Method and system for 4D flow MRI visualization of heart valve regurgitation vortex

Abstract

The present application relates to medical image processing technical field, especially to heart valve regurgitation vortex field 4D Flow MRI visualization method and system, based on heart 4D Flow MRI data, fusion four-chamber heart function sequence, mitral valve flow velocity coding and three geometry curve image, construct two-dimensional regurgitation heart grid model, use geometry curve to determine valve leaflet position and flow line start and end point, identify vortex structure through characteristic vector space analysis, extract vortex core line combined with multi-start point tracking, and adopt non-tracking seed algorithm to reconstruct vortex core body, calculate vortex momentum based on core body parameter, establish momentum damage prediction model, finally through three-dimensional flow line, color coding and dynamic animation and other multi-dimensional visualization technology, clearly present heart valve regurgitation vortex field characteristics, realize high-precision registration fusion of heart four-chamber heart function sequence image, mitral valve flow velocity coding image and mitral valve three geometry curve image.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical image processing, in particular to a heart valve regurgitation vortex field 4D Flow MRI visualization method and system, which is used to realize high-precision identification, extraction, visualization and momentum damage assessment of heart valve regurgitation vortex field, especially mitral valve, and provide technical support for diagnosis and treatment of heart valve diseases. BACKGROUND

[0002] Heart valve regurgitation is a common heart valve disease, among which mitral valve regurgitation is particularly common. When the mitral valve regurgitates, the blood in the left ventricle of the heart flows back to the left atrium during the systole, forming a complex vortex field structure. These vortex field structures are closely related to the severity of the disease and the prognosis. Accurate identification and assessment of these vortex field structures have important value for guiding clinical treatment.

[0003] Traditional heart function assessment methods, such as echocardiography and conventional MRI, can only provide limited anatomical structure and blood flow information, making it difficult to fully depict the complex hemodynamic characteristics in the heart. In recent years, the development of 4D Flow MRI technology has provided new possibilities for observing and analyzing blood flow in the heart. This technology can simultaneously obtain three-dimensional spatial information and temporal information of blood flow, and can fully capture the changes of blood flow in the heart during the cardiac cycle.

[0004] However, the existing 4D Flow MRI data analysis methods mainly have the following shortcomings: first, the vortex identification method is simple, and a single threshold is usually used for judgment, which lacks precision; second, there is a lack of accurate extraction technology for vortex core structure, making it difficult to quantify vortex characteristics; third, the visualization form is single, making it difficult to intuitively display the complex dynamic characteristics of the vortex field; fourth, there is a lack of quantitative correlation analysis between vortex dynamics characteristics and tissue damage.

[0005] Therefore, there is an urgent need for a technical method that can accurately identify heart valve regurgitation vortex field, extract vortex core structure, multi-dimensionally visualize vortex field characteristics, and quantitatively evaluate its dynamics impact, providing support for accurate diagnosis and personalized treatment of heart valve diseases. SUMMARY

[0006] The purpose of the present application is to provide a heart valve regurgitation vortex field 4D Flow MRI visualization method and system, which aims to overcome the problems of insufficient vortex identification precision, inaccurate core structure extraction, single visualization performance, and lack of quantitative evaluation in the prior art, and to realize high-precision identification, accurate extraction, multi-dimensional visualization, and quantitative assessment of momentum damage of heart valve regurgitation vortex field.

[0007] The present application proposes a heart valve regurgitation vortex field 4D Flow MRI visualization method, which comprises:

[0008] Acquire cardiac 4D Flow MRI data, which includes cardiac four-chamber functional sequence images, mitral valve velocity-coded images, and mitral valve trigeometric curve images;

[0009] By using cardiac anatomical registration technology, the cardiac four-chamber functional sequence image, mitral valve velocity-coded image, and mitral valve three-geometric curve image are fused to obtain the image fusion result.

[0010] Based on the image fusion results, a two-dimensional contour of the heart surface is extracted, and a two-dimensional regurgitation heart mesh model is constructed.

[0011] The position of the leaflets and the starting and ending points of the streamlines are determined using the aforementioned mitral valve trigeminal curves.

[0012] Vortex structures are identified based on the feature vector space analysis method, which includes constructing a velocity gradient tensor, calculating eigenvalues, and performing multi-threshold adaptive vortex determination.

[0013] The vortex coreline is extracted using a multi-starting-point vortex coreline tracing technique, which includes a starting point selection mechanism, a bidirectional tracing strategy, and a path merging technique.

[0014] Based on the vortex core line, the vortex core is reconstructed using a non-tracking seed algorithm, which includes seed point placement, region growth control, and adaptive mesh subdivision.

[0015] Based on the geometric and flow field parameters of the vortex core, the vortex momentum is calculated and a momentum damage prediction model is established.

[0016] The regurgitation vortex field of the heart valve is displayed through multi-dimensional visualization rendering technology, which includes three-dimensional streamline visualization, color coding mechanism and dynamic evolution animation.

[0017] Preferably, the acquisition of cardiac 4D Flow MRI data specifically includes:

[0018] 4D Free Breath Stack of Slices Cine MRI sequence images of the heart were acquired using an MRI scanner;

[0019] Based on the aforementioned 4D Free Breath Stack of Slices Cine MRI sequence images, cardiac four-chamber functional sequence images and mitral valve velocity-encoded images were obtained;

[0020] Based on the cardiac four-chamber functional sequence image and the mitral valve velocity encoding image, a mitral valve three-geometric curve image is obtained;

[0021] The time resolution of the cardiac four-chamber functional sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image is 1 second, and the spatial resolution is 1*1.25*1.25 mm.

