Spatial registration and interpolation calculation method and system for cardiac electrophysiological parameters

Abstract

The invention discloses a spatial registration and interpolation calculation method and system for cardiac electrophysiological parameters. The method comprises the following steps: segmenting heart magnetic resonance image data by using a three-dimensional DRLSE algorithm, and establishing a heart three-dimensional anatomical model; aligning the electrophysiological parameter data to a grid structure of the heart three-dimensional anatomical model, and performing anatomical model registration to obtain a registered electrophysiological parameter point set; on the registered electrophysiological parameter point set, anisotropic kernel function interpolation is designed, and multi-temporal electrophysiological data is formed; and introducing space-time coupled dynamic interpolation to the multi-temporal electrophysiological data to obtain a continuous three-dimensional space-time electrophysiological parameter field which is reconstructed in real time. The system comprises a three-dimensional anatomical model establishment module, a registration module, a high-precision static electrophysiological field generation module and a space-time coupling module. According to the method, high-precision non-rigid registration of the electrophysiological data and the grid model is realized through anatomical mark points.

Description

Technical Field

[0001] The present invention relates to the field of cardiac electrophysiological parameter data processing, and particularly to a method and system for spatial registration and interpolation calculation of cardiac electrophysiological parameters. Background Art

[0002] Accurate analysis of cardiac electrophysiological parameters is the core basis for studying the mechanism of arrhythmia and guiding interventional treatment. With the rapid development of medical imaging technologies (such as cardiac magnetic resonance CMR, electrophysiological mapping systems), the clinical demand for high-resolution and anatomically consistent cardiac electrophysiological parameter fields is increasing.

[0003] Currently, the mainstream method realizes parameter field reconstruction through the following process: (1) cardiac anatomical modeling, extracting endocardial, epicardial, and myocardial structures based on image segmentation technology; (2) electrophysiological data registration, aligning discrete mapping points to the anatomical model; (3) parameter field interpolation, generating static or dynamic electrophysiological parameter distributions. Existing technologies still have significant bottlenecks in multi-modal data fusion, spatio-temporal consistency modeling, etc. The traditional level set method is vulnerable to noise interference in the segmentation of complex anatomical structures (such as papillary muscles, trabeculae), resulting in calculation errors in the myocardial thickness field; registration algorithms based on rigid transformation or local affine cannot adapt to the non-uniform deformation of the heart, resulting in spatial mismatch between electrophysiological points and anatomical structures; conventional interpolation methods (such as inverse distance weighting, radial basis function) ignore anatomical characteristics such as myocardial thickness and curvature, introducing non-physiological mutations during interpolation across regions (such as from endocardium to epicardium); dynamic interpolation mostly adopts phase-independent processing, lacking a spatio-temporal coupling mechanism, resulting in discontinuous parameter evolution and being difficult to reflect the true propagation characteristics of cardiac electrical activity.

[0004] With the popularization of precise cardiac interventional treatment (such as radiofrequency ablation), the market has put forward higher requirements for the following capabilities: high-precision three-dimensional anatomical models, seamless fusion of multi-modal data, and real-time dynamic field reconstruction. However, in existing segmentation methods, traditional Snake models or geometric active contour algorithms cannot stably process gray-scale inhomogeneous regions in CMR images, resulting in blurred or broken myocardial boundaries (such as the free wall of the right ventricle); registration based on feature points (such as the ICP algorithm) relies on manual marking, with low efficiency and being vulnerable to local distortions; isotropic kernel functions (such as Gaussian kernels) are over-smoothed in regions with curvature mutations, losing key electrophysiological features (such as conduction block areas); phase-independent interpolation results in jumps in the parameter field at adjacent time points, unable to depict the continuous process of electrical signal propagation. Summary of the Invention

[0005] In view of this, the purpose of the embodiments of the present invention is to provide a method and system for spatial registration and interpolation calculation of cardiac electrophysiological parameters, which can achieve high-precision non-rigid registration of electrophysiological data and a grid model through anatomical landmark points, and through an anisotropic interpolation mechanism, achieve compliance with anatomical structure characteristics, and accurately reflect the electrical signal propagation process by constructing a continuous four-dimensional parameter field.

[0006] The embodiments of the present invention are implemented as follows:

[0007] A method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, comprising:

[0008] Using a three-dimensional DRLSE algorithm to segment cardiac magnetic resonance imaging data, extract the endocardium, epicardium, and myocardial regions, and establish a three-dimensional cardiac anatomical model.

[0009] Align the electrophysiological parameter data to the grid structure of the three-dimensional cardiac anatomical model, perform anatomical model registration, and obtain a registered electrophysiological parameter point set.

[0010] On the registered electrophysiological parameter point set, design an anisotropic kernel function interpolation to obtain a high-precision static electrophysiological field under anatomical constraints, and form multi-temporal electrophysiological data.

[0011] For the multi-temporal electrophysiological data, introduce spatio-temporal coupled dynamic interpolation to obtain a continuously reconstructed three-dimensional spatio-temporal electrophysiological parameter field in real time.

[0012] In a preferred embodiment of the present invention, in the above method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, the step of using a three-dimensional DRLSE algorithm to segment cardiac magnetic resonance imaging data, extract the endocardium, epicardium, and myocardial regions, and establish a three-dimensional cardiac anatomical model includes:

[0013] Collect cardiac magnetic resonance imaging to obtain three-dimensional cardiac magnetic resonance volume data I(x, y, z).

[0014] Perform gray normalization on the original three-dimensional cardiac magnetic resonance volume data I(x, y, z) to eliminate brightness differences and obtain normalized image data

[0015] Construct an initial level set function where Ω is the spatial range of the three-dimensional cardiac magnetic resonance volume data to be segmented, Ω0 is the initial seed region, is the boundary of the initial seed region, and ρ is a constant.

[0016] For the normalized image data Perform three-dimensional edge detection factor calculation to obtain an edge indicator function based on image gradient where Gσ is a three - dimensional Gaussian kernel with standard deviation σ, is a three - dimensional gradient operator, and * is a three - dimensional convolution operation.

