LVAD multi-objective optimization method, system, equipment, storage medium and product

Abstract

The present invention discloses a LVAD multi-objective optimization method, system, device, storage medium and product, the method comprising constructing a 3D model of ventricles and aorta, constructing a ventricular motion function; adding an LVAD structure to the 3D model of ventricles and aorta according to each set of sample parameters, obtaining multiple 3D models of ventricles and aorta containing LVAD structures; meshing the 3D model of ventricles and aorta containing LVAD structures to generate a network model; based on the ventricular motion function, controlling ventricular contraction and rotational motion, calculating platelet shear stress history and platelet retention time; constructing a Kriging model according to the platelet shear stress history and platelet retention time of each set of sample parameters; and using an optimization algorithm to find the optimal set of structural parameters from the prediction results of the Kriging model. The present invention uses multiple structural parameters as optimization variables, effectively solving the problem that thrombosis cannot be effectively reduced after LVAD implantation.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical engineering, and in particular relates to a multi-objective optimization method, system, device, storage medium and product for LVAD cannula structure and flow based on parametric modeling. Background Art

[0002] Heart failure is a common cardiovascular disease that affects the quality of life and life expectancy of millions of people. The left ventricular assisted device (LVAD) is a mechanical circulatory support device that can be used to treat patients with end-stage heart failure and improve their survival rate and quality of life. However, LVAD still has some serious complications, such as stroke and device thrombosis, which limit its clinical effectiveness and safety. LVAD-related complications are related to multiple risk factors, including patient factors, clinical factors, and hemodynamic factors. Hemodynamic factors mainly include unfavorable blood flow patterns, high shear stress, stagnant areas, and nonphysiological inflow and outflow. Platelets are the main cellular components of thrombus formation. They sense abnormal hemodynamic stimulation and initiate the coagulation cascade. The structural parameters that directly affect LVAD hemodynamics include LVAD insertion position, insertion depth, insertion angle, cannula diameter, etc. At present, there are no clear medical guidelines or recommendations for the above parameters.

[0003] The formation of thrombus is closely related to platelet shear stress history (SH) and platelet residence time (RT). Platelet shear stress history SH is equal to the integral of the shear force on the platelet over time, which can evaluate the shear-induced platelet activation level related to the size of the ventricle; platelet residence time RT refers to the time that platelets stay in the left ventricular area. Higher platelet shear stress history SH and longer platelet residence time RT will increase the risk of thrombosis. With the help of computed tomography (CT) images, some researchers have carried out non-steady-state numerical calculations (CFD) of ventricular reconstruction and intraventricular blood flow patterns, and studied the influence of single factors such as LVAD insertion position, insertion depth, insertion angle, and cannula diameter on left ventricular blood flow patterns and platelet activation. However, research on a single factor cannot provide the optimal solution combination for a specific goal, such as the optimal solution combination of LVAD insertion position, insertion depth, insertion angle, insertion diameter and other parameters corresponding to the minimum platelet shear stress history SH or platelet retention time RT.

[0004] In the LVAD device, the parameters related to the risk of thrombosis include: LVAD insertion position, insertion depth, insertion angle, cannula diameter, platelet shear stress history, platelet retention time, and these parameters can be quantified. The patent document with the authorization announcement number CN114098692B discloses a left ventricular assist device implantation method based on blood flow distribution optimization, which specifically includes: calculating the patient's cardiac output; obtaining the patient's heart physiological parameters by reconstructing the aorta and coronary vessels; using the blood flow and resistance of the normal population to calculate the blood flow velocity of the normal population; simulating the LAVD installation process, importing the patient's heart physiological parameters, and calculating the patient's blood flow velocity; according to the error function, obtaining the optimal parameter value. After obtaining the optimal parameter value according to the above method, the above optimal parameter value can be used to guide the implantation and use of LAVD, so that the blood supply to each blood vessel after LAVD implantation can be optimized, the formation of vortices can be reduced, and then the formation of thrombus can be reduced, thereby reducing the chance of complications such as pulmonary embolism and cerebral infarction. However, this method still has the following problems:

[0005] (1) It is impossible to automatically realize the parametric modeling and automatic meshing of the left ventricle and different LVAD input tube shapes, insertion positions and angles. For the study of LVAD input tube shapes, insertion positions and angles, the existing numerical simulation methods require manual construction of different calculation models, and meshing them separately, and then perform numerical solutions. On this basis, a large amount of calculation data is analyzed to obtain the influence mechanism of related factors. If you want to obtain more accurate results, or obtain the optimal combination of multiple structural parameters at the same time, you often need to carry out orthogonal simulation and build more calculation models. A single simulation may take hours or even days to complete. Simply relying on manual construction of different calculation models is not only a huge workload and inefficient, but also difficult to ensure accuracy. Therefore, automatic parametric modeling and automatic meshing of the left ventricle and different LVAD input tube shapes, insertion positions and angles are the prerequisites for optimization research.

