first database; the first database includes: a batch of three-dimensional cardiac models of patients, solid mechanical simulation results of each three-dimensional cardiac model, a first mapping network, and individualized patient data; the first mapping network is trained from the solid mechanical simulation results of each three-dimensional cardiac model in the first database, the batch of three-dimensional cardiac models of patients, and the individualized patient data;

[0046] The cardiac systolic and diastolic function parameter acquisition module is used to obtain the patient's cardiac systolic and diastolic function parameters based on the trained first mapping network;

[0047] The coronary artery 0-3D coupling model construction module is used to construct a coronary artery 0-3D coupling model based on the coronary artery 3D model and the assigned human blood circulation system lumped parameter model.

[0048] The second simulation module performs hemodynamic simulation based on the 0-3D coupled model and the patient's personalized cardiac systolic and diastolic function parameters, generating high-precision flow field feature data and low-density lipoprotein vessel wall polarization data.

[0049] The second database construction module is used to construct a second database. The second database includes: individualized parameters corresponding to patient physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, low-density lipoprotein vessel wall polarization feature data, and a second mapping network. The second mapping network is trained from the individualized parameters corresponding to patient physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, and low-density lipoprotein vessel wall polarization feature data in the second database.

[0050] A personalized coronary flow field information acquisition module is used to obtain the patient's personalized coronary flow field information based on a trained second mapping network; the personalized coronary flow field information includes pressure field, velocity field, and concentration polarization characteristics of low-density lipoprotein on the vessel wall.

[0051] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the coronary hemodynamic information prediction method described in any one of the above-mentioned methods.

[0052] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the coronary hemodynamic information prediction method described in any one of the above descriptions.

[0053] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the coronary hemodynamic information prediction method described in any one of the above-mentioned methods.

[0054] According to the specific embodiments provided in this application, this application has the following technical effects:

[0055] This application provides a method, device, equipment, medium, and product for predicting coronary hemodynamic information. Through a two-level physical constraint neural network, namely a first mapping network and a second mapping network, more reliable and comprehensive mathematical modeling can be performed on clinically collected patient-specific information to achieve higher diagnostic accuracy for coronary heart disease. By introducing a myocardial mechanics model, the fidelity of the patient-specific model is enhanced. By designing a substance transport model, the model can predict the deposition effect of lipid substances on the blood vessel wall, and combined with blood flow information, they can serve as diagnostic indicators for coronary heart disease. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 A flowchart illustrating a method for predicting coronary hemodynamic information according to an embodiment of this application;

[0058] Figure 2 This is a schematic diagram of a closed-loop circulation system according to an embodiment of this application;

[0059] Figure 3 This is a schematic diagram of the second mapping network according to an embodiment of this application;

[0060] Figure 4 This is a schematic diagram of the structure of a computer device provided in one embodiment of this application;

[0061] Figure 5 This application proposes a dual-mapping network architecture. Detailed Implementation

[0062] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0063] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0064] In one exemplary embodiment, such as Figure 1 As shown, a method for predicting coronary hemodynamic information is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, it includes the following steps 101 to 113. Wherein:

[0065] Step 101: Obtain clinical data; the clinical data includes: coronary CTA imaging data and individualized data; the individualized data includes: physiological and pathological data of patient heart rate, blood pressure, pulse wave velocity, blood low-density lipoprotein concentration, and patient personal information.

[0066] Step 102: Segment the coronary CTA imaging data to obtain a three-dimensional model of the coronary vessels and a three-dimensional model of the heart.

[0067] The specific segmentation process is as follows:

[0068] S1021: Import all of the patient's CTA imaging data into open-source image processing software, such as 3D Slicer;

[0069] S1022: Reconstruct 3D images;

[0070] S1023: Circle the location of the myocardial tissue in the three views of the reconstructed three-dimensional image, and segment the heart region in the three-dimensional image;

[0071] S1024: Threshold segmentation identifies the complete coronary artery region;

[0072] S1025: Use a 3D ball brush to mark the location of coronary arteries, and use tools such as scissors to trim and refine the labeling results of the coronary artery markings;

[0073] S1026: Outputs the segmentation results of the heart and coronary arteries.