[0017] Define the three - dimensional DRLSE energy functional \(E(\varphi)=\mu R(\varphi)+\lambda L(\varphi)+\alpha A(\varphi)\), where \(R(\varphi)\) is the regularization term, \(\mu\) is the regularization term weight coefficient, \(L(\varphi)\) is the edge term, \(\lambda\) is the edge term weight coefficient, \(A(\varphi)\) is the area term, and \(\alpha\) is the area term weight coefficient.

[0018] Take the variation of the energy functional \(E(\varphi)\) to obtain the gradient descent flow equation where, is the rate of change of the level - set function with respect to time, is the Laplacian operator of the level - set function, div is the divergence operator, is the normalized direction of the level - set gradient, \(\delta(\varphi)\) is the Dirac function, and \(g\) is the edge - indicating value of the image gradient.

[0019] Use the finite - difference method to discretize and iterate the initial level - set function \(\varphi_0(x,y,z)\) to obtain the iterative update formula where the termination condition is \(\|\varphi\) n+1 -\(\varphi\) n \|^2\lt\epsilon\), \(n\) is the number of iterations, \(\epsilon\) is a preset threshold, \(t\) is the time step, the converged level - set function is \(\varphi(x,y,z)\), and \(\varphi\) is a three - dimensional scalar function.

[0020] Extract the isosurfaces from the converged level - set function \(\varphi(x,y,z)\) to obtain the endocardial boundary \(\varphi(x,y,z) = 0\) and the epicardial boundary \(\varphi(x,y,z)=k\). Use the Marching Cubes algorithm to generate the triangular - mesh surface of the three - dimensional cardiac anatomical model \(M = \{V,E,F\}\), where \(V=\{v\) i |v i =(x i ,y i ,z i )}\) is the mesh vertex, \(v\) i is each vertex, \(i = 1,2,\cdots,N\), \(E\) is the mesh edge, and \(F\) is the mesh face.

[0021] For each vertex of the triangular mesh, calculate the myocardial thickness field in the three - dimensional cardiac anatomical model through the Euclidean distance where, is the mesh vertex coordinate of the endocardium, is the mesh vertex coordinate of the epicardium.

[0022] Its technical effects are as follows: The DRLSE algorithm constructs an edge indicator function based on image gradients. It can effectively avoid irregular deformation or numerical instability of the level set function during the evolution process while maintaining the complex boundaries of the heart structure. It has strong boundary sensitivity and high convergence, and is applicable to scenarios such as thin myocardial walls, blurred gray-scale transitions, and low signal-to-noise ratios in cardiac MRI data, and can accurately extract the endocardial and epicardial boundaries. By introducing a distance regularization term, the distance function form of the level set function is guaranteed, greatly reducing the computational amount and improving the algorithm stability and operation speed. Compared with traditional two-dimensional slice-level segmentation methods, three-dimensional DRLSE can perform unified evolution in the entire volume data space, making the segmentation results maintain structural continuity and consistency in three dimensions, so as to obtain a three-dimensional model of the heart with coherent anatomical structure and complete spatial topology, which is conducive to subsequent spatial analysis tasks such as myocardial thickness calculation, electrophysiological data mapping and interpolation.

[0023] In a preferred embodiment of the present invention, in the above method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, in the three-dimensional DRLSE energy functional E(φ),

[0024] The regularization term R(φ) is used to constrain the level set gradient modulus to be close to 1 to ensure numerical stability.

[0025] The edge term L(φ) is used to drive the curve to evolve towards the target boundary. where δ(·) is the Dirac function.

[0026] The area term A(φ) is used to control the evolution direction to contract inwards or outwards, A(φ) = ∫ Ω gH(-φ)dxdydz, where H(·) is the Heaviside step function.

[0027] Its technical effects are as follows: By introducing the regularization term, the stability and controllability of the evolution process are improved, ensuring that the segmentation results are coherent and unbroken in space; by introducing the edge term, the curve is driven to evolve towards the image edge to achieve accurate segmentation; by introducing the area term, the curve evolution direction is controlled to enhance the flexibility of target region extraction.

[0028] In a preferred embodiment of the present invention, in the above method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, after obtaining the three-dimensional anatomical model of the heart, interlayer offset correction is further performed, including:

[0029] Calculating the centroid coordinates of each layer section of the three-dimensional anatomical model M of the heart where N z is the number of vertices in the z-th layer.

[0030] The three-dimensional line equation l(z)=cz + f is obtained by least-squares fitting, where c is the slope and f is a constant.

[0031] For the offset layer, the vertex coordinates are translated, and the translation value Δv of the vertex coordinates = c z′ -l(z′) to make the centroid regress to the fitted three-dimensional line.

[0032] Its technical effect is that by fitting the centroid trajectory of each layer into a straight line and making the centroid of each layer regress to this trajectory, the irregular interlayer misalignment phenomenon is effectively eliminated. The generated three-dimensional model structure is more unified, smooth and continuous in space, providing a stable geometric structure basis for subsequent registration and parameter calculation. After correcting the interlayer offset, the spatial position of each layer is consistent with its anatomical meaning, thereby significantly improving the accuracy and stability of spatial interpolation and ensuring the spatial consistency and anatomical correspondence between the electrophysiological parameters and the real heart structure.

[0033] In a preferred embodiment of the present invention, in the above method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, the alignment of electrophysiological parameter data to the grid structure of the cardiac three-dimensional anatomical model for anatomical model registration to obtain the registered electrophysiological parameter point set includes:

[0034] Select anatomical landmark points P = {p k ∣p k ∈V} on the triangular mesh surface of the cardiac three-dimensional anatomical model M.

[0035] Collect electrophysiological parameter data to generate an electrophysiological point set Q = {q j |q j =(u j ,U j )}, where u j =(x j ,y j ,z j ) is the three-dimensional coordinate of the electrophysiological measurement point, and U j is the electrophysiological parameter value of the corresponding point. Match the point set S = {s k |s k ∈Q} corresponding to the anatomical landmark points P in the electrophysiological point set Q to make the anatomical position of the anatomical landmark points consistent with the electrophysiological parameter data.

[0036] Construct a thin plate spline transformation model where is the translation vector, is the affine transformation matrix, is the non-rigid deformation weight, φ(·)=· 2 log· is the radial basis function, and u is the original three-dimensional coordinate of any point in the electrophysiological point set Q.

