Left ventricular compliance assessment method and system based on medical image

Abstract

The invention relates to the technical field of medical image analysis, in particular to a left ventricular compliance assessment method and system based on medical image.The method comprises the following steps that a planar three-dimensional coordinate from a left ventricular apex to a mitral annulus is obtained through a CT image, a thin-plate spline interpolation algorithm is called to establish a left ventricular long-axis model, the ultrasonic end-stage internal diameter is obtained synchronously, and a left ventricular compliance assessment result is obtained. And inputting the left greenhouse long axis model and the ultrasonic diastole end-stage inner diameter into an Euclidean distance calculation function to generate a left greenhouse length-diameter ratio. According to the method, a left greenhouse long axis model is constructed by integrating CT three-dimensional coordinates and ultrasonic inner diameter, the length-diameter ratio is calculated by combining Euclidean distance, point cloud registration is carried out, a myocardial wall area is segmented, local deformation is quantized by utilizing Gaussian curvature gradient, a correction factor is generated by inputting a radial basis function network, and a parameter coupling mechanism is established by fusing an aorta root angle. And optimizing the weight of the pressure-volume equation, establishing a dynamic model, and realizing a quantitative evaluation system of spatial heterogeneity and time continuity.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image analysis, and in particular to a left ventricular compliance assessment method and system based on medical images. Background Art

[0002] The field of medical image analysis technology includes a technical system that uses medical imaging technology to obtain information on human anatomical structure and physiological function and conduct quantitative analysis. Its core content involves key technical links such as data acquisition of image acquisition equipment, image reconstruction algorithm optimization, anatomical structure feature extraction, and pathophysiological parameter calculation. It focuses on solving the subjective limitations of traditional medical image qualitative interpretation and achieving accurate quantitative evaluation from morphology to function. In the cardiovascular field, it is mainly used in clinical diagnostic support scenarios such as cardiac chamber volume measurement, myocardial motion tracking, and hemodynamic parameter calculation.

[0003] Among them, the left ventricular compliance assessment method refers to a technical solution for quantitatively analyzing the left ventricular diastolic function characteristics based on preoperative medical imaging data. This solution integrates the left ventricular end-diastolic internal diameter and relative wall thickness parameters measured by echocardiography, the aortic valve calcification score and valve ring diameter data obtained by CT scan, and combines basic clinical characteristics such as patient age and gender to establish a standardized parameter assignment system. A multivariate regression analysis method is used to construct a mathematical model of the left ventricular diastolic pressure-volume relationship, and finally a graded and quantifiable left ventricular stiffness assessment index is formed. Its technical implementation relies on the systematic coupling analysis of medical imaging measurement data and clinical parameters.

[0004] Existing technologies rely on single-dimensional analysis of ultrasound inner diameter and wall thickness parameters, and lack CT three-dimensional spatial coordinates to model the long-axis morphology of the heart, resulting in planarization errors in the evaluation of ventricular geometric reconstruction. Traditional methods use a single physiological parameter to construct a mathematical model, but fail to achieve systematic fusion of multi-source imaging features. The parameter assignment system is limited by the inherent spatial resolution deviation of single-modality imaging. The existing evaluation system uses a holistic ventricular wall thickness indicator, ignoring the difference in curvature gradient between the anterior septum and the lateral wall, resulting in insufficient sensitivity in the evaluation of local stiffness. The influence of the anatomical structure of the aortic root on ventricular filling pressure is not included in the calculation model, resulting in systematic deviations in the evaluation of diastolic function. The existing regression analysis method uses a fixed weight coefficient and cannot dynamically respond to the deformation process of the heart during diastole, reducing the ability to identify early functional abnormalities. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a left ventricular compliance assessment method and system based on medical imaging.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for evaluating left ventricular compliance based on medical imaging, comprising the following steps: S1: Obtain the three-dimensional coordinates of the plane from the left ventricular apex to the mitral valve annulus through CT imaging, call the thin plate spline interpolation algorithm to establish a left ventricular long axis model, synchronously obtain the ultrasound end-diastolic internal diameter, and input the left ventricular long axis model and the ultrasound end-diastolic internal diameter into the Euclidean distance calculation function to generate the left ventricular aspect ratio; S2: Based on the left ventricular aspect ratio, an adaptive volume compensation model is used to register the CT left ventricular cavity point cloud data, the anterior septum and lateral wall regions are segmented using a region growing algorithm, and the curvature gradient difference value is extracted using a Gaussian curvature calculation model; S3: performing a ratio operation on the curvature gradient difference value and the ultrasound ventricular septum thickness and side wall thickness, inputting the calculated values ​​into a radial basis function network to generate a regional stiffness correction factor, and simultaneously obtaining the aortic root inclination angle of CT angiography; S4: Based on the regional stiffness correction factor and the aortic root inclination angle, the ultrasound E / e' ratio is input into a multi-parameter coupling model, and the weight coefficient of the left ventricular filling pressure equation is iteratively calculated through a gradient descent optimization algorithm to output a left ventricular compliance assessment value.