[0074] Step 103: Construct a lumped parameter model of the human blood circulation system.

[0075] Specifically, the lumped parameter model of the human circulatory system is a 0-dimensional model; see [link to specific structure] for details. Figure 2 The part represented by symbols such as impedance and capacitance is used to construct a population-averaged blood circulation system model.

[0076] Step 104: Use a parameter optimization algorithm to convert the individualized data into model parameters.

[0077] Specifically, the lumped parameter model of the human circulatory system can open a parameter input port. Combined with the gradient descent method, based on the numerical calculation results of the model and the individualized data obtained in step S101, a parameter optimization objective function is established. Through numerical iteration, the individualized calibration of the model's parameters is achieved, so that the cardiovascular function parameters of the model are individually matched with the patient. The above objective function is optimized and solved by the modified gradient descent method to obtain the patient's individualized model parameters (that is, the patient's personalized indicators such as blood pressure and heart rate collected in S101 can be converted into model parameters through the parameter optimization algorithm, assigned to this model, and then hemodynamic simulation is performed).

[0078] Step 105: Assign the model parameters to the human circulatory system lumped parameter model to obtain the assigned patient-specific human circulatory system lumped parameter model.

[0079] Step 106: Construct a 0-3D myocardial mechanical coupling model based on the aforementioned three-dimensional cardiac model and the assigned value of the patient's personalized human blood circulation system lumped parameter model.

[0080] Step 107: Based on the aforementioned 0-3D myocardial mechanics coupling model, perform finite element simulation of the myocardium to generate solid mechanics simulation results for the three-dimensional model of the heart.

[0081] Step 108: Construct a first database; the first database includes: a batch of three-dimensional cardiac models of patients, solid mechanical simulation results of each three-dimensional cardiac model, a first mapping network, and individualized patient data; the first mapping network is trained from the solid mechanical simulation results of each three-dimensional cardiac model in the first database, the batch of three-dimensional cardiac models of patients, and individualized patient data.

[0082] The specific process is as follows: A first database is constructed, which contains: segmented cardiac models of a batch of patients (i.e., three-dimensional cardiac models), personalized physiological and pathological information of patients (collected from S101), and solid mechanics simulation results for each three-dimensional cardiac model. These simulation results include: cardiac contractile function parameters (E... lva ) and diastolic function parameters (E) lvb The purpose of performing solid mechanics simulation on a three-dimensional model of the patient's heart is to calculate and obtain cardiac systolic function parameters (E). lva ) and diastolic function parameters (E) lvb The calculation method for these two parameters includes the following steps:

[0083] S1: A passive constitutive model is constructed based on the Holazapfel-Ogden (HO) model. The discrete fiber differentiation method is used for fiber dispersion modeling. Simulation is performed using an open-source solver. The dynamic equations are:

[0084] Where M is the mass matrix, x is the displacement field, C is the damping matrix, K(x) is a nonlinear matrix that depends on the current configuration, and F ext The external force is represented by the damping matrix: C = aM, where a is the damping coefficient.

[0085] S2: The cardiovascular system is simplified into four parts: heart, systemic circulation, pulmonary circulation and coronary circulation. A limited number of model parameters are used to express the cardiovascular function corresponding to each part, which is the lumped parameter model of the circulatory system constructed in S105.

[0086] S3: Embed the three-dimensional model of the heart into the lumped parameter model of the circulatory system to form a closed-loop circulatory system (see details). Figure 2 It should be noted that the figure shows only a three-dimensional model of the left ventricle. Depending on the needs, the model can also be a three-dimensional model of a two-chamber or a four-chamber ventricle.

[0087] S4: Couple the active contraction mechanical model based on myocardial fiber structure and the passive stretching constitutive model of myocardium based on fiber distribution with the cardiac chamber model for solution;

[0088] S5: Get E lva and E lvb , as the parameter assignment for the 0-3D coupling model of the coronary artery.

[0089] S6: Construct a mapping network based on deep learning methods, and obtain E lva E lvb The parameters are mapped and trained with a three-dimensional cardiac model and patient physiological and pathological parameters to obtain a first mapping network. This first mapping network can be used to analyze E. lva E lvb Predicting parameters significantly improves efficiency.