[0022] Preferably, the 4D Flow MRI sequence adopted by the magnetic resonance scanner is a multi-echo velocity encoding gradient echo image acquisition module, which obtains the cardiac four-chamber functional sequence image and the mitral valve flow velocity encoding image at a velocity encoding of 1.5 m / s.

[0023] The image parameters of the cardiac four-chamber functional sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image include: matrix 512*304, slice number 20, layer thickness 6 mm, TR 3.08 ms, TE 1.16 ms, inversion angle 25°, bandwidth ±12.5 kHz.

[0024] Preferably, the cardiac anatomical structure registration technology specifically includes:

[0025] Through the cardiac 4D Flow MRI data processing system, the image registration is performed using the synchronous acquisition window between the cardiac four-chamber functional sequence image and the mitral valve three-geometry curve image.

[0026] A hierarchical registration strategy is adopted to sequentially perform global alignment based on the overall contour of the heart, regional alignment based on the heart cavity and main blood vessel structure, and local fine alignment based on the mitral valve and its surrounding structure.

[0027] The image registration error between the cardiac four-chamber functional sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image is less than 1 mm.

[0028] Preferably, the construction of the two-dimensional regurgitant heart mesh model specifically includes:

[0029] Different heart surface contours are extracted from the cardiac four-chamber functional sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image, respectively, wherein the left ventricular surface contour is extracted from the cardiac four-chamber functional sequence image, the mitral valve surface contour is extracted from the mitral valve flow velocity encoding image, and the left atrial surface contour is extracted from the mitral valve three-geometry curve image.

[0030] Based on the extracted mitral valve geometry surface contour, the mitral valve geometry surface contour is grid line encrypted.

[0031] The encrypted lines are smoothly connected using a curve smoothing connection technology to form a two-dimensional mesh heart model, wherein the curve smoothing connection technology realizes the continuity and smoothness of the curve through control point interpolation.

[0032] As preferred, the feature vector space analysis method specifically includes:

[0033] Calculate the spatial gradient of the velocity field on the spatial discrete points to form a velocity gradient tensor;

[0034] Perform eigenvalue decomposition on the velocity gradient tensor of each point to obtain eigenvalues and eigenvectors;

[0035] Construct a decision condition based on the mathematical relationship between the eigenvalues, and determine that the discrete point is a vortex structure point when the eigenvalues meet the set condition;

[0036] Dynamically adjust the decision threshold according to the local flow field characteristics to form a vortex possibility level of multiple thresholds;

[0037] Comprehensively determine the final vortex region by combining the eigenvalue relationship and the rotation intensity.

[0038] As preferred, the multi-startpoint vortex core line tracking technology specifically includes:

[0039] Select multiple potential starting points based on vortex intensity, and simultaneously track from each starting point in both positive and negative directions;

[0040] During the tracking process, dynamically adjust the tracking step size according to the local curvature, with smaller step sizes in areas with large curvatures and larger step sizes in areas with small curvatures;

[0041] Combine the trajectory of the previous steps, the local flow field characteristics, and the vortex eigenvalue direction to comprehensively predict the next direction;

[0042] Real-time correct cumulative errors during the tracking process to ensure the accuracy of long-path tracking;

[0043] Intelligently combine the paths tracked from different starting points to form a complete vortex core line.

[0044] As preferred, the non-tracking seed algorithm for reconstructing the vortex core body specifically includes:

[0045] Take the points on the vortex core line as seed points and expand outward to form an initial core body region;

[0046] Control the region expansion boundary based on the vortex determination condition to ensure that the expanded region meets the vortex characteristics;

[0047] Non-uniformly subdivide the space, with higher grid density in key areas;

[0048] Calculate the flow distribution in the expanded region to ensure the integrity and accuracy of the core body;

[0049] Through the iterative subdivision and evaluation process, finally form an accurate vortex core body.

[0050] As preferred, the calculating the swirl momentum and establishing the momentum injury prediction model specifically comprises:

[0051] calculating the geometric parameters of the swirl core body, including area, perimeter and volume;

[0052] calculating the flow field characteristic parameters in the swirl core body, including average velocity and rotation intensity;

[0053] calculating the comprehensive momentum index based on the geometric parameters and the flow field characteristic parameters;

[0054] dividing the comprehensive momentum index into different injury grades, including normal, mild, moderate and severe;

[0055] determining the critical value of each grade based on clinical data, and establishing the correlation model of the momentum value and the tissue injury degree;

[0056] determining the energy parameters and treatment position of radiofrequency ablation according to the momentum injury prediction model, and realizing the personalized treatment plan.

[0057] A heart valve regurgitation vortex field 4D Flow MRI visualization system, comprising:

[0058] a data acquisition module for acquiring heart 4D Flow MRI data, the heart 4D Flow MRI data including heart four-chamber function sequence images, mitral valve flow velocity encoding images and mitral valve three-geometry curve images;

[0059] an image preprocessing module for image fusion of the heart four-chamber function sequence images, the mitral valve flow velocity encoding images and the mitral valve three-geometry curve images through heart anatomical structure registration technology;

[0060] a mesh construction module for extracting a heart surface two-dimensional contour and constructing a two-dimensional regurgitation heart mesh model based on the image fusion result;

[0061] a valve positioning module for determining valve leaflet position and flow line starting point and ending point by using the mitral valve three-geometry curve;

[0062] a vortex identification module for identifying vortex structure based on a feature vector space analysis method, the feature vector space analysis method including constructing a velocity gradient tensor, calculating a feature value and performing a multi-threshold adaptive vortex judgment;

[0063] a core line tracking module for extracting a vortex core line through a multi-start point vortex core line tracking technology, the multi-start point vortex core line tracking technology including a start point selection mechanism, a bidirectional tracking strategy and a path merging technology;

[0064] a core body reconstruction module, configured to reconstruct a vortex core body based on the vortex core line by using a non-tracking seed algorithm, the non-tracking seed algorithm comprising seed point arrangement, region growth control and adaptive mesh subdivision;

[0065] a momentum analysis module, configured to calculate vortex momentum and establish a momentum injury prediction model according to geometric and flow field parameters of the vortex core body;

[0066] a visualization module, configured to display the heart valve regurgitation vortex flow field by using a multi-dimensional visualization rendering technology, the multi-dimensional visualization rendering technology comprising three-dimensional flow line visualization, color coding mechanism and dynamic evolution animation.

