LVAD inlet sleeve structure optimization method and device, storage medium and product

Abstract

The invention discloses an LVAD inlet cannula structure optimization method and device, a storage medium and a product, and the method comprises the steps: constructing a heart-inlet cannula grid model under each group of structure parameters, and carrying out the simulation calculation, and obtaining the blocking area and inlet pressure loss of an inlet cannula, and the area of a thrombus formation region; constructing a response surface model according to the blockage area and the inlet pressure loss of the inlet cannula under each group of structural parameters and the area of the thrombosis region; predicting the blockage area and inlet pressure loss of the inlet cannula under different structural parameters and the area of a thrombus formation region by using the response surface model; the optimal structural parameters are found from the predicted blockage area and inlet pressure loss of the inlet cannula under different structural parameters and the area of the thrombus formation area by adopting an optimization algorithm, so that the structural optimization design of the inlet cannula is realized, and the thrombus formation is reduced.

Description

Technical Field

[0001] This invention belongs to the field of biomedical engineering, and in particular relates to a method, device, storage medium and product for optimizing the structure of the LVAD inlet cannula considering the geometric uncertainties caused by thrombosis. Background Technology

[0002] A left ventricular assist device (LVAD) is a mechanical circulatory support device that can partially replace the pumping function of the left ventricle. Specifically, an LVAD is connected to the left ventricle through an inlet cannula, drawing blood out of the left ventricle and then pumping it to the aorta and other sites through an outlet cannula. This mechanically assists the heart in pumping blood, providing a new treatment option for patients with end-stage heart failure.

[0003] However, despite the significant success of LVADs in improving the prognosis of patients with heart failure, numerous blood compatibility-related complications remain during their use, with pump thrombosis, bleeding, and stroke being particularly prominent. These complications can not only damage organ function, such as renal insufficiency and liver damage, but may even be life-threatening. Stroke, in particular, as a serious neurological complication, remains prevalent among LVAD patients, severely impacting their survival and quality of life.

[0004] Currently, third-generation left ventricular assist devices (LVADs) have been optimized in design to prevent pump thrombus formation. For example, the use of contactless bearing technologies (such as magnetic levitation and liquid suspension) reduces mechanical wear and blood component contamination, lowering the risk of in-pump thrombus formation. However, clinical observations have revealed that patients still face a high risk of adverse events such as bleeding and stroke post-operatively. Studies show that thrombus formation and dislodgement around the LVAD inlet cannula may be one of the main causes of stroke, such as... Figure 1 As shown. Because thrombi can travel through the bloodstream to blood vessels in the brain, causing blockages and triggering ischemic strokes, or due to the instability of the thrombus, causing hemorrhagic strokes, this issue has attracted great attention from clinicians and researchers.

[0005] Thrombosis is closely related to hemodynamic factors. Abnormal blood flow patterns, high shear stress, areas of blood flow stagnation, and non-physiological blood inflow and outflow all increase the risk of thrombosis. Under high shear stress, platelets are easily activated, begin to aggregate, and trigger the coagulation cascade. As the main cellular component of thrombosis, platelets can sense abnormal changes in hemodynamics and aggregate on damaged vascular endothelium or artificial material surfaces. The structural design of the inlet cannula is one of the key factors affecting the hemodynamic characteristics and thrombosis risk of LVADs. The shape, diameter, surface roughness, and insertion angle of the inlet cannula all affect the local blood flow pattern. However, there are currently no clear guidelines or recommendations regarding the structural design and use of inlet cannulas, which leads to varying degrees of thrombosis risk in actual clinical applications depending on the LVAD and surgical procedure. In addition, thrombosis can alter the morphology of the flow channel, reduce the flow area, and thus change the originally rational blood flow pattern. This change may lead to abnormal pressure-flow curves, decreased pump efficiency, uneven flow field distribution, and further damage to blood components. Thrombosis is a major source of geometric uncertainty during LVAD use, potentially leading to cannula blockage, increased inlet pressure loss, and a larger high-thrombosis zone between the endocardium and the cannula. This can prevent the LVAD from achieving its expected performance targets. In practice, performance failures still occur frequently. This not only reduces the effectiveness of the LVAD but also poses potential risks to patients and increases the incidence of postoperative complications.

[0006] Patent document CN202920776U discloses an inflow conduit for an axial-flow left ventricular assist pump. The inflow conduit (i.e., the inlet cannula) employs a flared, bell-shaped structure, and the implanted conduit adheres closely to the ventricular wall, effectively preventing conduit adsorption and reducing blood damage within the ventricle and conduit. The suture ring on the outer wall of the conduit effectively reduces the risk of cannula dislodgement and bleeding at the suture site. However, it cannot quantify the impact of the conduit's structural parameters (cannula radius, insertion depth, outlet cross-sectional inclination angle, etc.) on the output (blockage area of ​​the left ventricular assist device inlet cannula, inlet pressure loss of the left ventricular assist device inlet cannula, area of ​​the high thrombosis zone between the endocardium and the inlet cannula, etc.); it also cannot assess the abnormal pressure-flow curve caused by thrombosis during use, the decrease in pump efficiency, the uneven flow field distribution, and further damage to blood components. Summary of the Invention

[0007] To address the aforementioned deficiencies, this invention provides a method, device, storage medium, and product for optimizing the LVAD inlet sleeve structure. By combining computational fluid dynamics to simulate thrombosis, it predicts blood flow patterns and potential risks under different design schemes, thereby optimizing the shape, size, and surface characteristics of the inlet sleeve. This results in an LVAD inlet sleeve structure that maintains relatively stable performance parameters such as pressure-flow curves, pump efficiency, and flow field distribution under the geometric uncertainties caused by thrombosis, thus reducing thrombosis.

