Atrial fibrillation thrombus prediction method, device and readable storage medium

Abstract

This invention discloses a method, device, and readable storage medium for predicting atrial fibrillation thrombosis, relating to the fields of medical image processing and biofluid dynamics numerical simulation. The method includes: acquiring a three-dimensional fluid dynamics model of the left atrium including the left atrial appendage, and blood rheological characteristics based on reconstructed cardiac images of the individual under test. The blood rheological characteristics include hematocrit and whole blood viscosity at at least three different shear rates. Based on the blood rheological characteristics, a constitutive relation characterizing the individual's blood viscosity as a function of shear rate is determined. The constitutive relation is applied to the model and solved to obtain hemodynamic indices including at least the shear rate distribution and the blood residence time distribution within the left atrial appendage. Thrombosis risk indices are determined for each region of the left atrial appendage based on these indices to assess the risk of thrombosis formation in the left atrial appendage. This invention reduces the deviation of the solved flow field from the individual's actual flow field, improves the stability of the assessment results, and provides a spatial distribution of risk.

Description

Technical Field

[0001] This invention relates to the field of medical image processing and biofluid dynamics numerical simulation technology, specifically to a method, device, and readable storage medium for predicting atrial fibrillation thrombosis. Background Technology

[0002] Atrial fibrillation is a common arrhythmia in clinical practice, and its serious complication is thromboembolic events. Thromboembolic events are often associated with non-valvular atrial fibrillation, and thrombi often originate from the left atrial appendage. Therefore, assessing the risk of left atrial appendage thrombosis is a fundamental technical requirement in related clinical work.

[0003] In the existing technology, there are two main methods for assessing the risk of left atrial appendage thrombosis.

[0004] The first approach scores individuals based on common clinical characteristics such as age, gender, and medical history, and assigns a qualitative risk classification according to rules pre-determined by population statistics.

[0005] The second approach involves reconstructing a three-dimensional geometric model of the left atrium based on cardiac images of the individual being evaluated. Computational fluid dynamics is then applied to this model to obtain the flow field within the left atrium. A risk assessment result is then given based on a specific hemodynamic quantity within this flow field. In this approach, the blood viscosity is determined by a pre-defined non-Newtonian constitutive equation with fixed parameters. These fixed parameters are derived from literature or population statistics and are independent of the individual's own blood rheological state.

[0006] However, the above two methods have the following technical drawbacks:

[0007] The above methods usually rely on a single hemodynamic quantity as the basis for assessment. Local fluctuations in this single quantity can cause significant changes in the assessment results, resulting in insufficient stability of the assessment results. Furthermore, the assessment results are mostly holistic qualitative conclusions and cannot provide information on the spatial distribution of risk within the left atrial appendage.

[0008] Therefore, how to reduce the deviation of the risk assessment results of left atrial appendage thrombosis from the actual blood flow status of individuals, improve the stability of the assessment results, and provide the spatial distribution of the risk are technical problems that urgently need to be solved in this field. Summary of the Invention

[0009] The technical problem to be solved by this invention is to reduce the deviation of the risk assessment results of left atrial appendage thrombosis from the actual blood flow state of an individual, improve the stability of the assessment results, and provide the spatial distribution of the risk in the left atrial appendage.

[0010] To solve the above-mentioned technical problems, the present invention adopts the following technical solution.

[0011] The first aspect of this invention provides a method for predicting thrombosis in atrial fibrillation, comprising the following steps: acquiring a three-dimensional hydrodynamic model of the left atrium including the left atrial appendage and blood rheological characteristics based on cardiac imaging reconstruction of an individual under test, wherein the blood rheological characteristics include hematocrit and whole blood viscosity at at least three different shear rates; determining a constitutive relation characterizing the change of blood viscosity with shear rate in the individual under test based on the acquired blood rheological characteristics; applying the constitutive relation to the three-dimensional hydrodynamic model of the left atrium and solving it to obtain hemodynamic indices within the left atrial appendage, wherein the hemodynamic indices include at least shear rate distribution and residence time distribution of blood within the left atrial appendage; and determining a thrombosis risk index for each region of the left atrial appendage based on the hemodynamic indices to assess the risk of thrombosis formation in the left atrial appendage.

[0012] Furthermore, the step of applying the constitutive relation to the three-dimensional fluid dynamics model of the left atrium and solving it to obtain hemodynamic parameters within the left atrial appendage includes: using the constitutive relation as the viscosity constitutive of blood in the three-dimensional fluid dynamics model of the left atrium, and numerically solving the model in combination with preset boundary conditions to obtain the flow field within the left atrial appendage; determining the shear rate distribution based on the flow field; and determining the residence time distribution by tracking blood particles based on the flow field.

[0013] Furthermore, the hemodynamic parameters also include the time-averaged wall shear stress and the oscillatory shear index of the left atrial appendage wall. The time-averaged wall shear stress represents the average level of wall shear stress within a cardiac cycle, and the oscillatory shear index represents the degree to which the wall shear stress deviates from the main blood flow direction within a cardiac cycle.

[0014] Furthermore, the hemodynamic parameters also include endothelial cell activation potential, which is determined based on the oscillatory shear index and the time-averaged wall shear stress, and is used to characterize the tendency of thrombosis caused by endothelial damage to the left atrial appendage wall.

[0015] Further, determining the thrombosis risk index for each region of the left atrial appendage includes: obtaining the blood flow velocity distribution within the left atrial appendage through the solution; obtaining the degree of non-Newtonian blood properties based on the shear rate distribution and the blood viscosity distribution determined by the constitutive relation; obtaining the degree of blood stasis based on the blood flow velocity distribution and the residence time distribution; obtaining the degree of endothelial damage based on the endothelial cell activation potential; and determining the thrombosis risk index for each region of the left atrial appendage based on the degree of non-Newtonian blood properties, the degree of blood stasis, and the degree of endothelial damage.