[0067] The present application has the following beneficial effects:

[0068] 1. By using a multi-modal medical image acquisition and fusion technology, high-precision registration and fusion of a heart four-chamber heart function sequence image, a mitral valve flow velocity encoding image and a mitral valve three-geometry curve image are realized, thereby providing a high-quality data basis for subsequent analysis.

[0069] 2. The vortex structure is identified by using a feature vector space analysis method, and a multi-threshold adaptive judgment mechanism is combined, so that the precision and robustness of vortex identification are greatly improved, and the limitation of a traditional single threshold method is overcome.

[0070] 3. The multi-start vortex core line tracking technology and the non-tracking seed algorithm are innovatively introduced, so that accurate extraction and reconstruction from the vortex core line to the core body are realized, thereby providing a basis for comprehensive quantification of vortex characteristics.

[0071] 4. Based on geometric and flow field parameters of the vortex core body, a quantitative correlation model of vortex momentum and tissue injury is established, thereby providing a scientific basis for clinical evaluation and treatment decision.

[0072] 5. By using a multi-dimensional visualization rendering technology, intuitive and comprehensive display of vortex flow field characteristics is realized, thereby enhancing the feasibility and practicality of clinical application.

[0073] 6. The overall technical scheme forms a complete technical chain from data acquisition, processing and analysis to visualization evaluation, and has good systematicness and practicality. BRIEF DESCRIPTION OF DRAWINGS

[0074] Figure 1 a flow chart of the heart valve regurgitation vortex flow field 4D Flow MRI visualization method of the present application;

[0075] Figure 2 a structural block diagram of the heart valve regurgitation vortex flow field 4D Flow MRI visualization system of the present application. DETAILED DESCRIPTION

[0076] Please refer to Figure 1 - Figure 2 The application will be described in detail below with reference to the accompanying drawings and specific embodiments. Those skilled in the art should understand that these embodiments are only used to illustrate the application, and are not used to limit the scope of the application. It should be noted that for those skilled in the art, without departing from the principles of the application, a number of improvements and modifications can be made to the application, and these improvements and modifications also fall within the protection scope of the claims of the application.

[0077] As shown in Figure 1 , the heart valve regurgitation vortex flow field 4D Flow MRI visualization method provided by the application comprises the following steps:

[0078] Heart 4D Flow MRI data acquisition: This step mainly realizes the acquisition of high-quality heart 4D Flow MRI data, including heart four-chamber function sequence images, mitral valve flow velocity encoding images and mitral valve three-geometry curve images.

[0079] Preferably, a magnetic resonance scanner is used to collect heart 4D Free Breath Stack of Slices Cine MRI sequence images. Specifically, a multi-echo velocity encoding gradient echo image acquisition module is used to obtain heart four-chamber function sequence images and mitral valve flow velocity encoding images at a velocity encoding of 1.5 m / s. Here, the velocity encoding of 1.5 m / s is selected because the heart valve regurgitation blood flow velocity is usually in the range of 0.8-1.2 m / s, and setting it to 1.5 m / s can fully cover the regurgitation velocity range and leave a margin to avoid velocity aliasing artifacts.

[0080] In an embodiment of the application, the time resolution of the collected images is set to 1 second, and the spatial resolution is 1x1.25x1.25 mm. This resolution setting can cover the complete cardiac cycle (usually 0.8-1.0 seconds) while ensuring image quality, and capture the dynamic change process of the regurgitation vortex.

[0081] In addition, to obtain the best image quality, the specific image parameters are set as follows: matrix 512x304, slice number 20, layer thickness 6 mm, TR (repetition time) 3.08 ms, TE (echo time) 1.16 ms, matrix number 1-50, layer number 1-20, FOV (field of view) 360x360 mm, inversion angle 25°, bandwidth ±12.5 kHz, NEX (number of signal averages) 2. The selection of these parameters is based on a large amount of clinical experience, and can obtain the best balance of signal-to-noise ratio and spatio-temporal resolution under the current clinical MRI equipment conditions.

[0082] Heart anatomical structure registration technology: This step mainly realizes the accurate registration and fusion of multi-sequence images, providing a basis for subsequent analysis.

[0083] Preferably, the image registration is performed by the heart 4D Flow MRI data processing system using the synchronized acquisition window between the heart four-chamber functional sequence images and the mitral valve three-geometry profile images. The registration process adopts a hierarchical strategy, sequentially performing global alignment, regional alignment, and local fine alignment.

[0084] Specifically, the global alignment is based on the overall contour of the heart, using a rigid transformation model including translation, rotation, and scaling operations to preliminarily align the overall position of the images. The regional alignment is based on the heart cavities and main blood vessel structures, using an elastic transformation model to focus on aligning the ventricle, atrium, and main blood vessel regions. The local fine alignment focuses on the mitral valve and its surrounding structures, using a nonlinear transformation model to achieve accurate alignment of the valve region.

[0085] In the present application, the registration accuracy is controlled within less than 1 mm, which is achieved by setting appropriate iteration times (usually 200-300 times) and convergence thresholds (usually 0.01 mm). High-precision registration ensures the accuracy of subsequent analysis, which is particularly important for accurate positioning and characterization of the valve regurgitation vortex field.

