A method and system for evaluating abnormalities based on cerebral vascular images

Abstract

The application provides a kind of abnormality evaluation method and system based on cerebral vascular image, which comprises the following steps: obtaining the time dimension dynamic enhancement vascular imaging sequence of target cerebral vascular region, establishing three-dimensional time-varying flow field, to calculate the first characteristic parameter distribution map and the second characteristic parameter distribution map;Identify feature abnormal area, for each feature abnormal area, construct energy function, and drive the initial tube wall interface grid model to deform iteration in local search space, to obtain the optimized tube wall interface;With the optimized tube wall interface as the boundary, the blood flow dynamic stimulation cumulative amount of each sampling point on the tube wall interface is calculated. It can represent the space-time pattern and cumulative stimulation effect of complex abnormal blood flow from two dimensions of rotation intensity and direction disorder, and finally output a direct and quantitative "blood flow dynamic stimulation cumulative amount" distribution map, which improves the accuracy of risk assessment of cerebral vascular disease.

Description

Technical Field

[0001] This invention relates to the field of medical image data processing technology, and in particular to a method and system for abnormal assessment based on cerebrovascular images. Background Technology

[0002] Cerebrovascular diseases, such as intracranial aneurysms and atherosclerotic stenosis, are important causes of stroke. With the development of medical imaging technology, computed tomography angiography (CTA) and magnetic resonance angiography (MRA) have become the main means of assessing the morphology of cerebral blood vessels.

[0003] In existing technologies, computational fluid dynamics (CFD) should be based on individualized patient imaging data to simulate and quantify hemodynamic parameters, facilitating the understanding of disease occurrence and development mechanisms and the assessment of treatment strategies. Current technical processes typically involve "image segmentation - CFD calculation - parameter analysis," and their accuracy heavily relies on the accuracy of the initial vessel segmentation. In lesion areas with blurred vessel wall boundaries, such as the neck of an aneurysm or the distal end of a stenosis, automatic segmentation algorithms are prone to errors. CFD simulations based on geometric models of these errors amplify and propagate these errors, leading to distortion in all subsequent hemodynamic parameter calculations and resulting in low assessment accuracy. Secondly, existing methods mostly provide transient or periodically averaged mechanical parameters, lacking quantitative models of the long-term cumulative effects of abnormal hemodynamic stimuli. However, pathological remodeling of the vessel wall (such as aneurysm growth and rupture) is precisely a process of long-term cumulative mechanical stimulation leading to progressive damage to biological tissues. Therefore, ensuring the accuracy of the vascular geometric model upon which the analysis is based and developing a quantitative indicator system that can more comprehensively characterize the spatiotemporal patterns of abnormal blood flow and its cumulative harm are critical technical bottlenecks that urgently need to be overcome in this field. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide an abnormal assessment method based on cerebral vascular imaging, which aims to solve the technical problem of low assessment accuracy in the prior art.

[0005] To achieve the above objectives, in a first aspect, the present invention provides: a method for abnormal assessment based on cerebral vascular imaging, comprising the following steps: A time-dimensional dynamic enhanced vascular imaging sequence of the target cerebral vascular region is obtained, and a three-dimensional time-varying flow velocity field is reconstructed based on the time-dimensional dynamic enhanced vascular imaging sequence through computational fluid dynamics simulation. The time-dimensional dynamically enhanced vascular imaging sequence is initially segmented to obtain an initial vessel wall interface mesh model, and the computational fluid dynamics simulation is performed with the initial vessel wall interface mesh model as the boundary condition to obtain an initial three-dimensional time-varying velocity field. Based on the initial three-dimensional time-varying velocity field, a first characteristic parameter distribution map, which is associated with the near-wall fluid rotation intensity, is calculated and mapped onto the initial pipe wall interface mesh model; a second characteristic parameter distribution map, which characterizes the time-varying turbulence degree of the wall shear force direction, is also calculated. Based on the first feature parameter distribution map and the second feature parameter distribution map, feature anomaly regions are identified on the initial pipe wall interface mesh model; For each of the aforementioned abnormal feature regions, an energy function is constructed that integrates image gradient information with hemodynamic feature gradient information constructed based on the gradients of the first and second feature parameter distribution maps. This function then drives the initial vessel wall interface mesh model to undergo deformation iteration in the local search space to obtain an optimized vessel wall interface. Using the optimized tube wall interface as the boundary, the computational fluid dynamics simulation is re-executed to obtain the optimized velocity field, and the cumulative hemodynamic stimulation at each sampling point on the tube wall interface is calculated based on the optimized velocity field.