[0016] Furthermore, determining the thrombosis risk index of each region of the left atrial appendage based on the degree of non-Newtonian blood properties, the degree of blood stasis, and the degree of endothelial damage includes: normalizing the degree of non-Newtonian blood properties, the degree of blood stasis, and the degree of endothelial damage respectively, and weighting and fusing the normalized results to obtain the thrombosis risk index of each region of the left atrial appendage.

[0017] Furthermore, the solution includes: applying pulsating boundary conditions characterizing atrial fibrillation blood flow to the three-dimensional hydrodynamic model of the left atrium, wherein the pulsating boundary conditions include setting inlet conditions for the pulmonary vein opening and setting outlet conditions for the mitral valve opening, and solving the full flow field of the model based on the pulsating boundary conditions.

[0018] Furthermore, the at least three different shear rates include a first shear rate, a second shear rate, and a third shear rate that increase sequentially. The first shear rate, the second shear rate, and the third shear rate constitute the low-shear rate, the medium-shear rate, and the high-shear rate in the whole blood viscosity test, respectively. The ratio of the second shear rate to the first shear rate is not less than 5, and the ratio of the third shear rate to the second shear rate is not less than 3, so that the first shear rate, the second shear rate, and the third shear rate are located in different segments of the blood viscosity versus shear rate curve.

[0019] Furthermore, the first shear rate is 1s. -1 up to 5s -1 The second shear rate is 30s. -1 up to 60s -1 The third shear rate is 180s. -1 up to 220s -1 .

[0020] Further, determining the constitutive relation characterizing the change of blood viscosity of the test individual with shear rate includes: constructing a general formula for a non-Newtonian constitutive equation, substituting the whole blood viscosity of the test individual at at least three different shear rates and the hematocrit into the general formula, and fitting the constitutive relation of the test individual; the general formula is: ; Where η is the dynamic viscosity of blood, γ is the shear rate, used to describe the magnitude of the velocity gradient in the fluid, and ε RBC is hematocrit, which represents the percentage of red blood cells in the blood volume; coefficients a, λ, and b are undetermined coefficients characterizing the shear thinning properties of blood, and coefficients k and k1 are coefficients characterizing the effect of hematocrit on blood viscosity.

[0021] Furthermore, the fitting process employs a nonlinear least squares method to determine the undetermined coefficients in the general formula.

[0022] Furthermore, the blood rheological characteristics also include the erythrocyte aggregation index and the erythrocyte deformability index; the coefficients k and k1 are determined according to the erythrocyte aggregation index and the erythrocyte deformability index according to a preset empirical relationship, and the coefficients a, λ, and b are determined by fitting the whole blood viscosity at at least three different shear rates.

[0023] Furthermore, the empirical relationship states that the coefficient k increases with the increase of the erythrocyte aggregation index, and the coefficient k1 decreases with the increase of the erythrocyte deformability index.

[0024] Furthermore, the blood rheological characteristics also include at least one of plasma fibrinogen concentration and plasma viscosity, as alternatives to or supplements to the erythrocyte aggregation index and the erythrocyte deformability index, used to determine the coefficient k and the coefficient k1.

[0025] Furthermore, in the solution, the blood is represented by a single-phase homogeneous non-Newtonian blood model, and the viscosity of the blood is determined by the constitutive relation.

[0026] A second aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the atrial fibrillation thrombosis prediction method described in any of the preceding claims.

[0027] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the atrial fibrillation thrombosis prediction method described in any of the preceding claims.

[0028] The beneficial effects of this invention are as follows:

[0029] First, by obtaining the hematocrit of the test individual and the whole blood viscosity at at least three different shear rates, and determining the constitutive relation characterizing the blood viscosity of the test individual as a function of shear rate based on these blood rheological characteristics, the viscosity variation with shear rate used in the numerical solution is consistent with the measured rheological data of the test individual, rather than being taken from a fixed-parameter constitutive equation unrelated to the individual. Furthermore, the fitting is performed on the premise that the number of measured points is not less than the number of undetermined coefficients fitted, and the obtained constitutive relation is not underdetermined. Therefore, in the low and medium shear rate range of the left atrium and left atrial appendage under atrial fibrillation, the viscosity characterization deviation is significantly reduced compared with the use of a fixed-parameter constitutive equation, and the deviation of the solved flow field from the actual flow field of the individual is reduced accordingly.

[0030] Secondly, since the at least three different shear rates constitute the low-shear, medium-shear, and high-shear rates in the whole blood viscosity test, and the difference between adjacent rates reaches a predetermined multiple, they are located in different segments of the blood viscosity versus shear rate curve. This allows for simultaneous constraints on the parameters characterizing the viscosity level at low shear rates, the parameters characterizing the initiation position of shear thinning, and the parameters characterizing the shear thinning rate in the constitutive relation. This avoids situations where the number of measurement points is insufficient, or where, although there are three measurement points, they are concentrated in the same segment, resulting in non-unique fitting results and distortion of the obtained constitutive relation in other segments. This improves the stability of the determined constitutive relation within the range of shear rates involved in the tested individual. Furthermore, the three shear rates mentioned above correspond to the three test levels typically given in clinical whole blood viscosity test reports, eliminating the need for additional testing items to implement this invention.