[0006] (2) It is impossible to simulate ventricular motion, and thus it is impossible to consider the effect of the rotational flow of blood in the ventricles and aorta on thrombus formation. The cardiac cycle refers to the process that the heart goes through to complete a pumping of blood. When the heart relaxes, the wall expands and blood flows from the atria to the ventricles; when the heart contracts, the ventricular wall shrinks and blood flows from the left ventricle to the aorta. The movement of the left ventricular wall can be roughly divided into three types: (a) radial movement, (b) axial twisting, and (c) circumferential movement. Accurately calculating the volume change of the ventricle during the cardiac cycle is a prerequisite for calculating the aortic blood flow and the blood flow distribution at the arterial inlet. The rotational movement of blood is closely related to the shear stress of platelets and blood vessel walls. Therefore, accurately calculating the circumferential motion speed of the ventricle is a prerequisite for calculating the platelet shear stress history SH and the minimum platelet retention time RT.

[0007] (3) LVAD insertion position, insertion depth, insertion angle, and cannula diameter are important factors that affect thrombosis, and the above method does not involve the optimization of the above variables. Current research mostly focuses on the influence mechanism of single factors such as insertion position, insertion depth, insertion angle, and cannula diameter on thrombosis. The parameter range studied is limited, and only the influence law of a single influencing factor within a certain range of variation can be obtained, but the optimal solution for the combination of influencing factors cannot be given. To obtain the optimal combination of multiple parameters, a large number of multi-factor orthogonal simulations are required, and the amount of calculation is huge. The above patent cannot realize computer automatic parametric modeling, automatic grid division, and automatic calculation. Therefore, it is impossible to realize large-scale multi-factor orthogonal simulation and obtain the LVAD multiple parameter combinations corresponding to the minimum thrombosis risk.

[0008] (4) The above method predicts the risk of thrombosis by analyzing the changes in flow loss. However, the changes in flow loss have little correlation with the risk of thrombosis. The main factors causing thrombosis are platelet shear stress history (SH) and platelet retention time (RT). The above method does not optimize the relevant parameters. Summary of the invention

[0009] The purpose of the present invention is to provide a LVAD multi-objective optimization method, system, device, storage medium and product to solve the problem that the prior art fails to simultaneously optimize multiple important factors affecting thrombosis, resulting in the inability to effectively reduce thrombosis after LVAD implantation.

[0010] The present invention solves the above technical problems through the following technical solutions: a LVAD multi-objective optimization method, comprising the following steps:

[0011] Construct a three-dimensional model of the ventricle and aorta based on cardiac CT images;

[0012] Based on the three-dimensional model of the ventricle and aorta, construct a ventricular motion function;

[0013] Selecting different groups of sample parameters from the value range of the LVAD structure parameters, adding a corresponding LVAD structure to the 3D model of the ventricle and aorta according to each group of sample parameters, obtaining a 3D model of the ventricle and aorta containing the corresponding LVAD structure, and further obtaining multiple 3D models of the ventricle and aorta containing the corresponding LVAD structure;

[0014] Mesh each 3D model of the ventricle and aorta containing the LVAD structure to generate a corresponding network model;

[0015] Based on the ventricular motion function, the ventricular contraction motion is controlled by controlling the motion of the ventricular wall nodes, and an angular velocity is applied to the ventricular wall to control the rotational motion of the ventricular wall, and the platelet shear stress history and platelet retention time under each set of sample parameters are calculated;

[0016] The Kriging model was constructed based on the platelet shear stress history and platelet retention time under different groups of sample parameters;

[0017] Different groups of structural parameters of the LVAD structure were input into the Kriging model to obtain the platelet shear stress history and platelet retention time under different groups of structural parameters;

[0018] An optimization algorithm is used to find the optimal group structure parameters from the platelet shear stress history and platelet retention time under different group structure parameters.

[0019] Furthermore, based on the three-dimensional model of the ventricle and aorta, a ventricular motion function is constructed, including:

[0020] Tracking the changes in the positions of multiple spots in different ventricular cross-sections in the three-dimensional model of the ventricle and aorta over time by two-dimensional speckle tracking echocardiography technology, and calculating the volume change of the ventricle and the angular velocity of rotation of the ventricle;

[0021] The ventricular motion function is fitted according to the volume change of the ventricle; wherein the ventricular motion function includes a ventricular contraction function and a ventricular expansion function, and the expression of the ventricular contraction function is:

[0022] M(t)=0.5*a*sin((10*π / 3)*(t+0.15));

[0023] The expression of the ventricular expansion function is:

[0024] M(t)=0.5*a*sin((2*π)*(t-0.55));

[0025] Among them, M(t) represents the volume change of the ventricle, that is, the amount of contraction or expansion; t represents time; a represents the correction coefficient; π represents pi.

[0026] Furthermore, Latin hypercube sampling method is used to select different groups of sample parameters from the value range of LVAD structural parameters.

