A method and system for evaluating the status of an arteriovenous fistula

Abstract

The application relates to the technical field of vascular fistula state evaluation, and discloses an arteriovenous fistula state evaluation method and system, which comprises the following steps: acquiring multiple frames of ultrasonic images and a blood flow matrix, stripping a global offset vector through cross-frame matching and mean value calculation to obtain a local deformation displacement; combining a discrete tangent vector cross product module length, aggregating a spatial average geometric twist factor, and generating a dynamic strain energy index according to a physical strain ratio; taking strain energy time integration as a driving force, constructing a reverse time mapping inverse function, resampling and projecting a flow velocity, and constructing a velocity gradient tensor; extracting a deformation rate tensor to calculate a viscous dissipation density, multiplying the viscous dissipation density by the aforementioned twist factor to generate a weighted dissipation density, generating a fluid-structure coupling dissipation index under double definite integration, and outputting an evaluation index through theoretical threshold mapping; the technical scheme avoids time dislocation interference, and objectively quantifies a fluid-structure dissipation state.

Description

Technical Field

[0001] This invention relates to the field of arteriovenous fistula status assessment technology, specifically to a method and system for assessing the status of arteriovenous fistulas. Background Technology

[0002] Arteriovenous fistulas (AVFs) are the fundamental vascular access for maintaining hemodialysis. Conventional ultrasound detection methods for AVFs often rely on extracting peak blood flow velocity or measuring cross-sectional width. This assessment method, based on static anatomical structure or apparent flow velocity, is difficult to objectively depict the complex hemodynamic evolution process inside the blood vessel. In scenarios where AVFs are subjected to high-pressure blood flow impact for a long time, the vessel wall is prone to local morphological distortion, which in turn leads to shear friction and energy dissipation of internal blood microclusters. Existing ultrasound post-processing solutions lack the means to physically and topologically correlate vessel wall geometric distortion with fluid internal friction mechanisms, and cannot output unified quantitative parameters reflecting the energy consumption state inside the system.

[0003] When extracting the pulsating displacement features of the vessel wall, the ultrasound probe is affected by the fluctuations of human breathing and muscle tremors during the detection process. Existing technologies usually cannot separate this low-frequency global offset interference from the real elastic deformation of the vessel wall, resulting in motion artifacts mixed in the extracted displacement sequence. At the same time, for the irregular geometric shape at the anastomosis site of the fistula, the conventional second derivative curvature calculation model is prone to numerical divergence at the bifurcation singularity, resulting in the loss of numerical stability in the quantitative assessment of the local morphological orientation.

[0004] In addition, the impact of pulsatile blood flow has unsteady and asynchronous expansion and contraction characteristics. Existing ultrasound blood flow analysis mostly uses an absolute physical time axis to mechanically align multi-frame cross-sectional data. This fixed time mapping benchmark is prone to causing cross-frame spatial misalignment and phase disorder of flow field data when facing complex cardiac cycles. This makes it difficult for the extracted velocity gradient tensor to accurately correspond to the real fluid stretching and shear deformation laws.

[0005] Therefore, there is an urgent need for a method and system for assessing the status of arteriovenous fistulas to solve the problems of difficulty in decoupling global motion interference, being limited by physical time misalignment, lack of fluid-structure dissipation coupling quantification parameters, and limited assessment methods in the existing technologies. Summary of the Invention

[0006] This invention provides a method and system for assessing the status of arteriovenous fistulas, which helps to solve the problems mentioned in the background art.

[0007] This invention provides the following technical solution: a method for assessing the status of arteriovenous fistulas, comprising:

[0008] Acquire multi-frame grayscale images and blood flow matrix of the arteriovenous fistula, locate boundary markers based on gradient extrema, and construct a discrete vascular boundary point coordinate sequence;

[0009] By combining the arc length topological difference and Euclidean distance, cross-frame boundary matching is performed, and the global offset vector is stripped by mean calculation to obtain the local deformation displacement.

[0010] Introducing a small constant, combining the local discrete tangent vectors with adjacent points, performing dimensionality-upgrading and zero-padding on the tangent vectors to calculate their cross product magnitude, and then performing aggregation processing on the cross product magnitude to extract the geometric torsion factor;

[0011] The physical strain ratio is calculated based on the displacement difference and initial length, and a quadratic polymerization is performed in combination with the elastic modulus. The strain energy index is generated by superimposing the physiological basis constant.

[0012] Driven by the time integral of strain energy, discrete sample points are extracted, and the inverse time mapping function is constructed by cubic spline interpolation with phase as the independent variable.

[0013] The velocity matrix and direction parameters are called, the velocity is resampled using the inverse function, and the projection components are extracted and the velocity gradient tensor is constructed after symbolic truncation angle compensation.

[0014] The symmetric component of the velocity gradient is extracted, the deformation rate tensor is constructed, the dissipation density is calculated by combining the viscosity coefficient, and the weighted dissipation density is generated by multiplying it by the aforementioned torsional factor.

[0015] The weighted dissipation density is subjected to a double definite integral over the reference area and the full phase domain to generate a fluid-structure interaction dissipation index. After mapping with the theoretical threshold index, the evaluation index is output.

[0016] Optionally, the step of acquiring multi-frame grayscale images and blood flow matrices of the arteriovenous fistula, locating boundary markers based on gradient extrema, and constructing a discrete vascular boundary point coordinate sequence specifically includes:

[0017] An ultrasound probe array is attached to the target area to simultaneously acquire multiple frames of grayscale image matrix and blood flow velocity matrix.

[0018] Initialize the data acquisition timestamps and establish the physical calculation origin;

[0019] The region of interest is delineated using the grayscale image at the initial time, and the reference cross-sectional area at the initial time is extracted;

[0020] Calculate the partial derivatives of the grayscale image matrix in the two-dimensional spatial coordinate system to obtain the rate of change of local pixel grayscale values;

[0021] Within a preset local search neighborhood, the extreme points of the rate of change are selected as geometric markers at the tissue interface, and the initial boundary point set coordinate sequence of the vascular intima is extracted.

[0022] Optionally, the step of combining arc length topological differences and Euclidean distance to perform cross-frame boundary matching, and then calculating the mean to strip the global offset vector and obtain local deformation displacement specifically includes:

[0023] The difference in arc length topology between adjacent boundary points is obtained as a penalty weight term and added to the Euclidean shortest distance calculation process to perform cross-frame nearest neighbor matching.

[0024] When a local match is lost, the coordinates of the effective boundary points of the adjacent successfully matched points are extracted as interpolation nodes. Under the smooth constraint of continuous local second derivative, spatial cubic spline interpolation compensation is performed.

[0025] For the registered coordinate sequence, extract the difference between the coordinate vector of the boundary point at the current time and the initial reference coordinate vector, and calculate the arithmetic mean of the difference over the total number of successfully paired boundary points to generate the global organization displacement vector at the current time.

[0026] Calculate the difference vector between the coordinate vector of the boundary point at the current moment and the initial reference coordinate vector, and subtract the global organization displacement vector from the difference vector to obtain the local pipe wall deformation displacement vector of the boundary point at the current moment.

[0027] Optionally, the process of introducing a small constant, combining the solution of local discrete tangent vectors with neighboring points, performing dimensionality-upgrading and zero-padding on the tangent vectors to calculate their cross product magnitude, and then performing aggregation processing on the cross product magnitude to extract the geometric torsion factor, specifically includes:

[0028] Within a preset effective node range, extract the coordinate vectors of adjacent boundary points and perform subtraction to obtain the coordinate difference vector, and calculate the norm of the coordinate difference vector;

[0029] The norm is compared with a preset small regularization constant, and the maximum value of the two is extracted as the denominator constraint term.

[0030] Divide the coordinate difference vector by the denominator constraint term to obtain the discrete tangent vector at the boundary point;

[0031] The discrete tangent vector in the two-dimensional coordinate system is supplemented with zero components along the third spatial dimension to construct a three-dimensional tangent vector. The cross product of adjacent three-dimensional tangent vectors is performed to obtain the normal difference vector, and the scalar magnitude of the normal difference vector is calculated and defined as the local geometric torsion index.

[0032] The local geometric torsion index within the node interval of the limited computational domain is summed and divided by the total number of valid nodes after removing out-of-bounds nodes to obtain the spatial average geometric torsion factor.

[0033] Optionally, the step of calculating the physical strain ratio based on the displacement difference and initial length, performing quadratic polymerization in conjunction with the elastic modulus, and superimposing the physiological basis constant to generate a strain energy index specifically includes:

[0034] Extract the local pipe wall deformation displacement vectors of adjacent nodes and perform subtraction processing to obtain the deformation difference vector, and calculate the norm of the deformation difference vector;

[0035] Extract the initial boundary coordinate vectors of adjacent nodes and perform subtraction to obtain the initial length difference vector, and calculate the norm of the initial length difference vector;

[0036] The local physical strain ratio is obtained by dividing the norm of the deformation difference vector by the norm of the initial length difference vector.