[0008] This invention solves the above-mentioned technical problems through the following technical solution: a method for optimizing the structure of an LVAD inlet sleeve, comprising:

[0009] Construct a parameter-controllable geometric model of the heart-entry cannula;

[0010] The range of structural parameters of the inlet sleeve is sampled to generate N sets of first sample parameters; each set of first sample parameters includes multiple structural parameters of the inlet sleeve.

[0011] The inlet cannula structural parameters in the heart-inlet cannula geometric model are changed according to the parameters of the first sample in different groups, and the mesh is divided to obtain the heart-inlet cannula mesh model under the parameters of the first sample in each group.

[0012] Simulation calculations were performed based on the heart-inlet cannula mesh model under the parameters of the first sample in each group to obtain the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula.

[0013] A response surface model was constructed based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under the parameters of the first sample in different groups.

[0014] The range of structural parameters of the inlet sleeve is sampled to generate M sets of second sample parameters; each set of second sample parameters includes multiple structural parameters of the inlet sleeve, and M is much larger than N;

[0015] The response surface model was used to obtain the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the thrombosis zone area between the endocardium and the inlet cannula, under different groups of second sample parameters.

[0016] An optimization algorithm was used to find the optimal set of second sample parameters from the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under different sets of second sample parameters, so as to obtain the optimal structural parameters of the inlet cannula.

[0017] Furthermore, the construction of a parameter-controllable heart-entry cannula geometric model specifically includes:

[0018] Construct a solid cardiac model based on cardiac CT images;

[0019] A three-dimensional coordinate system is constructed with the apex of the heart model as the origin.

[0020] A circular ring is constructed with the origin of the three-dimensional coordinate system as the center. The ring is stretched to form an inlet cannula, which is then connected to the apex of the heart, generating a heart-inlet cannula geometric model with controllable parameters.

[0021] Furthermore, the Latin hypercube sampling method is used to sample the range of structural parameters of the inlet sleeve, generating N sets of first sample parameters or M sets of second sample parameters.

[0022] Furthermore, simulation calculations were performed based on the heart-inlet cannula mesh model under the parameters of each group of first samples to obtain the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, specifically including:

[0023] Based on the blood flow pulsation velocity waveform at the patient's LVAD inlet, the inlet boundary conditions of the heart-inlet cannula mesh model are set to simulate blood flow motion;

[0024] Calculate the relative retention time and time-averaged wall shear stress of platelets during blood flow;

[0025] After running multiple cardiac cycles, thrombus components were added, and coagulation factors and bound platelets were calculated.

[0026] Determine whether a thrombus has completely formed based on clotting factors;

[0027] Once the thrombus has fully formed, calculate the blockage area of ​​the inlet cannula, the inlet pressure loss, and the area of ​​the thrombus formation zone between the endocardium and the inlet cannula.

[0028] Furthermore, the formulas for calculating the relative retention time and time-averaged wall shear stress of the platelets are as follows:

[0029]

[0030] Where RRT represents the relative retention time of platelets; T represents the duration of the cardiac cycle; τ w The instantaneous shear stress represents the stress at different time points; t represents time; TAWSS represents the time-averaged wall shear stress.

[0031] Furthermore, the formula for calculating the bound platelets is as follows:

[0032]

[0033]

[0034] Where BP represents the concentration of bound platelets; t represents time; k BP Φ represents the rate constant for the formation of bound platelets; BP The saturation function representing the aggregation of bound platelets during thrombus formation; Represents the shearing adjustment function; C AP Indicates the concentration of activated platelets; C represents coagulation factors; D ceff Indicates the effective diffusion coefficient; ΔC represents the diffusion term of the coagulation factor; k c Φ represents the rate constant of coagulation factor production; C c represents the concentration of the clotting factor; c represents the saturation concentration; c t Characteristic values ​​representing saturation concentration; RRT t The characteristic value representing the relative retention time of platelets; RRT represents the relative retention time of platelets; D C Indicates the molecular diffusion coefficient of coagulation factors; BP t Characteristic values ​​representing the concentration of bound platelets; Indicates shear rate; Characteristic values ​​representing shear rate.

[0035] Furthermore, a response surface model was constructed based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, and the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under the parameters of the first sample in different groups. This model included:

[0036] Preprocessing was performed on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the thrombosis zone area between the endocardium and the inlet cannula, under different group first sample parameters.

[0037] A sample dataset was constructed based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under the parameters of the first sample of different groups after preprocessing.

[0038] Construct a response surface model based on the aforementioned sample dataset;

[0039] Verify the accuracy of the response surface model. If the accuracy requirement is met, output the response surface model. If the accuracy requirement is not met, optimize the strategy for constructing the response surface model. Repeat the steps of obtaining the blockage area and inlet pressure loss of the inlet cannula after thrombosis, the thrombosis area between the endocardium and the inlet cannula, and constructing the response surface model under different groups of first sample parameters, until the response surface model meets the accuracy requirement.

[0040] 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 inlet sleeve structure optimization method as described above.

[0041] Based on the same concept, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the LVAD inlet sleeve structure optimization method as described above.