[0027] Further, according to each set of sample parameters, a corresponding LVAD structure is added to the 3D model of the ventricle and aorta, including:

[0028] Importing the 3D model of the ventricle and aorta into UG NX software;

[0029] Extracting the ventricular surface from the 3D model of the ventricle and aorta, and dividing the cannulation area at the apex of the heart;

[0030] Determine the reference point of the LVAD structure cross-sectional primitive change, and construct a local polar coordinate system with the apex of the heart as the origin;

[0031] constructing an LVAD structural primitive in the cannulation area according to each set of sample parameters;

[0032] Setting a composite inclination angle of the LVAD structure; wherein the composite inclination angle refers to the angle between the LVAD structure and the Z direction, where the Z direction is the span direction;

[0033] Boolean operations are performed on the LVAD structure and the imported three-dimensional model of the ventricle and aorta to obtain a three-dimensional model of the ventricle and aorta including the LVAD structure.

[0034] Furthermore, the platelet shear stress history and platelet retention time under each set of sample parameters are calculated, including:

[0035] Calculate the blood flow in the ventricle and aorta at the current time step;

[0036] Release thrombus particles and calculate the movement of thrombus particles at the current time step;

[0037] Moving network nodes within the ventricular wall and fluid regions;

[0038] Calculate the blood flow conditions and thrombus particle movement conditions in the ventricle and aorta at the next time step until all thrombus particles leave the ventricle and aorta;

[0039] The platelet shear stress history and platelet retention time were statistically analyzed.

[0040] Furthermore, a Kriging model was constructed based on the platelet shear stress history and platelet retention time under different groups of sample parameters, including:

[0041] Constructing a sample data set according to the platelet shear stress history and platelet retention time under different groups of sample parameters; wherein the sample data set includes a plurality of samples, each sample includes an input amount and an output amount, the input amount is a group of sample parameters, and the output amount is the platelet shear stress history and platelet retention time under the group of sample parameters;

[0042] Construct a Kriging model and use it to calculate the distance matrix between samples to determine the spatial correlation between input quantities;

[0043] estimating parameters of the Kriging model based on the spatial correlation between input quantities;

[0044] The input amount of the sample is input into the Kriging model, and whether the Kriging model meets the accuracy requirements is determined based on the output result of the Kriging model and the output amount of the sample; if so, the Kriging model is output; if not, the number of groups of sample parameters selected from the value range of the LVAD structural parameters is increased until the Kriging model meets the accuracy requirements.

[0045] Furthermore, the optimization algorithm is a multi-objective genetic algorithm.

[0046] Based on the same concept, the present invention provides a LVAD multi-objective optimization system, the system comprising:

[0047] A first construction unit is used to construct a three-dimensional model of the ventricle and aorta according to the cardiac CT image;

[0048] A second construction unit is used to construct a ventricular motion function based on the three-dimensional model of the ventricle and aorta;

[0049] A third construction unit is used to select different groups of sample parameters from the value range of the LVAD structure parameters, and add a corresponding LVAD structure to the 3D model of the ventricle and aorta according to each group of sample parameters to obtain a 3D model of the ventricle and aorta containing the corresponding LVAD structure, and then obtain multiple 3D models of the ventricle and aorta containing the corresponding LVAD structure;

[0050] A meshing unit, used for meshing each ventricle and aorta three-dimensional model including the LVAD structure to generate a corresponding network model;

[0051] a first calculation unit, configured to control the ventricular contraction movement by controlling the movement of the ventricular wall nodes based on the ventricular motion function, and to apply an angular velocity to the ventricular wall to control the rotational movement of the ventricular wall, and calculate the platelet shear stress history and platelet retention time under each set of sample parameters;

[0052] a fourth construction unit, for constructing a Kriging model according to the platelet shear stress history and platelet retention time under different groups of sample parameters;

[0053] A second calculation unit is used to input different sets of structural parameters of the LVAD structure into the Kriging model to obtain the platelet shear stress history and platelet retention time under different sets of structural parameters;

[0054] The optimization unit is used to find the optimal group structure parameters from the platelet shear stress history and platelet retention time under different group structure parameters by using an optimization algorithm.

[0055] Based on the same concept, the present invention provides an electronic device, including a memory, a processor, and a computer program / instructions stored in the memory, wherein the processor executes the computer program / instructions to implement the LVAD multi-objective optimization method as described above.

[0056] Based on the same concept, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon, which implements the LVAD multi-objective optimization method as described above when the computer program / instruction is executed by a processor.

[0057] Based on the same concept, the present invention provides a computer program product, including a computer program / instruction, which implements the LVAD multi-objective optimization method as described above when executed by a processor.