[0006] According to one aspect of the above technical solution, the steps for reconstructing the three-dimensional time-varying velocity field specifically include: The time-dimensional dynamically enhanced vascular imaging sequence is segmented into vascular cavities to obtain a three-dimensional vascular cavity mask at each time point. The three-dimensional vascular cavity mask sequences corresponding to each time point are spatiotemporally aligned and interpolated to construct a time-varying fluid domain computational grid; The blood flow velocity profile was set by using the timing of the contrast agent reaching the inlet of the target blood vessel as the inflow boundary condition; The transient Navier-Stokes equations are solved on the time-varying fluid domain grid to obtain the three-dimensional time-varying velocity field.

[0007] According to one aspect of the above technical solution, the step of calculating the first feature parameter specifically includes: Obtain the unit normal vector of the sampling point on the pipe wall interface, determine the nearest point where the normal vector points into the inside of the pipe cavity, and construct a vector; Calculate the relative velocity circulation on a tiny circular loop with a vector as the axis and a preset radius at the target time, and then calculate the relative velocity circulation density, using the relative velocity circulation density as the first feature parameter.

[0008] According to one aspect of the above technical solution, the calculation expression for the first feature parameter is: ; In the formula, The first characteristic parameter at sampling point s and time t represents the local rotation intensity of the fluid particle near the wall. This represents the velocity vector at point q on the integral loop C. This represents the velocity vector at sampling point s. Let be the radius of the tiny circular loop. Let be the infinitesimal vector of the line integral along loop C.

[0009] According to one aspect of the above technical solution, the step of calculating the second feature parameter specifically includes: Calculate the wall shear stress vector at the sampling point on the pipe wall interface at the target time; The wall shear stress vector is projected onto the tangential plane of the pipe wall interface at the sampling point to obtain the tangential vector, and its azimuth angle is calculated. Phase expansion is performed on the direction angle sequence sampled in continuous time to obtain a monotonically continuous expanded phase angle sequence, thereby eliminating the jump between 0 and 2π; Within a sliding time window of a preset length, the incremental standard deviation of the unfolded phase angle sequence is calculated; The second characteristic parameter is obtained by averaging the incremental standard deviation over multiple consecutive time windows.

[0010] According to one aspect of the above technical solution, the calculation expression for the second feature parameter is as follows: ; In the formula, Let be the second characteristic parameter at sampling point s and time t, representing the temporal disorder of the wall shear force direction, M be the total number of sliding time windows, and N be the number of sampling points within each time window. For the increment of the i-th unfolded phase angle within the j-th time window, This represents the average value of all increments within the j-th time window.

[0011] According to one aspect of the above technical solution, after obtaining the cumulative hemodynamic stimulation at each sampling point, the method further includes: The cumulative amount of hemodynamic stimulation is mapped onto the optimized vessel wall interface to generate a cloud map, and high-risk areas are marked based on a preset threshold to generate a structured assessment report containing spatial location, parameter values, and trends of change.

[0012] According to one aspect of the above technical solution, the steps of driving the initial pipe wall interface mesh model to undergo deformation iteration in the local search space specifically include: Mark the grid vertices within the aforementioned abnormal feature region as active vertices and fix the inactive vertices; Construct a total energy function for the position of the active vertex, and minimize the total energy function to update the position of the active vertex; Based on the updated pipe wall interface, the characteristic parameters are recalculated, and it is determined whether the average displacement or the number of iterations meets the convergence condition, so as to output the optimized pipe wall interface.

[0013] According to one aspect of the above technical solution, the expression for calculating the total energy function is as follows: ; ; ; ; ; ; In the formula, Represents the total energy function. This represents an adaptive weighting function that varies with the spatial sampling point s. This represents the image energy term based on the image grayscale gradient. This represents the hemodynamic feature energy term constructed based on the gradients of the first and second feature parameter distribution maps. and These are the normalized first and second characteristic parameters. Based on image weights, This represents the weight decay rate. This refers to the characteristic anomaly region. and Let be the spatial gradients of the first and second feature parameters at sampling point s, respectively. and Let x1 be the potential function. , The characteristic gradient constraint strength coefficient, The target gradient threshold, The feature gradient sensitivity coefficient, and To dynamically enhance the image grayscale value and its spatial gradient at sampling point s in the time dimension of the vascular imaging sequence, Let be the potential function for the image gradient. = , The intensity coefficient of the image gradient constraint. The desired ideal edge gradient magnitude threshold, This is the tolerance coefficient for the gradient magnitude.