[0037] Calculate the square of the local physical strain ratio, sum the squares of all local nodes, multiply the sum by the preset equivalent elastic modulus coefficient, and halve the product to generate an intermediate aggregation index.

[0038] Based on the intermediate aggregation index, a preset physiological basis regularization constant with equivalent pressure dimension is added to generate a discrete aggregation type dynamic strain energy index.

[0039] Optionally, the step of extracting discrete sample points driven by strain energy time integration, constructing an inverse time mapping function with phase as the independent variable through cubic spline interpolation, specifically includes:

[0040] For discrete aggregate dynamic strain energy index, integral processing is performed within the virtual time interval from the initial time to the current time to obtain the current cumulative integral value of strain energy;

[0041] The duration of the complete cardiac cycle is obtained, and the discrete-aggregate dynamic strain energy index is integrated over the duration of the complete cardiac cycle to obtain the baseline periodic strain energy integral value.

[0042] Divide the current cumulative integral value of strain energy by the integral value of strain energy during the reference period to generate the normalized pulsation phase mapping factor corresponding to the current moment.

[0043] Extract each physical time frame and its corresponding normalized pulsation phase mapping factor, construct discrete sample point pairs, use the normalized pulsation phase mapping factor as the independent variable and the corresponding physical time frame as the dependent variable, perform discrete strictly monotonic cubic spline interpolation, and construct the time mapping inverse function that maps the phase to physical time.

[0044] Optionally, the step of calling the velocity matrix and direction parameters, resampling the velocity using an inverse function, compensating for the symbolic truncation angle, extracting the projection components, and constructing the velocity gradient tensor specifically includes:

[0045] Retrieve the pre-stored ultrasound detection beam, the Doppler angle matrix of blood vessel direction, and the spatial unit vector matrix of blood vessel direction;

[0046] Based on the inverse time mapping function, the elements of the original blood flow velocity matrix are extracted by time domain resampling;

[0047] Calculate the cosine value of the elements of the Doppler angle matrix, extract the maximum value between the absolute value of the cosine value and the pre-stored safety truncation constant threshold, multiply it by the sign determination value corresponding to the cosine value, and construct the symbolic truncation compensation denominator.

[0048] Divide the elements of the blood flow velocity matrix extracted by the resampling by the symbolic truncated compensation denominator to obtain the amplitude-compensated flow velocity;

[0049] Extract the directional components of the spatial unit vector matrix of the blood vessel orientation in each two-dimensional spatial coordinate system base dimension, and multiply the amplitude compensation flow velocity by each of the directional components to generate the projected velocity field vector components.

[0050] For the projected velocity field vector components, partial derivatives are calculated along each basis variable dimension of the two-dimensional spatial coordinate system, and the results of the partial derivative calculations are used to construct a spatial velocity gradient tensor.

[0051] Optionally, the step of extracting the symmetric component of the velocity gradient, constructing a deformation rate tensor, calculating the dissipation density using the viscosity coefficient, and multiplying it by the aforementioned torsional factor to generate a weighted dissipation density specifically includes:

[0052] Extract the corresponding elements symmetrical about the main diagonal from the spatial velocity gradient tensor, add the symmetrical elements and halve them to construct the deformation rate tensor elements;

[0053] The squared values ​​of the deformation rate tensor elements in each dimension are calculated. The squared values ​​of each dimension in the two-dimensional spatial coordinate system are double-summed and accumulated. The accumulated result is multiplied by the pre-acquired blood dynamic viscosity constant and multiplication factor to obtain the local viscous dissipation density index.

[0054] Obtain a preset semi-empirical topology magnification factor, multiply the spatial average geometric torsion factor by the semi-empirical topology magnification factor and add 1 to construct a nonlinear topology magnification weight;

[0055] The local viscous dissipation density index is multiplied by the nonlinear topological amplification weight to generate a torsional weighted dissipation density index.

[0056] Optionally, the step of performing a double definite integral on the weighted dissipation density over the reference area and the full phase domain to generate a fluid-structure interaction dissipation index, and then mapping it to a theoretical threshold index to output an evaluation index, includes:

[0057] Divide the torsional weighted dissipation density index by the pre-acquired absolute blood temperature constant to obtain the quotient integrand function;

[0058] Using the extracted blood vessel reference cross-sectional area as the spatial two-dimensional integration domain and the normalized pulsation phase as the phase integration variable, a double definite integral is calculated on the integrand function of the quotient value to generate a two-dimensional fluid-structure interaction phase dissipation integral index.

[0059] Obtain the average reference velocity constant of the inflow channel space, and calculate the square value of the average reference velocity constant of the inflow channel space;

[0060] Obtain a preset hydrodynamic analytical constant, multiply the preset pi constant, blood dynamic viscosity constant, the hydrodynamic analytical constant and the square value in sequence, and divide the product by the blood absolute temperature constant to generate the reference theoretical limit threshold constant;

[0061] Divide the two-dimensional fluid-structure interaction phase dissipation integral index by the reference theoretical limit threshold constant, and take the negative value of the division result to generate the mapped variable.

[0062] Perform natural index operations on the mapped variables to generate a dimensionless state assessment index.

[0063] A system for implementing the arteriovenous fistula status assessment method, comprising:

[0064] Image acquisition module: used to acquire grayscale images and blood flow matrix of the arteriovenous fistula;

[0065] Boundary extraction module: used to construct a sequence of vessel boundary point coordinates based on gradient extrema;

[0066] Displacement decoupling module: used to match cross-frame boundaries and strip centroid offset to obtain local deformation displacement;

[0067] Torsion calculation module: used to calculate the cross product magnitude of the tangent vector and perform aggregation calculation on the cross product magnitude to generate the spatial average geometric torsion factor;

[0068] Strain energy construction module: Based on the physical strain ratio, it aggregates and generates dynamic strain energy indices;

[0069] Time mapping module: Based on the time integral of strain energy, an inverse time mapping function is constructed by interpolation;

[0070] Flow field reconstruction module: used to resample the flow velocity based on the inverse mapping function and construct the velocity gradient tensor;

[0071] Dissipation weighting module: Combines the deformation rate tensor and the torsion factor to generate a weighted dissipation density;

[0072] Evaluation output module: used to perform double definite integral on the weighted dissipation density and output the evaluation index via exponential mapping.

[0073] The present invention has the following beneficial effects:

[0074] 1. To address the challenges of flow field misalignment and difficulty in quantifying internal friction caused by asynchronous deformation and spatial morphological distortion of the vessel wall in arteriovenous fistulas under unsteady pulsating blood flow, this solution extracts boundary coordinates from multiple images to remove global offset, extracts geometric torsion factors using discrete tangent vectors, and generates dynamic strain energy indices based on physical strain ratios. Subsequently, a time-mapping inverse function is constructed using the strain energy integral as the driving force. Flow velocity is resampled and a velocity gradient tensor is constructed, and the viscous dissipation density corresponding to the deformation rate tensor is calculated. This density is then fused with the torsion factor to generate a weighted dissipation density. Finally, a double definite integral and theoretical threshold mapping are performed in the two-dimensional space and the full-phase domain to output an evaluation index. These technical steps break through the limitations of appearance morphology measurement and... The independent analysis barrier of single-point blood flow detection establishes a topological feedback mechanism between structural deformation and fluid work at the underlying mechanism level. By reconstructing the inverse function of time mapping using the cumulative strain energy of the vessel wall, it overcomes the spatial matching deviation of conventional absolute physical time in asynchronous pulsation state and achieves phase alignment between high-speed blood flow information and dynamic vessel wall characteristics in the energy dimension. At the same time, the scheme directly embeds the geometric torsion degree, which characterizes macroscopic structural anomalies, as a nonlinear variable into the internal friction calculation model of micro-fluid dynamics. This cross-dimensional feature coupling mechanism unifies the invisible flow field boundary separation risk and the objectively existing vascular tortuous configuration into the kinetic energy attenuation parameter under the fluid-structure interaction system, providing an evaluation scale independent of subjective visual experience.

[0075] 2. By simultaneously acquiring multimodal ultrasound images in the target evaluation area and extracting the baseline flow velocity and geometric direction prior parameters, the pixel grayscale change rate is calculated using the partial derivatives on the spatial coordinate components of the image. Within the set local search range, the spatial extreme points of this change rate are accurately selected, which serve as objective geometric markers of the tissue interface. A continuous and independent discrete coordinate extraction sequence of the vascular wall intima is generated. This processing logic effectively avoids the subjective judgment error caused by manual visual delineation, establishes an objective spatial physical constraint surface for subsequent dynamic deformation assessment, and eliminates the interference caused by non-target echoes from complex physiological background tissues on the identification of the inner wall boundary. This improves the extraction accuracy and numerical stability of the underlying basic morphological data, and lays a solid baseline origin condition for the overall fluid dynamics calculation.