[0058] Beneficial Effects

[0059] Compared with the prior art, the advantages of the present invention are:

[0060] The present invention uses insertion position, insertion depth, insertion angle, and LVAD structure diameter as optimization variables, and platelet shear stress history and platelet retention time as target parameters, combines the Kriging model with the optimization algorithm, and constructs a multi-objective optimization method for LVAD shape and position parameters with the lowest LVAD thrombosis risk. The method can solve the optimal combination of LVAD insertion position, insertion depth, insertion angle, and structure diameter with the lowest thrombosis risk, provide guidance for patient preoperative planning, optimize the biocompatibility of LVAD treatment, and effectively solve the problem that thrombosis cannot be effectively reduced after LVAD implantation. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only one embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 is a flow chart of a multi-objective optimization method for LVAD in an embodiment of the present invention;

[0063] Figure 2 is a three-dimensional model diagram of the ventricle and aorta in an embodiment of the present invention;

[0064] Figure 3 is a flow chart of adding an LVAD structure to a three-dimensional model of a ventricle and aorta in an embodiment of the present invention;

[0065] Figure 4 is a diagram defining the location of the LVAD structure in an embodiment of the present invention;

[0066] Figure 5 is a diagram of the LVAD structure angle positioning in an embodiment of the present invention;

[0067] Figure 6 is a schematic diagram of grid division in an embodiment of the present invention;

[0068] Figure 7 is a flow chart for calculating the platelet shear stress history and platelet retention time under each set of sample parameters in an embodiment of the present invention;

[0069] Figure 8 It is a flow chart of constructing a Kriging model in an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The following is a clear and complete description of the technical solutions in the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0072] Example 1

[0073] like Figure 1 As shown, an LVAD multi-objective optimization method provided by an embodiment of the present invention includes the following steps:

[0074] Step S1: constructing a three-dimensional model of the ventricle and aorta based on the cardiac CT image.

[0075] Based on the computed tomography images of the patient's heart, a three-dimensional model of the ventricle and aorta was constructed using the medical modeling software E3D. Figure 2 As shown, the 3D model of the ventricle and aorta is saved as an stl format file.

[0076] Step S2: constructing a ventricular motion function based on the three-dimensional model of the ventricle and aorta.

[0077] In a specific embodiment of the present invention, the specific process of constructing the ventricular motion function is: the ventricle is divided into 24 sections by two-dimensional speckle tracking echocardiography technology, a number of spots are selected on each ventricular section, and the motion trajectories of these spots are obtained by tracking the changes in the positions of these spots over time, and the real-time movement and deformation of the myocardial tissue are reconstructed, thereby obtaining the volume change (i.e., contraction movement) and rotational angular velocity of the ventricle.

[0078] The contraction movement of the ventricle is generally considered to be approximately sinusoidal movement, so the ventricular motion function can be fitted according to the volume change of the ventricle. The ventricular motion function includes the ventricular contraction function and the ventricular expansion function. The expression of the ventricular contraction function is:

[0079] M(t)=0.5*a*sin((10*π / 3)*(t+0.15)) (1)

[0080] The expression of ventricular expansion function is:

[0081] M(t)=0.5*a*sin((2*π)*(t-0.55)) (2)

[0082] Among them, M(t) represents the volume change of the ventricle, that is, the amount of contraction or expansion; t represents time; a represents the correction coefficient; π represents pi.

[0083] The ventricular contraction movement is realized through the ventricular contraction function. During the ventricular contraction movement, a rotational angular velocity is applied to the ventricular wall to realize the rotational movement or circular movement of the ventricular wall (Global Circumferential Strain, GCS), such as Figure 1 shown.

[0084] Step S3: Select different groups of sample parameters from the value range of LVAD structural parameters, add the corresponding LVAD structure to the three-dimensional model of the ventricle and aorta according to each group of sample parameters, obtain the three-dimensional model of the ventricle and aorta containing the corresponding LVAD structure, and then obtain multiple three-dimensional models of the ventricle and aorta containing the corresponding LVAD structure.

[0085] In this embodiment, the LVAD structural parameters include insertion position, insertion depth, insertion angle and insertion diameter. The value range of each structural parameter is first determined according to the medical guidelines, and then in the Ansys Workbench software, the Latin hypercube sampling method is used to select different groups of sample parameters from the value range of the LVAD structural parameters. Each group of sample parameters includes specific values ​​of the insertion position, insertion depth, insertion angle and insertion diameter. One group of sample parameters corresponds to one LVAD structure, and different groups of sample parameters correspond to different LVAD structures. The number of sample parameter groups selected by the Latin hypercube sampling method is the same as the number of LVAD structures. By adding an LVAD structure to the three-dimensional model of the ventricles and aorta, a three-dimensional model of the ventricles and aorta containing the LVAD structure is obtained. By changing the sample parameters, another three-dimensional model of the ventricles and aorta containing the LVAD structure is obtained, and then a three-dimensional model of the ventricles and aorta containing different LVAD structures is obtained.

[0086] In a specific embodiment of the present invention, Figure 3 As shown, according to each set of sample parameters, a corresponding LVAD structure is added to the three-dimensional model of the ventricle and aorta to obtain a three-dimensional model of the ventricle and aorta containing the corresponding LVAD structure, including:

[0087] Step S3.1: Export the 3D model of the ventricle and aorta from the medical modeling software E3D, and then import the 3D model of the ventricle and aorta into UG NX software (Unigraphics NX).

[0088] Step S3.2: Extract the ventricular surface from the 3D model of the ventricle and aorta, and divide the cannulation area at the apex.

[0089] Step S3.3: Determine the reference point of the change of the primitive of the LVAD structure cross section, and construct a local polar coordinate system with the apex of the heart as the origin.