[0014] According to one aspect of the above technical solution, the calculation expression for the cumulative amount of hemodynamic stimulation is as follows: ; ; In the formula, This represents the cumulative amount of hemodynamic stimulation at sampling point s. Here, T is the normalization coefficient, and T is the duration of the cardiac cycle. This represents the instantaneous intensity of the composite stimulus at sampling point s and time t. The attenuation coefficient is... As a reference time constant, and These are the first characteristic parameter and the second characteristic parameter, respectively. For adjustment coefficients, This is the preset threshold for the second feature parameter.

[0015] On the other hand, this solution also provides an abnormality assessment system based on cerebrovascular imaging, including: The acquisition module is used to acquire a time-dimensional dynamic enhanced vascular imaging sequence of the target cerebral vascular region; to perform initial image segmentation on the time-dimensional dynamic enhanced vascular imaging sequence to acquire an initial vessel wall interface mesh model; and to perform computational fluid dynamics simulation with the initial vessel wall interface mesh model as the boundary condition to obtain an initial three-dimensional time-varying velocity field. The parameter distribution module is used to calculate, based on the initial three-dimensional time-varying velocity field, a first characteristic parameter distribution map mapped onto the initial pipe wall interface mesh model and associated with the near-wall fluid rotation intensity, and a second characteristic parameter distribution map characterizing the time-varying turbulence degree of the wall shear force direction; The identification module is used to identify characteristic abnormal regions on the initial pipe wall interface mesh model based on the first feature parameter distribution map and the second feature parameter distribution map. The iterative optimization module is used to construct an energy function for each of the aforementioned abnormal feature regions, which integrates image gradient information with hemodynamic feature gradient information constructed based on the gradients of the first feature parameter distribution map and the second feature parameter distribution map, and drives the initial vessel wall interface mesh model to undergo deformation iteration in the local search space to obtain the optimized vessel wall interface. The calculation module is used to re-execute the computational fluid dynamics simulation with the optimized tube wall interface as the boundary, obtain the optimized flow velocity field, and calculate the cumulative amount of hemodynamic stimulation at each sampling point on the tube wall interface based on the optimized flow velocity field.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: It utilizes the preliminarily calculated hemodynamic characteristics as an internal feedback signal to drive the initial vascular wall interface to undergo self-correcting deformation in the abnormal region, fundamentally solving the problem of subsequent calculation distortion caused by inaccurate initial segmentation in traditional methods, and providing a high-fidelity geometric basis for analysis. Simultaneously, through the defined first and second characteristic parameters and their nonlinear weighted cumulative model, it can more comprehensively and physiologically characterize the spatiotemporal patterns and cumulative stimulation effects of complex abnormal blood flow from two dimensions: rotational intensity and directional disorder. Ultimately, it outputs an intuitive and quantitative distribution map of "cumulative hemodynamic stimulation," improving the accuracy of risk assessment for cerebrovascular diseases. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the abnormal assessment method based on cerebral vascular imaging in the first embodiment of the present invention. Figure 2 This is a structural block diagram of the abnormal assessment system based on cerebral vascular imaging in the second embodiment of the present invention; The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0018] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0019] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] Example 1 Please see Figure 1The figure shows a flowchart of the abnormal assessment method based on cerebral vascular imaging in the first embodiment of the present invention. As shown in the figure, the method includes the following steps: Step S100: Obtain the time-dimensional dynamic enhanced vascular imaging sequence of the target cerebral vascular region; perform initial image segmentation on the time-dimensional dynamic enhanced vascular imaging sequence to obtain an initial vessel wall interface mesh model, and perform computational fluid dynamics simulation with the initial vessel wall interface mesh model as the boundary condition to obtain an initial three-dimensional time-varying velocity field.

[0022] Specifically, in some application scenarios of this embodiment, a time-resolved dynamic enhancement scan of the target cerebral vascular region (such as intracranial arteries) of a patient is performed using a magnetic resonance angiography (MRI) device to obtain a three-dimensional image sequence containing multiple time points (e.g., 20-30 phases) within a complete cardiac cycle. The sequence undergoes necessary preprocessing such as noise reduction and motion artifact correction to improve data quality. For the three-dimensional image at each time point, threshold segmentation, region growing, or machine learning algorithms are used to distinguish the vascular lumen from the surrounding tissue, generating a three-dimensional vascular lumen mask corresponding to each time point. This mask is essentially binary three-dimensional volumetric data that identifies the internal space of the blood vessel.