[0042] Based on the same concept, the present invention provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the LVAD inlet sleeve structure optimization method as described above.

[0043] Compared with the prior art, the advantages of the present invention are as follows:

[0044] The LVAD inlet cannula structure optimization method provided by this invention first uses numerical simulation to obtain the blockage area, inlet pressure loss, and thrombus formation area under different structural parameters. Then, based on the blockage area, inlet pressure loss, and thrombus formation area under different structural parameters, a response surface model is constructed. The response surface model is used to fit the relationship between structural parameters and blockage area, inlet pressure loss, and thrombus formation area. Furthermore, the response surface model is used to predict the blockage area, inlet pressure loss, and thrombus formation area under a large number of structural parameters. Finally, the optimal structural parameters are found from the blockage area, inlet pressure loss, and thrombus formation area under a large number of structural parameters, thereby optimizing the LVAD inlet cannula structure and reducing thrombus formation.

[0045] This invention utilizes numerical simulation methods to construct an accurate response surface model based on the blockage area, inlet pressure loss, and thrombus formation area under a small number of different structural parameters. This response surface model is then used to predict the blockage area, inlet pressure loss, and thrombus formation area under a large number of different structural parameters, enabling optimization among these parameters. Compared to using numerical simulation methods to obtain these parameters under a large number of different structural parameters, this invention significantly simplifies the optimization calculation process and improves the efficiency of optimization design. Attached Figure Description

[0046] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only one embodiment of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart of the LVAD inlet sleeve structure optimization method in an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram illustrating the construction of the heart-entry cannula geometric model in an embodiment of the present invention;

[0049] Figure 3 This is a simulation calculation flowchart in an embodiment of the present invention;

[0050] Figure 4 This is a schematic diagram of thrombosis in an embodiment of the present invention;

[0051] Figure 5 This is a schematic diagram of the thrombus formation dynamic grid in an embodiment of the present invention. Detailed Implementation

[0052] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0054] Example 1

[0055] like Figure 1 As shown, the LVAD inlet sleeve structure optimization method provided in this embodiment of the invention includes the following steps:

[0056] Step S1: Construct a parameter-controllable heart-entry cannula geometric model.

[0057] In a specific embodiment of the present invention, constructing a parameter-controllable heart-entry cannula geometric model specifically includes:

[0058] Step S1.1: Construct a solid cardiac model based on cardiac CT images.

[0059] Based on computed tomography (CT) images of the patient's heart, a three-dimensional model of the ventricle and aorta was constructed using the medical modeling software E3D, and the three-dimensional model of the ventricle and aorta was saved as an STL file. The STL file was then imported into Ansys Design Modeler (DM) and converted into a solid heart model.

[0060] CT scans essentially use X-rays to penetrate tissue and obtain tomographic images at different levels. Each slice represents a section of tissue at a different depth, and each pixel in these slices contains different density information, such as bone, soft tissue, and fluid. Using this density information, the three-dimensional coordinates of each pixel are calculated, and these slices are stitched together layer by layer to reconstruct the three-dimensional structure of the tissue. Medical modeling software E3D can process CT images to generate a precise three-dimensional model (e.g., a three-dimensional model of the ventricles and aorta). This three-dimensional model not only realistically reflects the internal anatomical structure of the tissue but also provides doctors with more detailed and comprehensive information about lesions through observation from different angles. For patients requiring implanted medical devices, such as LVADs or stents, 3D reconstruction can also help doctors better plan the installation location and angle of the device, minimizing surgical risks.

[0061] Step S1.2: Construct a three-dimensional coordinate system with the apex of the heart model as the origin.

[0062] In this embodiment, a three-dimensional coordinate system is constructed with the apex of the heart model as the origin, the vertically upward direction as the Z-axis, and the plane parallel to the ground as the XY plane.

[0063] Step S1.3: Construct a ring with the origin of the three-dimensional coordinate system as the center, stretch the ring to form an inlet cannula and connect the inlet cannula to the apex of the heart, generating a heart-inlet cannula geometric model with controllable parameters.

[0064] like Figure 2 As shown, a ring is constructed with the origin of the three-dimensional coordinate system as the center. The ring is stretched using the extrusion operation and Boolean operations of DM to form an inlet cannula, with one end inserted into the heart model and the other end extending out. Simultaneously with the ring construction and stretching operations, the structural parameters of the inlet cannula are processed to generate a parameter-controllable heart-inlet cannula geometric model. Changing the structural parameters of the inlet cannula generates a corresponding heart-inlet cannula geometric model. In this embodiment, the structural parameters of the inlet cannula include its radius R, insertion depth h (i.e., the length inserted into the heart model), and outlet plane tilt angle α (i.e., the angle between the cross-section of the inlet cannula and the XY plane of the three-dimensional coordinate system). Using the structural parameters of the inlet cannula as variables, an objective function is constructed: the occlusion area A1 of the inlet cannula, the inlet pressure loss ΔP, and the thrombus formation area A3 between the endocardium and the inlet cannula. Figure 2 In this context, γ represents the angle between the inclined cylinder axis and the Z-axis, used to describe the degree of inclination; θ represents the azimuth angle in the XY plane, that is, the rotation angle of the inclined cylinder axis in the XY plane.