[0007] As a further solution of the present invention, the left ventricular aspect ratio is specifically the geometric ratio of the long axis distance to the end-diastolic internal diameter, the curvature gradient difference value includes the anterior septal curvature distribution characteristics and the side wall curvature distribution characteristics, the regional stiffness correction factor specifically refers to the product term of the ratio of the ventricular septum to the side wall thickness and the curvature gradient, the aortic root inclination angle includes the coronal plane projection angle and the sagittal plane projection angle, and the left ventricular compliance assessment value includes the filling pressure weight coefficient and the multimodal parameter coupling value.

[0008] As a further solution of the present invention, the thin plate spline interpolation algorithm satisfies the continuity condition of the first-order derivative at the nodes; The Euclidean distance calculation function is defined as: ; in, The left ventricular long axis model The three-dimensional coordinates of the vertices, represents the normalized spatial coordinates corresponding to the end-diastolic diameter of ultrasound, is the total number of model vertices; During the calculation of the curvature gradient difference value, the CT point cloud spatial coordinate unit and the ultrasonic thickness measurement value are uniformly converted into millimeter dimensions; A normalization layer is set before the ratio calculation to eliminate the dimension differences of the curvature gradient difference value and the ventricular septum thickness and the side wall thickness.

[0009] As a further solution of the present invention, the step of obtaining the left ventricular aspect ratio is specifically as follows: S101: Acquire CT image data, detect spatial positioning points of the left ventricular apex and the anterior and posterior mitral valves based on the left ventricular apex and mitral valve annulus planes in the CT image, organize the three-dimensional coordinates of multiple positioning points, summarize the coordinate data into a spatial sequence, and generate a spatial coordinate sequence; S102: Calling a thin plate spline interpolation algorithm to perform long axis curve fitting based on the multiple three-dimensional coordinate data of the spatial coordinate sequence to establish a three-dimensional structure of the long axis of the left ventricle, collecting the left ventricular internal diameter at end-diastole of the ultrasound, and grouping the three-dimensional structural parameters and the inner diameter values ​​into a parameter group to generate a three-dimensional structural parameter group; S103: Based on the three-dimensional structural parameter group, a Euclidean distance calculation function is called to combine the left ventricular long axis model with the ultrasound end-diastolic internal diameter value, perform Euclidean distance calculation between each parameter, and obtain the left ventricular aspect ratio.

[0010] As a further solution of the present invention, the step of obtaining the curvature gradient difference value is specifically as follows: S201: Based on the left ventricular aspect ratio, register the left ventricular cavity point cloud data acquired by CT using an adaptive volume compensation model, match the compensation model parameters with the left ventricular aspect ratio value according to the three-dimensional spatial distribution of the point cloud, adjust the spatial offset between multiple point clouds, and generate a registration spatial coordinate set; S202: Based on the registration spatial coordinate set, a region growing algorithm is applied to identify the front septum region and the side wall region, a region growing starting point is selected, and based on the density and topological structure characteristics of the spatial coordinates, the partitions to which the multiple points belong are determined, the point clouds are grouped into spatial partitions, and a partitioned point cloud set is obtained; S203: Based on the partitioned point cloud set, using the Gaussian curvature calculation model, select the three-dimensional spatial neighborhood of multiple points in the partition, analyze the trend of surface normal changes between the point sets, and perform curvature gradient calculation through the numerical distribution of adjacent normal angles between points to obtain the curvature gradient difference value.

[0011] As a further solution of the present invention, in the Gaussian curvature calculation model, the curvature gradient is associated with the spatial distribution characteristics of the myocardial fiber orientation.