[0023] The aforementioned series of 3D vascular lumen masks arranged chronologically are spatiotemporally aligned to ensure that the same spatial location corresponds at different time points. Due to the limited number of acquisition time points, spatiotemporal interpolation is required between masks at adjacent time points to generate vascular lumen shapes under a denser time series. Based on this series of shapes, a time-varying fluid domain computational mesh that can deform with time is constructed using mesh generation techniques (such as tetrahedral or finite volume mesh). Then, blood flow simulation boundary conditions are set. Based on the concentration-time curve of contrast agent reaching the target blood vessel inlet in the dynamic enhanced image, the blood flow velocity waveform (velocity profile) at that location is calculated and set as the inflow boundary condition; the outlet boundary adopts pressure outlet or flow resistance boundary conditions, set based on the impedance of the downstream blood vessel, and the blood vessel wall interface is regarded as a no-slip boundary. On the constructed time-varying fluid domain grid, blood is defined as an incompressible Newtonian fluid, and its density and viscosity parameters are assigned accordingly. Numerical simulation is performed by solving the fundamental physical equations describing fluid motion—the transient Navier-Stokes equations. This calculation is performed iteratively at multiple time steps, ultimately outputting the velocity vector of each computational grid node at each time point, i.e., the complete three-dimensional time-varying velocity field.

[0024] Step S200: Based on the initial three-dimensional time-varying velocity field, calculate the distribution map of the first characteristic parameter, which is associated with the near-wall fluid rotation intensity, mapped onto the initial pipe wall interface mesh model, and the distribution map of the second characteristic parameter, which characterizes the degree of turbulence in the time-varying direction of the wall shear force. Specifically, For each triangular facet (or vertex) on the initial pipe wall interface mesh model, perform the following operations: Determine the spatial location of the center point (or vertex) of the patch and its unit normal vector; In the opposite direction of the normal vector (i.e., pointing towards the inside of the lumen), at a distance δ from the vessel wall (e.g., δ = 0.1 mm, approximately 1%-5% of the vessel radius), define a near-wall sampling point p. The selection of the distance δ must ensure that the sampling point is located within the blood flow boundary layer; From the initial three-dimensional time-varying velocity field v(x,t), the velocity vector v(p,t) of the sampling point p at each time point in the entire cardiac cycle, as well as the velocity of the tube wall point s, are obtained by spatial interpolation. The steps for calculating the first feature parameter specifically include: Obtain the unit normal vector of the sampling point on the pipe wall interface, determine the nearest point where the normal vector points into the inside of the pipe cavity, and construct a vector; Calculate the relative velocity circulation on a tiny circular loop with a vector as the axis and a predetermined radius at the target time, and then calculate the relative velocity circulation density. Use the relative velocity circulation density as the first feature parameter, and iterate through all pipe wall points to calculate their respective... And assign a grid model to generate the first feature parameter distribution map.

[0025] The expression for calculating the first feature parameter is: ; In the formula, The first characteristic parameter at sampling point s and time t represents the local rotation intensity of the fluid particle near the wall. This represents the velocity vector at point q on the integral loop C. This represents the flow velocity vector at sampling point s. Let be the radius of the tiny circular loop. Let be the infinitesimal vector of the line integral along loop C.

[0026] The specific steps for calculating the second characteristic parameter include: Calculate the wall shear stress vector at the sampling point on the pipe wall interface at the target time; The wall shear stress vector is projected onto the tangential plane of the pipe wall interface at the sampling point to obtain the tangential vector, and its azimuth angle is calculated. Phase expansion is performed on the direction angle sequence sampled in continuous time to obtain a monotonically continuous expanded phase angle sequence, thereby eliminating the jump between 0 and 2π; Within a sliding time window of a preset length, the incremental standard deviation of the unfolded phase angle sequence is calculated; The incremental standard deviation is averaged over multiple consecutive time windows to obtain the second characteristic parameter. This process is then repeated for all pipe wall points to calculate their respective... And assign a grid model to generate the first feature parameter distribution map.