[0031] Third, since the three-dimensional fluid dynamics model of the left atrium is obtained by segmenting and reconstructing the heart image of the individual being tested, its geometry and the blood rheological properties characterized by the constitutive relation belong to the same individual being tested. The two complement each other in the solution: the geometry of the left atrial appendage determines the spatial location and range of the low shear rate region inside it, while the constitutive relation maps the local shear rate to the local viscosity, and the non-Newtonian properties of blood are most significant in the low shear rate region; therefore, when the geometry and the constitutive relation are taken from the same individual being tested, the location of the low shear rate region and the characterization of the viscosity in that region match each other, and the deviation of the obtained thrombosis risk index is further reduced compared to when either one is individually personalized.

[0032] Fourth, since the hemodynamic parameters include both shear rate distribution and residence time distribution, and may further include time-averaged wall shear stress, oscillatory shear index, and endothelial cell activation potential, the determined thrombosis risk index simultaneously reflects the physical causes of blood non-Newtonian properties, blood stasis, and endothelial injury. Compared to assessment methods based solely on a single hemodynamic quantity, the impact of local fluctuations in a single parameter on the assessment results is significantly reduced, thus improving the stability of the assessment results.

[0033] Fifth, since the thrombosis risk index is output on a per-region basis for each area of ​​the left atrial appendage, the assessment result is the spatial distribution of risk within the left atrial appendage rather than a single overall conclusion, which can locate high-risk areas within the left atrial appendage.

[0034] To make the above and other objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a flowchart illustrating the atrial fibrillation thrombosis prediction method provided in Embodiment 1 of the present invention;

[0037] Figure 2 This is a curve showing the change of apparent viscosity of blood with shear rate, and a schematic diagram showing the correspondence between the three shear rate levels (low shear, medium shear, and high shear) for whole blood viscosity testing and the different segments of the curve.

[0038] Figure 3 A schematic diagram showing the microscopic state of red blood cells under different shear rates;

[0039] Figure 4 This is a schematic diagram of the structure of the electronic device provided in Embodiment 3 of the present invention.

[0040] Explanation of reference numerals in the attached figures:

[0041] 100 - Electronic device; 101 - Memory; 102 - Processor; 103 - Communication bus. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0043] It should be noted that in the description of this invention, the terms "first," "second," etc., are used only for descriptive purposes and to distinguish similar objects; there is no order between them, nor should they be construed as indicating or implying relative importance. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0044] It should also be noted that the processing objects of the method described in this invention are already obtained cardiac imaging data and blood rheology detection data, and its execution process does not include the step of obtaining the above data directly from a living human or animal body; its output result is a quantitative indicator characterizing the blood flow state in the left atrial appendage, namely the thrombosis risk index. This indicator is an intermediate technical quantity obtained from hemodynamic calculations and is not a disease diagnosis result.

[0045] Example 1

[0046] See Figure 1 This embodiment provides a method for predicting atrial fibrillation thrombosis, including the following steps S1 to S4.

[0047] S1. Obtain a three-dimensional hydrodynamic model of the left atrium including the left atrial appendage and blood rheological characteristics based on the reconstruction of the heart image of the individual to be tested. The blood rheological characteristics include hematocrit and whole blood viscosity at at least three different shear rates.

[0048] Specifically, step S1 includes the following sub-steps S11 and S12.

[0049] S11. The cardiac images of the individual to be tested are segmented, reconstructed in three dimensions, and meshed to obtain a three-dimensional hydrodynamic model of the left atrium including the left atrial appendage. The cardiac images can be enhanced computed tomography images of the heart, and the reconstruction slice thickness should preferably not exceed 1 mm to ensure the distinguishability of small structures such as the pectinate muscle of the left atrial appendage; the cardiac images can also be cardiac magnetic resonance images or three-dimensional transesophageal echocardiogram images.

[0050] The segmentation can be based on the grayscale difference between the contrast agent-filled area in the left atrial cavity and the surrounding myocardial tissue. Threshold segmentation combined with region growing is used to extract the blood flow domains corresponding to the left atrial cavity and left atrial appendage layer by layer from the cardiac image, and the segmentation results are manually verified.

[0051] The three-dimensional geometric reconstruction can involve converting the segmented blood flow domain mask into a surface geometric model composed of triangular facets, and then smoothing and repairing the surface geometric model to eliminate the step-like artifacts caused by layered segmentation. Furthermore, the end faces of each pulmonary vein opening and the mitral valve opening can be planarized and extended along the normal direction of each end face by a predetermined length, for example, 3 to 5 times the equivalent diameter of the opening, to reduce the impact of the applied boundary conditions on the flow field inside the left atrial cavity.

[0052] The mesh generation can be achieved by discretizing the fluid domain enclosed by the aforementioned three-dimensional geometric model to obtain the three-dimensional hydrodynamic model of the left atrium. In this embodiment, a tetrahedral mesh can be used to discretize the interior of the fluid domain, and a boundary layer mesh can be generated near the wall. The boundary layer mesh can have, for example, 5 layers, with a first layer thickness of 0.05 mm, a growth rate of 1.2, and a total mesh count of, for example, 1.5 million to 4 million. After the mesh generation is completed, mesh independence verification can be performed. That is, the periodic average flow velocity at the cross-section of the left atrial appendage is used as the monitoring quantity. If the relative change of this monitoring quantity is less than 2% after doubling the number of meshes, the mesh generation is considered to meet the calculation accuracy requirements.

[0053] S12. Obtain the blood rheological characteristics of the individual to be tested. The blood rheological characteristics include hematocrit and whole blood viscosity at at least three different shear rates. The whole blood viscosity at each shear rate is obtained by measuring the whole blood sample of the individual to be tested using a rotational viscometer; the hematocrit represents the volume percentage of red blood cells in the blood and can be obtained by routine blood tests or centrifugation.