[0012] As a further solution of the present invention, the acquisition step of S3 is specifically as follows: S301: obtaining the original data of the curvature gradient difference value, detecting the thickness measurement values ​​of the ventricular septum and the lateral wall in the ultrasound image, calculating the ratio of the curvature gradient difference value to the ventricular septum thickness and the lateral wall thickness respectively using a point-by-point division operation, establishing a ventricular septum thickness ratio parameter and a lateral wall thickness ratio parameter, fusing the two parameters using a weighted average algorithm to generate a ventricular wall thickness ratio parameter; S302: calling the wall thickness ratio parameter as a network input layer node, constructing a radial basis function network hidden layer Gaussian kernel function, calculating the Euclidean distance between the input parameter and the hidden layer center point, activating the hidden layer nodes through an exponential function, calculating the output layer weight coefficient using the least squares method, performing a linear weighted sum operation, and generating a regional stiffness correction factor; S303: Calling DICOM format image data of CT angiography, extracting the aortic root contour through a threshold segmentation algorithm, establishing a three-dimensional rectangular coordinate system, and calculating the angle between the aortic root central axis and the spatial vector of the horizontal plane to generate the aortic root inclination angle.

[0013] As a further solution of the present invention, the step of obtaining the left ventricular compliance evaluation value is specifically as follows: S401: Obtain the regional stiffness correction factor and the aortic root inclination angle, input the ultrasound E / e' ratio into a multi-parameter coupling model, establish a parameter correlation matrix through matrix determinant operations, use the least squares method to eliminate collinear interference between parameters, and generate an initial value of the filling pressure weight; S402: calling the initial value of the filling pressure weight, constructing the residual function of the left ventricular filling pressure equation, calculating the partial derivative of the weight coefficient using a gradient descent algorithm, updating the weight parameter along the negative gradient direction, terminating the iteration when the residual sum of squares is less than a convergence threshold, and obtaining the optimized value of the weight coefficient; S403: Substitute the optimized value of the weight coefficient into the left ventricular filling pressure equation, calculate the convolution integral of the myocardial stress tensor and the ventricular volume change rate, extract the principal strain component through eigenvalue decomposition, and output the left ventricular compliance evaluation value.

[0014] As a further embodiment of the present invention, the myocardial stress tensor is defined as a stress distribution matrix of myocardial tissue in three-dimensional space; The ventricular volume change rate is the rate of change of the left ventricular cavity volume per unit time.