[0027] The expression for calculating the second characteristic parameter is: ; In the formula, Let be the second characteristic parameter at sampling point s and time t, representing the temporal disorder of the wall shear force direction, M be the total number of sliding time windows, and N be the number of sampling points within each time window. For the increment of the i-th unfolded phase angle within the j-th time window, This represents the average value of all increments within the j-th time window.

[0028] Step S300: Based on the first and second feature parameter distribution maps, identify characteristic anomaly regions on the initial pipe wall interface mesh model. Specifically, the first and second feature parameter maps first need to be normalized. ; ; min( ) and max( These refer to the minimum and maximum values ​​calculated over the entire target vessel wall in the current case, respectively. After normalization, and The range of values ​​is within the interval [0, 1].

[0029] Then, a composite characteristic index that can comprehensively reflect the rotation intensity and directional disorder is defined. This indicator employs a non-linear fusion approach to highlight regions where both are abnormal: ; In the formula, γ is the adjustment coefficient (preferably 5). The preferred reference threshold for orientation disorder is 0.5. Its physical meaning is that when the orientation disorder... At lower levels, the composite index depends primarily on the rotational intensity. ;when Exceeding the threshold At that time, this term will be amplified non-linearly. The contribution indicates that "disorder" exacerbates the potential adverse effects of "rotation".

[0030] Calculate the composite characteristic index of all vertices on the entire pipe wall interface. mean and standard deviation Define the anomaly detection threshold: ; In the formula, k is a constant factor, taken as 1.5. 3.0, preferably k=2.0. This method is based on statistical principles, classifying regions with eigenvalues ​​significantly higher than the population average as abnormal. The threshold k can be determined using historical healthy case data or expert experience.

[0031] Traverse each vertex of the initial pipe wall interface mesh model; if its composite characteristic index satisfies... ≥ If so, then mark the vertex as a "candidate anomalous vertex"; Discrete vertices that are only labeled may correspond to noise or sporadic anomalies, and need to be aggregated to form meaningful continuous regions. In this embodiment, each "candidate anomaly vertex" is used as a seed point, and the same conditions are found in its first-order neighborhood (adjacent vertices sharing the same edge). ≥ The vertex is selected and included in the same region. This process is performed recursively until no new vertices are added. This step aggregates all spatially connected anomalous vertices into several independent candidate anomalous regions {R1, R2, ..., R...}. m}; Set a minimum area threshold A min (For example, 1% of the total area of ​​the target blood vessel wall). For each candidate abnormal region R i Calculate the total area of ​​the grid cells it contains. If its area is Area(R) i ) min If a region is found to be noise or a computational artifact, it is considered to be removed. The remaining regions are then defined as feature anomaly regions.

[0032] Step S400: For each of the aforementioned abnormal feature regions, an energy function is constructed that fuses image gradient information with hemodynamic feature gradient information constructed based on the gradients of the first and second feature parameter distribution maps. This function then drives the initial vessel wall interface mesh model to undergo deformation iteration in the local search space, resulting in an optimized vessel wall interface. Specifically, the step of driving the initial vessel wall interface mesh model to undergo deformation iteration in the local search space includes: Mark the grid vertices within the aforementioned abnormal feature region as active vertices and fix the inactive vertices; Construct a total energy function for the position of the active vertex, and minimize the total energy function to update the position of the active vertex; Based on the updated pipe wall interface, the characteristic parameters are recalculated, and it is determined whether the average displacement or the number of iterations meets the convergence condition, so as to output the optimized pipe wall interface.​

[0033] For each feature anomaly region, a composite energy function is constructed to drive the deformation at its active vertices: ; ; ; ; ; ; In the formula, Represents the total energy function. This represents an adaptive weighting function that varies with the spatial sampling point s. This represents the image energy term based on the image grayscale gradient. This represents the hemodynamic feature energy term constructed based on the gradients of the first and second feature parameter distribution maps. and These are the normalized first and second characteristic parameters. Based on image weights, This represents the weight decay rate. This refers to the characteristic abnormal region. and Let be the spatial gradients of the first and second feature parameters at sampling point s, respectively. and Let x1 be the potential function. , The characteristic gradient constraint strength coefficient, The target gradient threshold, The feature gradient sensitivity coefficient, and To dynamically enhance the image grayscale value and its spatial gradient at sampling point s in the time dimension of the vascular imaging sequence, Let be the potential function with respect to the image gradient. = , The intensity coefficient of the image gradient constraint. The desired ideal edge gradient magnitude threshold, This is the tolerance coefficient for the gradient magnitude.