[0054] In this step, the blood rheological characteristics also include the erythrocyte aggregation index and the erythrocyte deformability index. The erythrocyte aggregation index characterizes the ability of erythrocytes to aggregate and form rouleaux structures under low shear conditions, and the erythrocyte deformability index characterizes the ability of erythrocytes to change their shape under shear. Both can be directly obtained from whole blood rheology test reports.

[0055] Optionally, the blood rheological characteristics may further include at least one of plasma fibrinogen concentration and plasma viscosity, as a substitute for or supplement to the erythrocyte aggregation index and the erythrocyte deformability index. The plasma fibrinogen concentration can be obtained by coagulation function testing, and the plasma viscosity can be obtained directly by plasma viscosity measurement, or calculated from the plasma fibrinogen concentration according to a preset empirical relationship. The above items in step S2 are used to determine the coefficients characterizing the effect of hematocrit on viscosity in the general formula.

[0056] The selection of the at least three different shear rates is explained below. See [link / reference] Figure 2 and Figure 3 The viscosity of blood changes with shear rate in three segments: On the side with a low shear rate, red blood cells aggregate into rouleaux structures, and the viscosity is close to zero shear viscosity, with a gradual change as the shear rate increases; this segment is the low shear rate segment. As the shear rate continues to increase, the red blood cell aggregates deaggregate and gradually orient and deform, and the viscosity decreases significantly with the increase of the shear rate; this segment is the shear-thinning transition segment. On the side with a high shear rate, the orientation and deformation of red blood cells tend to saturate, and the viscosity tends to be an infinite shear viscosity plateau, with a gradual change again with the increase of the shear rate; this segment is the high shear rate segment.

[0057] To ensure that the determined constitutive relation matches the measured data across the range of shear rates involved in the tested individual, the at least three different shear rates include a first shear rate, a second shear rate, and a third shear rate that increase sequentially. The ratio of the second shear rate to the first shear rate is not less than 5, and the ratio of the third shear rate to the second shear rate is not less than 3. This places the three rates in different segments of the curve, representing low-shear, medium-shear, and high-shear rates, respectively. If the three shear rate measurement points are close to each other and concentrated in the same segment, the curve's trajectory in other segments will not be constrained by the measured data, and the determined constitutive relation will be significantly distorted when extrapolated to other segments. The correspondence between each measurement point and the coefficients in the general formula described in step S2 is detailed in step S2.

[0058] In this embodiment, the first shear rate is 1s. -1 up to 5s -1 The second shear rate is 30s. -1 up to 60s -1 The third shear rate is 180s. -1 up to 220s -1 The three intervals mentioned above correspond to the low-shear, medium-shear, and high-shear test levels typically given in clinical whole blood viscosity test reports, thus eliminating the need for additional testing items. More specifically, the first shear rate is 1 second. -1 3s -1 Or 5s -1 The second shear rate is 30s. -1 Or 50s -1 The third shear rate is 200 s. -1 In this embodiment, the first shear rate is taken as 1s. -1 The second shear rate is 50s. -1 The third shear rate is 200s. -1 .

[0059] For some individuals to be tested, see [link to test results]. Figure 2 The first shear rate at the aforementioned low shear rate setting may still be on the low shear rate side of the shear-thinning transition zone and has not yet entered the region where viscosity changes gradually. Since the function of this measuring point is to constrain the value of the general formula on the low shear rate side, the purpose of this invention can be achieved as long as the three measuring points are located in different sections of the curve according to the aforementioned multiple relationship.

[0060] It should be noted that "acquiring" in this step includes both directly reading the above cardiac images and blood rheological features by the subject executing this method, and receiving the above cardiac images and blood rheological features from external systems such as image archiving and communication systems and test information systems by the subject executing this method. This invention does not limit the acquisition process of the above data itself.

[0061] S2. Based on the acquired blood rheological characteristics, determine the constitutive relation characterizing the change of blood viscosity with shear rate in the individual under test.

[0062] In this embodiment, step S2 includes: constructing a general formula for a non-Newtonian constitutive equation, substituting the whole blood viscosity of the individual under test at at least three different shear rates into the general formula, and fitting the constitutive relation of the individual under test. The general formula is: ; In the formula, η is the dynamic viscosity of blood; γ is the shear rate, used to describe the magnitude of the velocity gradient in the fluid. ε RBC is the hematocrit, which represents the percentage of red blood cells in the blood volume, and its value is the hematocrit obtained in step S12; a, λ, b, k, and k1 are coefficients. The general formula is constructed by introducing a correction factor with hematocrit as the independent variable based on the cross-type shear-thinning constitutive equation.

[0063] In the general formula, the first term characterizes the shear-thinning property of blood viscosity, which decreases with increasing shear rate. Here, 'a' represents the viscosity level at low shear rates, 'λ' represents the characteristic time corresponding to shear-thinning, and 'b' represents the rate of shear-thinning. The second term characterizes the nonlinear amplification effect of hematocrit on blood viscosity, where 'k' and 'k1' are the quadratic and linear coefficients of the hematocrit. Since hematocrit and whole blood viscosity at low, medium, and high shear rates vary among individuals, these coefficients must be determined based on the measured data for each individual.

[0064] As can be seen from the general formula, in step S12, the measuring points located on the low shear rate side mainly constrain the coefficient 'a', the measuring points located in the shear-thinning transition section mainly constrain the coefficients 'λ' and 'b', and the measuring points located on the high shear rate side are used to constrain the value of the general formula on that side, preventing the fitted curve from decreasing excessively in that interval. Therefore, when the three measuring points are located in different segments of the curve, the above coefficients can all be effectively constrained. If the three measuring points are concentrated in the same segment, at least one of the above coefficients will not be effectively constrained, and the fitting result will not be unique.