[0090] The first mapping network may include: an input layer, a first fully connected layer, a second fully connected layer, a third fully connected layer, a fourth fully connected layer, and an output layer; the input layer, the first fully connected layer, the second fully connected layer, the third fully connected layer, the fourth fully connected layer, and the output layer are connected sequentially. The input consists of a three-dimensional cardiac model and patient-specific parameters, and the output is E. lva E lvb parameter.

[0091] The loss function used when training this first mapping network is:

[0092] Loss = L data1 +L physicsU +L physicsT ;

[0093] Where Loss is the total loss of the first mapping network; L data1 For data loss; L physicsU For the loss of strain energy; L physicsT This is stress loss;

[0094] Efficient prediction can be achieved through the first mapping network described above in this application. If E is obtained using the finite element simulation method... lva and E lvb Indicators require a lot of computing resources, but by using deep learning, once the network is trained, the above indicators can be predicted quickly. These indicators are all patient-specific indicators that characterize the patient's cardiac function and can be input into the S105 model for coronary hemodynamic simulation.

[0095] Step 109: Obtain the patient's personalized cardiac systolic and diastolic function parameters based on the trained first mapping network.

[0096] Step 110: Construct a 0-3D coupled model of the coronary artery based on the three-dimensional model of the coronary artery and the assigned lumped parameter model of the human blood circulation system.

[0097] The construction process is as follows:

[0098] The three-dimensional model of the coronary artery (based on S102 segmentation and reconstruction) is coupled with the zero-dimensional model of the cardiovascular system (based on S105 calibration) to form a 0-3 dimensional coupled model.

[0099] The 0-3D coupled model is used to accurately simulate the hemodynamic environment of the coronary artery, while the 0D part (the 0D model in this application, also known as the lumped parameter model, including the heart, systemic circulation, and pulmonary circulation) provides system-level blood flow boundary conditions for the 3D model.

[0100] The parameters assigned to the 0-dimensional part of the model are consistent with the corresponding parameters of the patient-individualized model in S105, which have been corrected by the optimization algorithm.

[0101] Step 111: Based on the coronary 0-3D coupling model and the patient's personalized cardiac systolic and diastolic function parameters, perform hemodynamic simulation to generate high-precision flow field feature data and low-density lipoprotein vessel wall polarization data.

[0102] That is, based on the 0-3 coupling model obtained from S110, batch personalized coronary hemodynamic simulations are performed to generate high-precision flow field characteristic data and low-density lipoprotein vessel wall polarization data, as detailed below:

[0103] 1) Utilize open-source simulation software such as OpenFoam to perform two-order format three-dimensional hemodynamic calculations;

[0104] 2) Perform hemodynamic simulation to obtain hemodynamic parameters of the three-dimensional flow field of the coronary artery, including velocity field, pressure field, vorticity, helicity, and LDL (low-density lipoprotein) concentration polarization region.

[0105] Step 112: Construct a second database; the second database includes: individualized parameters corresponding to the patient's physiological and pathological indicators, three-dimensional models of coronary vessels of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, low-density lipoprotein vessel wall polarization feature data, and a second mapping network; the second mapping network is trained from the individualized parameters corresponding to the patient's physiological and pathological indicators, three-dimensional models of coronary vessels of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, and low-density lipoprotein vessel wall polarization feature data in the second database.

[0106] A data mapping architecture based on deep learning algorithms, namely the second mapping network, is constructed. Unlike existing neural network model architectures, this application not only uses the geometric features of the patient's three-dimensional coronary artery model as input to the second mapping network, but also uses the individualized physiological and pathological information of the patient collected by S101 as model input to train the second mapping network. The second mapping network uses the residuals of the Navier-Stokes equation (continuity equation + momentum equation) and the convection-diffusion equation of low-density lipoprotein in blood vessels as loss function terms, as shown in the following formula:

[0107] Loss = L data2 +ω*L physcis

[0108]

[0109] L data The error between the predicted and actual values ​​obtained from the hemodynamic data (velocity and pressure) trained by the second mapping network, L physics The error (physical information loss error) is obtained by incorporating the predicted value into the hemodynamic equation, with weighting coefficients ω and N. data e represents the number of randomly selected data points. i (x j , t j The results are obtained by subtracting the left side from the right side of the governing equation.