[0086] Constructing a two-dimensional regurgitant heart mesh model: This step mainly realizes the extraction of the two-dimensional contour of the heart surface and the construction of the two-dimensional regurgitant heart mesh model.

[0087] First, different heart surface contours are extracted from the registered heart four-chamber functional sequence images, mitral valve flow velocity encoding images, and mitral valve three-geometry profile images. Specifically, the left ventricular surface contour is extracted from the heart four-chamber functional sequence images, the mitral valve surface contour is extracted from the mitral valve flow velocity encoding images, and the left atrial surface contour is extracted from the mitral valve three-geometry profile images.

[0088] The contour extraction adopts a gradient-based edge detection method, and preferably uses the Canny edge detection algorithm, whose mathematical expression is:

[0089] ,

[0090] wherein: is the image gradient, representing the gradient value at the coordinate ; is the Gaussian filter, is the standard deviation of the Gaussian filter (usually set to 1.5-2.0) for controlling the smoothing degree; is the original image, representing the pixel value at the coordinate ; * represents the convolution operation for image filtering; represents the gradient operator, calculating the image gradient in and derivative in the direction.

[0091] Then, based on the extracted mitral valve geometric surface profile, the profile is grid line encrypted. Preferably, the encryption adopts an equidistant interpolation method to insert new control points between the original profile points, so that the distance between adjacent control points does not exceed 0.5 mm, which can ensure that the subsequent constructed grid model has sufficient accuracy.

[0092] Finally, the encrypted lines are smoothly connected using a curve smoothing connection technique to form a two-dimensional grid heart model. Preferably, the curve smoothing connection technique is used to realize the continuity and smoothness of the curve through control point interpolation. The control point interpolation here adopts a cubic spline interpolation method to ensure that the generated curve has continuous first and second derivatives at the control points, avoiding the occurrence of sharp edges or discontinuous points.

[0093] Determine the position of the leaflet and the start and end points of the flow line: This step mainly uses the three geometric curves of the mitral valve to determine the position of the leaflet and the start and end points of the flow line, providing a basis for subsequent vortex identification and visualization.

[0094] Preferably, the mitral valve geometric surface profile is used as the initial position for displaying the mitral regurgitation vortex, and the three geometric curves provide three-dimensional reference positions and reference point positions for the mitral regurgitation vortex flow field. Specifically, the mitral valve geometric surface profile line is the boundary line of the two-dimensional regurgitation heart grid model, and the three-dimensional space points of the regurgitation vortex flow field are constructed based on this coordinate line; the three geometric curves are points on the mitral valve surface in the four-chamber heart function sequence image, the mitral valve flow velocity encoding image and the mitral valve three geometric curve image, providing three-dimensional space points for the regurgitation vortex flow field.

[0095] In actual application, the flow line start point is usually selected on the mitral valve orifice plane, and 16-24 points are evenly distributed along the valve ring circumference; the flow line end point is automatically determined according to the blood flow direction, and the maximum tracking length of 100 mm or the maximum tracking time of 500 ms is usually set as the termination condition. This setting can ensure that the flow line completely captures the regurgitation vortex structure, while avoiding visual confusion caused by too long flow lines.

[0096] Identify the vortex structure based on the feature vector space analysis method: This step is one of the core innovative points of the present application, which realizes high-precision vortex structure identification.

[0097] The feature vector space analysis method first calculates the spatial gradient of the velocity field on the spatial discrete points to form a velocity gradient tensor. The calculation formula of the velocity gradient tensor is as follows:

[0098] ,

[0099] Wherein: , , are the components of the velocity field in the x, y, z directions, respectively, in m / s; , , are the components of the velocity field in the x, y, z directions, respectively, in m / s; denotes the partial derivative of the velocity component in the x direction, in s -1; , or ; is the 3x3 velocity gradient tensor, which describes the rate of change of the flow field in space.

[0100] Then, the velocity gradient tensor of each point is decomposed to obtain eigenvalues , , and corresponding eigenvectors. The process of eigen decomposition is to solve the characteristic equation:

[0101] ,

[0102] where: is the 3x3 identity matrix; denotes the matrix determinant operation; is the eigenvalue, in s -1 , which represents the strength of local rotation and deformation of the flow field. The eigenvalues are arranged in the order of .

[0103] Based on the mathematical relationship between the eigenvalues, a vortex determination condition is constructed. Preferably, when the eigenvalues satisfy the following conditions, the discrete point is determined to be a vortex structure point:

[0104] ,

[0105] where: , , are the three eigenvalues of the velocity gradient tensor, in s -1 ; is the determination threshold, in s -1 , which is dynamically adjusted according to the local flow field characteristics. In practical applications, is usually set to 5% to 15% of , and the specific value is adaptively adjusted according to the local flow field noise level. ≈ indicates that the two eigenvalues are approximately equal, with a difference less than the set threshold.

[0106] ​​The present invention innovatively introduces a multi-threshold adaptive vortex determination mechanism, dynamically adjusts the determination threshold according to the local flow field characteristics, and forms a vortex possibility level with multiple thresholds. Specifically, three levels of thresholds are set , and (5%, 10% and 15% of λ1 respectively), and the vortex structure points are divided into three levels: high possibility ( − <Tc1), medium possibility ( ≤ − <Tc2) and low possibility ([[ID=१७]] ≤ − <Tc3).

[0107] In addition, a comprehensive determination is carried out by combining the eigenvalue relationship and the rotation intensity to determine the final vortex region. The calculation formula of the rotation intensity is:

[0108] ,

[0109] where: Ω is the rotation intensity, with the unit of s -1 , representing the intensity of local rotation of the fluid; |∇v| represents the norm of the velocity gradient tensor, with the unit of s -1 ; represents the transpose matrix of the velocity gradient tensor; represents the norm of the transpose matrix, with the unit of s -1 . The matrix norm adopts the Frobenius norm, and the calculation formula is the square root of the sum of the squares of all elements of the matrix.