[0076] 3. By incorporating the arc length topological difference between adjacent boundary points as a penalty weight term into the shortest distance matching operation, combined with the spatial spline interpolation compensation mechanism under continuous smooth constraints, the risk of in-situ conflicts and matching loss caused by local occlusion is effectively eliminated, and the independent index consistency of cross-frame coordinate sequence evolution is maintained. Furthermore, the coordinate difference between the registration sequence at the current time and the initial reference is extracted, and the spatial arithmetic mean is calculated to generate a tissue displacement component representing low-frequency global offset. Subtracting this global tissue displacement component from the current absolute coordinate difference successfully decouples the non-target overall offset interference caused by environmental factors such as external breathing fluctuations or probe jitter, restores the elastic deformation characteristics after stripping external motion artifacts, and improves the fidelity of local deformation displacement data.

[0077] 4. By calculating the norm of the difference vector between adjacent boundary coordinates within the effective node interval, comparing it with a preset small regularization constant, and extracting the maximum value as the denominator constraint term, the risk of division by zero divergence triggered in extreme structural closure or collapse states is effectively avoided, the original physical distribution law of the discrete tangent vector direction is maintained, the tangent vector is supplemented with zero components along the third spatial dimension and spatial cross product processing is performed, the scalar magnitude of the normal difference vector is extracted to construct the local geometric torsion index, and then the dimensionless spatial summation and averaging operation is performed on the local torsion index within the effective computational domain, objectively quantifying the degree of abnormal spatial curvature and geometric distortion caused by long-term pressure, overcoming the numerical instability defect of conventional curvature algorithms at irregular bifurcation singularities;

[0078] 5. By extracting the local tube wall deformation displacement difference and comparing it with the initial length difference, a physical strain ratio is constructed. The square of this strain ratio is summed and aggregated over the entire domain. Combined with the equivalent elastic modulus coefficient, the mechanical energy absorbed by the structure is calculated. The two-dimensional spatial displacement characteristics of the tube wall are transformed into dynamic energy parameters that characterize the tensile resistance of the tissue. At the same time, based on the aggregated calculation results, a physiological base regularization constant with equivalent pressure dimension is added to endow the system with the basic hydrostatic ground state energy during the diastolic stagnation period. This prevents the output value from returning to zero due to deformation stagnation, ensures the numerical continuity of subsequent time mapping and integral calculation, and objectively reflects the dynamic energy reserve state and physical deformation compensation capacity of the tube wall under the impact of pulsating blood flow.

[0079] 6. By performing time integration processing on the extracted discrete aggregate dynamic strain energy index within the current virtual time span and the complete cardiac cycle, the ratio of the accumulated strain energy integral values ​​is calculated to generate the normalized pulsation phase factor corresponding to the current moment. Using this normalized phase factor as the independent variable and the corresponding physical time frame as the dependent variable, discrete spline interpolation is performed under monotonicity constraints to construct an inverse function that maps the structural evolution phase inversely to physical time. By utilizing the historical energy accumulation of the structural response, the traditional fixed physical time reference is replaced, which effectively improves the spatial matching misalignment phenomenon, ensures the phase alignment stability of the multidimensional flow field data within the complex pulsation cycle, and improves the spatiotemporal synchronization accuracy of cross-frame resampling.

[0080] 7. By retrieving the pre-stored Doppler angle parameters and spatial direction components, the blood flow velocity is resampled in the time domain according to the inverse time mapping function. The absolute value of the cosine of the sound beam angle is calculated and compared with the safe truncation threshold to extract the maximum value. Combined with the sign determination condition, a truncation compensation denominator is constructed, which effectively avoids the risk of amplitude division by zero divergence caused when the sound beam is nearly perpendicular to the blood flow, and prevents the velocity value distortion caused by the detection angle jitter. The amplitude-compensated flow velocity is multiplied with the spatial unit direction component to obtain the projected velocity field vector component. The partial derivatives are calculated along each dimension of the spatial coordinate system to construct the velocity gradient tensor, which truly restores the flow field distribution matrix with energy phase alignment characteristics. This provides a reliable data foundation for the subsequent stripping of the spatial shear deformation and rotation trend inside the complex fluid.

[0081] 8. By extracting the main diagonal symmetric elements of the spatial velocity gradient tensor and performing addition and halving, the rigid body rotation component in the motion of fluid micro-clusters is removed, while the pure physical tensor of stretching and shear deformation rate that actually induces internal friction work is retained. The square values ​​of each dimension of the deformation rate are calculated and double summation is performed. Combined with the dynamic viscosity parameter, the local viscous dissipation density is obtained. Then, the pre-extracted spatial average geometric torsion factor and semi-empirical topological coefficient are introduced to construct a nonlinear amplification weight. The topological weight is multiplied with the local viscous dissipation density, and the macroscopic three-dimensional spatial geometric distortion is directly coupled to the microscopic fluid internal friction calculation model. This objectively quantifies the physical amplification effect of irregular spatial configuration on fluid vortex dissipation and improves the sensitivity of the overall evaluation system.

[0082] 9. By dividing the torsional weighted dissipation density by the absolute temperature constant, the integrand of the quotient is obtained. Under the two-dimensional physical closed-loop constraint of the reference cross-sectional area and the normalized pulsation phase, a double definite integral is performed to calculate the fluid-structure interaction dissipation integral index for evaluating the overall cross-sectional energy degradation. Based on the laminar flow model principle, a theoretical limit threshold with the same physical dimensions is constructed using the spatial average reference velocity and the hydrodynamic analytical constant. The actual dissipation integral index is divided by this theoretical limit threshold, and a natural exponential mapping operation is performed to generate an independent state evaluation parameter that removes dimensional interference. Through this state evaluation parameter, a quantitative result that objectively reflects the overall quality of hemodynamics within the system is output, reducing the dependence of the evaluation result on visual interpretation experience. Attached Figure Description

[0083] Figure 1 This is a schematic diagram of the main process of the evaluation method of the present invention.

[0084] Figure 2 This is a schematic diagram illustrating the deformation displacement decoupling and geometric torsion extraction principle of the present invention.

[0085] Figure 3 This is a schematic diagram illustrating the principle of discrete boundary point gradient optimization extraction in this invention.

[0086] Figure 4 This is a schematic diagram of the Doppler blood flow vector truncation compensation and spatial projection decomposition of the present invention. Detailed Implementation

[0087] 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.

[0088] Example 1, as Figure 1 As shown, a method for assessing the status of arteriovenous fistulas includes:

[0089] Acquire multi-frame grayscale images and blood flow matrix of the arteriovenous fistula, locate boundary markers based on gradient extrema, and construct a discrete vascular boundary point coordinate sequence;

[0090] By combining the arc length topological difference and Euclidean distance, cross-frame boundary matching is performed, and the global offset vector is stripped by mean calculation to obtain the local deformation displacement.

[0091] Introducing a small constant, combining the local discrete tangent vectors with adjacent points, performing dimensionality-upgrading and zero-padding on the tangent vectors to calculate their cross product magnitude, and then performing aggregation processing on the cross product magnitude to extract the geometric torsion factor;

[0092] The physical strain ratio is calculated based on the displacement difference and initial length, and a quadratic polymerization is performed in combination with the elastic modulus. The strain energy index is generated by superimposing the physiological basis constant.

[0093] Driven by the time integral of strain energy, discrete sample points are extracted, and the inverse time mapping function is constructed by cubic spline interpolation with phase as the independent variable.

[0094] The velocity matrix and direction parameters are called, the velocity is resampled using the inverse function, and the projection components are extracted and the velocity gradient tensor is constructed after symbolic truncation angle compensation.

[0095] The symmetric component of the velocity gradient is extracted, the deformation rate tensor is constructed, the dissipation density is calculated by combining the viscosity coefficient, and the weighted dissipation density is generated by multiplying it by the aforementioned torsional factor.

[0096] The weighted dissipation density is subjected to a double definite integral over the reference area and the full phase domain to generate a fluid-structure interaction dissipation index. After mapping with the theoretical threshold index, the evaluation index is output.

[0097] The process of acquiring multi-frame grayscale images and blood flow matrices of the arteriovenous fistula, locating boundary markers based on gradient extrema, and constructing a discrete vascular boundary point coordinate sequence specifically includes:

[0098] An ultrasound probe array is attached to the target area to simultaneously acquire multiple frames of grayscale image matrix and blood flow velocity matrix.

[0099] Initialize the data acquisition timestamps and establish the physical calculation origin;

[0100] The region of interest is delineated using the grayscale image at the initial time, and the reference cross-sectional area at the initial time is extracted;

[0101] Calculate the partial derivatives of the grayscale image matrix in the two-dimensional spatial coordinate system to obtain the rate of change of local pixel grayscale values;

[0102] Within a preset local search neighborhood, the extreme points of the rate of change are selected as geometric markers at the tissue interface, and the initial boundary point set coordinate sequence of the vascular intima is extracted.

[0103] The process of combining arc-length topological differences and Euclidean distance to perform cross-frame boundary matching, extracting the global offset vector through mean calculation, and obtaining local deformation displacement specifically includes:

[0104] The difference in arc length topology between adjacent boundary points is obtained as a penalty weight term and added to the Euclidean shortest distance calculation process to perform cross-frame nearest neighbor matching.