[0065] The occlusion area A1 of the inlet cannula refers to the projected area of ​​the thrombus volume onto the cross-section of the inlet cannula. It is calculated by projecting the thrombus volume onto the cross-section of the inlet cannula and then calculating the area it occupies. A smaller occlusion area A1 indicates a smaller impact of the thrombus on blood flow.

[0066] The inlet pressure loss ΔP is calculated by monitoring the pressure at the inlet sleeve cross-section. During simulation, the pressure at this monitoring surface is read in real-time, using the pressure before thrombus formation as a baseline. After the thrombus calculation is completed, the pressure at this monitoring surface is read again; the difference between this pressure value and the baseline is the inlet pressure loss ΔP. ​​A smaller inlet pressure loss ΔP indicates a smaller impact of the thrombus on blood flow.

[0067] After thrombosis calculation, the number and area of ​​cells with a time-averaged wall shear stress (TAWSS) below 0.2 Pa and a platelet (BP) concentration below 200 nM were counted. The number of cells multiplied by the cell area is defined as the thrombosis formation area A3 between the endocardium and the inlet cannula. The smaller the area of ​​the high-thrombosis formation zone, the lower the risk of thrombosis.

[0068] Step S2: Sample the range of structural parameters of the inlet sleeve to generate N sets of first sample parameters; where each set of first sample parameters includes multiple structural parameters of the inlet sleeve.

[0069] In this embodiment, Latin Hypercube Design (LHD) sampling is used to sample the range of structural parameters of the inlet sleeve, generating N sets of first sample parameters. The structural parameters in this embodiment include the radius R of the inlet sleeve, the insertion depth h, and the outlet plane tilt angle α. The radius R of the inlet sleeve ranges from 15 to 20 mm, the insertion depth h ranges from 10 to 30 mm, and the outlet plane tilt angle α ranges from -14° to 14°. The range of each structural parameter is divided into n equal intervals. For each structural parameter, a value is randomly selected from each interval as a sample point. By randomly arranging the sample points of each structural parameter, a k-dimensional hypercube sample is formed. Adjusting the arrangement order of the sample points ensures that the samples are uniformly distributed in the high-dimensional space, avoiding excessive concentration or duplication of samples.

[0070] In this embodiment, the number of structural parameters is 3. Considering the high complexity of the optimization problem or the rapid change of the objective function, n can be increased to 20. Therefore, the number of sample points is 60.

[0071] Step S3: Change the inlet cannula structural parameters in the heart-inlet cannula geometric model according to the parameters of the first sample in different groups, and perform mesh generation to obtain the heart-inlet cannula mesh model under the parameters of the first sample in each group.

[0072] Based on the parameters of each group of first samples, the structural parameters of the inlet cannula can be changed to generate the corresponding heart-inlet cannula geometric model (corresponding geometry file). The heart-inlet cannula geometric model under each group of first sample parameters is meshed to generate the heart-inlet cannula mesh model under each group of first sample parameters, thus obtaining the mesh file for each group of first sample parameters.

[0073] Step S4: Perform simulation calculations based on the heart-inlet cannula mesh model under the parameters of the first sample in each group to obtain the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the thrombosis area between the endocardium and the inlet cannula.

[0074] In a specific embodiment of the present invention, such as Figure 3 As shown, simulation calculations are performed based on the heart-entry cannula mesh model under the parameters of each group of first samples, specifically including:

[0075] Step S4.1: Model import.

[0076] Import the heart-inlet cannula mesh model and check the mesh quality: Skewness < 0.8, Aspect Ratio < 5. Set the blood density ρ to 1050 kg / m³. 3 The viscosity model chosen is the Carreau non-Newton model. Since the Newton model in the numerical calculation software Fluent is no longer applicable, a compiled user-defined function is used to define the viscosity of blood.

[0077] Step S4.2: Based on the blood flow pulsation velocity waveform at the patient's LVAD inlet, set the inlet boundary conditions of the heart-inlet cannula mesh model to simulate blood flow motion.

[0078] Transient blood flow pulsation velocity waveforms at the patient's LVAD inlet were acquired to obtain a time-velocity comparison table (CSV / TXT format). The time-velocity data was smoothed to eliminate noise interference and ensure the velocity time series was continuous and differentiable. The velocity time series was written into a User-Defined Function (UDF) as an array in Fluent software, using the DEFINE_PROFILE type. The data file was saved to the Fluent working directory, and the Profile file was read into Fluent software to load the transient velocity waveforms, which were then used to simulate blood flow at the patient's LVAD inlet.

[0079] At the LVAD inlet, the inlet concentration of unactivated platelet RP is 20 (normalized units), and the outlet is set as a pressure outlet. The wall surface is set to a no-slip boundary condition, and the coagulation factor (C) flux is triggered conditionally, i.e., when the time-averaged variable shear stress (TAWSS) of the vessel wall is less than 0.2 Pa and the platelet-bound (BP) concentration is less than 200 nM, the wall surface triggers the generation of coagulation factor (C).

[0080] Step S4.3: Calculate the relative retention time and time-averaged wall shear stress of platelets during blood flow.

[0081] The transient solver was used for the solution. The calculation lasted for 3 cardiac cycles, with each cardiac cycle being 0.8 s and a time step increment of 0.01 s. To avoid errors caused by non-convergence, the shear rate, time-mean wall shear stress (TAWSS), and relative retention time (RRT) of platelets during blood flow were monitored starting from the third cardiac cycle in the heart-inlet cannula solid model. Subsequently, thrombus components were introduced, and the concentrations of coagulation factor (C) and bound platelet (BP) were calculated. The thrombus formation rate was calculated on units with a TAWSS below 0.2 Pa and a BP concentration below 200 nM (nanomoles per liter), simulating thrombus formation for 1 s. The above calculation process was repeated 5 times in all models, with a total thrombus accumulation simulation of 5 s.