[0110] In practical applications, the rotation intensity threshold Ωmin is usually set to 5 - 10 s -1 , and only the points that simultaneously satisfy the eigenvalue condition and the rotation intensity Ω > Ωmin are finally determined as vortex structure points. This comprehensive determination method greatly improves the accuracy and robustness of vortex recognition.

[0111] Extract the vortex core line through the multi-start vortex core line tracking technology: This step realizes the accurate extraction of the vortex core line and lays a foundation for the reconstruction of the core body.

[0112] The multi-start vortex core line tracking technology first selects multiple potential starting points based on the vortex intensity. Preferably, among the identified vortex structures, the top 10 - 20 points with the highest rotation intensity are selected as potential starting points. Tracking is carried out simultaneously in both forward and reverse directions from each starting point, and this two-way tracking strategy can improve the integrity of core line extraction.

[0113] In the tracking process, the tracking step is dynamically adjusted according to the local curvature. Preferably, the calculation formula of the step s is as follows:

[0114] ,

[0115] Wherein: is the current tracking step, with the unit of mm; is the minimum step (usually 0.2 mm), with the unit of mm; is the maximum step (usually 1.0 mm), with the unit of mm; is the local curvature, with the unit of , indicating the bending degree of the curve at the point; is the adjustment parameter (usually set to 2.0), dimensionless, controlling the influence degree of the curvature on the step; is the base number of natural logarithm, approximately equal to 2.71828. In this way, a smaller step is used in the area with large curvature to improve the accuracy, and a larger step is used in the area with small curvature to improve the efficiency.

[0116] The prediction of the tracking direction is the key to the core line extraction. The present application comprehensively predicts the next step direction by combining the track of the previous steps, the local flow field characteristics and the vortex eigenvalue direction. Specifically, the calculation formula of the next step direction is as follows:

[0117] ,

[0118] Wherein: is the predicted next tracking direction, which is a three-dimensional unit vector; is the previous step direction, which is a three-dimensional unit vector; is the local flow field average velocity direction, which is a three-dimensional unit vector, obtained by normalizing the flow field velocity vector around the current point; is the eigenvalue corresponding to the eigenvector direction, which is a three-dimensional unit vector, indicating the direction of the local rotation axis of the vortex; and are weight factors (usually set to 0.5, 0.3 and 0.2), dimensionless, satisfying , used to control the influence degree of each factor.

[0119] In order to ensure the accuracy of long path tracking, the present application corrects the cumulative error in real time during the tracking process. Specifically, every 10-15 steps of tracking, the vortex characteristics of the current point are recalculated, and if it deviates from the vortex region, it is backtracked to the last confirmed point and the direction is adjusted for re-tracking. This error correction mechanism effectively avoids the drift problem in the tracking process.

[0120] Finally, the paths tracked from different starting points are intelligently merged to form a complete vortex core line. During the merging process, if the minimum distance between two paths is less than a preset threshold (usually 1.0 mm) and the direction angle is less than a preset angle (usually 30°), it is considered that the two paths belong to the same vortex structure and are merged. The merged core line is smoothed to ensure geometric continuity.

[0121] Reconstructing the vortex core body based on the non-tracking seed algorithm: This step realizes the accurate expansion from the vortex core line to the core body, and fully describes the three-dimensional morphology of the vortex.

[0122] The non-tracking seed algorithm first takes the points on the vortex core line as seed points and expands outward to form an initial core body region. Preferably, 20-30 seed points are uniformly selected on the core line to ensure coverage of the entire core line.

[0123] Then, based on the vortex judgment condition, the region expansion boundary is controlled to ensure that the expanded region meets the vortex characteristics. Specifically, from each seed point, expand in 6 main directions (±x, ±y, ±z), expand one grid unit (usually 0.5-1.0 mm) at a time, and if the newly expanded point meets the vortex judgment condition, it is added to the core body region and used as a new expansion starting point; otherwise, stop expanding in that direction.

[0124] The present application innovatively uses a non-uniform spatial subdivision strategy to non-uniformly subdivide the space, with higher grid density in key areas. Specifically, within a 2mm range around the core line, the grid resolution is set to 0.25mm; within a 2-5mm range, the resolution is 0.5mm; and beyond 5mm, the resolution is 1.0mm. This non-uniform subdivision strategy greatly reduces the computational load while ensuring the accuracy of the core region.

[0125] To ensure the integrity and accuracy of the core body, the present application calculates the flow distribution in the expanded region. The formula for calculating the flow Q is:

[0126] ,

[0127] Where: Q is the flow, with a unit of m 3 / s; V is the core body volume, with a unit of m 3 ; is the flow velocity vector, with a unit of m / s; is the surface normal vector, which is a unit vector representing the vertical direction of each point on the surface; dS is the surface infinitesimal, with a unit of m 2 . The integral represents the total flow of the flow velocity along the normal direction on the entire core body surface.

[0128] In practical applications, a flow threshold is usually set To be 10% of the flow at the core line, when the flow at a certain point is lower than the threshold, stop expanding in that direction. This flow-based control strategy can more accurately reflect the physical boundaries of the vortex.

[0129] Through the iterative subdivision and evaluation process, the final vortex core body is formed. Usually 10-15 iterations can reach a state of convergence, and the surface of the final core body is triangulated for subsequent visualization and analysis.

[0130] Calculate the vortex momentum and establish a momentum damage prediction model: This step realizes the quantitative evaluation of the dynamic characteristics of the vortex, providing a scientific basis for clinical decision-making.