[0034] Preferably, the convergence conditions include: the average displacement of the active vertex in two consecutive iterations is less than a preset threshold (preferably 0.01 mm); the relative change in total energy is less than a threshold (preferably 10). 4 ); reach the maximum number of iterations (preferably 100 times).

[0035] Optimization is performed sequentially on each feature anomaly region. Once a region is optimized, its vertex position is fixed and used as input for the optimization of the next adjacent region. After all regions are optimized, the globally optimized pipe wall interface is obtained.

[0036] Step S500: Using the optimized pipe wall interface as the boundary, re-execute the computational fluid dynamics simulation to obtain the optimized velocity field. Based on this optimized velocity field, calculate the cumulative hemodynamic stimulation at each sampling point on the pipe wall interface. Specifically, use the optimized pipe wall cross-section as the final simulated fluid domain, and set the same boundaries and physical adjustments as described above. Similarly, perform simulations for one to several cardiac cycles by solving the transient Navier-Stokes equations until the flow field reaches periodic stability. Record the optimized three-dimensional time-varying velocity field for all time steps within the last complete cardiac cycle, and simultaneously calculate the corresponding first and second characteristic parameters. Calculate the cumulative hemodynamic stimulation according to the following formula: ; ; In the formula, This represents the cumulative amount of hemodynamic stimulation at sampling point s. Here, T is the normalization coefficient, and T is the duration of the cardiac cycle. This represents the instantaneous intensity of the composite stimulus at sampling point s and time t. The attenuation coefficient is... As a reference time constant, and These are the first characteristic parameter and the second characteristic parameter, respectively. For adjustment coefficients, This is the preset threshold for the second feature parameter.

[0037] This is the activation threshold for orientation disorder. When the second feature parameter is less than the activation threshold, the orientation is relatively ordered, the Sigmoid function outputs -1, and S ≈ The stimulus primarily originates from rotation; when the second feature parameter is greater than the activation threshold, the orientation becomes highly disordered, the Sigmoid function output is >1, and at this point, S > η indicates that directional disorder amplifies the effect of rotational stimulation. η controls the steepness of the transition of the amplification effect.

[0038] Preferably, after obtaining the cumulative hemodynamic stimulation at each sampling point, the method further includes: The cumulative amount of hemodynamic stimulation is mapped onto the optimized vessel wall interface to generate a cloud map, and high-risk areas are marked based on a preset threshold to generate a structured assessment report containing spatial location, parameter values, and trends of change.

[0039] In summary, the abnormal assessment method based on cerebrovascular imaging in the above embodiments of the present invention utilizes the preliminarily calculated hemodynamic characteristics as an internal feedback signal to drive the initial vascular wall interface to undergo self-correcting deformation in the characteristic abnormal region. This fundamentally solves the problem of subsequent calculation distortion caused by inaccurate initial segmentation in traditional methods, providing a high-fidelity geometric basis for analysis. At the same time, through the defined first characteristic parameter, second characteristic parameter, and their nonlinear weighted cumulative model, the spatiotemporal patterns and cumulative stimulation effects of complex abnormal blood flow can be characterized more comprehensively and physiologically from the two dimensions of rotational intensity and directional disorder. Finally, an intuitive and quantitative distribution map of "hemodynamic stimulation accumulation" is output, improving the accuracy of risk assessment for cerebrovascular diseases.

[0040] Example 2 like Figure 2 As shown, the second embodiment of the present invention provides an abnormality assessment system based on cerebral vascular imaging, comprising: The acquisition module 100 is used to acquire a time-dimensional dynamic enhanced vascular imaging sequence of the target cerebral vascular region; perform initial image segmentation on the time-dimensional dynamic enhanced vascular imaging sequence to obtain an initial vessel wall interface mesh model; and perform computational fluid dynamics simulation with the initial vessel wall interface mesh model as the boundary condition to obtain an initial three-dimensional time-varying velocity field. The parameter distribution module 200 is used to calculate, based on the initial three-dimensional time-varying velocity field, a first characteristic parameter distribution map mapped onto the initial pipe wall interface mesh model and associated with the near-wall fluid rotation intensity, and a second characteristic parameter distribution map characterizing the time-varying turbulence degree of the wall shear force direction; The identification module 300 is used to identify feature anomaly regions on the initial pipe wall interface mesh model based on the first feature parameter distribution map and the second feature parameter distribution map. The iterative optimization module 400 is used to construct an energy function for each of the aforementioned abnormal feature regions, which integrates image gradient information and hemodynamic feature gradient information constructed based on the gradients of the first feature parameter distribution map and the second feature parameter distribution map, and drives the initial vessel wall interface mesh model to undergo deformation iteration in the local search space to obtain the optimized vessel wall interface. The calculation module 500 is used to re-execute the computational fluid dynamics simulation with the optimized tube wall interface as the boundary, obtain the optimized flow velocity field, and calculate the cumulative amount of hemodynamic stimulation at each sampling point on the tube wall interface based on the optimized flow velocity field.