[0065] It should be noted that, in this embodiment, the coefficients k and k1 are predetermined as follows: the coefficients k and k1 are determined by the erythrocyte aggregation index and erythrocyte deformability index obtained in step S12 according to a preset empirical relationship, and the coefficient k increases with the increase of the erythrocyte aggregation index, while the coefficient k1 decreases with the increase of the erythrocyte deformability index, and the hematocrit correction factor is greater than zero. The empirical relationship can be pre-calibrated from multiple samples containing different hematocrits. When the erythrocyte aggregation index and the erythrocyte deformability index cannot be obtained, the coefficients k and k1 can also be determined by at least one of the plasma fibrinogen concentration and the plasma viscosity according to a preset empirical relationship.

[0066] After the coefficients k and k1 are predetermined in the manner described above, the whole blood viscosity of the individual to be tested at the at least three different shear rates is substituted into the general formula to obtain a set of equations with a, λ, and b as unknowns. The number of unknowns corresponds to the number of measurement points at the at least three different shear rates. The set of equations is well-determined or overdetermined.

[0067] In this embodiment, the fitting process uses a nonlinear least squares method to determine the undetermined coefficients in the general formula. Specifically, the objective function is the sum of squared residuals between the measured whole blood viscosity at each shear rate and the calculated value of the general formula. The Levenburg-Marquardt iterative algorithm is used to solve for the undetermined coefficients that minimize this objective function. The initial iteration value can be set as follows: 'a' is the ratio of the measured whole blood viscosity of the individual under the low shear rate to the hematocrit correction factor, λ is 1 s, and b is 0.6. The coefficients k and k1 do not participate in the iteration.

[0068] Substituting the obtained undetermined coefficients back into the general formula yields a constitutive relation applicable only to the individual being tested, characterizing the change in blood viscosity with shear rate. Validation on clinical samples shows that the constitutive relation determined in this manner achieves a goodness of fit of over 0.99 to the measured whole blood viscosity points; furthermore, the fitting is performed with a number of measurement points no less than the number of undetermined coefficients being fitted, and the resulting constitutive relation is not underdetermined.

[0069] It is understandable that when there are more than three shear rate measurement points corresponding to the obtained whole blood viscosity, each measurement point can participate in the fitting of coefficients a, λ, and b in the above manner. The more measurement points there are, the higher the characterization accuracy of the obtained constitutive relation in the corresponding shear rate range. At this time, the coefficients k and k1 are still predetermined in the above manner.

[0070] S3. Apply the constitutive relation to the three-dimensional hydrodynamic model of the left atrium and solve it to obtain the hemodynamic parameters in the left atrial appendage. The hemodynamic parameters include at least the shear rate distribution and the residence time distribution of blood in the left atrial appendage.

[0071] Specifically, step S3 includes the following sub-steps S31 to S35.

[0072] S31. The constitutive relation is used as the viscosity constitutive relation of blood in the three-dimensional hydrodynamic model of the left atrium. That is, in each iteration of the numerical solution, the viscosity of blood in each grid cell is calculated in real time according to the current shear rate of that grid cell by the constitutive relation, rather than being taken as a constant value. In this embodiment, the blood in the solution adopts a single-phase homogeneous non-Newtonian blood model, that is, whole blood is regarded as a single homogeneous fluid with a constant density, for example, 1055 kg / m³. 3 Its viscosity is determined by the constitutive relation.

[0073] S32. Apply pulsating boundary conditions characterizing atrial fibrillation blood flow to the three-dimensional hydrodynamic model of the left atrium. The pulsating boundary conditions include setting inlet conditions for the openings of the left superior pulmonary vein, right superior pulmonary vein, left inferior pulmonary vein, and right inferior pulmonary vein, and setting outlet conditions for the mitral valve opening. In this embodiment, each pulmonary vein opening can be set as a pressure inlet, and the mitral valve opening can be set as a velocity outlet. The velocity magnitude of the velocity outlet changes with time according to the blood flow velocity waveform at the mitral valve opening within one cardiac cycle, thereby characterizing the pulsating blood flow in atrial fibrillation. The walls of the left atrial cavity and the left atrial appendage can be set as rigid, non-slip walls.

[0074] S33. Solve the full flow field of the model based on the pulsating boundary conditions. In this embodiment, the flow model is laminar; the cardiac cycle is 0.8s; the time step is 0.005s; 10 cardiac cycles are calculated continuously, and the total simulation time is 8s. The continuity equation and momentum equation are discretized and solved iteratively using the finite volume method. The convergence criterion in each time step is that the root mean square residual of each equation is less than 1×10⁻⁶. -4 The calculation result of the last cardiac cycle is used as the analysis object to eliminate the influence of the initial conditions on the result. The flow field in the left atrial appendage is obtained from the solution, and the flow field includes the blood velocity and pressure at each time and spatial location in the left atrial appendage.

[0075] S34. Determine the shear rate distribution based on the flow field: Calculate the shear rate of each grid cell from the velocity gradient at each grid cell, and take the time average of the calculation results over a cardiac cycle to obtain the periodic average shear rate distribution in the left atrial appendage.

[0076] S35. Determine the residence time distribution by tracking blood particles based on the flow field: Release tracer particles that move with the flow field, integrate the trajectory of each tracer particle over time using the Lagrangian method, and calculate the residence time of each tracer particle in the left atrial appendage. In this embodiment, the residence time distribution is characterized by relative residence time; the larger the relative residence time, the longer the blood stays at that location.