[0037] Preliminarily align the electrophysiological point set Q with the parameter values in the thin plate spline transformation model F(u) to obtain u'. j = a + Bu j (j = 1, 2,..., N), superimpose the radial basis function correction term to obtain the registration coordinates

[0038] Project the registration coordinates u' j ' onto the mesh surface of the three-dimensional cardiac anatomical model M to obtain the registered electrophysiological parameter point set V'. j = {v' j |v' j ∈ V ∪ F}.

[0039] Its technical effect is as follows: Non-rigid spatial alignment is performed through the thin plate spline transformation model, while ensuring one-to-one correspondence between anatomical landmark points, accurately aligning electrophysiological parameters with the cardiac structure, realizing structure-function coupling, and being able to truly reflect myocardial electrical activity. The TPS transformation can handle rigid pose changes through the affine term and local deformations through the radial basis function term, support individualized modeling and time series dynamic registration, and can be used in scenarios such as pathological comparison and pre-operative and post-operative change analysis. Through mesh surface projection, the transformed electrophysiological points are made to fit the three-dimensional cardiac surface, avoiding abnormalities such as points falling outside the heart or penetrating the heart wall; enhancing the geometric rationality and visualization effect of the registered point set, which is conducive to subsequent visual analysis and clinical intervention of electrophysiological data.

[0040] In a preferred embodiment of the present invention, in the above method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, after constructing the thin plate spline transformation model F(u), it further includes setting the constraint conditions of the thin plate spline transformation model and solving to obtain the parameter values in the thin plate spline transformation model F(u).

[0041] The constraint conditions include minimizing the bending energy Satisfying the interpolation condition F(s k ) = p k (k = 1, 2,..., K) and additional constraints Solving the spatial transformation Such that F(s k ) = p k .

[0042] Its technical effects are as follows: minimizing bending energy can significantly suppress severe deformation caused by a small amount of corresponding point errors or uneven distribution, generating a continuous, smooth, and natural transformation surface, making the spatial registration more conform to the true physical characteristics of cardiac tissue deformation. The interpolation condition ensures that each source point after transformation is accurately matched to the target point, improving the local geometric alignment accuracy and ensuring the one-to-one correspondence between key anatomical sites and electrophysiological parameters. The additional constraint restricts that the non-rigid part does not contain translational or rotational components, separates the rigid and non-rigid transformation components, reduces the coupling of redundant parameters, and improves the solution stability and the interpretability of the results.

[0043] In a preferred embodiment of the present invention, in the method for spatial registration and interpolation calculation of the above cardiac electrophysiological parameters, on the electrophysiological parameter point set after registration, an anisotropic kernel function interpolation is designed to obtain a high-precision static electrophysiological field under anatomical constraints, and the formation of multi-temporal electrophysiological data includes:

[0044] Define the anisotropic kernel function Where is the spatial distance term, is the thickness difference term, v is the coordinate of the point to be interpolated, and σ d is the spatial distance bandwidth, and σ t is the myocardial thickness difference weight.

[0045] For each vertex v on the mesh surface i ∈V, calculate the interpolation parameter

[0046] Integrate the static fields at M time points for multi-temporal data, unify the time axis through electrocardiogram gating markers, divide the time phases into groups according to functional stages, and obtain a high-precision static electrophysiological field under anatomical constraints v i ∈V, m = 1, 2,..., M, where is the static electrophysiological parameter field at the m-th time point, and multi-temporal electrophysiological data is formed.

[0047] Its technical effects are as follows: the anisotropic kernel function introduces the "thickness difference term" of the myocardium as a weighting factor, making the interpolation consider not only the spatial distance (geometric proximity) but also the myocardial structure characteristics (thickness heterogeneity). The interpolation result is more in line with the anatomical structure changes of the myocardium, avoiding the unrealistic cross-layer propagation caused by only geometric distance, realizing anisotropic propagation modeling, and conforming to the myocardial electrical activity path. Compared with the traditional isotropic Gaussian kernel, the anisotropic kernel function has better adaptability on the irregular cardiac surface mesh. The interpolation calculation for each vertex considers its local thickness characteristics, avoids the interference of non-structural points, adapts to the complex three-dimensional mesh structure, improves the interpolation stability, and can accurately restore electrophysiological information on the three-dimensional reconstructed myocardial surface.

[0048] In a preferred embodiment of the present invention, in the method for spatial registration and interpolation calculation of the above cardiac electrophysiological parameters, the introduction of spatio-temporal coupled dynamic interpolation for the multi-phase electrophysiological data to obtain a continuously reconstructed real-time three-dimensional spatio-temporal electrophysiological parameter field includes:

[0049] Define a time label t for the multi-phase electrophysiological data m ∈[0,T], where t m is the absolute time at the m-th time point, and T is the total duration of the cardiac cycle. Map the time label t m to the normalized cycle ratio

[0050] Define a spatio-temporal coupling kernel function K(v, v′ j ; t, t m ) = K(v, v′ j )·K time (t, t m ), where K(v, v′ j ) is an anisotropic spatial kernel function, and K time (t, t m ) is a time kernel function.

[0051] The time kernel function σ τ is the time bandwidth.

[0052] For any time point (v, t), calculate the electrophysiological parameter value through dynamic interpolation For each t, only load the data within the time window [t - 2στ, t + 2στ], where represents the weighted parameter sum of adjacent spatio-temporal points in all phases, and represents the normalization factor.

[0053] Perform anisotropic spatio-temporal smoothing on the electrophysiological parameter value to obtain a continuously reconstructed real-time three-dimensional spatio-temporal electrophysiological parameter field.

[0054] Its technical effects are as follows: By defining time tags and mapping them to standardized periodic ratios, data at different time phases are incorporated into a unified time framework, breaking the time discretization limit of static data and enabling the estimation of electrophysiological values at any moment. According to spatial anatomical features and temporal evolution laws, a continuous and fine-grained three-dimensional + time electrophysiological distribution field is constructed, which is conducive to truly restoring the process of electrical signal conduction. By loading data at nearby moments through a local time window and using a dynamic kernel interpolation method to quickly calculate the electrophysiological state at any moment, a continuous three-dimensional cardiac electrophysiological field with real-time reconstruction is achieved. Spatiotemporal coupling interpolation improves the temporal consistency and spatial coherence of data, enabling potential abnormalities such as premature beats and ectopic pacing at a certain time point to be continuously mapped spatially, and the activity curve of a certain spatial region on the time axis to be stably estimated, improving the spatial localization accuracy and temporal judgment ability of electrophysiological abnormalities, and providing a reliable basis for arrhythmia diagnosis, electroablation path planning, etc.