[0090] like Figure 4 As shown in the figure, a local polar coordinate system (m, r) is constructed with the apex as the origin. The cross-sectional center point f of any LVAD structure on the ventricular surface can be expressed by the local polar coordinate system as (m f ,r f ). The parametric design of the spatial relationship of the LVAD structure in the local polar coordinate system is as follows Figure 5 As shown, Figure 5The three-dimensional coordinate system in the figure takes the apex as the origin, the direction perpendicular to the inner surface of the ventricle as the X-axis, and the tangential direction at any point on the outer surface of the ventricle as the Y-axis. The Y-axis is perpendicular to the central axis of the LVAD structure. The Z-axis is determined according to the right-hand rule. The incident angle of the LVAD structure is θ (the angle between the LVAD structure and the Y direction), and the compound inclination angle is α (that is, the angle between the LVAD structure and the Z direction). In order to achieve parametric modeling, the incident angle θ and the compound inclination angle α in the three-dimensional coordinate system are converted to the local polar coordinate system to obtain the incident angle θ in the local polar coordinate system. f and the composite inclination angle α f Therefore, the geometric parameters of the LVAD structure can be expressed as (m f ,r f ,θ f ,α f ).

[0091] Step S3.4: Automatically construct LVAD structural primitives in the cannulation area according to each set of sample parameters.

[0092] Step S3.5: Setting the composite inclination angle α of the LVAD structure f .

[0093] Step S3.6: Perform Boolean operation on the LVAD structure and the imported three-dimensional model of the ventricle and aorta to obtain the three-dimensional model of the ventricle and aorta including the LVAD structure.

[0094] Since the effect of the LVAD structure insertion segment on the probability of thrombosis is not considered, the imported 3D model of the ventricle and aorta is subjected to Boolean operation with the LVAD structure drawn in the UG NX software to obtain the 3D model of the ventricle and aorta containing the LVAD structure and save it in the stp general format. In step S3.4, by changing the sample parameters, different LVAD structure primitives are constructed, and different LVAD structures can be obtained.

[0095] The present invention realizes the parametric modeling of the insertion position, insertion depth, insertion angle and insertion diameter of the LVAD structure through the secondary development function provided by UG NX software, and then obtains the three-dimensional model of the ventricle and aorta containing the corresponding LVAD structure through Boolean operation.

[0096] Step S4: Mesh each ventricle and aorta three-dimensional model including the LVAD structure to generate a corresponding network model.

[0097] The 3D models of each ventricle and aorta containing the LVAD structure were imported into the ANSYS Workbench software for automatic meshing. Due to the complex structures of the left ventricle and aorta, unstructured meshing was used for the left ventricle and aorta containing the LVAD structure. The meshing results are shown in Figure 2. Figure 6 shown.

[0098] Step S5: Based on the ventricular motion function (Formula (1) and Formula (2)), the ventricular contraction motion is controlled by controlling the motion of the ventricular wall nodes, and an angular velocity is applied to the ventricular wall to control the rotational motion of the ventricular wall, and the platelet shear stress history and platelet retention time under each set of sample parameters are calculated.

[0099] Each mesh model was imported into Ansys Fluent software, the ventricle was set as the motion area, the ventricular wall was set as the moving boundary, and the ventricular motion function was constructed through user-defined functions. The ventricular contraction movement was controlled by controlling the motion of the ventricular wall nodes. At the same time, an angular velocity was applied to the ventricular wall through the UDF tool (i.e., user-defined tool) to control the rotational movement of the ventricular wall. The contraction and rotational movement of the ventricle were realized at the same time, simulating the transportation and generation of various intermediate components in the process of thrombus formation. During the contraction and rotational movement of the ventricle, the platelet shear stress history and platelet retention time were calculated. A set of sample parameters corresponds to a network model, and a network model corresponds to a set of platelet shear stress history and platelet retention time, so the platelet shear stress history and platelet retention time under different sets of sample parameters were obtained.

[0100] The non-steady-state calculation method is used to calculate the blood flow in the ventricle. In one cardiac cycle, the ventricular contraction period takes 0.3 seconds and the ventricular diastole period takes 0.5 seconds. The contraction and rotation of the ventricle are defined by the user-defined function UDF, and the time step is uniformly set to 0.05. In a specific embodiment of the present invention, Figure 7 As shown, the platelet shear stress history and platelet retention time under each set of sample parameters are calculated, including:

[0101] Step S5.1: Calculate the blood flow in the ventricle and aorta at the current time step;

[0102] Step S5.2: releasing thrombus particles and calculating the movement of thrombus particles at the current time step;

[0103] Step S5.3: Move the network nodes in any boundary and fluid region through the DEFINE_GRID_MOTION macro in the user-defined function UDF; after the wall grid nodes move, smooth the grid to improve the grid quality;

[0104] Step S5.4: Calculate the blood flow conditions in the ventricle and aorta and the movement conditions of thrombus particles in the next time step until all thrombus particles leave the ventricle and aorta;

[0105] Step S5.5: After the calculation is completed, the platelet shear stress history and platelet retention time are counted.