[0077] Furthermore, the hemodynamic parameters also include the time-averaged wall shear stress and the oscillatory shear index of the left atrial appendage wall. The time-averaged wall shear stress is the average value of the wall shear stress over time within one cardiac cycle, representing the average level of wall shear stress within one cardiac cycle. The oscillatory shear index represents the degree to which the wall shear stress deviates from the main blood flow direction within one cardiac cycle, and its value ranges from 0 to 0.5. A larger value indicates a more drastic change in the direction of the wall shear stress within one cardiac cycle.

[0078] Furthermore, the hemodynamic parameters also include endothelial cell activation potential (EVP), which is determined based on the oscillatory shear index and the time-averaged wall shear stress. In this embodiment, the EVP is taken as the ratio of the oscillatory shear index to the time-averaged wall shear stress. A larger EVP indicates that the wall surface is simultaneously in a state of low wall shear stress and high oscillation, and the tendency for thrombosis due to endothelial injury is higher.

[0079] Furthermore, the blood stasis factor can be statistically analyzed from the flow field. The blood stasis factor is the ratio of the duration during which the shear rate of the left atrial appendage wall remains below a preset threshold to the duration of the cardiac cycle. The preset threshold is, for example, 5 seconds. -1 The higher the blood stasis factor, the more severe the blood stasis in that area.

[0080] S4. Based on the hemodynamic parameters, determine the thrombosis risk index for each region of the left atrial appendage to assess the risk of thrombosis in the left atrial appendage.

[0081] In this embodiment, the left atrial appendage is first divided into several regions. For example, the division is to divide the left atrial appendage into several segments along the direction from the mouth to the tip, or to use each grid unit on the wall of the left atrial appendage as a region. Then, the following sub-steps S41 to S44 are performed on each region.

[0082] S41. The blood flow velocity distribution within the left atrial appendage is obtained through the solution. Based on the shear rate distribution and the blood viscosity distribution determined by the constitutive relation, the degree of non-Newtonian properties of the blood is obtained. In this embodiment, the periodic average viscosity of this region is used as a characterizing quantity of the degree of non-Newtonian properties of the blood. The periodic average viscosity is calculated by substituting the periodic average shear rate of this region into the constitutive relation. The higher the periodic average viscosity, the stronger the promoting effect of the blood's non-Newtonian properties on thrombus formation in this region.

[0083] S42. The degree of blood stasis is obtained based on the blood flow velocity distribution and the residence time distribution. In this embodiment, the periodic average flow velocity and relative residence time of the region are used to characterize the degree of blood stasis: the lower the periodic average flow velocity and the longer the relative residence time, the higher the degree of blood stasis in the region.

[0084] S43. The degree of endothelial damage is obtained based on the endothelial cell activation potential. In this embodiment, the average area of ​​the endothelial cell activation potential on the wall surface of this region is used as a characterizing measure of the degree of endothelial damage.

[0085] S44. Normalize the degree of non-Newtonian characteristics of the blood, the degree of blood stasis, and the degree of endothelial damage respectively, and then weight and fuse the normalized results to obtain the thrombosis risk index of each region of the left atrial appendage.

[0086] The normalization process ensures that the values ​​of each characteristic quantity fall within the range of 0 to 1, with higher normalized values ​​indicating a higher risk of thrombosis. Specifically, for characteristic quantities where higher values ​​indicate higher risk, such as periodic average viscosity, relative residence time, and endothelial cell activation potential, normalization is performed using the formula "(current value - minimum value) / (maximum value - minimum value)". For characteristic quantities where higher values ​​indicate lower risk, such as periodic average flow velocity, normalization is performed using the formula "(maximum value - current value) / (maximum value - minimum value)". The maximum and minimum values ​​can be taken from the extreme values ​​of the characteristic quantity in all regions of the left atrial appendage of the same individual, or from the extreme values ​​of the characteristic quantity in a pre-established sample library. The former is used to give the spatial distribution of the relative risk levels among different regions within the left atrial appendage of the individual, while the latter ensures that the thrombosis risk indices obtained by normalization using the same sample library fall on the same scale.

[0087] The weighted fusion can be achieved by fusing the normalized blood non-Newtonian property degree N, blood stasis degree S, and endothelial injury degree E according to the following formula to obtain the thrombosis risk index TRI for that region: TRI = w1 × N + w2 × S + w3 × E; where w1, w2, and w3 are weighting coefficients, and w1 + w2 + w3 = 1. In this embodiment, w1 = w2 = w3 = 1 / 3, that is, the blood non-Newtonian property degree, the blood stasis degree, and the endothelial injury degree are fused with equal weights.

[0088] After obtaining the thrombosis risk index for each region of the left atrial appendage, the risk of each region can be further classified according to the thrombosis risk index. For example, a thrombosis risk index less than 0.33 is considered low risk, not less than 0.33 and not greater than 0.66 is considered medium risk, and greater than 0.66 is considered high risk. The distribution of the thrombosis risk index on the wall of the left atrial appendage can be output and displayed in the form of a cloud map, thereby giving the spatial distribution of risk in the left atrial appendage.

[0089] See Figure 2 Curve I represents the viscosity-shear rate curve corresponding to the general non-Newtonian constitutive equation with fixed parameters used in existing technologies. Curves II and III represent the curves corresponding to the constitutive relations determined in step S2 for two different individuals. The hollow dots in the figure represent the whole blood viscosity measurement points for these two individuals at low, medium, and high shear rates. Figure 2 It can be seen that the viscosity-shear rate curves of different individuals show significant differences in the low and medium shear rate ranges; and the viscosity given by the general non-Newtonian constitutive equation in the low shear rate range is significantly higher than the measured values ​​of the two individuals mentioned above, with its zero-shear viscosity being several times higher than that determined by the measured data. Analysis of clinical samples using the method described in this embodiment shows that, compared with using a non-Newtonian constitutive equation with fixed parameters, using the constitutive relation determined in step S2 results in significant differences in the periodic average viscosity distribution, periodic average velocity distribution, and relative residence time distribution within the left atrial appendage. This indicates that individual differences in the variation of blood viscosity with shear rate have a substantial impact on the flow field within the left atrial appendage; the endothelial cell activation potential in the left atrial appendage region is about two orders of magnitude higher than in other regions of the left atrium, with the high-value area concentrated at the tip of the left atrial appendage; individuals with higher whole blood viscosity have lower periodic average velocity, higher relative residence time, and higher endothelial cell activation potential within the left atrial appendage, resulting in a correspondingly higher thrombosis risk index.