[0110] The input to the second mapping network is the spatial coordinates (x, y, z...) of the point cloud, time (t), patient blood pressure, heart rate, pulse wave velocity, and blood component indicators (such as...). Figure 3 As shown in the figure, the output is the coronary flow field information.

[0111] Step 113: Obtain personalized coronary flow field information of the patient based on the trained second mapping network; the personalized coronary flow field information includes pressure field, velocity field and concentration polarization characteristics of low-density lipoprotein on the vessel wall.

[0112] Implementing steps 101 to 113 above, through a two-level physical constraint neural network (see...) Figure 5 The first and second mapping networks can perform more reliable and comprehensive mathematical modeling of personalized patient information collected clinically, so as to achieve higher diagnostic accuracy for coronary heart disease. By introducing a myocardial mechanics model, the fidelity of the personalized patient model is enhanced. By designing a material transport model, the model can predict the deposition effect of lipids on the blood vessel wall, and combined with blood flow information, they can serve as diagnostic indicators for coronary heart disease.

[0113] Based on the same inventive concept, this application also provides a coronary hemodynamic information prediction device for implementing the coronary hemodynamic information prediction method described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the coronary hemodynamic information prediction device provided below can be found in the limitations of the coronary hemodynamic information prediction method described above, and will not be repeated here.

[0114] In one exemplary embodiment, a coronary hemodynamic information prediction device is provided, comprising:

[0115] The data acquisition module is used to acquire clinical data, which includes coronary CTA imaging data and individualized data. The individualized data includes physiological and pathological data of patient heart rate, blood pressure, pulse wave velocity, blood low-density lipoprotein concentration, and patient personal information.

[0116] The segmentation module is used to segment the coronary CTA imaging data to obtain a three-dimensional model of the coronary arteries and a three-dimensional model of the heart.

[0117] The module for constructing a lumped parameter model of the human circulatory system is used to construct a lumped parameter model of the human circulatory system.

[0118] The parameter conversion module is used to convert the individualized data into model parameters using a parameter optimization algorithm;

[0119] The parameter assignment module is used to assign the model parameters to the human blood circulation system lumped parameter model to obtain the assigned patient-personalized human blood circulation system lumped parameter model.

[0120] The 0-3D myocardial mechanics model construction module is used to construct a 0-3D myocardial mechanics coupling model based on the aforementioned three-dimensional cardiac model and the assigned patient-personalized human blood circulation system lumped parameter model.

[0121] The first simulation module is used to generate solid mechanics simulation results of a three-dimensional cardiac model based on the 0-3D myocardial mechanics coupling model.

[0122] A first database construction module is used to construct a first database; the first database includes: a batch of three-dimensional cardiac models of patients, solid mechanical simulation results of each three-dimensional cardiac model, a first mapping network, and individualized patient data; the first mapping network is trained from the solid mechanical simulation results of each three-dimensional cardiac model in the first database, the batch of three-dimensional cardiac models of patients, and the individualized patient data;

[0123] The cardiac systolic and diastolic function parameter acquisition module is used to obtain the patient's cardiac systolic and diastolic function parameters based on the trained first mapping network;

[0124] The coronary artery 0-3D coupling model construction module is used to construct a coronary artery 0-3D coupling model based on the coronary artery 3D model and the assigned human blood circulation system lumped parameter model.

[0125] The second simulation module performs hemodynamic simulation based on the coronary 0-3D coupled model and the patient's personalized cardiac systolic and diastolic function parameters, generating high-precision flow field feature data and low-density lipoprotein vessel wall polarization data.

[0126] The second database construction module is used to construct a second database. The second database includes: individualized parameters corresponding to patient physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, low-density lipoprotein vessel wall polarization feature data, and a second mapping network. The second mapping network is trained from the individualized parameters corresponding to patient physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, and low-density lipoprotein vessel wall polarization feature data in the second database.

[0127] A personalized coronary flow field information acquisition module is used to obtain the patient's personalized coronary flow field information based on a trained second mapping network; the personalized coronary flow field information includes pressure field, velocity field, and concentration polarization characteristics of low-density lipoprotein on the vessel wall.