[0015] A left ventricular compliance assessment system based on medical imaging, wherein the left ventricular compliance assessment system based on medical imaging is used to implement the above-mentioned left ventricular compliance assessment method based on medical imaging, and the system comprises: a long-axis modeling module for obtaining the three-dimensional coordinates of the left ventricular apex and the mitral valve annulus plane through CT images, constructing a left ventricular long-axis geometric model using a thin-plate spline interpolation algorithm, synchronously obtaining ultrasound end-diastolic internal diameter data, inputting the vertex coordinates of the long-axis geometric model and the ultrasound internal diameter data into a Euclidean distance calculation function to generate a left ventricular aspect ratio, and transmitting the left ventricular aspect ratio to a curvature registration module; a curvature registration module, configured to perform adaptive volume compensation model registration on the CT left ventricular cavity point cloud data using the left ventricular aspect ratio, segment the anterior septum and lateral wall anatomical regions using a region growing algorithm, extract the curvature distribution characteristics of the anterior septum and lateral wall using a Gaussian curvature calculation model, calculate a curvature gradient difference value between the two regions, and transmit the curvature gradient difference value to the stiffness correction module; a stiffness correction module, configured to perform a ratio operation on the curvature gradient difference value and the ultrasonic ventricular septum thickness and side wall thickness, input the operation result into a radial basis function network to generate a regional stiffness correction factor, simultaneously obtain the aortic root inclination angle measurement value during CT angiography, and transmit the regional stiffness correction factor and the aortic root inclination angle to the compliance coupling module; a compliance coupling module for weighting the ultrasound E / e' ratio using the regional stiffness correction factor, constructing a multi-parameter coupling equation based on the aortic root inclination angle, invoking a gradient descent optimization algorithm to iteratively solve the weight coefficients of early diastolic pressure and atrial systolic pressure in the left ventricular filling pressure equation, and outputting a quantitative assessment value of left ventricular compliance; The input parameters of the multi-parameter coupling equation correspond one-to-one to the regional stiffness correction factor, the aortic root inclination angle, and the ultrasound E / e' ratio.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by integrating CT three-dimensional coordinates and ultrasound inner diameter data, a left ventricular long axis model is constructed by a thin plate spline interpolation algorithm, and the aspect ratio is calculated in combination with the Euclidean distance function to establish a quantitative association between spatial geometric parameters and functional parameters. Based on the adaptive volume compensation model, CT point cloud data is dynamically aligned, the regional growing algorithm is used to segment the myocardial wall area, and the Gaussian curvature gradient difference value is used to quantify the local deformation characteristics to achieve spatial analysis of myocardial motion heterogeneity. The curvature gradient difference and the ultrasound chamber wall thickness ratio are input into the radial basis function network to generate a regional correction factor, and the aortic root inclination angle is synchronously fused to construct a multimodal parameter coupling mechanism to enhance the correlation analysis between anatomical structure and hemodynamics. The weight coefficient of the filling pressure equation is iteratively optimized by the gradient descent algorithm, and a dynamic pressure-volume relationship model is established to solve the problem that the static regression model is insufficient in describing the time-varying characteristics, forming a quantitative evaluation system with both spatial heterogeneity and temporal continuity. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the workflow of the present invention; Figure 2 Flowchart of the steps for obtaining the left ventricular aspect ratio of the present invention; Figure 3 Flowchart of the steps for obtaining the curvature gradient difference value of the present invention; Figure 4 Flowchart of the acquisition step of S3 of the present invention; Figure 5 Flowchart of the steps for obtaining the left ventricular compliance evaluation value of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0019] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0020] Example 1 See also Figure 1 The present invention provides a technical solution: a method for evaluating left ventricular compliance based on medical imaging, comprising the following steps: S1: The three-dimensional coordinates of the plane from the left ventricular apex to the mitral annulus are obtained through CT imaging. The thin plate spline interpolation algorithm is used to establish the left ventricular long axis model. The ultrasound end-diastolic diameter is simultaneously obtained. The left ventricular long axis model and the ultrasound end-diastolic diameter are input into the Euclidean distance calculation function to generate the left ventricular aspect ratio. S2: Based on the left ventricular aspect ratio, an adaptive volume compensation model is used to register the CT left ventricular cavity point cloud data. The anterior septum and lateral wall regions are segmented using a region growing algorithm, and the curvature gradient difference value is extracted using a Gaussian curvature calculation model. S3: The curvature gradient difference value is ratioed with the ultrasound ventricular septum thickness and side wall thickness, and the result is input into the radial basis function network to generate the regional stiffness correction factor, and the aortic root inclination angle of CT angiography is simultaneously obtained; S4: Based on the regional stiffness correction factor and the aortic root inclination angle, the ultrasound E / e' ratio is input into the multi-parameter coupling model, and the weight coefficient of the left ventricular filling pressure equation is iteratively calculated using the gradient descent optimization algorithm to output the left ventricular compliance assessment value.

[0021] The left ventricular aspect ratio is specifically the geometric ratio of the long axis distance to the end-diastolic internal diameter. The curvature gradient difference value includes the anterior septal curvature distribution characteristics and the lateral wall curvature distribution characteristics. The regional stiffness correction factor specifically refers to the product of the ratio of the ventricular septum to the lateral wall thickness and the curvature gradient. The aortic root inclination angle includes the coronal plane projection angle and the sagittal plane projection angle. The left ventricular compliance assessment value includes the filling pressure weight coefficient and the multimodal parameter coupling value.

[0022] The thin plate spline interpolation algorithm satisfies the continuity condition of the first-order derivative at the nodes; The Euclidean distance calculation function is defined as: ; in, The left ventricular long axis model The three-dimensional coordinates of the vertices, represents the normalized spatial coordinates corresponding to the end-diastolic diameter of ultrasound, is the total number of model vertices; In the process of calculating the curvature gradient difference value, the spatial coordinate units of the CT point cloud and the ultrasonic thickness measurement value are uniformly converted into millimeter dimensions; A normalization layer was set before the ratio calculation to eliminate the dimensionless differences in the curvature gradient difference and the ventricular septum thickness and lateral wall thickness.