[0090] Example 2

[0091] The difference between this embodiment and Embodiment 1 lies in the following aspects; the rest are the same as in Embodiment 1 and will not be repeated here.

[0092] The number and value of the at least three different shear rates can be varied, as long as there are three shear rates that are divided into the low shear rate segment, the shear-thinning transition segment, and the high shear rate segment according to the aforementioned multiple relationship, for example, 1s. -1 10s -1 50s -1 100s -1 and 200s -1A total of five shear rate measurement points were used, and the whole blood viscosity at each measurement point was included in the fitting. The more measurement points, the higher the accuracy of the constitutive relation within the corresponding shear rate range.

[0093] The general formula for the non-Newtonian constitutive equation can also take other forms that can simultaneously characterize the shear-thinning properties and the effect of hematocrit, such as an expression based on the Carlo-Jasoda form or power law form with an added correction factor using hematocrit as the independent variable, or an expression that includes a constant term characterizing infinite shear viscosity in addition to the shear-thinning term. This is as long as the general formula contains undetermined coefficients that can be determined by fitting the whole blood viscosity at the at least three different shear rates.

[0094] In addition to using the nonlinear least squares method, the fitting process can also employ maximum likelihood estimation, genetic algorithms, or particle swarm optimization to determine the undetermined coefficients in the general formula.

[0095] The hemodynamic parameters may include only the shear rate distribution and the residence time distribution of blood in the left atrial appendage. In this case, the thrombosis risk index is obtained by weighted fusion of the normalized degree of blood non-Newtonian properties and the degree of blood stasis, and the degree of endothelial damage and the time-averaged wall shear stress, oscillatory shear index, and endothelial cell activation potential on which it is based are omitted accordingly.

[0096] In addition to being characterized by relative residence time, the residence time distribution can also be characterized by the average residence time of tracer particles in the left atrial appendage, or the time required for the tracer concentration in the left atrial appendage to decay to half of its initial concentration.

[0097] In addition to being given in the manner described in Example 1, the pulsating boundary conditions can also be obtained by extracting the actual mitral valve orifice blood flow velocity profile from the individual under test using echocardiography or four-dimensional flow magnetic resonance imaging, and using this velocity profile as the outlet condition.

[0098] Example 3

[0099] See Figure 4 This embodiment provides an electronic device 100, including a memory 101 and a processor 102. The memory 101 stores a computer program, and the processor 102 executes the computer program to implement the atrial fibrillation thrombosis prediction method described in Embodiment 1 or Embodiment 2. The memory 101 and the processor 102 are communicatively connected via a communication bus 103.

[0100] The memory 101 can be a read-only memory, random access memory, magnetic disk, or optical disk, etc.; the processor 102 can be a central processing unit, graphics processing unit, digital signal processor, field-programmable gate array, or application-specific integrated circuit, etc. Since the numerical solution of the three-dimensional hydrodynamic model of the left atrium in step S3 involves a large amount of computation, the processor 102 can be a processing unit with parallel computing capabilities, or a computing cluster composed of multiple computing nodes. The electronic device 100 may also include a display unit for displaying the distribution of the thrombosis risk index on the wall of the left atrial appendage.

[0101] This embodiment also provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the atrial fibrillation thrombosis prediction method described in Embodiment 1 or Embodiment 2. The computer-readable storage medium includes various media capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory, random access memory, magnetic disk, or optical disk.

[0102] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0103] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for predicting atrial fibrillation thrombosis, characterized in that, Includes the following steps: A three-dimensional hydrodynamic model of the left atrium, including the left atrial appendage, and blood rheological characteristics were obtained based on the reconstruction of cardiac images of the individual under test. The blood rheological characteristics included hematocrit and whole blood viscosity at at least three different shear rates. Based on the acquired blood rheological characteristics, a constitutive relation characterizing the change of blood viscosity with shear rate in the tested individual was determined; The constitutive relation is applied to the three-dimensional hydrodynamic model of the left atrium and solved to obtain hemodynamic parameters in the left atrial appendage. The hemodynamic parameters include at least the shear rate distribution and the residence time distribution of blood in the left atrial appendage. Based on the hemodynamic parameters, the thrombosis risk index of each region of the left atrial appendage was determined to assess the risk of thrombosis in the left atrial appendage.

2. The atrial fibrillation thrombosis prediction method according to claim 1, characterized in that, The process of applying the constitutive relation to the three-dimensional hydrodynamic model of the left atrium and solving it to obtain hemodynamic parameters within the left atrial appendage includes: The constitutive relation is used as the viscosity constitutive relation of blood in the three-dimensional fluid dynamics model of the left atrium. The model is numerically solved in combination with preset boundary conditions to obtain the flow field in the left atrial appendage. The shear rate distribution is determined based on the flow field, and the residence time distribution is determined by tracking blood particles based on the flow field.