[0105] When a local match is lost, the coordinates of the effective boundary points of the adjacent successfully matched points are extracted as interpolation nodes. Under the smooth constraint of continuous local second derivative, spatial cubic spline interpolation compensation is performed.

[0106] For the registered coordinate sequence, extract the difference between the coordinate vector of the boundary point at the current time and the initial reference coordinate vector, and calculate the arithmetic mean of the difference over the total number of successfully paired boundary points to generate the global organization displacement vector at the current time.

[0107] Calculate the difference vector between the coordinate vector of the boundary point at the current moment and the initial reference coordinate vector, and subtract the global organization displacement vector from the difference vector to obtain the local pipe wall deformation displacement vector of the boundary point at the current moment.

[0108] The process involves introducing a small constant, solving for the local discrete tangent vector by combining neighboring points, performing dimensionality-upgrading and zero-padding on the tangent vector to calculate its cross product magnitude, and then performing aggregation processing on the cross product magnitude to extract the geometric torsion factor. Specifically, this includes:

[0109] Within a preset effective node range, extract the coordinate vectors of adjacent boundary points and perform subtraction to obtain the coordinate difference vector, and calculate the norm of the coordinate difference vector;

[0110] The norm is compared with a preset small regularization constant, and the maximum value of the two is extracted as the denominator constraint term.

[0111] Divide the coordinate difference vector by the denominator constraint term to obtain the discrete tangent vector at the boundary point;

[0112] The discrete tangent vector in the two-dimensional coordinate system is supplemented with zero components along the third spatial dimension to construct a three-dimensional tangent vector. The cross product of adjacent three-dimensional tangent vectors is performed to obtain the normal difference vector, and the scalar magnitude of the normal difference vector is calculated and defined as the local geometric torsion index.

[0113] The local geometric torsion index within the node interval of the limited computational domain is summed and divided by the total number of valid nodes after removing out-of-bounds nodes to obtain the spatial average geometric torsion factor.

[0114] The process of calculating the physical strain ratio based on the displacement difference and initial length, performing quadratic polymerization in conjunction with the elastic modulus, and superimposing the physiological basis constant to generate a strain energy index specifically includes:

[0115] Extract the local pipe wall deformation displacement vectors of adjacent nodes and perform subtraction processing to obtain the deformation difference vector, and calculate the norm of the deformation difference vector;

[0116] Extract the initial boundary coordinate vectors of adjacent nodes and perform subtraction to obtain the initial length difference vector, and calculate the norm of the initial length difference vector;

[0117] The local physical strain ratio is obtained by dividing the norm of the deformation difference vector by the norm of the initial length difference vector.

[0118] Calculate the square value of the local physical strain ratio, sum the square values ​​of all local nodes, multiply the sum by the preset equivalent elastic modulus coefficient, and halve the product to generate an intermediate aggregation index.

[0119] Based on the intermediate aggregation index, a preset physiological basis regularization constant with equivalent pressure dimension is added to generate a discrete aggregation type dynamic strain energy index.

[0120] The process of extracting discrete sample points using strain energy time integration as the driving force, constructing an inverse time mapping function with phase as the independent variable through cubic spline interpolation, specifically includes:

[0121] For discrete aggregate dynamic strain energy index, integral processing is performed within the virtual time interval from the initial time to the current time to obtain the current cumulative integral value of strain energy;

[0122] The duration of the complete cardiac cycle is obtained, and the discrete-aggregate dynamic strain energy index is integrated over the duration of the complete cardiac cycle to obtain the integral value of the baseline periodic strain energy.

[0123] Divide the current cumulative integral value of strain energy by the integral value of strain energy during the reference period to generate the normalized pulsation phase mapping factor corresponding to the current moment.

[0124] Extract each physical time frame and its corresponding normalized pulsation phase mapping factor, construct discrete sample point pairs, use the normalized pulsation phase mapping factor as the independent variable and the corresponding physical time frame as the dependent variable, perform discrete strictly monotonic cubic spline interpolation, and construct the time mapping inverse function that maps the phase to physical time.

[0125] The process of calling the velocity matrix and direction parameters, resampling the velocity using an inverse function, compensating for the symbolic truncation angle, extracting the projection components, and constructing the velocity gradient tensor specifically includes:

[0126] Retrieve the pre-stored ultrasound detection beam, the Doppler angle matrix of blood vessel direction, and the spatial unit vector matrix of blood vessel direction;

[0127] Based on the inverse time mapping function, the elements of the original blood flow velocity matrix are extracted by time domain resampling;

[0128] Calculate the cosine value of the elements of the Doppler angle matrix, extract the maximum value between the absolute value of the cosine value and the pre-stored safety truncation constant threshold, multiply it by the sign determination value corresponding to the cosine value, and construct the symbolic truncation compensation denominator.

[0129] Divide the elements of the blood flow velocity matrix extracted by the resampling by the symbolic truncated compensation denominator to obtain the amplitude-compensated flow velocity;

[0130] Extract the directional components of the spatial unit vector matrix of the blood vessel orientation in each two-dimensional spatial coordinate system base dimension, and multiply the amplitude compensation flow velocity by each of the directional components to generate the projected velocity field vector components.

[0131] For the projected velocity field vector components, partial derivatives are calculated along each basis variable dimension of the two-dimensional spatial coordinate system, and the results of the partial derivative calculations are used to construct a spatial velocity gradient tensor.

[0132] The extraction of the symmetric component of the velocity gradient, the construction of the deformation rate tensor, the calculation of the dissipation density using the viscosity coefficient, and the multiplication by the aforementioned torsional factor to generate a weighted dissipation density, specifically includes:

[0133] Extract the corresponding elements symmetrical about the main diagonal from the spatial velocity gradient tensor, add the symmetrical elements and halve them to construct the deformation rate tensor elements;

[0134] The squared values ​​of the deformation rate tensor elements in each dimension are calculated. The squared values ​​of each dimension in the two-dimensional spatial coordinate system are double-summed and accumulated. The accumulated result is multiplied by the pre-acquired blood dynamic viscosity constant and multiplication factor to obtain the local viscous dissipation density index.

[0135] Obtain a preset semi-empirical topology magnification factor, multiply the spatial average geometric torsion factor by the semi-empirical topology magnification factor and add 1 to construct a nonlinear topology magnification weight;

[0136] The local viscous dissipation density index is multiplied by the nonlinear topological amplification weight to generate a torsional weighted dissipation density index.

[0137] The process involves performing a double definite integral on the weighted dissipation density over the reference area and the full phase domain to generate a fluid-structure interaction dissipation index. This index is then mapped using a theoretical threshold index to output an evaluation index, including:

[0138] Divide the torsional weighted dissipation density index by the pre-acquired absolute blood temperature constant to obtain the quotient integrand function;

[0139] Using the extracted blood vessel reference cross-sectional area as the spatial two-dimensional integration domain and the normalized pulsation phase as the phase integration variable, a double definite integral is calculated on the integrand function of the quotient value to generate a two-dimensional fluid-structure interaction phase dissipation integral index.

[0140] Obtain the average reference velocity constant of the inflow channel space, and calculate the square value of the average reference velocity constant of the inflow channel space;

[0141] Obtain a preset hydrodynamic analytical constant, multiply the preset pi constant, blood dynamic viscosity constant, the hydrodynamic analytical constant and the square value in sequence, and divide the product by the blood absolute temperature constant to generate the reference theoretical limit threshold constant;

[0142] Divide the two-dimensional fluid-structure interaction phase dissipation integral index by the reference theoretical limit threshold constant, and take the negative value of the division result to generate the mapped variable.

[0143] Perform natural index calculation on the mapped variable to generate a dimensionless state assessment index;

[0144] To address the challenges of asynchronous vessel wall deformation and spatial morphological distortion in arteriovenous fistulas (AVFs) under unsteady pulsatile blood flow impact, which can easily lead to flow field misalignment and difficulties in quantifying internal friction, this scheme extracts boundary coordinates from multiple image frames to remove global offset, extracts geometric torsion factors using discrete tangent vectors, and generates dynamic strain energy indices based on physical strain ratios. Subsequently, a time-mapping inverse function is constructed using the strain energy integral as the driving force. Flow velocity is resampled and a velocity gradient tensor is constructed, and the viscous dissipation density corresponding to the deformation rate tensor is calculated. This density is then fused with the torsion factor to generate a weighted dissipation density. Finally, a double definite integral and theoretical threshold mapping are performed in the two-dimensional space and the full-phase domain to output an evaluation index. These technical steps break through the limitations of appearance morphology measurement and single-frame analysis. The independent analysis barrier of point blood flow detection establishes a topological feedback mechanism between structural deformation and fluid work at the underlying mechanism level. By reconstructing the inverse function of time mapping using the cumulative strain energy of the vessel wall, it overcomes the spatial matching deviation of conventional absolute physical time in asynchronous pulsation state and achieves phase alignment between high-speed blood flow information and dynamic vessel wall characteristics in the energy dimension. At the same time, the scheme directly embeds the geometric torsion degree, which characterizes macroscopic structural anomalies, as a nonlinear variable into the internal friction calculation model of micro-fluid dynamics. This cross-dimensional feature coupling mechanism unifies the invisible flow field boundary separation risk and the objectively existing vascular tortuous configuration into the kinetic energy attenuation parameter under the fluid-structure interaction system, providing an evaluation scale independent of subjective visual experience.