[0041] The embodiments described above merely illustrate specific implementation methods of this application, and while the descriptions are detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for abnormal assessment based on cerebral vascular imaging, characterized in that, Includes the following steps: A time-dimensional dynamic enhanced vascular imaging sequence of the target cerebral vascular region is acquired; the time-dimensional dynamic enhanced vascular imaging sequence is initially segmented to obtain an initial vessel wall interface mesh model, and computational fluid dynamics simulation is performed with the initial vessel wall interface mesh model as boundary conditions to obtain an initial three-dimensional time-varying velocity field. Based on the initial three-dimensional time-varying velocity field, a first characteristic parameter distribution map, which is associated with the near-wall fluid rotation intensity, is calculated and mapped onto the initial pipe wall interface mesh model; a second characteristic parameter distribution map, which characterizes the time-varying turbulence degree of the wall shear force direction, is also calculated. Based on the first feature parameter distribution map and the second feature parameter distribution map, feature anomaly regions are identified on the initial pipe wall interface mesh model; For each of the aforementioned abnormal feature regions, an energy function is constructed that integrates image gradient information with hemodynamic feature gradient information constructed based on the gradients of the first and second feature parameter distribution maps. This function then drives the initial vessel wall interface mesh model to undergo deformation iteration in the local search space to obtain an optimized vessel wall interface. Using the optimized tube wall interface as the boundary, the computational fluid dynamics simulation is re-executed to obtain the optimized velocity field, and the cumulative hemodynamic stimulation at each sampling point on the tube wall interface is calculated based on the optimized velocity field.

2. The abnormal assessment method based on cerebral vascular imaging according to claim 1, characterized in that, The specific steps for reconstructing a three-dimensional time-varying velocity field include: The time-dimensional dynamically enhanced vascular imaging sequence is segmented into vascular cavities to obtain a three-dimensional vascular cavity mask at each time point. The three-dimensional vascular cavity mask sequences corresponding to each time point are spatiotemporally aligned and interpolated to construct a time-varying fluid domain computational grid; The blood flow velocity profile was set by using the timing of the contrast agent reaching the inlet of the target blood vessel as the inflow boundary condition; The transient Navier-Stokes equations are solved on the time-varying fluid domain grid to obtain the three-dimensional time-varying velocity field.

3. The abnormal assessment method based on cerebral vascular imaging according to claim 1, characterized in that, The steps for calculating the first feature parameter specifically include: Obtain the unit normal vector of the sampling point on the pipe wall interface, determine the nearest point where the normal vector points into the inside of the pipe cavity, and construct a vector; Calculate the relative velocity circulation on a tiny circular loop with a vector as the axis and a preset radius at the target time, and then calculate the relative velocity circulation density, using the relative velocity circulation density as the first feature parameter.

4. The abnormal assessment method based on cerebral vascular imaging according to claim 3, characterized in that, The expression for calculating the first feature parameter is: ； In the formula, The first characteristic parameter at sampling point s and time t represents the local rotation intensity of the fluid particle near the wall. This represents the velocity vector at point q on the integral loop C. This represents the flow velocity vector at sampling point s. Let be the radius of the tiny circular loop. Let be the infinitesimal vector of the line integral along loop C.

5. The method for abnormal assessment based on cerebral vascular imaging according to claim 1, characterized in that, The specific steps for calculating the second characteristic parameter include: Calculate the wall shear stress vector at the sampling point on the pipe wall interface at the target time; The wall shear stress vector is projected onto the tangential plane of the pipe wall interface at the sampling point to obtain the tangential vector, and its azimuth angle is calculated. Phase expansion is performed on the direction angle sequence sampled in continuous time to obtain a monotonically continuous expanded phase angle sequence, thereby eliminating the jump between 0 and 2π; Within a sliding time window of a preset length, the incremental standard deviation of the unfolded phase angle sequence is calculated; The second characteristic parameter is obtained by averaging the incremental standard deviation over multiple consecutive time windows.