[0023] See also Figure 2 , the steps for obtaining the left ventricular length-to-diameter ratio are as follows: S101: Acquire CT image data, detect spatial positioning points of the left ventricular apex and the anterior and posterior mitral valves based on the left ventricular apex and mitral valve annulus planes in the CT image, organize the three-dimensional coordinates of multiple positioning points, summarize the coordinate data into a spatial sequence, and generate a spatial coordinate sequence; When implementing this step, first, the original image data obtained by the patient at the end of diastole using a 64-row or higher-row computed tomography (CT) device is retrieved from the picture archiving and communication system (PACS). The data is in DICOM format, with a layer thickness set to 0.625 mm and a reconstruction matrix of 512x512 pixels to ensure that the image quality meets the requirements of subsequent three-dimensional reconstruction and precise measurement. The image data is then imported into a dedicated medical image post-processing workstation.

[0024] On the post-processing workstation, the multi-planar reconstruction (MPR) function is used to simultaneously observe the axial, sagittal, and coronal images of the heart to identify and locate the left ventricular apex. The left ventricular apex is determined by its tip structure that gradually narrows at the end of the cardiac cavity in continuous tomographic images. The most distal point of the cardiac endocardium at the apex is selected and its three-dimensional coordinates are recorded. The coordinates of this apex point are determined to be (X:132.5, Y:98.2, Z:215.0), all in millimeters (mm). Subsequently, the mitral annulus plane is identified, which is the plane between the left atrium and the left ventricle. The anatomical boundaries of the chamber were determined. In the MPR view, the positions of the roots of the anterior and posterior mitral leaflets attached to the valve annulus were clearly identified. The spatial positioning points of the midpoints of the attachment points of the anterior and posterior mitral leaflets were detected respectively, and their three-dimensional coordinates were recorded as the anterior leaflet point (X: 135.8, Y: 120.5, Z: 160.3) mm and the posterior leaflet point (X: 125.2, Y: 123.8, Z: 160.1) mm, respectively. To more accurately describe the morphology of the mitral valve annulus, at least four additional anatomical landmarks were selected along the circumference of the valve annulus, and their coordinates were recorded together.

[0025] The three-dimensional coordinate data of all detected positioning points, including the left ventricular apex point and the various positioning points on the mitral valve annulus, are systematically sorted and these coordinate data are connected to the specific sequence points of the mitral valve annulus in the logical order of the anatomical structure, starting from the apex point, and then sequentially formed into an ordered spatial coordinate sequence. This sequence is , where the ellipsis represents the coordinates of other annular positioning points, thereby generating a spatial coordinate sequence for subsequent model construction.

[0026] Table 1 Three-dimensional coordinate data of left ventricular anatomical positioning points

[0027] As shown in Table 1, the table lists in detail the three-dimensional coordinates of key anatomical positioning points of the left ventricle obtained in a CT image analysis. These data are the basis for subsequent calculation of the long axis length of the left ventricle and morphological analysis.

[0028] S102: Calling a thin plate spline interpolation algorithm to perform long axis curve fitting based on multiple three-dimensional coordinate data of the spatial coordinate sequence to establish the long axis three-dimensional structure of the left ventricle, collect the left ventricular internal diameter at end-diastole of the ultrasound, and group the three-dimensional structural parameters and the inner diameter values ​​into a parameter group to generate a three-dimensional structural parameter group; Based on the spatial coordinate sequence generated in step S101, which includes the three-dimensional coordinates of the left ventricular apex point and multiple positioning points of the mitral valve annulus, the three-dimensional structure of the left ventricular long axis is accurately reconstructed. This process uses thin plate spline interpolation technology, and uses each three-dimensional coordinate point in the spatial coordinate sequence as the control point of interpolation. By solving a mathematical problem that minimizes the bending energy of the interpolation function, a smooth three-dimensional curve passing through all control points is generated. This curve is the three-dimensional geometric representation of the left ventricular long axis, which accurately reflects the actual spatial path from the apex to the center of the mitral valve annulus. By calculating the distance from the apex point on this three-dimensional curve to the center point of the mitral valve annulus (the center point is obtained by averaging the coordinates of multiple positioning points on the valve annulus), For example, the coordinates of the midpoint of the anterior leaflet attachment point, the midpoint of the posterior leaflet attachment point, auxiliary point 1, and auxiliary point 2 in Table 1 are used to calculate the average value, and the coordinates of the center point of the mitral valve annulus are obtained as follows: (135.8+125.2+130.5+128.9) / 4, (120.5+123.8+115.2+128.5) / 4, (160.3+160.1+161.0+159.8) / 4, that is, (130.10, 122.00, 160.30) mm), and the left ventricular long axis length (L_LAX_model) is obtained. In this case, the calculated L_LAX_model is 98.5 mm.