[0145] Example 2, a method for assessing the status of arteriovenous fistulas, further includes the following steps:

[0146] Step 1: Simultaneously acquire a matrix of consecutive two-dimensional B-mode grayscale images and a color Doppler blood flow velocity matrix, including the anastomosis and the inflow or outflow tract vascular segment, in the targeted area of ​​the patient's arteriovenous fistula.

[0147] Initialize data acquisition timestamps and establish the physical calculation origin. ;

[0148] Using the grayscale image at the initial moment, the region of interest of the static blood vessel wall is delineated, and the reference cross-sectional area at the initial moment is extracted as the absolute spatial physical constraint surface for subsequent two-dimensional integration.

[0149] The average reference velocity constant of the inflow channel in the Doppler flow field is extracted, and the equipment parameters of the ultrasound equipment are analyzed. The sound beam emission angle is extracted to construct the Doppler angle matrix for blood flow detection. At the same time, based on the spatial orientation of the blood vessel boundary in the initial frame grayscale image, the central axis fitting algorithm is used to calculate and generate the spatial unit vector matrix of the blood vessel orientation.

[0150] like Figure 3 As shown, to accurately extract the initial boundary point set of the vascular intima, it is necessary to calculate the partial derivatives of the grayscale image matrix in the two-dimensional spatial coordinate system to obtain the rate of change of local pixel grayscale values, and then find the extreme points of this rate of change within a preset local search neighborhood as geometric markers at the tissue interface, thereby constructing a discrete coordinate extraction formula:

[0151]

[0152] in, Indicates at time The extracted first The coordinate vectors of the boundary points represent the target quantity that is currently required. Indicates the preset first A local search neighborhood, the size of which is set according to the physical spatial resolution of the ultrasound probe, preferably set as a dynamic pixel matrix window with a side length of 3 to 5 mm centered on the initial set point; Indicates at time coordinate The pixel intensity of the B-mode grayscale image at that location is a known quantity that can be directly measured; and For two-dimensional spatial coordinate system components; The sign for partial derivatives;

[0153] By simultaneously acquiring multimodal ultrasound images in the target evaluation area and extracting the baseline flow velocity and geometric direction prior parameters, the pixel grayscale change rate is calculated using the partial derivatives on the spatial coordinate components of the image. Within a set local search range, the spatial extreme points of this change rate are accurately selected, which serve as objective geometric markers of the tissue interface. This generates a continuous and independent discrete coordinate extraction sequence of the vascular wall intima. This processing logic effectively avoids subjective judgment errors caused by manual visual delineation, establishes an objective spatial physical constraint surface for subsequent dynamic deformation assessment, and eliminates interference from non-target echoes of complex physiological background tissues on the identification of the inner wall boundary. This improves the extraction accuracy and numerical stability of the underlying basic morphological data, and lays a solid baseline origin condition for the overall fluid dynamics calculation.

[0154] Step Two: Refer to Figure 2 The difference in arc length topology between adjacent boundary points is obtained and added as a penalty weight to the Euclidean shortest distance calculation process to perform cross-frame nearest neighbor matching.

[0155] When a local match is lost, the coordinates of the effective boundary points of the adjacent successfully matched points are extracted as interpolation nodes. Under the smooth constraint of continuous local second derivative, spatial cubic spline interpolation compensation is performed to resolve the same-distance conflict and ensure that the continuously evolving independent index closed loop is absolutely constant.

[0156] To decouple low-frequency global offset interference between the probe and blood vessel relative positions caused by patient breathing or muscle tremors, the coordinate difference between the registered boundary points at the current time and the initial time is extracted, and the arithmetic mean of this difference over the total number of successfully paired boundary points is calculated to generate the global tissue displacement vector at the current time.

[0157]

[0158] in, Indicates at time The global organization displacement vector is the intermediate calculation amount required at present; This represents the total number of known and successfully paired constant boundary points; For the first obtained through prior calculation The boundary points at time... The coordinate vector; Initial time The first time A reference coordinate vector for each boundary point;

[0159] To eliminate the aforementioned global low-frequency motion interference, the absolute coordinate difference vector of the boundary point at the current moment is subtracted from the global tissue displacement vector, and the local pipe wall deformation displacement vector is obtained through differential calculation:

[0160]

[0161] in, Indicates the first The boundary points at time... The local pipe wall deformation displacement vector is the target quantity that is currently required; and The known boundary point coordinate vector; The global organization displacement vector obtained from the preceding calculations;

[0162] By incorporating the arc length topological difference between adjacent boundary points as a penalty weight term into the shortest distance matching operation, combined with the spatial spline interpolation compensation mechanism under continuous smooth constraints, the risk of in-situ conflicts and matching loss caused by local occlusion is effectively eliminated, and the independent index consistency of cross-frame coordinate sequence evolution is maintained. Furthermore, the coordinate difference between the registration sequence at the current time and the initial reference is extracted, and the spatial arithmetic mean is calculated to generate a tissue displacement component representing low-frequency global offset. Subtracting this global tissue displacement component from the current absolute coordinate difference successfully decouples the non-target overall offset interference caused by environmental factors such as external breathing fluctuations or probe jitter, restores the elastic deformation characteristics after stripping external motion artifacts, and improves the fidelity of local deformation displacement data.

[0163] Step 3: To quantify the geometric tortuosity of the blood vessel wall, the coordinates of adjacent boundary points are extracted based on the effective node intervals within the vessel and subtracted to construct a coordinate difference vector reflecting the local morphological orientation. The norm of this vector is calculated and compared with a preset small regularization constant. The maximum value is extracted as a denominator constraint term with zero-distance protection applied, thereby obtaining the discrete tangent vector at the boundary points.

[0164]

[0165] in, Indicates at time And the index range is limited to The The discrete tangent vectors at the boundary points are the intermediate quantities currently being calculated; and The coordinate vector of the adjacent boundary points extracted earlier; The Euclidean norm operator for vectors; Screening operators for extreme values; This is a preset, small regularization constant;

[0166] For this small regularization constant, the preferred value is recommended to be set to... to The logic behind its value setting is as follows: using the lower limit of floating-point arithmetic precision to avoid the risk of division by zero caused by boundary collapse, while maintaining the original physical authenticity of the tangent vector direction; if the value is too large, the magnitude of the tangent vector will be forcibly compressed, causing severe distortion of local geometric features; if the value is too small, division by zero may still be triggered under extreme pathological deformation.

[0167] To avoid data divergence at bifurcation singularities, the discrete tangent vectors in the two-dimensional coordinate system are supplemented with zero components along the third spatial dimension to construct three-dimensional tangent vectors. Cross products are then performed on adjacent three-dimensional tangent vectors to obtain the normal difference vectors, and their scalar magnitude is calculated and defined as the local geometric torsion index.

[0168]

[0169] in, Indicates at time And the index range is strictly limited to The local geometric torsion index is the intermediate value obtained at this time; and The discrete tangent vector obtained beforehand; The cross product operator for three-dimensional vectors;

[0170] To form a macroscopic evaluation parameter reflecting the overall vascular segment anomaly status, the local geometric torsion indices within a defined computational domain are summed and divided by the total number of valid nodes after removing out-of-bounds nodes. Dimensionless spatial aggregation is then performed to obtain the spatially averaged geometric torsion factor.

[0171]

[0172] in, Indicates at time The spatial average geometric torsion factor is the target quantity that is currently required. This represents the total number of valid computational domain nodes after removing out-of-bounds nodes. The local geometric torsional index is obtained beforehand;

[0173] By calculating the norm of the difference vector between adjacent boundary coordinates within the effective node interval, comparing it with a preset small regularization constant, and extracting the maximum value as the denominator constraint term, the risk of division-by-zero divergence triggered in extreme structural closure or collapse states is effectively avoided, and the original physical distribution law of the discrete tangent vector direction is maintained. The tangent vector is supplemented with zero components along the third spatial dimension and spatial cross product processing is performed to extract the scalar magnitude of the normal difference vector to construct the local geometric torsion index. Subsequently, the local torsion index is subjected to dimensionless spatial summation and averaging operations within the effective computational domain, which objectively quantifies the degree of abnormal spatial curvature and geometric distortion caused by long-term pressure, and overcomes the numerical instability defect of conventional curvature algorithms at irregular bifurcation singularities.