[0082] Three cardiac cycles were run (cycle duration 0.8s, time step 0.01s). Data from the third cycle were used to calculate the shear rate, time-mean wall shear stress, and relative retention time of platelets to avoid the influence of initial transients. TAWSS, RRT, and other data were exported as interpolation data for iterative calculations of thrombus growth. Fluent software can monitor TAWSS, RRT, and other data at arbitrary cross-sections. The specific formula for calculating TAWSS is as follows:

[0083]

[0084] Where TAWSS represents the time-averaged wall shear stress; T represents the duration of the cardiac cycle; τ w The instantaneous shear stress represents the stress at different points in time; t represents time.

[0085] The specific formula for calculating RRT is as follows:

[0086]

[0087] RRT represents the relative retention time of platelets.

[0088] Step S4.4: After running multiple cardiac cycles, add thrombus components to simulate thrombus formation and calculate coagulation factors and bound platelets.

[0089] After three cardiac cycles, thrombus components were added to simulate thrombus formation for one second, and this process was repeated five times, simulating a total thrombus accumulation of 5 seconds. Bound platelets (BP), activated platelets (AP), and coagulation factors (C) all participate in thrombus formation. Relative retention time (RRT) and time-mean wall shear stress (TAWSS) directly affect the BP formation process, and the reaction process is as follows: Figure 4 As shown.

[0090] Platelet-bound (BP) and activated platelet (AP) are defined as diluent chemicals transported by fluids. Their local concentrations describe platelet activity. High AP levels indicate a higher probability of activated platelets in that area, a key factor in thrombosis. The concentrations of BP and AP can be calculated using the convection-diffusion transport equation.

[0091]

[0092] Among them, C i This represents the concentration of bound platelets (BP) or activated platelets (AP). When i = BP, C BP This represents the concentration of bound platelets (BP); when i = AP, C AP The concentration of activated platelets (AP) is represented by t; time is represented by t. This represents the gradient operator, used to represent spatial derivatives or divergence operations. The term represents the convection term, reflecting the spatial changes in the concentration of bound platelets (BP) or activated platelets (AP), and participates in the convection-diffusion process; u represents blood flow velocity; D p This represents the diffusion coefficient of platelets, taking into account the shear enhancement effect. 0.5|BP||AP| indicates the activating effect of activated platelets on bound platelets; This indicates the activating effect of shear stress on bound platelets; μ represents the shear rate; μ represents blood viscosity.

[0093] Coagulation factors (C) form at sites of lower wall shear stress, involving a cascade of platelet aggregation. High shear rates promote platelet activation, while low shear rates promote coagulation factor stagnation. Higher concentrations of activated platelets and coagulation factors result in slower blood flow and a greater rate of bound platelet (BP) formation. Bound platelets (BP) represent thrombus formation. The formula for calculating bound platelets (BP) is:

[0094]

[0095]

[0096] Where BP represents the concentration of bound platelets; t represents time; k BP Φ represents the rate constant of platelet-binding cell production; this parameter determines the rate at which platelets are produced. BP It represents the saturation function of platelet aggregation during thrombus formation, reflecting the degree of platelet aggregation; This represents the shear regulation function; high shear promotes diffusion; C AP C represents the concentration of activated platelets, which promotes the production of bound platelets; D represents coagulation factor; ceff The effective diffusion coefficient is represented by ΔC, which is the diffusion rate under shear rate regulation; ΔC represents the diffusion term of the coagulation factor; k c The rate constant representing the formation rate of coagulation factors determines the rate at which coagulation factors are formed; Φ C c represents the concentration of the clotting factor; c represents the saturation concentration; c t Characteristic values ​​representing saturation concentration; RRT t The characteristic value representing the relative retention time of platelets is used to normalize the saturation function; RRT represents the relative retention time of platelets; D C Indicates the molecular diffusion coefficient of coagulation factors; BP t Characteristic values ​​representing the concentration of bound platelets; Indicates shear rate; Characteristic values ​​representing shear rate.

[0097] In this embodiment, k c =200nM / s; k BP =2×10 -5 nM / s, BP t =20nM; c t =1nM; RRT t =0.9; D C =10 -8 m 2 / s.

[0098] Step S4.5: Determine whether the thrombus has completely formed based on clotting factors.

[0099] Coagulation factor (C) promotes the aggregation of activated platelets (AP) and the formation of bound platelets (BP). Thrombi will only begin to grow from the wall in cells where the time-mean wall shear stress (TAWSS) is below 0.2 Pa and the platelet-bound (BP) concentration is less than 200 nM. When a thrombus is fully formed, the boundary input of coagulation factor (C) becomes zero.

[0100] The formula for calculating coagulation factor (C) flux is:

[0101]

[0102] Where n represents the direction of the unit normal vector perpendicular to the blood vessel wall, i.e., the direction of the wall normal.