[0131] First, calculate the geometric parameters of the vortex core body, including area A, perimeter P, and volume V. These parameters are directly calculated through the triangulated mesh of the core body.

[0132] Then, calculate the flow field characteristic parameters in the vortex core body, including average velocity and rotational intensity . The formula for calculating the average velocity is:

[0133] ,

[0134] Where: is the average velocity, with units of m / s; is the core body volume, with units of m³; is the flow rate, with units of m / s; is the volume element, with units of m³. The integral represents the average value of the flow rate over the entire core body volume.

[0135] The formula for calculating the average value of the rotational intensity is:

[0136] ,

[0137] Where: is the average rotational intensity, with units of s⁻¹; is the core body volume, with units of m³; is the local rotational intensity, with units of s⁻¹; is the volume element, with units of m³. The integral represents the average value of the rotational intensity over the entire core body volume.

[0138] Based on the geometric parameters and flow field characteristic parameters, calculate the comprehensive momentum index . The formula for calculating the momentum index is:

[0139] ,

[0140] Where: is the integrated momentum indicator, with the unit of kg·m² / s²; is the blood density (usually taken as 1060 kg / m³), with the unit of kg / m³; is the core volume, with the unit of m³; is the average velocity, with the unit of m / s; is the average rotational intensity, with the unit of s⁻¹.

[0141] The present application innovatively uses the integrated momentum indicator to divide the damage into four grades, including normal , mild , moderate , and severe . In practical applications, based on a large amount of clinical data analysis, the critical values are usually set as kg·m² / s², kg·m² / s², kg·m² / s².

[0142] The critical values of each grade are determined based on clinical data, and a correlation model between the momentum value and the degree of tissue damage is established. In practical applications, mild damage usually corresponds to a slight decrease in heart function (5-10% reduction in ejection fraction), moderate damage corresponds to a moderate decrease in heart function (10-20% reduction in ejection fraction), and severe damage corresponds to a severe decrease in heart function (>20% reduction in ejection fraction).

[0143] According to the momentum damage prediction model, the energy parameters and treatment positions of radiofrequency ablation can be determined to achieve a personalized treatment plan. Preferably, the normal damage grade does not require treatment, the mild damage uses low-energy radiofrequency ablation (usually 20-30W), the moderate damage uses moderate energy (30-40W), and the severe damage uses high energy (40-50W). The treatment position is preferably selected in the momentum concentration area, usually the center of the vortex core.

[0144] The multi-dimensional visualization rendering technology displays the vortex field of the heart valve regurgitation: this step realizes the intuitive and comprehensive display of the characteristics of the vortex field, and enhances the feasibility of clinical application.

[0145] The multi-dimensional visualization rendering technology first realizes three-dimensional streamline visualization. The generation of streamlines uses the fourth-order Runge-Kutta integral method, which starts from the preset starting point and integrates along the velocity field direction to generate. In order to clearly display the vortex structure, usually only 10-20 representative streamlines are displayed to avoid visual confusion.

[0146] The present application adopts a color coding mechanism to represent hemodynamic parameters with different colors. Preferably, a rainbow color spectrum (from blue to red) is used to map flow velocity magnitude, with blue representing low velocity (typically <0.2 m / s) and red representing high velocity (typically >1.0 m / s); another color spectrum (from green to purple) is used to map rotation intensity, with green representing low rotation intensity (typically <0.2 m / s) and purple representing high rotation intensity (typically >1.0 m / s).

[0147] To highlight key structures, the present application adopts a transparency mapping technique. Preferably, the core body region is set to be semi-transparent (transparency 0.3-0.5) so that internal streamlines are visible; the surrounding region is set to high transparency (transparency 0.8-0.9) as a reference background only. This transparency setting can effectively highlight the vortex core structure while maintaining the visibility of the overall spatial relationship.

[0148] The present application also implements a dynamic evolution animation to demonstrate the changes of the vortex over the cardiac cycle. Preferably, a frame rate of 10-15 frames / s is used to cover the complete cardiac cycle (typically 10-15 frames). The animation can be played in a loop or paused for detailed observation at key phases (e.g. mid-systole, end-systole).

[0149] To facilitate clinical application, the present application provides an interactive exploration and analysis interface, supporting functions such as multi-view observation, phase control, parameter adjustment, and region zooming. These interactive functions greatly enhance the practicality of the system, allowing doctors to flexibly adjust the observation method and in-depth analyze the characteristics of the vortex field according to their needs.

[0150] As shown in Figure 2 , the present application also provides a heart valve regurgitation vortex field 4D Flow MRI visualization system, including the following functional modules:

[0151] The data acquisition module 1 is used to obtain heart 4D Flow MRI data, including heart four-chamber function sequence images, mitral valve flow velocity encoding images, and mitral valve three-geometry curve images. This module interfaces with the MRI device, receives raw data and performs preliminary processing, including signal denoising, image reconstruction, and format conversion operations.

[0152] The image preprocessing module 2 is used to perform image fusion of the heart four-chamber function sequence images, mitral valve flow velocity encoding images, and mitral valve three-geometry curve images through heart anatomical structure registration technology. This module realizes accurate registration of multi-modal medical images, ensuring that the data basis for subsequent analysis is accurate and reliable.

[0153] ​​The mesh construction module 3 is used to extract the two-dimensional contour of the heart surface based on the image fusion result, and construct a two-dimensional regurgitation heart mesh model. This module is responsible for the geometric expression of the heart anatomy, providing a spatial reference framework for subsequent vortex field analysis.

[0154] The valve positioning module 4 is used to determine the position of the valve leaflet and the start and end points of the flow line using the three geometric curves of the mitral valve. The role of this module is to accurately position the mitral valve and its surrounding area, providing key anatomical reference points for vortex field analysis.