[0174] Step 4: To assess the physical tensile effect of blood flow impact on the vessel wall, the local vessel wall deformation displacement vectors and initial boundary coordinate vectors of adjacent nodes are extracted, and a subtraction operation is performed to calculate the corresponding deformation difference vector norm and the initial length difference vector norm. The local physical strain ratio is calculated by dividing the two:

[0175]

[0176] in, Indicates time node The local physical strain ratio is an intermediate value obtained at this time; and This is the local pipe wall deformation displacement vector obtained beforehand; and The initial boundary coordinate vector is known;

[0177] Calculate the square of the local physical strain ratio at each node, perform a global quadratic summation, multiply it by the preset equivalent elastic modulus coefficient and halve it, then add the preset physiological basis regularization constant with equivalent pressure dimensions to generate a discrete aggregated dynamic strain energy index for overall absorption:

[0178]

[0179] in, Indicates time The discrete aggregate dynamic strain energy index is an intermediate value obtained at present; For the local physical strain ratio obtained beforehand; The preset equivalent elastic modulus coefficient of the arteriovenous fistula wall; This is a pre-defined physiological basis regularization constant with equivalent pressure dimensions;

[0180] Regarding this equivalent elastic modulus The preferred value is recommended to be set as follows: to The optimal value can be obtained through experimental statistical testing: perform in vitro tensile mechanical testing on the vein wall of mature arteriovenous fistulas and take the statistical median of the experimental test results; if the value of this parameter is too large, it will lead to a smaller overall strain energy index, which will compress the subsequent phase mapping interval and lose the high-frequency pulsation characteristics of the flow field; if the value of this parameter is too small, it will abnormally amplify the tiny ultrasonic detection noise, making it difficult for the time phase distortion function to converge smoothly.

[0181] For this physiological base regularization constant The preferred value is set to the baseline value of the central venous pressure measurement of the arteriovenous fistula. to The logical basis for setting the value is: to characterize the hydrostatic ground state energy of the fluid and to ensure that the output of the system is non-zero in any cardiac arrest diastolic state; if the value of this parameter is too large, it will mask the characteristic weight of the dynamic pulsation component, causing the subsequent phase resampling homogenization to fail; if the value of this parameter is too small, it will not be able to play its role in preventing zero overflow through mathematical regularization, and will face the risk of inverse mapping collapse.

[0182] By extracting the local tube wall deformation displacement difference and comparing it with the initial length difference, a physical strain ratio is constructed. The square of this strain ratio is summed and aggregated over the entire domain. Combined with the equivalent elastic modulus coefficient, the mechanical energy absorbed by the structure is calculated. The two-dimensional spatial displacement characteristics of the tube wall are transformed into dynamic energy parameters that characterize the tensile resistance of the tissue. At the same time, based on the aggregated calculation results, a physiological base regularization constant with equivalent pressure dimension is accumulated, which gives the system the basic hydrostatic ground state energy during the diastolic stagnation period. This prevents the output value from returning to zero due to deformation stagnation, ensures the numerical continuity of subsequent time mapping and integral calculation, and objectively reflects the dynamic energy reserve state and physical deformation compensation capacity of the tube wall under the impact of pulsating blood flow.

[0183] Step 5: To improve the spatial mismatch caused by asynchronous deformation of the tube wall and ensure the uniqueness of the phase space mapping, the discrete aggregate dynamic strain energy index is integrally processed within the virtual time interval from the initial moment to the current moment and within the complete cardiac cycle, respectively, to obtain the current cumulative integral value and the baseline cycle integral value, and the two are divided to generate the normalized pulsation phase mapping factor corresponding to the current moment:

[0184]

[0185] in, Indicates time The corresponding normalized pulsation phase mapping factor is the target quantity that needs to be obtained at present; The discrete aggregate dynamic strain energy index function is obtained beforehand; For integration, the virtual time variable; The duration of a single complete cardiac cycle as measured;

[0186] Extract each physical time frame and its corresponding normalized pulsation phase mapping factor Construct a set of discrete sample point pairs Using the phase mapping factor as the independent variable, and under the monotonicity constraint that ensures the first derivative is always positive, discrete cubic spline interpolation is performed on the set of point pairs to construct an inverse time mapping function that continuously maps arbitrary phases back to physical time. ;

[0187] By performing time integration processing on the extracted discrete aggregate dynamic strain energy index within the current virtual time span and the complete cardiac cycle, the ratio of the accumulated strain energy integral values ​​is calculated to generate the normalized pulsation phase factor corresponding to the current moment. Using this normalized phase factor as the independent variable and the corresponding physical time frame as the dependent variable, discrete spline interpolation is performed under monotonicity constraints to construct an inverse function that maps the structural evolution phase inversely to physical time. By utilizing the historical energy accumulation of the structural response, the traditional fixed physical time reference is replaced, which effectively improves the spatial matching misalignment phenomenon, ensures the phase alignment stability of the multidimensional flow field data within the complex pulsation cycle, and improves the spatiotemporal synchronization accuracy of cross-frame resampling.

[0188] Step 6: Retrieve the pre-acquired ultrasound detection beam, the Doppler angle matrix of the blood vessel direction, and the spatial unit vector matrix of the blood vessel direction;

[0189] Based on the inverse time mapping function, the elements of the original blood flow velocity matrix are extracted by time domain resampling;

[0190] Combination Figure 4 As shown, the cosine values ​​of the elements of the Doppler angle matrix are calculated. The maximum value between the absolute value of the cosine value and the pre-stored safety truncation constant threshold is extracted. Combined with the sign determination value, a symbolic truncation compensation denominator is constructed. The resampled flow velocity is divided by this denominator for amplitude compensation, and then multiplied by the direction components of the spatial unit vector matrix in each basis dimension to generate projected velocity field vector components with phase alignment characteristics.

[0191]

[0192] in, Indicates a unified phase The projected velocity field under the following conditions, in the first... The vector components of the direction are the target quantities that are currently required; To extract the inverse function of the time map constructed by cubic spline interpolation The original known Doppler blood flow velocity matrix elements are directly retrieved; The known Doppler angle matrix; For the pre-acquired unit vector matrix of the vessel orientation space in the th... Component of direction; Indicates zero-point output as Symbolic operators; Screening operators for extreme values; The preset safety truncation constant threshold;

[0193] Regarding this safety cutoff constant threshold The preferred value is recommended to be set as follows: to The equivalent numerical range; the logic for the value is: to physically avoid the Doppler division-to-zero singularity caused when the ultrasound detection beam is nearly perpendicular to the actual blood flow direction; if the value of this parameter is too large, it will cause the actual flow velocity at the edge to be excessively suppressed and severely distorted under the condition of large detection angle; if the value of this parameter is too small, it will not be able to effectively intercept the risk of numerical divergence caused by slight hand tremors or angle jitter.

[0194] To further analyze the internal frictional mechanisms within the fluid, partial derivatives of the projected velocity field vector components are calculated along each dimension of the basis variables in the two-dimensional spatial coordinate system, thereby constructing the spatial velocity gradient tensor:

[0195]

[0196] in, Indicates phase Below, the spatial velocity gradient tensor in the th... Line 1 The tensor elements of the column represent the target quantity that is currently required; The components of the projected velocity field vector obtained beforehand; The first representing the spatial coordinate system One basic variable; The sign for partial derivatives;

[0197] By retrieving pre-stored Doppler angle parameters and spatial orientation components, and resampling the blood flow velocity in the time domain according to the inverse time mapping function, the absolute value of the cosine of the sound beam angle is calculated and compared with the safe truncation threshold to extract the maximum value. Combined with the sign determination condition, a truncation compensation denominator is constructed, which effectively avoids the risk of amplitude division by zero divergence caused when the sound beam is nearly perpendicular to the blood flow, and prevents the velocity value distortion caused by the jitter of the detection angle. The amplitude-compensated flow velocity is multiplied with the spatial unit direction component to obtain the projected velocity field vector component, and the partial derivatives are calculated along each dimension of the spatial coordinate system to construct the velocity gradient tensor. The flow field distribution matrix with energy phase alignment characteristics is realistically restored, providing a reliable data foundation for subsequent stripping of the spatial shear deformation and rotation trend inside the complex fluid.

[0198] Step 7: To isolate the pure shear deformation portion, extract the corresponding elements symmetric about the main diagonal in the spatial velocity gradient tensor, perform summation and halving symmetry processing, and construct the deformation rate tensor elements:

[0199]

[0200] in, The specific elements of the fluid deformation rate tensor matrix are intermediate values ​​obtained at this time. and These are all elements at the corresponding positions of the spatial velocity gradient tensor obtained through prior calculation;

[0201] By calculating the squared values ​​of the deformation rate tensor elements in each dimension, performing double summation, and then multiplying the summation result by the pre-acquired blood dynamic viscosity constant and the multiplication factor 2, the local viscous dissipation density index is calculated.

[0202]

[0203] in, Indicates phase and spatial coordinates The local viscous dissipation density index at the location is an intermediate value obtained at this time; The fluid deformation rate tensor element obtained beforehand; The target is the blood dynamic viscosity constant to be obtained;

[0204] To characterize the risk caused by geometric spatial distortion, a pre-defined semi-empirical topological amplification factor is obtained. This factor is then multiplied by the aforementioned spatial average geometric torsion factor and incremented by 1 to construct a nonlinear topological amplification weight. Finally, the local viscous dissipation density index is multiplied by this weight to generate a torsion-weighted dissipation density index.