[0055] In a preferred embodiment of the present invention, in the above method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, the anisotropic spatio-temporal smoothing process for the electrophysiological parameter values includes:

[0056] Among them, is the spatial smoothing term, λ s is the median of the spatial gradient, is the temporal smoothing direction, λ t is the median of the temporal gradient.

[0057] A system for spatial registration and interpolation calculation of cardiac electrophysiological parameters, which includes:

[0058] A three-dimensional anatomical model establishment module, which is used to segment cardiac magnetic resonance imaging data using the three-dimensional DRLSE algorithm, extract the endocardium, epicardium, and myocardial regions, and establish a three-dimensional cardiac anatomical model.

[0059] A registration module, which is used to align electrophysiological parameter data to the grid structure of the three-dimensional cardiac anatomical model for anatomical model registration to obtain a registered set of electrophysiological parameter points.

[0060] A high-precision static electrophysiological field generation module, which is used to design an anisotropic kernel function interpolation on the registered set of electrophysiological parameter points to obtain a high-precision static electrophysiological field under anatomical constraints and form multi-phase electrophysiological data.

[0061] A spatiotemporal coupling module, which is used to introduce spatiotemporal coupling dynamic interpolation for the multi-phase electrophysiological data to obtain a continuously reconstructed three-dimensional spatiotemporal electrophysiological parameter field in real time.

[0062] The beneficial effects of the embodiments of the present invention are:

[0063] The present invention adopts a cardiac MRI segmentation method based on three-dimensional DRLSE (Distance Regularized Level Set Evolution). Through the design of the regularization term, edge term, and area term in the energy functional, the level set contour is guided to accurately converge to the endocardial, epicardial, and myocardial boundaries. The DRLSE method has a built-in distance regularization term, eliminating the need for repeated re-initialization of the traditional level set, effectively improving the numerical stability of the evolution process. After segmentation, isosurface reconstruction of the results can generate a closed and continuous cardiac surface mesh model, ensuring the integrity of the model structure without omission. By obtaining the complete structures of the heart chambers and heart walls through three-dimensional segmentation and combining with the energy functional design to make the boundaries closely adhere to the real anatomical interfaces, the generated three-dimensional cardiac model is outstanding in terms of structural integrity, boundary clarity, and numerical stability.

[0064] The present invention combines anatomical landmark guidance with Thin-Plate Spline (TPS) non-rigid transformation to achieve high-precision mapping of multi-source electrophysiological data points to a three-dimensional cardiac mesh. Key anatomical landmarks such as valve annulus, apex, catheter inlet, etc. are selected to establish initial corresponding relationships, and the EAM is mapped to MRI / CT. Usually, it starts from manually annotating anatomical landmark pairs, and then the TPS is used to perform smooth interpolation deformation on the entire point set. As an interpolation function based on the bending energy of a curved plate, the TPS can maintain the smoothness of the deformation as much as possible while passing through all the landmarks. The present invention also sets interpolation constraints for local structural features (such as neighborhood relationships, myocardial thickness, etc.), and further improves the accuracy and consistency of registration by optimizing the constraint conditions.

[0065] For the static electrophysiological data at each moment, the present invention designs an interpolation model based on an anisotropic kernel function, incorporating both spatial distance and myocardial thickness differences into the interpolation weights. Traditional interpolation usually only considers spatial distance, but the electrical conductivity characteristics of cardiac tissue are not isotropic. By taking the myocardial wall thickness as an additional factor of the kernel function, this method can reflect the influence of thickness differences on the electric field: existing studies have shown that thicker myocardial tissue will generate a larger potential amplitude. On this basis, the kernel function assigns different weights to points that are spatially close but have different thicknesses, improving the accuracy of interpolation. The present invention can construct a high-precision static electrophysiological field at multiple time phases, thus more realistically depicting the electrical behavior of each part of the heart.

[0066] The present invention introduces standardized time tags and time-domain kernel functions, couples the time dimension with spatial domain interpolation, and constructs a continuous spatio-temporal interpolation framework. By assigning time coordinates to spatial electrical signal points and adding time-distance weights to the kernel function, data between different time points support each other for interpolation. In this way, the electric field values at adjacent moments can transition smoothly, forming a continuous spatio-temporal field. Finally, the spatio-temporal coupling interpolation function can achieve three-dimensional electric field reconstruction varying with time in the entire heart model, demonstrating the dynamic evolution process of cardiac electrical activity. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0068] Figure 1 It is a flowchart of the method for spatial registration and interpolation calculation of cardiac electrophysiological parameters of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0070] Please refer to Figure 1 The first embodiment of the present invention provides a method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, which includes: using a three-dimensional DRLSE algorithm to segment cardiac magnetic resonance imaging data, extract the endocardium, epicardium, and myocardial regions, and establish a three-dimensional cardiac anatomical model; aligning electrophysiological parameter data to the grid structure of the three-dimensional cardiac anatomical model for anatomical model registration to obtain a registered set of electrophysiological parameter points; designing anisotropic kernel function interpolation on the registered set of electrophysiological parameter points to obtain a high-precision static electrophysiological field under anatomical constraints, forming multi-temporal electrophysiological data; introducing spatio-temporal coupling dynamic interpolation for the multi-temporal electrophysiological data to obtain a continuously reconstructed three-dimensional spatio-temporal electrophysiological parameter field in real time.