[0106] The present invention utilizes software and sampling methods to realize the entire process of automatic generation of sample parameters, parameterized modeling, grid division, numerical calculation and optimization, thereby greatly improving the calculation efficiency.

[0107] Step S6: Construct a Kriging model based on the platelet shear stress history and platelet retention time under different groups of sample parameters.

[0108] The Kriging model is an unbiased estimation model that predicts the response at an unknown test point through known test point information. The present invention can construct an unbiased estimation model of the response through the numerical calculation results of the known group sample parameters (i.e., platelet shear stress history and platelet retention time).

[0109] In a specific embodiment of the present invention, Figure 8 As shown in the figure, a Kriging model is constructed based on the platelet shear stress history and platelet retention time under different groups of sample parameters, including:

[0110] Step S6.1: Construct a sample data set based on the platelet shear stress history and platelet retention time under different groups of sample parameters; wherein the sample data set includes multiple samples, each sample includes an input and an output, the input is a group of sample parameters, and the output is the platelet shear stress history and platelet retention time under the group of sample parameters.

[0111] Before constructing the sample data set, each set of sample parameters is normalized and outliers are removed.

[0112] Step S6.2: Construct a Kriging model and use the Kriging model to calculate the distance matrix between samples to determine the spatial correlation between input quantities.

[0113] According to the data distribution characteristics of the present invention, a multinomial Kriging model is selected.

[0114] Step S6.3: Estimate the parameters of the Kriging model based on the spatial correlation between the input quantities.

[0115] Step S6.4: Input the sample input into the Kriging model, and the Kriging model outputs the result.

[0116] Step S6.5: Determine whether the Kriging model meets the accuracy requirements based on the output results of the Kriging model and the output of the sample; if so, output the Kriging model; if not, increase the number of groups of sample parameters selected from the value range of the LVAD structural parameters, go to step S3, execute steps S3 to S5, and obtain the platelet shear stress history and platelet retention time under more groups of sample parameters to improve the Kriging model until the Kriging model meets the accuracy requirements.

[0117] In this embodiment, an initial Kriging model is first constructed based on the platelet shear stress history and platelet retention time under 21 sets of sample parameters. The initial Kriging model does not meet the accuracy requirements, so a more accurate Kriging model is subsequently obtained by using the platelet shear stress history and platelet retention time under 61 sets of sample parameters. The Kriging model can be used to calculate the blood flow movement, platelet shear stress history and platelet retention time in the ventricle.

[0118] Step S7: inputting different groups of structural parameters of the LVAD structure into the Kriging model to obtain the platelet shear stress history and platelet retention time under different groups of structural parameters;

[0119] After the Kriging model was constructed, different groups of structural parameters of the LVAD structure were obtained. The structural parameters were the same as the sample parameters, including insertion position, insertion depth, insertion angle, and insertion diameter. Different groups of structural parameters were used as inputs to the Kriging model to obtain the platelet shear stress history and platelet retention time under different groups of structural parameters.

[0120] Step S8: Using an optimization algorithm to find the optimal group structure parameters from the platelet shear stress history and platelet retention time under different group structure parameters.

[0121] In this embodiment, the optimization algorithm is a multi-objective genetic algorithm, which uses a multi-objective genetic algorithm (reference: Zheng Anbo, Ma Handong, Luo Xiaoyun, et al. Multi-objective genetic algorithm based on Kriging [J]. Aeronautical Computing Technology, 2014 (2).) to find the Pareto optimal solution set and Pareto frontier in the platelet shear stress history and platelet retention time under different groups of structural parameters output by the Kriging model, and finally artificially obtain one to multiple models from the Pareto frontier as the optimal model. The objectives of multi-objective optimization problems are often different in dimension and conflict with each other, and it is difficult to directly compare the objective values ​​to determine the optimal solution like a single-objective optimization problem. In the present invention, the Pareto optimal solution set is the set of all non-inferior solutions within the feasible range of the input quantity (i.e., different groups of structural parameters), and the Pareto optimal frontier is the set of platelet shear stress history and platelet retention time corresponding to the Pareto optimal solution set. Multi-objective genetic algorithm is a method to evaluate the quality of multi-objective algorithm. This method can screen and retain non-dominated solutions by using fitness evaluation and selection mechanism, so as to find Pareto optimal solution set and Pareto frontier in the prediction results of Kriging model, and finally obtain one to multiple models from Pareto frontier as the optimal model. Based on CFD method, the optimal group structural parameters are calculated, the intraventricular blood flow, platelet shear stress history and platelet retention time are analyzed, and an analysis report is given to provide preoperative planning for LVAD implantation surgery for heart failure patients and optimize the biocompatibility of LVAD treatment.

[0122] Example 2

[0123] An LVAD multi-objective optimization system provided by an embodiment of the present invention includes: a first construction unit, a second construction unit, a third construction unit, a grid division unit, a first calculation unit, a fourth construction unit, a second calculation unit and an optimization unit.