[0205]

[0206] in, This represents the torsional weighted dissipation density index, which is the target quantity currently required. The local viscous dissipation density index is obtained beforehand; The spatial average geometric torsion factor obtained beforehand; This is a preset semi-empirical topology magnification factor;

[0207] Regarding the topological magnification factor The preferred value is recommended to be set as follows: to The optimal solution for this parameter is not absolutely fixed, but can be obtained within the above range through the following experimental fitting process: A three-dimensional computational fluid dynamics (CFD) simulation sample library is established for a large number of arteriovenous fistulas with different degrees of stenosis and tortuosity. The difference between the eddy current dissipation energy extracted from the actual CFD and the Poiseuille laminar flow reference dissipation energy is used as the objective function. The correlation between the local viscous dissipation density index and the spatial average geometric torsion factor is fitted nonlinearly using the least squares method. The weight coefficient at the convergence of the fitting is extracted as the optimal solution of the semi-empirical topological amplification coefficient. If the value of this range is too large, it will cause the system to misjudge the slight natural physiological curvature of the blood vessel. If the value is too small, it will mask the actual energy loss caused by the spatial spiral vortex dissipation, causing the model to lose its sensitivity.

[0208] By extracting the main diagonal symmetric elements of the spatial velocity gradient tensor and performing addition and halving, the rigid body rotation component in the motion of fluid micro-clusters is removed, while the pure physical tensor of stretching and shear deformation rate that actually induces internal friction work is retained. The square values ​​of each dimension of the deformation rate are calculated and double summation is performed. Combined with the dynamic viscosity parameter, the local viscous dissipation density is obtained. Then, the pre-extracted spatial average geometric torsion factor and semi-empirical topological coefficient are introduced to construct a nonlinear amplification weight. The topological weight is multiplied with the local viscous dissipation density, and the macroscopic three-dimensional spatial geometric distortion is directly coupled to the microscopic fluid internal friction calculation model. This objectively quantifies the physical amplification effect of irregular spatial configuration on fluid vortex dissipation and improves the sensitivity of the overall evaluation system.

[0209] Step 8: To evaluate the energy degradation degree of the arteriovenous fistula cross-section, the torsional weighted dissipation density index is divided by the pre-acquired absolute blood temperature constant to obtain the quotient integrand. A double definite integral is then performed on this integrand, using the reference cross-sectional area of ​​the blood vessel and the normalized pulsation phase as the integration domain variables, to generate a two-dimensional fluid-structure interaction phase dissipation integral index.

[0210]

[0211] in, This represents the two-dimensional fluid-structure interaction phase dissipation integral index, which is an intermediate value obtained at this time. The torsional weighted dissipation density index is obtained beforehand; The integral domain of the initially extracted blood vessel reference cross section is used to form a two-dimensional spatial physical closed-loop constraint; This is the normalized integral variable for the pulsation phase; The absolute temperature constant of the blood to be obtained;

[0212] Based on the Poiseuille laminar flow model principle, the squares of the preset constants for pi, blood dynamic viscosity, fluid dynamics analytical constants, and the average reference velocity constant in the inflow channel are multiplied sequentially and then divided by the absolute temperature constant of blood to generate the reference theoretical limit threshold constant.

[0213]

[0214] in, This represents the theoretical limit threshold, which is the intermediate value currently being calculated. Pi is a constant. The target is the blood dynamic viscosity constant to be obtained; The absolute temperature constant of the blood to be obtained; This is the known inflow channel space-averaged reference velocity constant extracted from the flow field;

[0215] The fluid dynamics analytical constant is 8, which is the analytical solution constant derived from the Navier-Stokes equations of fluid dynamics when performing radial integration on the Poiseuille laminar flow profile in an ideal circular straight pipe.

[0216] Divide the two-dimensional fluid-structure interaction phase dissipation integral index by the baseline theoretical limit threshold constant, and take the negative value to construct the mapped variable. Perform natural exponent operation on it to output a single-valued dimensionless state assessment index:

[0217]

[0218] in, This represents the final output arteriovenous fistula status assessment index, which is the final target output value currently required. The integral index for the two-dimensional fluid-structure interaction phase dissipation obtained beforehand; For mathematical natural exponent operators; The baseline theoretical limit threshold constant is obtained beforehand;

[0219] The numerical range of the state assessment index is: The larger the value, the better the evaluation status; conversely, the smaller the value, the worse the evaluation status.

[0220] By dividing the torsional weighted dissipation density by the absolute temperature constant, the integrand of the quotient is obtained. Under the two-dimensional physical closed-loop constraint of the reference cross-sectional area and the normalized pulsation phase, a double definite integral is performed to calculate the fluid-structure interaction dissipation integral index for evaluating the overall cross-sectional energy degradation. Based on the laminar flow model principle, a theoretical limit threshold with the same physical dimensions is constructed using the spatial average reference velocity and the hydrodynamic analytical constant. The actual dissipation integral index is divided by this theoretical limit threshold, and a natural exponential mapping operation is performed to generate an independent state evaluation parameter that removes dimensional interference. Through this state evaluation parameter, a quantitative result that objectively reflects the overall quality of hemodynamics within the system is output, reducing the dependence of the evaluation result on visual reading experience.

[0221] Example 3: A system for implementing the arteriovenous fistula status assessment method, comprising:

[0222] Image acquisition module: used to acquire grayscale images and blood flow matrix of the arteriovenous fistula;

[0223] Boundary extraction module: used to construct a sequence of vessel boundary point coordinates based on gradient extrema;

[0224] Displacement decoupling module: used to match cross-frame boundaries and strip centroid offset to obtain local deformation displacement;

[0225] Torsion calculation module: used to calculate the cross product magnitude of the tangent vector and perform aggregation calculation on the cross product magnitude to generate the spatial average geometric torsion factor;

[0226] Strain energy construction module: Based on the physical strain ratio, it aggregates and generates dynamic strain energy indices;

[0227] Time mapping module: Based on the time integral of strain energy, an inverse time mapping function is constructed by interpolation;

[0228] Flow field reconstruction module: used to resample the flow velocity based on the inverse mapping function and construct the velocity gradient tensor;

[0229] Dissipation weighting module: Combines the deformation rate tensor and the torsion factor to generate a weighted dissipation density;

[0230] Evaluation output module: used to perform double definite integral on the weighted dissipation density and output the evaluation index via exponential mapping.

[0231] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0232] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for assessing the status of arteriovenous fistulas, characterized in that, include: Acquire multi-frame grayscale images and blood flow matrix of the arteriovenous fistula, locate boundary markers based on gradient extrema, and construct a discrete vascular boundary point coordinate sequence; By combining the arc length topological difference and Euclidean distance, cross-frame boundary matching is performed, and the global offset vector is stripped by mean calculation to obtain the local deformation displacement. Introducing a small constant, combining the local discrete tangent vectors with adjacent points, performing dimensionality-upgrading and zero-padding on the tangent vectors to calculate their cross product magnitude, and then performing aggregation processing on the cross product magnitude to extract the geometric torsion factor; The physical strain ratio is calculated based on the displacement difference and initial length, and a quadratic polymerization is performed in combination with the elastic modulus. The strain energy index is generated by superimposing the physiological basis constant. Driven by the time integral of strain energy, discrete sample points are extracted, and the inverse time mapping function is constructed by cubic spline interpolation with phase as the independent variable. The velocity matrix and direction parameters are called, the velocity is resampled using the inverse function, and the projection components are extracted and the velocity gradient tensor is constructed after symbolic truncation angle compensation. The symmetric component of the velocity gradient is extracted, the deformation rate tensor is constructed, the dissipation density is calculated by combining the viscosity coefficient, and the weighted dissipation density is generated by multiplying it by the aforementioned torsional factor. The weighted dissipation density is subjected to a double definite integral over the reference area and the full phase domain to generate a fluid-structure interaction dissipation index. After mapping with the theoretical threshold index, the evaluation index is output.

2. The method for assessing the status of arteriovenous fistulas according to claim 1, characterized in that, The process of acquiring multi-frame grayscale images and blood flow matrices of the arteriovenous fistula, locating boundary markers based on gradient extrema, and constructing a discrete vascular boundary point coordinate sequence specifically includes: An ultrasound probe array is attached to the target area to simultaneously acquire multiple frames of grayscale image matrix and blood flow velocity matrix. Initialize the data acquisition timestamps and establish the physical calculation origin; The region of interest is delineated using the grayscale image at the initial time, and the reference cross-sectional area at the initial time is extracted; Calculate the partial derivatives of the grayscale image matrix in the two-dimensional spatial coordinate system to obtain the rate of change of local pixel grayscale values; Within a preset local search neighborhood, the extreme points of the rate of change are selected as geometric markers at the tissue interface, and the initial boundary point set coordinate sequence of the vascular intima is extracted.