[0071] Among them, using the three-dimensional DRLSE algorithm to segment cardiac magnetic resonance imaging data, extract the endocardium, epicardium and myocardial regions, and establish a three-dimensional cardiac anatomical model includes: acquiring cardiac magnetic resonance imaging to obtain three-dimensional cardiac magnetic resonance volume data I(x, y, z); performing gray normalization on the original three-dimensional cardiac magnetic resonance volume data I(x, y, z) to eliminate brightness differences and obtain normalized image data Construct an initial level set function Among them, Ω is the spatial range of the three-dimensional cardiac magnetic resonance volume data to be segmented, and Ω0 is the initial seed region, that is, the initial rough region of the target to be segmented preset by the user or algorithm, selected manually or automatically, is the boundary of the initial seed region, ρ is a constant, and ρ = 2 is taken; for the normalized image data Perform three-dimensional edge detection factor calculation to obtain an edge indicator function based on image gradient Among them, G σ is a three-dimensional Gaussian kernel with a standard deviation of σ, is a three-dimensional gradient operator, and * is a three-dimensional convolution operation; define the three-dimensional DRLSE energy functional E(φ) = μR(φ) + λL(φ) + αA(φ), where R(φ) is the regularization term, μ is the regularization term weight coefficient, L(φ) is the edge term, λ is the edge term weight coefficient, A(φ) is the area term, and α is the area term weight coefficient; take the variation of the energy functional E(φ) to obtain the gradient descent flow equation Among them, is the change rate of the level set function with time, is the Laplacian operator of the level set function, div is the divergence operator, is the normalized direction of the level set gradient, δ(φ) is the Dirac function, and g is the edge indication value of the image gradient; use the finite difference method to discretize and iterate the initial level set function φ0(x, y, z) to obtain an iterative update formula Among them, the termination condition is ||φ n+1 -φ n ||2 < ò, n is the number of iterations, ò is a preset threshold, t is the time step, the converged level set function is φ(x, y, z), and φ is a three-dimensional scalar function; perform isosurface extraction on the converged level set function φ(x, y, z) to obtain the endocardial boundary φ(x, y, z) = 0 and the epicardial boundary φ(x, y, z) = k, where k > 0, preset according to the myocardial thickness, and use the Marching Cubes algorithm to generate the triangular mesh surface of the three-dimensional cardiac anatomical model M = {V, E, F}, where V = {v i ∣v i =(x i ,y i ,zi )} is the grid vertex, v i For each vertex, i = 1, 2, ..., N, E is the edge of the grid, and F is the patch of the grid; for each vertex of the triangular grid, the myocardial thickness field in the three-dimensional anatomical model of the heart is calculated by the Euclidean distance Wherein, is the grid vertex coordinate of the endocardium, is the grid vertex coordinate of the epicardium.

[0072] Its technical effect is that: the DRLSE algorithm constructs an edge indicator function based on the image gradient, which can effectively avoid the irregular deformation or numerical instability of the level set function during the evolution process while maintaining the complex boundaries of the heart structure. It has strong boundary sensitivity and high convergence, and is suitable for scenarios such as thin myocardial walls, blurred gray-scale transitions, and low signal-to-noise ratios in cardiac MRI data, and can accurately extract the boundaries of the endocardium and epicardium. By introducing a distance regularization term, the distance function form of the level set function is guaranteed, greatly reducing the computational amount and improving the algorithm stability and operation speed. Compared with the traditional two-dimensional slice-level segmentation method, the three-dimensional DRLSE can be uniformly evolved in the entire volume data space, making the segmentation results maintain structural continuity and consistency in three dimensions, so as to obtain a three-dimensional model of the heart with coherent anatomical structure and complete spatial topology, which is beneficial to subsequent spatial analysis tasks such as myocardial thickness calculation, electrophysiological data mapping and interpolation.

[0073] Wherein, in the three-dimensional DRLSE energy functional E(φ), the regularization term R(φ) is used to constrain the level set gradient norm to be close to 1 to ensure numerical stability, the edge term L(φ) is used to drive the curve to evolve towards the target boundary, Wherein, δ(·) is the Dirac function; the area term A(φ) is used to control the evolution direction to contract inwards or outwards, A(φ) = ∫ Ω gH(-φ)dxdydz, wherein, H(·) is the Heaviside step function.

[0074] Its technical effect is that: by introducing a regularization term, the stability and controllability of the evolution process are improved, ensuring that the segmentation results are coherent and unbroken in space; by introducing an edge term, the curve is driven to evolve towards the image edge to achieve accurate segmentation; by introducing an area term, the evolution direction of the curve is controlled, enhancing the flexibility of target region extraction.

[0075] Wherein, after obtaining the three-dimensional anatomical model of the heart, interlayer offset correction is also performed, including: calculating the centroid coordinates of each layer section for the three-dimensional anatomical model M of the heart Wherein, N zis the number of vertices in the z-th layer; the three-dimensional line equation l(z)=cz + f is obtained by least squares fitting, where c is the slope and f is a constant; for the offset layer, the vertex coordinates are translated, and the translation value of the vertex coordinates Δv = c z′ -l(z′) to make the centroid return to the fitted three-dimensional line.

[0076] Its technical effect is as follows: by fitting the centroid trajectory of each layer into a straight line and making the centroid of each layer return to this trajectory, the irregular interlayer misalignment phenomenon is effectively eliminated, and the generated three-dimensional model structure is more unified, smooth and continuous in space, providing a stable basis for structure geometry for subsequent registration and parameter calculation. After correcting the interlayer offset, the spatial position of each layer is consistent with its anatomical meaning, thereby significantly improving the accuracy and stability of spatial interpolation and ensuring the consistency and anatomical correspondence of electrophysiological parameters with the real heart structure in space.

[0077] Among them, aligning the electrophysiological parameter data to the grid structure of the three-dimensional anatomical model of the heart and performing anatomical model registration to obtain the registered electrophysiological parameter point set includes: selecting anatomical landmark points P = {p k ∣p k ∈V} on the triangular mesh surface of the three-dimensional anatomical model M of the heart, and the apex point of the left ventricle, the midpoint of the interventricular septum or the attachment point of the base of the papillary muscle can be selected; collecting electrophysiological parameter data to generate an electrophysiological point set Q = {q j |q j =(u j ,U j )}, where u j =(x j ,y j ,z j ) is the three-dimensional coordinate of the electrophysiological measurement point, and U j is the electrophysiological parameter value corresponding to the point, including potential and conduction velocity, etc. Matching the point set S = {s k |s k ∈Q} corresponding to the anatomical landmark points P in the electrophysiological point set Q to make the anatomical position of the anatomical landmark points consistent with the electrophysiological parameter data; constructing a thin plate spline transformation model Among them, is the translation vector, is the affine transformation matrix, is the non-rigid deformation weight, φ(·)=· 2 log· is the radial basis function, and u is the original three-dimensional coordinate of any point in the electrophysiological point set Q; initially aligning the electrophysiological point set Q with the parameter values in the thin plate spline transformation model F(u) to u′ j =a + Bu j(j = 1, 2, ..., N), add a radial basis function correction term to obtain the registered coordinates Project the registered coordinate u″ j onto the mesh surface of the three-dimensional cardiac anatomical model M to obtain the registered electrophysiological parameter point set V j ′ = {v′ j |v′ j ∈V ∪ F}.