6. The abnormal assessment method based on cerebral vascular imaging according to claim 1, characterized in that, The expression for calculating the second characteristic parameter is: ； In the formula, Let be the second characteristic parameter at sampling point s and time t, representing the temporal disorder of the wall shear force direction, M be the total number of sliding time windows, and N be the number of sampling points within each time window. For the increment of the i-th unfolded phase angle within the j-th time window, This represents the average value of all increments within the j-th time window.

7. The abnormal assessment method based on cerebral vascular imaging according to claim 1, characterized in that, After obtaining the cumulative hemodynamic stimulation at each sampling point, the method further includes: The cumulative amount of hemodynamic stimulation is mapped onto the optimized vessel wall interface to generate a cloud map, and high-risk areas are marked based on a preset threshold to generate a structured assessment report containing spatial location, parameter values, and trends of change.

8. The abnormal assessment method based on cerebral vascular imaging according to claim 1, characterized in that, The specific steps for driving the initial pipe wall interface mesh model to undergo deformation iteration in the local search space include: Mark the grid vertices within the aforementioned abnormal feature region as active vertices and fix the inactive vertices; Construct a total energy function for the position of the active vertex, and minimize the total energy function to update the position of the active vertex; Based on the updated pipe wall interface, the characteristic parameters are recalculated, and it is determined whether the average displacement or the number of iterations meets the convergence condition, so as to output the optimized pipe wall interface.

9. The abnormal assessment method based on cerebral vascular imaging according to claim 1, characterized in that, The expression for calculating the total energy function is as follows: ； ； ； ； ； ； In the formula, Represents the total energy function. This represents an adaptive weighting function that varies with the spatial sampling point s. This represents the image energy term based on the image grayscale gradient. This represents the hemodynamic feature energy term constructed based on the gradients of the first and second feature parameter distribution maps. and These are the normalized first and second characteristic parameters. Based on image weights, This represents the weight decay rate. This refers to the characteristic anomaly region. and Let be the spatial gradients of the first and second feature parameters at sampling point s, respectively. and Let x1 be the potential function. , The characteristic gradient constraint strength coefficient, The target gradient threshold, The feature gradient sensitivity coefficient, and To dynamically enhance the image grayscale value and its spatial gradient at sampling point s in the time dimension of the vascular imaging sequence, Let be the potential function for the image gradient. = , The intensity coefficient of the image gradient constraint. The desired ideal edge gradient magnitude threshold, This is the tolerance coefficient for the gradient magnitude.

10. A system for assessing abnormalities based on cerebral vascular imaging, for implementing the method for assessing abnormalities based on cerebral vascular imaging as described in any one of claims 1-9, characterized in that, include: The acquisition module is used to acquire a time-dimensional dynamic enhanced vascular imaging sequence of the target cerebral vascular region; to perform initial image segmentation on the time-dimensional dynamic enhanced vascular imaging sequence to acquire an initial vessel wall interface mesh model; and to perform computational fluid dynamics simulation with the initial vessel wall interface mesh model as the boundary condition to obtain an initial three-dimensional time-varying velocity field. The parameter distribution module is used to calculate, based on the initial three-dimensional time-varying velocity field, a first characteristic parameter distribution map mapped onto the initial pipe wall interface mesh model and associated with the near-wall fluid rotation intensity, and a second characteristic parameter distribution map characterizing the time-varying turbulence degree of the wall shear force direction; The identification module is used to identify characteristic abnormal regions on the initial pipe wall interface mesh model based on the first feature parameter distribution map and the second feature parameter distribution map. The iterative optimization module is used to construct an energy function for each of the aforementioned abnormal feature regions, which integrates image gradient information with hemodynamic feature gradient information constructed based on the gradients of the first feature parameter distribution map and the second feature parameter distribution map, and drives the initial vessel wall interface mesh model to undergo deformation iteration in the local search space to obtain the optimized vessel wall interface. The calculation module is used to re-execute the computational fluid dynamics simulation with the optimized tube wall interface as the boundary, obtain the optimized flow velocity field, and calculate the cumulative amount of hemodynamic stimulation at each sampling point on the tube wall interface based on the optimized flow velocity field.

Images