[0103] Thrombi continuously accumulate on the vessel wall elements. Combining thrombus density with the calculation of thrombus volume growth rate, dynamic mesh displacement, and other related results can be obtained. Dynamic mesh deformation includes boundary recalculation and mesh redistribution. For example... Figure 5 As shown, for a triangular surface element, each surface cell stores the corresponding thrombus height. Subsequently, the thrombus height is calculated based on the surface cell normal vector and projected onto the x, y, and z directions in the global coordinate system. After each time step, all nodes on the thrombus surface are updated.

[0104]

[0105]

[0106] in, This represents the coordinate position of the i-th node at the current time step on the blood vessel wall; The coordinates of the i-th node at the next time step on the vessel wall represent the location of the thrombus after its growth; t0 represents the current time step; Δt represents the time step length; Δh represents the time step size. i θ represents the height increment of the i-th node, i.e., the height increase of the thrombus within the time step Δt; i The angle between the normal vector of the thrombus surface at the i-th node and the coordinate axis of the global coordinate system determines the projection length of the height increment on the coordinate axis. Formula (11) updates the coordinate position of the node by calculating the projection of the thrombus height increment on the global coordinate system, thereby simulating the growth process of the thrombus on the blood vessel wall.

[0107] Step S4.6: After the thrombus has fully formed, calculate the blockage area of ​​the inlet cannula, the inlet pressure loss, and the area of ​​the thrombus formation zone between the endocardium and the inlet cannula.

[0108] The continuously growing thrombus reduces the flow interface area of ​​the LVAD (Low Voltage Amplifier), thereby increasing flow resistance. The calculated thrombus volume is projected onto the inlet sleeve cross-section (i.e., the cross-section), and its occupied area is calculated; this is the blockage area A1 of the inlet sleeve. The smaller the blockage area A1, the less impact the thrombus has on blood flow. The inlet sleeve cross-section is set as the monitoring surface, and the pressure value of this cross-section is read in real time during the calculation. The pressure value of this cross-section before thrombus formation is used as a benchmark. After the thrombus calculation is completed, the pressure of this cross-section is read again. The difference between this pressure value and the benchmark is the inlet pressure loss ΔP before and after thrombus formation. The smaller the pressure loss ΔP, the less impact the thrombus has on blood flow.

[0109] After thrombosis calculation, the number and area of ​​cells with a time-averaged wall shear stress (TAWSS) below 0.2 Pa and a platelet count (BP) below 200 nM were counted. The number of cells multiplied by the cell area was defined as the thrombus formation area A3 between the endocardium and the inlet cannula. The smaller the thrombus formation area A3, the lower the risk of thrombosis.

[0110] By repeating the above calculation process, we can obtain the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the thrombosis area between the endocardium and the inlet cannula, under multiple sets of first sample parameters.

[0111] Step S5: Construct a response surface model based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under the parameters of the first sample in different groups.

[0112] Response surface methodology (RSM) is an experimental design and optimization technique that explores the complex relationships between multiple independent and dependent variables by constructing mathematical models, and then seeks the optimal parameters based on these models. RSM can rapidly predict responses while maintaining a certain level of accuracy, thereby accelerating processes such as design optimization, parameter analysis, and uncertainty quantification.

[0113] In a specific embodiment of the present invention, a response surface model is constructed based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis under different group first sample parameters, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula. Specifically, this includes:

[0114] Step S5.1: Preprocess the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the thrombosis area between the endocardium and the inlet cannula, under the parameters of the first sample in different groups.

[0115] In this embodiment, the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the thrombosis area between the endocardium and the inlet cannula, are standardized under different group first sample parameters to eliminate dimensional differences.

[0116] Step S5.2: Construct a sample dataset based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under the preprocessed parameters of the first sample of different groups.

[0117] A sample dataset was constructed based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under the parameters of the first sample from different groups after preprocessing. The sample dataset was then divided into a training set and a validation set. Typically, the number of samples in the validation set is 20% to 30% of the number of samples in the sample dataset.

[0118] Step S5.3: Construct a response surface model based on the sample dataset.

[0119] Choose an appropriate response surface model, such as the Kriging surrogate model. For each sample, use structural parameters (i.e., the radius R of the inlet cannula, the insertion depth h, and the tilt angle α of the outlet plane) as inputs, and the corresponding blockage area of ​​the inlet cannula, the inlet pressure loss, and the thrombus formation area between the endocardium and the inlet cannula as expected outputs to train and test the selected response surface model.

[0120] Step S5.4: Verify the accuracy of the response surface model. If the accuracy requirement is met, output the response surface model. If the accuracy requirement is not met, optimize the strategy for constructing the response surface model and go to step S2 until the response surface model meets the accuracy requirement.

[0121] The accuracy of the response surface model is verified using a validation set. Specifically, the structural parameters of each sample in the validation set are input into the response surface model to obtain the response output. Evaluation metrics are calculated based on the blockage area and inlet pressure loss of the inlet cannula corresponding to the response output in the validation set, as well as the thrombus formation area between the endocardium and the inlet cannula. These evaluation metrics are used to determine whether the accuracy of the response surface model meets the accuracy requirements. In this embodiment, the evaluation metrics include mean squared error, relative error, and coefficient of determination.

[0122] Strategies for optimizing response surface model construction include at least one of the following: increasing the number of parameter sets in the first sample, optimizing the sampling strategy (e.g., adaptive sampling for high error intervals), trying more complex response surface models (e.g., Gaussian process regression, deep neural networks), and adjusting the parameters of the response surface model. Adjusting the parameters of the response surface model can be done by adjusting the kernel function parameters in the Kriging surrogate model or adjusting the number of hidden layers in a deep neural network. Alternatively, a separate sub-response surface model can be constructed for high error intervals to improve local prediction accuracy.