[0078] Its technical effect is as follows: Through the thin plate spline transformation model for non-rigid spatial alignment, while ensuring one-to-one correspondence between anatomical landmark points, accurately aligning electrophysiological parameters with the cardiac structure, realizing structure-function coupling, and being able to truly reflect myocardial electrical activity. The TPS transformation can handle rigid pose changes through the affine term and local deformations through the radial basis function term, support individualized modeling and time series dynamic registration, and can be used in scenarios such as pathological comparison and preoperative and postoperative change analysis. Through mesh surface projection, the transformed electrophysiological points are fitted to the three-dimensional cardiac surface, avoiding abnormalities such as points falling outside the heart or penetrating the heart wall; enhancing the geometric rationality and visualization effect of the registered point set, which is beneficial for subsequent visual analysis and clinical intervention of electrophysiological data.

[0079] Among them, after constructing the thin plate spline transformation model F(u), it also includes setting the constraint conditions of the thin plate spline transformation model and solving to obtain the parameter values in the thin plate spline transformation model F(u); the constraint conditions include minimizing the bending energy satisfying the interpolation condition F(s k ) = p k (k = 1, 2, ..., K) and additional constraints Solving the spatial transformation such that F(s k ) = p k .

[0080] Its technical effect is as follows: Minimizing the bending energy can significantly suppress the drastic deformation caused by a small number of corresponding point errors or uneven distributions, generating a continuous, smooth, and natural transformation surface, making the spatial registration more in line with the true physical characteristics of cardiac tissue deformation. The interpolation condition ensures that each source point after transformation is accurately matched to the target point, improving the local geometric alignment accuracy and ensuring one-to-one correspondence between key anatomical sites and electrophysiological parameters. The additional constraints ensure that the non-rigid part does not contain translation or rotation components, separating the rigid and non-rigid transformation components, reducing redundant parameter coupling, and improving the solution stability and interpretability of the results.

[0081] Among them, on the registered electrophysiological parameter point set, designing an anisotropic kernel function interpolation to obtain a high-precision static electrophysiological field under anatomical constraints and forming multi-temporal electrophysiological data includes: defining an anisotropic kernel function Among them, is the spatial distance term, is the thickness difference term, v is the coordinates of the point to be interpolated, and σ d is the spatial distance bandwidth, and σ t is the myocardial thickness difference weight; for each vertex v on the mesh surface i ∈V, calculate the interpolation parameter Integrate the static fields at M time points for multi-temporal data, unify the time axis through electrocardiogram gating markers, divide the time phases into groups according to functional phases, and obtain a high-precision static electrophysiological field under anatomical constraints v i ∈V, m = 1, 2,..., M, where is the static electrophysiological parameter field at the m-th time point, forming multi-temporal electrophysiological data.

[0082] Its technical effect is that: the anisotropic kernel function introduces the "myocardial thickness difference term" as a weighting factor, making the interpolation consider not only the spatial distance (geometric proximity) but also the myocardial structural characteristics (thickness heterogeneity). The interpolation result is more in line with the anatomical structure changes of the myocardium, avoiding the unrealistic cross-layer propagation caused by relying solely on geometric distance, realizing anisotropic propagation modeling, and conforming to the myocardial electrical activity path. Compared with the traditional isotropic Gaussian kernel, the anisotropic kernel function has better adaptability on the irregular cardiac surface mesh. The interpolation calculation for each vertex considers its local thickness characteristics, avoiding the interference of non-structural points, adapting to the complex three-dimensional mesh structure, improving the interpolation stability, and accurately restoring electrophysiological information on the three-dimensional reconstructed myocardial surface.

[0083] Among them, introducing spatio-temporal coupled dynamic interpolation for the multi-temporal electrophysiological data to obtain a continuously reconstructed three-dimensional spatio-temporal electrophysiological parameter field includes: defining a time label t m ∈[0, T], where t m is the absolute time at the m-th time point, T is the total duration of the cardiac cycle, and mapping the time label t m to the normalized cycle ratio Define a spatio-temporal coupled kernel function K(v, v′ j ; t, t m ) = K(v, v′ j ) · K time (t, t m ), where K(v, v′ j ) is an anisotropic spatial kernel function, and K time (t, t m ) is a time kernel function; the time kernel function σ τ is the time bandwidth; for any time point (v, t), calculate the electrophysiological parameter value through dynamic interpolation For each \(t\), only the data within the time window \([t - 2\sigma\tau, t + 2\sigma\tau]\) is loaded, where represents the weighted parameter sum of adjacent spatio-temporal points in all time phases, represents the normalization factor; for the electrophysiological parameter values perform anisotropic spatio-temporal smoothing processing to obtain a continuously reconstructed three-dimensional spatio-temporal electrophysiological parameter field in real time.

[0084] Its technical effects are as follows: By defining time tags and mapping them to the standardized cycle ratio, data in different time phases are incorporated into a unified time framework, breaking the time discretization limit of static data and realizing the estimation of electrophysiological values at any moment. According to the spatial anatomical characteristics and time evolution laws, a continuous and fine-grained three-dimensional + time electrophysiological distribution field is constructed, which is conducive to truly restoring the process of electrical signal conduction. By loading data at nearby moments through a local time window and using a dynamic kernel interpolation method to quickly calculate the electrophysiological state at any moment, a continuously reconstructed three-dimensional cardiac electrophysiological field in real time is realized. Spatio-temporal coupling interpolation improves the time consistency and spatial coherence of data, enabling potential abnormalities such as premature beats and ectopic pacing at a certain time point to be continuously mapped spatially, and the activity curve of a certain spatial region on the time axis to be stably estimated, improving the spatial localization accuracy and chronological judgment ability of electrophysiological abnormalities, and providing a reliable basis for arrhythmia diagnosis, electroablation path planning, etc.

[0085] Among them, the anisotropic spatio-temporal smoothing processing of the electrophysiological parameter values includes: Among them, is the spatial smoothing term, \(\lambda\) s is the median of the spatial gradient, is the time smoothing direction, \(\lambda\) t is the median of the time gradient.