[0155] The vortex identification module 5 is used to identify vortex structures based on the feature vector space analysis method, which includes constructing a velocity gradient tensor, calculating eigenvalues, and performing a multi-threshold adaptive vortex judgment. This is the core analysis module of the system, responsible for accurately identifying vortex structures from complex flow fields.

[0156] The core line tracking module 6 is used to extract vortex core lines through a multi-startpoint vortex core line tracking technique, which includes a start point selection mechanism, a bidirectional tracking strategy, and a path merging technique. This module realizes the accurate extraction of vortex core lines, laying the foundation for core body reconstruction.

[0157] The core body reconstruction module 7 is used to reconstruct the vortex core body based on the vortex core line using a non-tracking seed algorithm, which includes seed point placement, region growth control, and adaptive mesh subdivision. This module expands the line structure to a body structure, fully depicting the three-dimensional morphology of the vortex.

[0158] The momentum analysis module 8 is used to calculate the momentum of the vortex and establish a momentum damage prediction model according to the geometry and flow field parameters of the vortex core body. This module realizes the quantitative evaluation of the dynamics of the vortex, providing a scientific basis for clinical decision-making.

[0159] The visualization module 9 is used to display the heart valve regurgitation vortex field through multi-dimensional visualization rendering technology, which includes three-dimensional flow line visualization, color coding mechanism, and dynamic evolution animation. This module is the output interface of the system, which converts complex analysis results into intuitive and understandable visual forms.

[0160] The data flow sequence between the modules is consistent with the method steps in Embodiment 1, forming a complete technical chain. The system software architecture adopts modular design, and each functional module can be independently updated and optimized, facilitating the maintenance and upgrade of the system.

[0161] In terms of hardware platform, the system can be deployed on a standard medical image workstation, with recommended configuration: Intel Xeon processor (8 cores or more), 32 GB or more memory, 8 GB or more professional graphics card, and Windows 10 Professional or Linux operating system. The system supports integration with the hospital PACS system, facilitating the import of clinical data and the archiving of results.

[0162] The heart valve regurgitation vortex field 4D flow MRI visualization system provided by the application has good expansibility, and new function modules such as an automatic diagnosis module based on machine learning and a treatment planning module can be added as needed, so that the clinical application value of the system is further enhanced.

[0163] The above merely provides a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any skilled person in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements shall be encompassed within the protection scope of the present application.

Claims

1. A method for 4D Flow MRI visualization of heart valve regurgitation vortex fields, characterized in that, The method comprises the following steps: acquiring heart 4D Flow MRI data, wherein the heart 4D Flow MRI data comprises a heart four-chamber function sequence image, a mitral valve flow velocity encoding image and a mitral valve three-geometry curve image; image fusion is performed on the heart four-chamber function sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image through a heart anatomical structure registration technology to obtain an image fusion result; based on the image fusion result, a heart surface two-dimensional contour is extracted to construct a two-dimensional regurgitation heart mesh model; the mitral valve three-geometry curve is used to determine a valve leaflet position and a flow line starting point and an ending point; a vortex structure is identified based on a feature vector space analysis method, wherein the feature vector space analysis method comprises constructing a velocity gradient tensor, calculating a feature value and performing a multi-threshold adaptive vortex judgment; a vortex core line is extracted through a multi-starting-point vortex core line tracking technology, wherein the multi-starting-point vortex core line tracking technology comprises a starting point selection mechanism, a bidirectional tracking strategy and a path merging technology; based on the vortex core line, a vortex core body is reconstructed using a non-tracking seed algorithm, wherein the non-tracking seed algorithm comprises seed point arrangement, region growth control and adaptive mesh subdivision; vortex momentum is calculated and a momentum damage prediction model is established according to the geometry and flow field parameters of the vortex core body; a heart valve regurgitation vortex flow field is displayed through a multi-dimensional visualization rendering technology, wherein the multi-dimensional visualization rendering technology comprises three-dimensional flow line visualization, a color coding mechanism and a dynamic evolution animation.

2. The method of visualizing cardiac valve regurgitation vortex flow fields 4D Flow MRI according to claim 1, characterized in that, The acquiring heart 4D Flow MRI data specifically comprises: acquiring a heart 4D Free Breath Stack of Slices Cine MRI sequence image through a magnetic resonance scanner; based on the heart 4D Free Breath Stack of Slices Cine MRI sequence image, a heart four-chamber function sequence image and a mitral valve flow velocity encoding image are obtained; based on the heart four-chamber function sequence image and the mitral valve flow velocity encoding image, a mitral valve three-geometry curve image is obtained; wherein the time resolution of the heart four-chamber function sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image is 1 second, and the spatial resolution is 1*1.25*1.25mm.

3. The method of visualizing cardiac valve regurgitation vortex flow fields 4D Flow MRI of claim 2, wherein, The 4D Flow MRI sequence used by the magnetic resonance scanner is a multi-echo velocity encoding gradient echo image acquisition module, which obtains the heart four-chamber function sequence image and the mitral valve flow velocity encoding image at a velocity encoding of 1.5m / s; wherein the image parameters of the heart four-chamber function sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image comprise: matrix 512*304, slice number 20, layer thickness 6mm, TR 3.08ms, TE 1.16ms, inversion angle 25°, bandwidth ±12.5kHz.

4. The method of visualizing cardiac valve regurgitation vortex flow fields 4D Flow MRI of claim 1, wherein, The heart anatomical structure registration technology specifically comprises: The image registration is performed by using a synchronous acquisition window between the cardiac four-chamber functional sequence image and the mitral valve three-geometry curve image through the cardiac 4D Flow MRI data processing system; A hierarchical registration strategy is adopted to sequentially perform global alignment based on the overall contour of the heart, regional alignment based on the heart cavity and main blood vessel structure, and local fine alignment based on the mitral valve and its surrounding structure; The image registration error between the cardiac four-chamber functional sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image is less than 1 mm.