[0123] Numerical simulation of the blood flow to thrombus process is time-consuming. Obtaining the optimal combination of different structural parameters of the inlet cannula often requires hundreds or thousands of calculations, resulting in significant computational inefficiency. To address this problem, this invention performs numerical simulations on a small number of first-sample parameters, constructs a response surface model based on the simulation results, and then uses this accurate model to predict and calculate a large number of second-sample parameters. This yields the blockage area and inlet pressure loss of the inlet cannula under numerous different sets of structural parameters, as well as the thrombus formation area between the endocardium and the inlet cannula, significantly reducing computational time and improving efficiency.

[0124] Step S6: Sample the range of structural parameters of the inlet sleeve to generate M sets of second sample parameters; wherein each set of second sample parameters includes multiple structural parameters of the inlet sleeve, and M is much larger than N.

[0125] The size of the number of sample parameters M is related to the optimization accuracy of the inlet sleeve structure. The larger M is, the higher the accuracy of the optimized inlet sleeve structure, and the longer the optimization time. Conversely, the smaller M is, the lower the accuracy of the optimized inlet sleeve structure, and the shorter the optimization time. The value of M is much larger than N. In this embodiment, "much larger than" means more than 10 times.

[0126] Step S7: Use response surface methodology to obtain the blockage area and inlet pressure loss of the inlet cannula, as well as the thrombus formation area between the endocardium and the inlet cannula, under different groups of second sample parameters.

[0127] The parameters of each group of second samples are input into the response surface model to obtain the corresponding blockage area and inlet pressure loss of the inlet cannula, as well as the thrombus formation area between the endocardium and the inlet cannula. In other words, the blockage area and inlet pressure loss of the inlet cannula, as well as the thrombus formation area between the endocardium and the inlet cannula, are obtained under the parameters of the second samples of group M.

[0128] Step S8: Using an optimization algorithm, find the optimal group of second sample parameters from the blockage area and inlet pressure loss of the inlet cannula under different group second sample parameters, as well as the thrombus formation area between the endocardium and the inlet cannula, to obtain the optimal structural parameters of the inlet cannula.

[0129] In this embodiment, the optimization algorithm employs a multi-objective genetic algorithm (MOGA). Specifically, the MOGA searches for the Pareto optimal solution set and Pareto front from the blockage area and inlet pressure loss of the inlet cannula, as well as the thrombus formation area between the endocardium and the inlet cannula, under M sets of second sample parameters. One or more sets of second sample parameters are then obtained from the Pareto front as the optimal structural parameters for the inlet cannula. This results in an LVAD inlet cannula structure that maintains relatively stable performance parameters such as the pressure-flow curve, pump efficiency, and flow field distribution under the geometric uncertainties caused by thrombus formation.

[0130] This invention accelerates the thrombus formation process by adjusting the diffusion coefficient and activation response coefficient of platelets and coagulation factors, thus achieving accelerated computation of the thrombosis process. The computation is performed using the geometry-mesh-fluent-optislang component in Ansys Workbench. Based on the input and output of the numerical simulation method, a fitting-like algorithm is generated to approximate the relationship between structural parameters and outputs (i.e., the blockage area of ​​the inlet cannula and the inlet pressure loss, as well as the thrombus formation area between the endocardium and the inlet cannula), i.e., the response surface model. A multi-objective genetic algorithm (MOGA) is used to find the Pareto optimal solution set and Pareto front in the prediction results of the response surface model. Finally, one or more sets of structural parameters are obtained from the Pareto front as the optimal structural parameters. In other words, this invention seeks a left ventricular assist system inlet cannula structure that maintains relatively stable performance parameters such as pressure-flow curves, pump efficiency, and flow field distribution under the geometric uncertainties caused by thrombosis.

[0131] Example 2

[0132] This invention also provides an electronic device, which includes a memory, a processor, and a computer program / instructions stored in the memory. The processor executes the computer program / instructions to implement the LVAD inlet sleeve structure optimization method of this invention.

[0133] 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 read-only memory (ROM) or loaded from a storage portion into random access memory (RAM). The processor can be a multi-core processor or may contain multiple processors. In some embodiments, the processor may include a general-purpose main processor and one or more specialized coprocessors, such as a central processing unit, graphics processing unit (GPU), neural network processor (NPU), digital signal processor (DSP), etc. Various programs and data required for device operation are also stored in RAM. The processor, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0134] The processor and memory described above are used together to execute programs / instructions stored in the memory. When the program / instructions are executed by the computer, they can implement the methods, steps, or functions described in the above embodiments.

[0135] Although not shown, embodiments of the present invention also provide a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the LVAD inlet sleeve structure optimization method of the present invention.

[0136] Storage media in embodiments of the present invention include articles that are permanent or non-permanent, removable or non-removable, and can store information by any method or technology. 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 technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information that can be accessed by a computing device.

[0137] Readable storage media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. 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 technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient media, such as modulated data signals and carrier waves.

[0138] Although not shown, embodiments of the present invention also provide a computer program product, including: a computer program / instructions that, when executed by a processor, implement the LVAD inlet sleeve structure optimization method of the present invention.