[0086] The second embodiment of the present invention provides a spatial registration and interpolation calculation system for cardiac electrophysiological parameters, which includes: a three-dimensional anatomical model establishment module for segmenting cardiac magnetic resonance imaging data using the three-dimensional DRLSE algorithm, extracting the endocardium, epicardium, and myocardial regions, and establishing a three-dimensional cardiac anatomical model; a registration module for aligning electrophysiological parameter data to the grid structure of the three-dimensional cardiac anatomical model, performing anatomical model registration, and obtaining a registered electrophysiological parameter point set; a high-precision static electrophysiological field generation module for designing an anisotropic kernel function interpolation on the registered electrophysiological parameter point set to obtain a high-precision static electrophysiological field under anatomical constraints, forming multi-time-phase electrophysiological data; a spatio-temporal coupling module for introducing spatio-temporal coupling dynamic interpolation for the multi-time-phase electrophysiological data to obtain a continuously reconstructed three-dimensional spatio-temporal electrophysiological parameter field.

[0087] A computer program product of a method and apparatus for spatial registration and interpolation calculation of cardiac electrophysiological parameters provided by an embodiment of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods in the foregoing method embodiments. For specific implementation, reference can be made to the method embodiments and will not be elaborated herein.

[0088] Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, etc. When the computer program on the storage medium is run, it can execute the above-mentioned method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, so as to achieve high-precision non-rigid registration of electrophysiological data and a grid model through anatomical landmark points, and through an anisotropic interpolation mechanism, achieve compliance with anatomical structure characteristics, and accurately reflect the electrical signal propagation process by constructing a continuous four-dimensional parameter field.

[0089] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or part of this technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc that can store program code.

[0090] Finally, it should be noted that the above-mentioned embodiments are only specific implementation manners of the present invention, used to illustrate the technical solutions of the present invention, rather than limiting them. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: any person skilled in the art within the technical scope disclosed by the present invention can still modify the technical solutions recorded in the foregoing embodiments or can easily think of changes, or perform equivalent replacements on some of the technical features; and these modifications, changes, or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention and should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for spatial registration and interpolation calculation of cardiac electrophysiological parameters, characterized in that Including: Using a three-dimensional DRLSE algorithm to segment cardiac magnetic resonance imaging data, extract the endocardium, epicardium, and myocardial regions, and establish a three-dimensional cardiac anatomical model; Aligning electrophysiological parameter data to the grid structure of the three-dimensional cardiac anatomical model, performing anatomical model registration, and obtaining a registered electrophysiological parameter point set; Designing an anisotropic kernel function interpolation on the registered electrophysiological parameter point set to obtain a high-precision static electrophysiological field under anatomical constraints and form multi-phase electrophysiological data; Introducing spatio-temporal coupled dynamic interpolation for the multi-phase electrophysiological data to obtain a continuously reconstructed real-time three-dimensional spatio-temporal electrophysiological parameter field.

2. The spatial registration and interpolation calculation method for cardiac electrophysiological parameters according to claim 1, wherein The step of using a three-dimensional DRLSE algorithm to segment cardiac magnetic resonance imaging data, extract the endocardium, epicardium, and myocardial regions, and establish a three-dimensional cardiac anatomical model includes: Collecting cardiac magnetic resonance imaging to obtain three-dimensional cardiac magnetic resonance volume data I(x, y, z); Perform gray normalization on the original cardiac magnetic resonance three-dimensional volume data I(x, y, z) to eliminate brightness differences and obtain normalized image data Construct the initial level set function where Ω is the spatial range of the three-dimensional volume data of the cardiac magnetic resonance to be segmented, and Ω0 is the initial seed region is the boundary of the initial seed region, and ρ is a constant Perform three-dimensional edge detection factor calculation on the normalized image data to obtain an edge indication function based on image gradients where G σ is a three-dimensional Gaussian kernel with standard deviation σ, is a three-dimensional gradient operator, and * is a three-dimensional convolution operation; Defining a three-dimensional DRLSE energy functional E(φ) = μR(φ) + λL(φ) + αA(φ), where R(φ) is a regularization term, μ is a regularization term weight coefficient, L(φ) is an edge term, λ is an edge term weight coefficient, A(φ) is an area term, and α is an area term weight coefficient; Taking the variation of the energy functional \(E(\varphi)\) gives the gradient descent flow equation where is the rate of change of the level set function with respect to time, is the Laplacian operator of the level set function, div is the divergence operator, is the normalized direction of the level set gradient, \(\delta(\varphi)\) is the Dirac function, and \(g\) is the edge indicator value of the image gradient; The initial level set function φ0(x, y, z) is discretely iterated using the finite difference method to obtain an iterative update formula where the termination condition is ||φ n+1 - φ n ||2 < δ, n is the number of iterations, δ is a preset threshold, t is the time step, the converged level set function is φ(x, y, z), and φ is a three-dimensional scalar function; Extract the isosurface from the converged level set function φ(x, y, z) to obtain the endocardial boundary φ(x, y, z) = 0 and the epicardial boundary φ(x, y, z) = k. Use the Marching Cubes algorithm to generate the triangular mesh surface of the three-dimensional cardiac anatomical model M = {V, E, F}, where V = {v i |v i =(x i , y i , z i )} is the mesh vertex, v i is each vertex, i = 1, 2,..., N, E is the edge of the mesh, and F is the face of the mesh; For each vertex of the triangular mesh, the myocardial thickness field in the three-dimensional anatomical model of the heart is calculated by Euclidean distance wherein, is the grid vertex coordinate of the endocardium, is the grid vertex coordinate of the epicardium.

3. The method for spatial registration and interpolation calculation of cardiac electrophysiological parameters according to claim 2, wherein In the three-dimensional DRLSE energy functional E(φ), The regularization term R(φ) is used to constrain the magnitude of the level set gradient to be close to 1 to ensure numerical stability. The edge term L(φ) is used to drive the curve to evolve towards the target boundary. where δ(·) is the Dirac function; The area term A(φ) is used to control the evolution direction to contract inward or outward, and A(φ) = ∫ Ω gH(−φ)dxdydz, where H(·) is the Heaviside step function.