3. The method for assessing the status of arteriovenous fistulas according to claim 1, characterized in that, The process of combining arc-length topological differences and Euclidean distance to perform cross-frame boundary matching, extracting the global offset vector through mean calculation, and obtaining local deformation displacement specifically includes: The difference in arc length topology between adjacent boundary points is obtained as a penalty weight term and added to the Euclidean shortest distance calculation process to perform cross-frame nearest neighbor matching. When a local match is lost, the coordinates of the effective boundary points of the adjacent successfully matched points are extracted as interpolation nodes. Under the smooth constraint of continuous local second derivative, spatial cubic spline interpolation compensation is performed. For the registered coordinate sequence, extract the difference between the coordinate vector of the boundary point at the current time and the initial reference coordinate vector, and calculate the arithmetic mean of the difference over the total number of successfully paired boundary points to generate the global organization displacement vector at the current time. Calculate the difference vector between the coordinate vector of the boundary point at the current moment and the initial reference coordinate vector, and subtract the global organization displacement vector from the difference vector to obtain the local pipe wall deformation displacement vector of the boundary point at the current moment.

4. The method for assessing the status of arteriovenous fistulas according to claim 1, characterized in that, The process involves introducing a small constant, solving for the local discrete tangent vector by combining neighboring points, performing dimensionality-upgrading and zero-padding on the tangent vector to calculate its cross product magnitude, and then performing aggregation processing on the cross product magnitude to extract the geometric torsion factor. Specifically, this includes: Within a preset effective node range, extract the coordinate vectors of adjacent boundary points and perform subtraction to obtain the coordinate difference vector, and calculate the norm of the coordinate difference vector; The norm is compared with a preset small regularization constant, and the maximum value of the two is extracted as the denominator constraint term. Divide the coordinate difference vector by the denominator constraint term to obtain the discrete tangent vector at the boundary point; The discrete tangent vector in the two-dimensional coordinate system is supplemented with zero components along the third spatial dimension to construct a three-dimensional tangent vector. The cross product of adjacent three-dimensional tangent vectors is performed to obtain the normal difference vector, and the scalar magnitude of the normal difference vector is calculated and defined as the local geometric torsion index. The local geometric torsion index within the node interval of the limited computational domain is summed and divided by the total number of valid nodes after removing out-of-bounds nodes to obtain the spatial average geometric torsion factor.

5. The method for assessing the status of arteriovenous fistulas according to claim 1, characterized in that, The process of calculating the physical strain ratio based on the displacement difference and initial length, performing quadratic polymerization in conjunction with the elastic modulus, and superimposing the physiological basis constant to generate a strain energy index specifically includes: Extract the local pipe wall deformation displacement vectors of adjacent nodes and perform subtraction processing to obtain the deformation difference vector, and calculate the norm of the deformation difference vector; Extract the initial boundary coordinate vectors of adjacent nodes and perform subtraction to obtain the initial length difference vector, and calculate the norm of the initial length difference vector; The local physical strain ratio is obtained by dividing the norm of the deformation difference vector by the norm of the initial length difference vector. Calculate the square value of the local physical strain ratio, sum the square values ​​of all local nodes, multiply the sum by the preset equivalent elastic modulus coefficient, and halve the product to generate an intermediate aggregation index. Based on the intermediate aggregation index, a preset physiological basis regularization constant with equivalent pressure dimension is added to generate a discrete aggregation type dynamic strain energy index.

6. The method for assessing the status of arteriovenous fistulas according to claim 1, characterized in that, The process of extracting discrete sample points using strain energy time integration as the driving force, constructing an inverse time mapping function with phase as the independent variable through cubic spline interpolation, specifically includes: For discrete aggregate dynamic strain energy index, integral processing is performed within the virtual time interval from the initial time to the current time to obtain the current cumulative integral value of strain energy; The duration of the complete cardiac cycle is obtained, and the discrete-aggregate dynamic strain energy index is integrated over the duration of the complete cardiac cycle to obtain the integral value of the baseline periodic strain energy. Divide the current cumulative integral value of strain energy by the integral value of strain energy during the reference period to generate the normalized pulsation phase mapping factor corresponding to the current moment. Extract each physical time frame and its corresponding normalized pulsation phase mapping factor, construct discrete sample point pairs, use the normalized pulsation phase mapping factor as the independent variable and the corresponding physical time frame as the dependent variable, perform discrete strictly monotonic cubic spline interpolation, and construct the time mapping inverse function that maps the phase to physical time.

7. The method for assessing the status of arteriovenous fistulas according to claim 1, characterized in that, The process of calling the velocity matrix and direction parameters, resampling the velocity using an inverse function, compensating for the symbolic truncation angle, extracting the projection components, and constructing the velocity gradient tensor specifically includes: Retrieve the pre-stored ultrasound detection beam, the Doppler angle matrix of blood vessel direction, and the spatial unit vector matrix of blood vessel direction; Based on the inverse time mapping function, the elements of the original blood flow velocity matrix are extracted by time domain resampling; Calculate the cosine value of the elements of the Doppler angle matrix, extract the maximum value between the absolute value of the cosine value and the pre-stored safety truncation constant threshold, multiply it by the sign determination value corresponding to the cosine value, and construct the symbolic truncation compensation denominator. Divide the elements of the blood flow velocity matrix extracted by the resampling by the symbolic truncated compensation denominator to obtain the amplitude-compensated flow velocity; Extract the directional components of the spatial unit vector matrix of the blood vessel orientation in each two-dimensional spatial coordinate system base dimension, and multiply the amplitude compensation flow velocity by each of the directional components to generate the projected velocity field vector components. For the projected velocity field vector components, partial derivatives are calculated along each basis variable dimension of the two-dimensional spatial coordinate system, and the results of the partial derivative calculations are used to construct a spatial velocity gradient tensor.

8. The method for assessing the status of arteriovenous fistulas according to claim 1, characterized in that, The extraction of the symmetric component of the velocity gradient, the construction of the deformation rate tensor, the calculation of the dissipation density using the viscosity coefficient, and the multiplication by the aforementioned torsional factor to generate a weighted dissipation density, specifically includes: Extract the corresponding elements symmetrical about the main diagonal from the spatial velocity gradient tensor, add the symmetrical elements and halve them to construct the deformation rate tensor elements; The squared values ​​of the deformation rate tensor elements in each dimension are calculated. The squared values ​​of each dimension in the two-dimensional spatial coordinate system are double-summed and accumulated. The accumulated result is multiplied by the pre-acquired blood dynamic viscosity constant and multiplication factor to obtain the local viscous dissipation density index. Obtain a preset semi-empirical topology magnification factor, multiply the spatial average geometric torsion factor by the semi-empirical topology magnification factor and add 1 to construct a nonlinear topology magnification weight; The local viscous dissipation density index is multiplied by the nonlinear topological amplification weight to generate a torsional weighted dissipation density index.

9. The method for assessing the status of arteriovenous fistulas according to claim 1, characterized in that, The process involves performing a double definite integral on the weighted dissipation density over the reference area and the full phase domain to generate a fluid-structure interaction dissipation index. This index is then mapped using a theoretical threshold index to output an evaluation index, including: Divide the torsional weighted dissipation density index by the pre-acquired absolute blood temperature constant to obtain the quotient integrand function; Using the extracted blood vessel reference cross-sectional area as the spatial two-dimensional integration domain and the normalized pulsation phase as the phase integration variable, a double definite integral is calculated on the integrand function of the quotient value to generate a two-dimensional fluid-structure interaction phase dissipation integral index. Obtain the average reference velocity constant of the inflow channel space, and calculate the square value of the average reference velocity constant of the inflow channel space; Obtain a preset hydrodynamic analytical constant, multiply the preset pi constant, blood dynamic viscosity constant, the hydrodynamic analytical constant and the square value in sequence, and divide the product by the blood absolute temperature constant to generate the reference theoretical limit threshold constant; Divide the two-dimensional fluid-structure interaction phase dissipation integral index by the reference theoretical limit threshold constant, and take the negative value of the division result to generate the mapped variable. Perform natural index operations on the mapped variables to generate a dimensionless state assessment index.

10. A system for implementing the arteriovenous fistula status assessment method of claim 1, characterized in that, include: Image acquisition module: used to acquire grayscale images and blood flow matrix of the arteriovenous fistula; Boundary extraction module: used to construct a sequence of vessel boundary point coordinates based on gradient extrema; Displacement decoupling module: used to match cross-frame boundaries and strip centroid offset to obtain local deformation displacement; Torsion calculation module: used to calculate the cross product magnitude of the tangent vector and perform aggregation calculation on the cross product magnitude to generate the spatial average geometric torsion factor; Strain energy construction module: Based on the physical strain ratio, it aggregates and generates dynamic strain energy indices; Time mapping module: Based on the time integral of strain energy, an inverse time mapping function is constructed by interpolation; Flow field reconstruction module: used to resample the flow velocity based on the inverse mapping function and construct the velocity gradient tensor; Dissipation weighting module: Combines the deformation rate tensor and the torsion factor to generate a weighted dissipation density; Evaluation output module: used to perform double definite integral on the weighted dissipation density and output the evaluation index via exponential mapping.

Images