[0139] The above description only discloses specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or modifications that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing the structure of an LVAD inlet sleeve, characterized in that, The optimization method includes: Construct a parameter-controllable geometric model of the heart-entry cannula; The range of structural parameters of the inlet sleeve is sampled to generate N sets of first sample parameters; each set of first sample parameters includes multiple structural parameters of the inlet sleeve. The inlet cannula structural parameters in the heart-inlet cannula geometric model are changed according to the parameters of the first sample in different groups, and the mesh is divided to obtain the heart-inlet cannula mesh model under the parameters of the first sample in each group. Simulation calculations were performed based on the heart-inlet cannula mesh model under the parameters of the first sample in each group to obtain the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula. A response surface model was constructed based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under the parameters of the first sample in different groups. The range of structural parameters of the inlet sleeve is sampled to generate M sets of second sample parameters; each set of second sample parameters includes multiple structural parameters of the inlet sleeve, and M is much larger than N; The response surface model was used to obtain the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the thrombosis zone area between the endocardium and the inlet cannula, under different groups of second sample parameters. An optimization algorithm was used to find the optimal set of second sample parameters from the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under different sets of second sample parameters, so as to obtain the optimal structural parameters of the inlet cannula.

2. The LVAD inlet sleeve structure optimization method according to claim 1, characterized in that, The construction of a parameter-controllable heart-entry cannula geometric model specifically includes: Construct a solid cardiac model based on cardiac CT images; A three-dimensional coordinate system is constructed with the apex of the heart model as the origin. A circular ring is constructed with the origin of the three-dimensional coordinate system as the center. The ring is stretched to form an inlet cannula, which is then connected to the apex of the heart, generating a heart-inlet cannula geometric model with controllable parameters.

3. The LVAD inlet sleeve structure optimization method according to claim 1, characterized in that, The Latin hypercube sampling method is used to sample the range of structural parameters of the inlet sleeve, generating N sets of first sample parameters or M sets of second sample parameters.

4. The LVAD inlet sleeve structure optimization method according to claim 1, characterized in that, Simulation calculations were performed based on the heart-inlet cannula mesh model under the parameters of the first sample in each group to obtain the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, specifically including: Based on the blood flow pulsation velocity waveform at the patient's LVAD inlet, the inlet boundary conditions of the heart-inlet cannula mesh model are set to simulate blood flow motion; Calculate the relative retention time and time-averaged wall shear stress of platelets during blood flow; After running multiple cardiac cycles, thrombus components were added, and coagulation factors and bound platelets were calculated. Determine whether a thrombus has completely formed based on clotting factors; Once the thrombus has fully formed, calculate the blockage area of ​​the inlet cannula, the inlet pressure loss, and the area of ​​the thrombus formation zone between the endocardium and the inlet cannula.

5. The LVAD inlet sleeve structure optimization method according to claim 4, characterized in that, The formulas for calculating the relative retention time and time-averaged wall shear stress of platelets are as follows: Where RRT represents the relative retention time of platelets; T represents the duration of the cardiac cycle; τ w The instantaneous shear stress represents the stress at different time points; t represents time; TAWSS represents the time-averaged wall shear stress.

6. The LVAD inlet sleeve structure optimization method according to claim 4, characterized in that, The formula for calculating the bound platelets is as follows: Where BP represents the concentration of bound platelets; t represents time; k BP Φ represents the rate constant for the formation of bound platelets; BP The saturation function representing the aggregation of bound platelets during thrombus formation; Represents the shearing adjustment function; C AP Indicates the concentration of activated platelets; C represents coagulation factors; D ceff Indicates the effective diffusion coefficient; ΔC represents the diffusion term of the coagulation factor; k c Φ represents the rate constant of coagulation factor production; C c represents the concentration of the clotting factor; c represents the saturation concentration; c t Characteristic values ​​representing saturation concentration; RRT t The characteristic value representing the relative retention time of platelets; RRT represents the relative retention time of platelets; D C Indicates the molecular diffusion coefficient of coagulation factors; BP t Characteristic values ​​representing the concentration of bound platelets; Indicates shear rate; Characteristic values ​​representing shear rate.

7. The LVAD inlet sleeve structure optimization method according to any one of claims 1 to 6, characterized in that, A response surface model was constructed based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under different group first sample parameters. This model included: Preprocessing was performed on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the thrombosis zone area between the endocardium and the inlet cannula, under different group first sample parameters. A sample dataset was constructed based on the blockage area and inlet pressure loss of the inlet cannula after thrombosis, as well as the area of ​​the thrombosis zone between the endocardium and the inlet cannula, under the parameters of the first sample of different groups after preprocessing. Construct a response surface model based on the aforementioned sample dataset; Verify the accuracy of the response surface model. If the accuracy requirement is met, output the response surface model. If the accuracy requirement is not met, optimize the strategy for constructing the response surface model. Repeat the steps of obtaining the blockage area and inlet pressure loss of the inlet cannula after thrombosis, the thrombosis area between the endocardium and the inlet cannula, and constructing the response surface model under different groups of first sample parameters, until the response surface model meets the accuracy requirement.

8. An electronic device comprising a memory, a processor, and a computer program / instructions stored in the memory, characterized in that, The processor executes the computer program / instructions to implement the LVAD inlet sleeve structure optimization method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instruction is executed by the processor, it implements the LVAD inlet sleeve structure optimization method as described in any one of claims 1 to 7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the LVAD inlet sleeve structure optimization method as described in any one of claims 1 to 7.

Images