4. The method for spatial registration and interpolation calculation of cardiac electrophysiological parameters according to claim 2, wherein After obtaining the three-dimensional cardiac anatomical model, interlayer offset correction is also performed, including: Calculate the centroid coordinates of each layer section for the three-dimensional cardiac anatomical model M where N z is the number of vertices in the z-th layer; Using the least squares method to fit to obtain a three-dimensional straight line equation l(z) = cz + f, where c is the slope and f is a constant; For the offset layer, the vertex coordinates are translated, and the translation value Δv of the vertex coordinates = c z′ -l(z′), so that the centroid returns to fit the three-dimensional line.

5. The spatial registration and interpolation calculation method of cardiac electrophysiological parameters according to claim 2, characterized in that The step of aligning electrophysiological parameter data to the grid structure of the three-dimensional cardiac anatomical model, performing anatomical model registration, and obtaining a registered electrophysiological parameter point set includes: Select anatomical landmark points \(P = \{p k |p k \in V\}\) on the triangular mesh surface of the three-dimensional cardiac anatomical model \(M\); Collect electrophysiological parameter data to generate an electrophysiological point set Q = {q j |q j =(u j , U j )}, where u j =(x j , y j , z j ) is the three-dimensional coordinate of the electrophysiological measurement point, and U j is the electrophysiological parameter value of the corresponding point. Match the point set S = {s k |s k ∈Q} corresponding to the anatomical landmark point P in the electrophysiological point set Q, so that the anatomical position of the anatomical landmark point is consistent with the electrophysiological parameter data; Construct a thin-plate spline transformation model Among them, is the translation vector, is the affine transformation matrix, is the non-rigid deformation weight, φ(·) = · 2 log· is the radial basis function, and u is the original three-dimensional coordinate of any point in the electrophysiological point set Q; Preliminarily align the electrophysiological point set Q with the parameter value in the thin plate spline transformation model F(u) to obtain u' j = a + Bu j (j = 1, 2,..., N), superimpose the radial basis function correction term to obtain the registration coordinates Project the registration coordinates u′ j ′ onto the mesh surface of the three-dimensional anatomical model M of the heart to obtain the registered electrophysiological parameter point set V j ′ = {v′ j |v′ j ∈V∪F}.

6. The method for spatial registration and interpolation calculation of cardiac electrophysiological parameters according to claim 5, characterized in that After constructing the thin plate spline transformation model F(u), it also includes setting the constraint conditions of the thin plate spline transformation model and solving to obtain the parameter values in the thin plate spline transformation model F(u); The constraint conditions include minimizing the bending energy satisfying the interpolation condition F(s k ) = p k (k = 1, 2,..., K) and additional constraints Solving the spatial transformation such that F(s k ) = p k .

7. The method for spatial registration and interpolation calculation of cardiac electrophysiological parameters according to claim 5, wherein The step of designing an anisotropic kernel function interpolation on the registered electrophysiological parameter point set to obtain a high-precision static electrophysiological field under anatomical constraints and form multi-phase electrophysiological data includes: Define the anisotropic kernel function where is the spatial distance term is the thickness difference term, v is the coordinates of the point to be interpolated, and σ d is the spatial distance bandwidth, and σ t is the myocardial thickness difference weight; For each vertex v of the mesh surface i ∈V, calculate the interpolation parameter Perform multi-temporal data integration on the static fields at M time points, unify the time axis through electrocardiogram gating markers, divide the time phase groups according to the functional phases, and obtain a high-precision static electrophysiological field under anatomical constraints v i ∈V, m = 1, 2,..., M, where is the static electrophysiological parameter field at the m-th time point, forming multi-temporal electrophysiological data.

8. The method for spatial registration and interpolation calculation of cardiac electrophysiological parameters according to claim 7, wherein The step of introducing spatio-temporal coupled dynamic interpolation for the multi-phase electrophysiological data to obtain a continuously reconstructed real-time three-dimensional spatio-temporal electrophysiological parameter field includes: Define a time tag t for the multi-temporal electrophysiological data m ∈[0,T], where t m is the absolute time of the m-th time point, and T is the total duration of the cardiac cycle. Map the time tag t m to the normalized cycle ratio Define the spatio-temporal coupling kernel function \(K(\mathbf{v},\mathbf{v}' j ;t,t' m ) = K(\mathbf{v},\mathbf{v}' j )\cdot K time (t,t' m ), where \(K(\mathbf{v},\mathbf{v}' j )\) is the anisotropic spatial kernel function and \(K time (t,t' m )\) is the temporal kernel function; The time kernel function σ τ is the time bandwidth; For any time point (v, t), calculate the electrophysiological parameter value through dynamic interpolation For each t, only load the data within the time window [t - 2στ, t + 2στ], where represents the weighted parameter sum of adjacent spatio-temporal points in all time phases represents the normalization factor Perform anisotropic spatio-temporal smoothing on the electrophysiological parameter values to obtain a continuously reconstructed real-time three-dimensional spatio-temporal electrophysiological parameter field.

9. The method for spatial registration and interpolation calculation of cardiac electrophysiological parameters according to claim 8, wherein The electrophysiological parameter value Performing anisotropic spatio-temporal smoothing processing includes: Among them, is the spatial smoothing term, λ s is the median value of the spatial gradient, is the temporal smoothing term, λ t is the median value of the temporal gradient.

10. A spatial registration and interpolation calculation system for cardiac electrophysiological parameters, characterized in that, Including: A three-dimensional anatomical model establishment module for using a three-dimensional DRLSE algorithm to segment cardiac magnetic resonance imaging data, extract the endocardium, epicardium, and myocardial regions, and establish a three-dimensional cardiac anatomical model; A registration module for aligning electrophysiological parameter data to the grid structure of the three-dimensional cardiac anatomical model, performing anatomical model registration, and obtaining a registered electrophysiological parameter point set; A high-precision static electrophysiological field generation module for designing an anisotropic kernel function interpolation on the registered electrophysiological parameter point set to obtain a high-precision static electrophysiological field under anatomical constraints and form multi-phase electrophysiological data; A spatio-temporal coupling module for introducing spatio-temporal coupled dynamic interpolation for the multi-phase electrophysiological data to obtain a continuously reconstructed real-time three-dimensional spatio-temporal electrophysiological parameter field.

Images