5. The method of visualizing heart valve regurgitation vortex flow field 4D Flow MRI of claim 1, wherein, The constructing the two-dimensional regurgitation heart mesh model specifically comprises: Different heart surface contours are extracted from the cardiac four-chamber functional sequence image, the mitral valve flow velocity encoding image and the mitral valve three-geometry curve image respectively, wherein the left ventricular surface contour is extracted from the cardiac four-chamber functional sequence image, the mitral valve surface contour is extracted from the mitral valve flow velocity encoding image, and the left atrial surface contour is extracted from the mitral valve three-geometry curve image; Based on the extracted mitral valve geometry surface contour, the mitral valve geometry surface contour is subjected to grid line densification; The densified lines are smoothly connected by using a curve smoothing connection technology to form a two-dimensional mesh heart model, wherein the curve smoothing connection technology realizes the continuity and smoothness of the curve by a control point interpolation method.

6. The method of visualizing heart valve regurgitation vortex flow field 4D Flow MRI of claim 1, wherein, The feature vector space analysis method specifically comprises: The spatial gradient of the velocity field is calculated on the spatial discrete points to form a velocity gradient tensor; The velocity gradient tensor of each point is subjected to feature decomposition to obtain eigenvalues and eigenvectors; A judgment condition is constructed based on the mathematical relationship between the eigenvalues, and when the eigenvalues satisfy the set condition, the discrete point is determined as a vortex structure point; A multi-level threshold vortex possibility level is formed by dynamically adjusting the judgment threshold according to the local flow field characteristics; The final vortex region is determined by comprehensively judging the eigenvalue relationship and the rotation intensity.

7. The method of visualizing heart valve regurgitation vortex flow fields 4D Flow MRI of claim 1, wherein, The multi-startpoint vortex core line tracking technology specifically comprises: Based on the vortex intensity, multiple potential starting points are selected, and tracking is simultaneously performed in two directions from each starting point; During the tracking process, the tracking step is dynamically adjusted according to the local curvature, and a step of 0.2 mm is used in the region with large curvature and a step of 1.0 mm is used in the region with small curvature; The direction of the next step is comprehensively predicted by combining the trajectories of the previous steps, the local flow field characteristics and the vortex eigenvalue direction; The cumulative error is corrected in real time during the tracking process to ensure the accuracy of long-path tracking; The paths tracked from different starting points are intelligently merged to form a complete vortex core line.

8. The method of visualizing heart valve regurgitation vortex flow field 4D Flow MRI of claim 1, wherein, The non-tracking seed algorithm for reconstructing the vortex core body specifically comprises: Points on the vortex core line are taken as seed points to form an initial core body region by outward expansion; The region expansion boundary is controlled based on the vortex judgment condition to ensure that the expanded region satisfies the vortex characteristics; The space is non-uniformly subdivided, and higher grid density is used in the key region; The flow distribution in the expanded region is calculated to ensure the integrity and accuracy of the core body; Through the iterative subdivision and evaluation process, an accurate vortex core body is finally formed.

9. The method of visualizing heart valve regurgitation vortex flow field 4D Flow MRI of claim 1, wherein, The calculation of the vortex momentum and the establishment of the momentum damage prediction model specifically comprise: The geometric parameters of the vortex core body are calculated, including the area, the perimeter and the volume; Calculate the flow field characteristic parameters in the vortex core body, including average velocity and rotational intensity; Calculate the comprehensive momentum index based on the geometric parameters and flow field characteristic parameters; Divide the comprehensive momentum index into different damage levels, including normal, mild, moderate and severe; Determine the critical value of each level based on clinical data, and establish the correlation model of momentum value and tissue damage degree; Determine the energy parameters and treatment position of radiofrequency ablation according to the momentum damage prediction model, and realize personalized treatment plan.

10. Heart valve regurgitation vortex flow field 4D Flow MRI visualization system for implementing the heart valve regurgitation vortex flow field 4D Flow MRI visualization method according to any one of claims 1 to 9, characterized in that, Comprise: A data acquisition module for acquiring heart 4D Flow MRI data, including heart four-chamber function sequence images, mitral valve flow velocity encoding images and mitral valve three-geometry curve images; An image preprocessing module for image fusion of the heart four-chamber function sequence images, mitral valve flow velocity encoding images and mitral valve three-geometry curve images through heart anatomical structure registration technology; A mesh construction module for extracting the two-dimensional contour of the heart surface based on the image fusion results, and constructing a two-dimensional regurgitation heart mesh model; A valve positioning module for determining the position of the valve leaflet and the starting point and ending point of the flow line by using the mitral valve three-geometry curve; A vortex identification module for identifying vortex structure based on a feature vector space analysis method, which includes constructing a velocity gradient tensor, calculating eigenvalues and performing a multi-threshold adaptive vortex judgment; A core line tracking module for extracting vortex core line through a multi-start vortex core line tracking technology, which includes a start point selection mechanism, a bidirectional tracking strategy and a path merging technology; A core body reconstruction module for reconstructing the vortex core body based on the vortex core line using a non-tracking seed algorithm, which includes seed point arrangement, region growth control and adaptive mesh subdivision; A momentum analysis module for calculating vortex momentum and establishing a momentum damage prediction model according to the geometric and flow field parameters of the vortex core body; A visualization module for displaying the heart valve regurgitation vortex flow field through a multi-dimensional visualization rendering technology, which includes three-dimensional flow line visualization, color coding mechanism and dynamic evolution animation.

Images