3. The atrial fibrillation thrombosis prediction method according to claim 1, characterized in that, The hemodynamic parameters also include the time-averaged wall shear stress and the oscillatory shear index of the left atrial appendage wall. The time-averaged wall shear stress represents the average level of wall shear stress within a cardiac cycle, and the oscillatory shear index represents the degree to which the wall shear stress deviates from the main blood flow direction within a cardiac cycle.

4. The atrial fibrillation thrombosis prediction method according to claim 3, characterized in that, The hemodynamic parameters also include endothelial cell activation potential, which is determined based on the oscillatory shear index and the time-averaged wall shear stress, and is used to characterize the tendency of thrombosis caused by endothelial damage to the left atrial appendage wall.

5. The atrial fibrillation thrombosis prediction method according to claim 4, characterized in that, The determination of thrombosis risk indices for different regions of the left atrial appendage includes: The blood flow velocity distribution within the left atrial appendage is obtained from the solution. The degree of non-Newtonian properties of blood is obtained based on the shear rate distribution and the blood viscosity distribution determined by the constitutive relation. The degree of blood stasis is obtained based on the blood flow velocity distribution and the residence time distribution; The degree of endothelial damage is determined based on the endothelial cell activation potential. The thrombosis risk index for each region of the left atrial appendage is determined based on the degree of non-Newtonian blood properties, the degree of blood stasis, and the degree of endothelial damage.

6. The atrial fibrillation thrombosis prediction method according to claim 5, characterized in that, The determination of thrombotic risk indices for different regions of the left atrial appendage based on the degree of non-Newtonian blood properties, the degree of blood stasis, and the degree of endothelial damage includes: The degree of non-Newtonian properties of the blood, the degree of blood stasis, and the degree of endothelial damage were normalized respectively. The normalized results were then weighted and fused to obtain the thrombosis risk index of each region of the left atrial appendage.

7. The atrial fibrillation thrombosis prediction method according to claim 1, characterized in that, The solution process includes: applying pulsatile boundary conditions characterizing atrial fibrillation blood flow to the three-dimensional hydrodynamic model of the left atrium, wherein the pulsatile boundary conditions include setting inlet conditions for the pulmonary vein opening and outlet conditions for the mitral valve opening, and solving the full flow field of the model based on the pulsatile boundary conditions.

8. The atrial fibrillation thrombosis prediction method according to claim 1, characterized in that, The at least three different shear rates include a first shear rate, a second shear rate, and a third shear rate that increase sequentially, wherein the first shear rate, the second shear rate, and the third shear rate constitute the low-shear rate, the medium-shear rate, and the high-shear rate in the whole blood viscosity test, respectively. Wherein, the ratio of the second shear rate to the first shear rate is not less than 5, and the ratio of the third shear rate to the second shear rate is not less than 3, so that the first shear rate, the second shear rate, and the third shear rate are respectively located in different segments of the blood viscosity versus shear rate curve.

9. The atrial fibrillation thrombosis prediction method according to claim 8, characterized in that, The first shear rate is 1s -1 up to 5s -1 The second shear rate is 30s. -1 up to 60s -1 The third shear rate is 180s. -1 up to 220s -1 .

10. The atrial fibrillation thrombosis prediction method according to claim 1, characterized in that, The determination of the constitutive relation characterizing the change of blood viscosity as a function of shear rate in the tested individual includes: A general formula for a non-Newtonian constitutive equation is constructed. The whole blood viscosity and hematocrit of the individual under test at at least three different shear rates are substituted into the general formula, and the constitutive relationship of the individual under test is obtained by fitting. The general formula is: ； Where η is the dynamic viscosity of blood, γ is the shear rate, used to describe the magnitude of the velocity gradient in the fluid, and ε RBC is hematocrit, which represents the percentage of red blood cells in the blood volume; coefficients a, λ, and b are undetermined coefficients characterizing the shear thinning properties of blood, and coefficients k and k1 are coefficients characterizing the effect of hematocrit on blood viscosity.

11. The atrial fibrillation thrombosis prediction method according to claim 10, characterized in that, The fitting process uses a nonlinear least squares method to determine the undetermined coefficients in the general formula.

12. The atrial fibrillation thrombosis prediction method according to claim 10, characterized in that, The blood rheological characteristics also include the erythrocyte aggregation index and the erythrocyte deformability index. The erythrocyte aggregation index characterizes the ability of erythrocytes to aggregate together to form a rouleaux structure under low shear conditions, and the erythrocyte deformability index characterizes the ability of erythrocytes to change their shape under shear. The coefficients k and k1 are determined according to a preset empirical relationship between the erythrocyte aggregation index and the erythrocyte deformability index, and the coefficients a, λ, and b are determined by fitting the whole blood viscosity at at least three different shear rates.

13. The atrial fibrillation thrombosis prediction method according to claim 12, characterized in that, The empirical relationship states that the coefficient k increases with the increase of the erythrocyte aggregation index, and the coefficient k1 decreases with the increase of the erythrocyte deformability index.

14. The atrial fibrillation thrombosis prediction method according to claim 12, characterized in that, The blood rheological characteristics also include at least one of plasma fibrinogen concentration and plasma viscosity, wherein the plasma fibrinogen concentration and plasma viscosity serve as alternatives to or supplements to the erythrocyte aggregation index and the erythrocyte deformability index, and are used to determine the coefficient k and the coefficient k1.

15. The atrial fibrillation thrombosis prediction method according to claim 1, characterized in that, The blood described in the solution is a single-phase homogeneous non-Newtonian blood model, and the viscosity of the blood is determined by the constitutive relation.

16. An electronic device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the atrial fibrillation thrombosis prediction method according to any one of claims 1 to 15.

17. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the atrial fibrillation thrombosis prediction method according to any one of claims 1 to 15.

Images