[0128] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 4 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores and processes data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When executed by the processor, the computer program implements a method for predicting coronary hemodynamic information.

[0129] Those skilled in the art will understand that Figure 4 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0130] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0131] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0132] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0133] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0134] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0135] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0136] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for predicting coronary hemodynamic information, characterized in that, The method for predicting coronary hemodynamic information includes: Acquire clinical data; the clinical data includes: coronary CTA imaging data and individualized data; the individualized data includes: physiological and pathological data of patient heart rate, blood pressure, pulse wave velocity, blood low-density lipoprotein concentration, and patient personal information; The coronary CTA imaging data is segmented to obtain a three-dimensional model of the coronary arteries and a three-dimensional model of the heart. Constructing a lumped parameter model of the human blood circulation system; The individualized data is converted into model parameters using a parameter optimization algorithm; The model parameters are assigned to the human circulatory system lumped parameter model to obtain the patient-personalized human circulatory system lumped parameter model after assignment. A 0-3D myocardial mechanical coupling model is constructed based on the aforementioned three-dimensional cardiac model and the assigned value of the patient's personalized human blood circulation system lumped parameter model. Based on the aforementioned 0-3D myocardial mechanics coupling model, finite element simulation of the myocardium is performed to generate solid mechanics simulation results for a three-dimensional model of the heart. A first database is constructed; the first database includes: three-dimensional cardiac models of a batch of patients, solid mechanical simulation results of each three-dimensional cardiac model, a first mapping network, and individualized patient data; the first mapping network is trained from the solid mechanical simulation results of each three-dimensional cardiac model in the first database, the three-dimensional cardiac models of a batch of patients, and the individualized patient data; Personalized cardiac systolic and diastolic function parameters of patients are obtained based on the trained first mapping network. A 0-3D coupled model of the coronary arteries is constructed based on the aforementioned 3D model of the coronary arteries and the assigned lumped parameter model of the human blood circulation system. Based on the 0-3D coupled model and the patient's personalized cardiac systolic and diastolic function parameters, hemodynamic simulation is performed to generate high-precision flow field feature data and low-density lipoprotein vessel wall polarization data. A second database is constructed; the second database includes: individualized parameters corresponding to the patient's physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, low-density lipoprotein vessel wall polarization feature data, and a second mapping network; the second mapping network is trained from the individualized parameters corresponding to the patient's physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, and low-density lipoprotein vessel wall polarization feature data in the second database; the individualized parameters corresponding to the patient's physiological and pathological indicators include: cardiac systolic function parameters and diastolic function parameters corresponding to the patient obtained from the first mapping network; Personalized coronary flow field information of patients is obtained based on a trained second mapping network; the personalized coronary flow field information includes pressure field, velocity field, and concentration polarization characteristics of low-density lipoprotein on the vessel wall.

2. The method for predicting coronary hemodynamic information according to claim 1, characterized in that, The process of segmenting the coronary CTA imaging data to obtain a three-dimensional model of the coronary arteries and a three-dimensional model of the heart specifically includes the following steps: The coronary CTA imaging data is imported into open-source image processing software for three-dimensional reconstruction to obtain the reconstructed three-dimensional image. The locations of the myocardial tissue are circled in the three views of the reconstructed three-dimensional image; The heart region is segmented in the reconstructed 3D image based on the location of myocardial tissue. Threshold segmentation was used to obtain the complete coronary artery region; The location of the coronary arteries was marked on the complete coronary artery region using a three-dimensional ball brush. The coronary artery locations are then refined and cropped to obtain the final segmentation results of the heart and coronary vessels.

3. The method for predicting coronary hemodynamic information according to claim 1, characterized in that, The coronary hemodynamics information prediction method further includes the following steps between "segmenting the coronary CTA imaging data to obtain a three-dimensional model of the coronary vessels and a three-dimensional model of the heart" and "constructing a lumped parameter model of the human circulatory system": The coronary artery 3D model and the heart 3D model are preprocessed, including format conversion and time-space discretization.

4. The method for predicting coronary hemodynamic information according to claim 1, characterized in that, The specific method for converting the individualized data into model parameters using the parameter optimization algorithm is as follows: Based on the patient's individualized data and the numerical calculation results of the lumped parameter model of the human circulatory system, a parameter optimization objective function is established. The global optimization algorithm is used to solve the parameter optimization objective function to obtain cardiovascular function parameters that correspond to the patient's actual physiological condition.