[0124] The first construction unit is used to construct a three-dimensional model of the ventricle and aorta according to the cardiac CT image; the second construction unit is used to construct a ventricular motion function based on the three-dimensional model of the ventricle and aorta; the third construction unit is used to select different groups of sample parameters from the value range of LVAD structural parameters, and add the corresponding LVAD structure to the three-dimensional model of the ventricle and aorta according to each group of sample parameters to obtain a three-dimensional model of the ventricle and aorta containing the corresponding LVAD structure, and then obtain multiple three-dimensional models of the ventricle and aorta containing the corresponding LVAD structure; the grid division unit is used to grid each three-dimensional model of the ventricle and aorta containing the LVAD structure to generate a corresponding network model; the first calculation ... based on the value range of LVAD structural parameters, and then obtain a three-dimensional model of the ventricle and aorta containing the corresponding LVAD structure The invention relates to a method for controlling the ventricular contraction movement by controlling the movement of the ventricular wall nodes, and applying an angular velocity to the ventricular wall to control the rotational movement of the ventricular wall, and calculating the platelet shear stress history and platelet retention time under each set of sample parameters; the fourth construction unit is used to construct a Kriging model according to the platelet shear stress history and platelet retention time under different sets of sample parameters; the second calculation unit is used to input different sets of structural parameters of the LVAD structure into the Kriging model to obtain the platelet shear stress history and platelet retention time under different sets of structural parameters; the optimization unit is used to use an optimization algorithm to find the optimal set of structural parameters from the platelet shear stress history and platelet retention time under different sets of structural parameters.

[0125] In some embodiments, the LVAD multi-objective optimization system may be combined with the method features of the LVAD multi-objective optimization method of any embodiment, and vice versa, which will not be elaborated here.

[0126] Example 3

[0127] An embodiment of the present invention also provides an electronic device, which includes: a memory, a processor, and a computer program / instructions stored in the memory, and the processor executes the computer program / instructions to implement the LVAD multi-objective optimization method in the embodiment of the present application.

[0128] Although not shown, the electronic device includes a processor that can perform various appropriate operations and processes based on programs and / or data stored in a read-only memory (ROM) or programs and / or data loaded from a storage portion into a random access memory (RAM). The processor can be a multi-core processor or can include multiple processors. In some embodiments, the processor can include a general-purpose main processor and one or more special coprocessors, such as a central processing unit, a graphics processing unit (GPU), a neural network processor (NPU), a digital signal processor (DSP), etc. In the RAM, there is also a

[0129] Stores various programs and data required for device operation. The processor, ROM, and RAM are connected to each other through the bus. The input / output (I / O) interface is also connected to the bus.

[0130] The processor and memory are used together to execute the program / instructions stored in the memory. When the program / instructions are executed by the computer, the methods, steps or functions described in the above embodiments can be implemented.

[0131] Although not shown, an embodiment of the present invention further provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the LVAD multi-objective optimization method in the embodiment of the present application.

[0132] Storage media in embodiments of the present invention include permanent and non-permanent, removable and non-removable items that can be implemented by any method or technology to store information. Examples of storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0133] Readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0134] Although not shown, an embodiment of the present invention further provides a computer program product, including: a computer program / instruction, which, when executed by a processor, implements the LVAD multi-objective optimization method in the embodiment of the present application.

[0135] What is disclosed above is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or modifications within the technical scope disclosed in the present invention, which should be covered within the protection scope of the present invention.

Claims

1. A multi-objective optimization method for LVAD, characterized in that: The optimization method comprises the following steps: Construct a three-dimensional model of the ventricle and aorta based on cardiac CT images; Based on the three-dimensional model of the ventricle and aorta, construct a ventricular motion function; Selecting different groups of sample parameters from the value range of the LVAD structure parameters, adding a corresponding LVAD structure to the 3D model of the ventricle and aorta according to each group of sample parameters, obtaining a 3D model of the ventricle and aorta containing the corresponding LVAD structure, and further obtaining multiple 3D models of the ventricle and aorta containing the corresponding LVAD structure; Mesh each 3D model of the ventricle and aorta containing the LVAD structure to generate a corresponding network model; Based on the ventricular motion function, the ventricular contraction motion is controlled by controlling the motion of the ventricular wall nodes, and an angular velocity is applied to the ventricular wall to control the rotational motion of the ventricular wall, and the platelet shear stress history and platelet retention time under each set of sample parameters are calculated; The Kriging model was constructed based on the platelet shear stress history and platelet retention time under different groups of sample parameters; Different groups of structural parameters of the LVAD structure were input into the Kriging model to obtain the platelet shear stress history and platelet retention time under different groups of structural parameters; An optimization algorithm is used to find the optimal group structure parameters from the platelet shear stress history and platelet retention time under different group structure parameters.