5. The method for predicting coronary hemodynamic information according to claim 1, characterized in that, The loss function of the first mapping network is: Loss1=L data1 +L physicsU +L physicsT ; Where Loss1 is the total loss of the first mapping network; L data1 For data loss; L physicsU For the loss of strain energy; L physicsT This refers to stress loss.

6. The method for predicting coronary hemodynamic information according to claim 1, characterized in that, The loss function of the second mapping network is: Loss2=Loss data2 +ω*L physcis Where Loss2 is the loss of the second mapping network; L data2 The loss of the data is represented by ω; the weighting coefficient is represented by L. physics N represents the residual term of the hemodynamic equation. data U represents the number of data points used in the training process. true The simulated hemodynamic values ​​are used as the gold standard for training; x j Spatial coordinates of the point cloud of the blood vessel wall; t j For time; U pred The values ​​of hemodynamic parameters predicted by the network; i is the governing equation number; e i This is the value obtained by subtracting both sides of the three-dimensional hemodynamic equation.

7. A coronary hemodynamics information prediction device, characterized in that, The coronary hemodynamic information prediction device includes: The data acquisition module is used to acquire clinical data, which includes coronary CTA imaging data and individualized data. The individualized data includes physiological and pathological data of patient heart rate, blood pressure, pulse wave velocity, blood low-density lipoprotein concentration, and patient personal information. The segmentation module is used to segment the coronary CTA imaging data to obtain a three-dimensional model of the coronary arteries and a three-dimensional model of the heart. The module for constructing a lumped parameter model of the human circulatory system is used to construct a lumped parameter model of the human circulatory system. The parameter conversion module is used to convert the individualized data into model parameters using a parameter optimization algorithm; The parameter assignment module is used to assign the model parameters to the human blood circulation system lumped parameter model to obtain the assigned patient-personalized human blood circulation system lumped parameter model. The 0-3D myocardial mechanics model construction module is used to construct a 0-3D myocardial mechanics coupling model based on the aforementioned three-dimensional cardiac model and the assigned patient-personalized human blood circulation system lumped parameter model. The first simulation module is used to generate solid mechanics simulation results of a three-dimensional cardiac model based on the 0-3D myocardial mechanics coupling model. A first database construction module is used to construct a first database; the first database includes: a batch of three-dimensional cardiac models of patients, solid mechanical simulation results of each three-dimensional cardiac model, a first mapping network, and individualized patient data; the first mapping network is trained from the solid mechanical simulation results of each three-dimensional cardiac model in the first database, the batch of three-dimensional cardiac models of patients, and the individualized patient data; The cardiac systolic and diastolic function parameter acquisition module is used to obtain the patient's cardiac systolic and diastolic function parameters based on the trained first mapping network; A 0-3D coupled model construction module is used to construct a 0-3D coupled model of the coronary artery based on the three-dimensional model of the coronary artery and the assigned lumped parameter model of the human blood circulation system. The second simulation module performs hemodynamic simulation based on the 0-3D coupled model and the patient's personalized cardiac systolic and diastolic function parameters, generating high-precision flow field feature data and low-density lipoprotein vessel wall polarization data. The second database construction module is used to construct a second database. The second database includes: individualized parameters corresponding to patient physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, low-density lipoprotein vessel wall polarization feature data, and a second mapping network. The second mapping network is trained from the individualized parameters corresponding to patient physiological and pathological indicators, three-dimensional models of coronary arteries of a batch of patients, high-precision flow field feature data obtained from hemodynamic simulation, and low-density lipoprotein vessel wall polarization feature data in the second database. A personalized coronary flow field information acquisition module is used to obtain the patient's personalized coronary flow field information based on a trained second mapping network; the personalized coronary flow field information includes pressure field, velocity field, and concentration polarization characteristics of low-density lipoprotein on the vessel wall.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the coronary hemodynamic information prediction method according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the coronary hemodynamic information prediction method according to any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the coronary hemodynamic information prediction method according to any one of claims 1-6.

Images