2. The LVAD multi-objective optimization method according to claim 1, characterized in that: Based on the three-dimensional model of the ventricle and aorta, a ventricular motion function is constructed, including: Tracking the changes in the positions of multiple spots in different ventricular cross-sections in the three-dimensional model of the ventricle and aorta over time by two-dimensional speckle tracking echocardiography technology, and calculating the volume change of the ventricle and the angular velocity of rotation of the ventricle; The ventricular motion function is fitted according to the volume change of the ventricle; wherein the ventricular motion function includes a ventricular contraction function and a ventricular expansion function, and the expression of the ventricular contraction function is: M(t)=0.5*a*sin((10*π / 3)*(t+0.15)); The expression of the ventricular expansion function is: M(t)=0.5*a*sin((2*π)*(t-0.55)); Among them, M(t) represents the volume change of the ventricle, that is, the amount of contraction or expansion; t represents time; a represents the correction coefficient; π represents pi.

3. The LVAD multi-objective optimization method according to claim 1, characterized in that: According to each set of sample parameters, a corresponding LVAD structure is added to the 3D model of the ventricle and aorta, including: Importing the 3D model of the ventricle and aorta into UG NX software; Extracting the ventricular surface from the 3D model of the ventricle and aorta, and dividing the cannulation area at the apex of the heart; Determine the reference point of the LVAD structure cross-sectional primitive change, and construct a local polar coordinate system with the apex of the heart as the origin; constructing an LVAD structural primitive in the cannulation area according to each set of sample parameters; Setting a composite inclination angle of the LVAD structure; wherein the composite inclination angle refers to the angle between the LVAD structure and the Z direction, where the Z direction is the span direction; Boolean operations are performed on the LVAD structure and the imported three-dimensional model of the ventricle and aorta to obtain a three-dimensional model of the ventricle and aorta including the LVAD structure.

4. The LVAD multi-objective optimization method according to claim 1, characterized in that: Calculate the platelet shear stress history and platelet retention time for each set of sample parameters, including: Calculate the blood flow in the ventricle and aorta at the current time step; Release thrombus particles and calculate the movement of thrombus particles at the current time step; Moving network nodes within the ventricular wall and fluid regions; Calculate the blood flow conditions and thrombus particle movement conditions in the ventricle and aorta at the next time step until all thrombus particles leave the ventricle and aorta; The platelet shear stress history and platelet retention time were statistically analyzed.

5. The LVAD multi-objective optimization method according to any one of claims 1 to 4, characterized in that: The Kriging model was constructed based on the platelet shear stress history and platelet retention time under different groups of sample parameters, including: Constructing a sample data set according to the platelet shear stress history and platelet retention time under different groups of sample parameters; wherein the sample data set includes a plurality of samples, each sample includes an input amount and an output amount, the input amount is a group of sample parameters, and the output amount is the platelet shear stress history and platelet retention time under the group of sample parameters; Construct a Kriging model and use it to calculate the distance matrix between samples to determine the spatial correlation between input quantities; estimating parameters of the Kriging model based on the spatial correlation between input quantities; The input amount of the sample is input into the Kriging model, and whether the Kriging model meets the accuracy requirements is determined based on the output result of the Kriging model and the output amount of the sample; if so, the Kriging model is output; if not, the number of groups of sample parameters selected from the value range of the LVAD structural parameters is increased until the Kriging model meets the accuracy requirements.

6. The LVAD multi-objective optimization method according to claim 1, characterized in that: The optimization algorithm is a multi-objective genetic algorithm.

7. A LVAD multi-objective optimization system, characterized in that: The system comprises: A first construction unit is used to construct a three-dimensional model of the ventricle and aorta according to the cardiac CT image; A second construction unit is used to construct a ventricular motion function based on the three-dimensional model of the ventricle and aorta; A third construction unit is used to select different groups of sample parameters from the value range of the LVAD structure parameters, and add a corresponding LVAD structure to the 3D model of the ventricle and aorta according to each group of sample parameters to obtain a 3D model of the ventricle and aorta containing the corresponding LVAD structure, and then obtain multiple 3D models of the ventricle and aorta containing the corresponding LVAD structure; A meshing unit, used for meshing each ventricle and aorta three-dimensional model including the LVAD structure to generate a corresponding network model; a first calculation unit, configured to control the ventricular contraction movement by controlling the movement of the ventricular wall nodes based on the ventricular motion function, and to apply an angular velocity to the ventricular wall to control the rotational movement of the ventricular wall, and calculate the platelet shear stress history and platelet retention time under each set of sample parameters; a fourth construction unit, for constructing a Kriging model according to the platelet shear stress history and platelet retention time under different groups of sample parameters; A second calculation unit is used to input different sets of structural parameters of the LVAD structure into the Kriging model to obtain the platelet shear stress history and platelet retention time under different sets of structural parameters; The optimization unit is used to find the optimal group structure parameters from the platelet shear stress history and platelet retention time under different group structure parameters by using an optimization algorithm.

8. An electronic device comprising a memory, a processor, and a computer program / instruction stored in the memory, characterized in that: The processor executes the computer program / instructions to implement the LVAD multi-objective optimization method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the LVAD multi-objective optimization method according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the LVAD multi-objective optimization method according to any one of claims 1 to 6 is implemented.

Images