[0029] At the same time, the left ventricular internal diameter at end-diastole (LVIDd) was collected from the patient's two-dimensional echocardiography examination. Based on the peak of the R wave on the electrocardiogram (representing the end-diastolic phase), the LVIDd value was obtained by measuring perpendicularly between the interventricular septum and the inner edge of the left ventricular posterior wall using M-mode ultrasound or two-dimensional ultrasound electronic calipers on the standard parasternal left ventricular long-axis view. In this case, the LVIDd was measured to be 47.0 mm. This measurement follows the standard method recommended by the American Society of Echocardiography.

[0030] Finally, the three-dimensional structural parameters, namely the calculated left ventricular long axis length L_LAX_model (98.5 mm), and the ultrasound-measured left ventricular internal diameter LVIDd value (47.0 mm) at end diastole are combined to form a parameter group containing these two key physiological parameters. The parameter group is specifically expressed as {L_LAX_model: 98.5 mm, LVIDd: 47.0 mm}, thereby generating a three-dimensional structural parameter group for subsequent calculations.

[0031] S103: Based on the three-dimensional structural parameter group, the Euclidean distance calculation function is called to combine the left ventricular long axis model with the ultrasound end-diastolic internal diameter value, complete the Euclidean distance calculation between each parameter, and obtain the left ventricular aspect ratio.

[0032] The three-dimensional structural parameter set generated according to step S102 includes the long axis length L_LAX_model calculated by the left ventricular long axis model, which is 98.5 mm, and the left ventricular internal diameter at end diastole LVIDd measured by ultrasound, which is 47.0 mm. The length of the "left ventricular long axis model" here is directly calculated by the apical coordinates. and the calculated coordinates of the mitral annulus center The spatial distance between the two is refined using the left ventricular apex coordinates in Table 1. mm, and the coordinates of the mitral annulus center calculated from the four points on the mitral annulus in Table 1 mm.

[0033] The exact length of the left ventricular long axis L_LAX is calculated using the three-dimensional Euclidean distance formula: ; Substitute specific values: mm; mm; mm; Calculated in millimeters mm, this calculated The value (59.70 mm) will replace the L_LAX_model value initially obtained in S102 as a more accurate left ventricular long axis length.

[0034] The long axis length of the left ventricle is calculated accurately (59.70 mm) is combined with the obtained ultrasound end-diastolic internal diameter value LVIDd (47.0 mm) to prepare for the ratio calculation and complete the left ventricular length-to-diameter ratio calculation, which is to calculate the left ventricular long axis length. Divide by the left ventricular end-diastolic diameter (LVIDd): Left ventricular length-to-diameter ratio = / LVIDd Substitute the value: Left ventricular aspect ratio = mm / The result of the millimeter calculation is approximately 1.27, which is a pure number without units. It objectively reflects the overall geometric shape of the left ventricle, thereby obtaining the left ventricular aspect ratio.

[0035] See also Figure 3 , the steps for obtaining the curvature gradient difference value are as follows: S201: Based on the left ventricular aspect ratio, an adaptive volume compensation model is used to register the left ventricular cavity point cloud data acquired by CT. According to the three-dimensional spatial distribution of the point cloud, the compensation model parameters are matched with the left ventricular aspect ratio value, and the spatial offset between multiple point clouds is adjusted to generate a registration space coordinate set. Based on the left ventricular aspect ratio obtained in step S103, which is 1.27, and using the three-dimensional point cloud data of the left ventricular inner wall segmented and extracted from the same CT examination (the point cloud data contains approximately 60,000 XYZ coordinate points, accurately depicting the morphology of the left ventricular cavity at the end of diastole), fine alignment of the point cloud data is performed. This alignment process is achieved through an adaptive morphological adjustment process rather than simply applying a specific model name. The process first refers to a left ventricular digital template point cloud with a standard physiological morphology (its inherent aspect ratio, such as 1.60), and compares the left ventricular aspect ratio of 1.27 calculated for the current case with the template aspect ratio of 1.60. Since the aspect ratio of the current case is smaller (1.27<1.60), it indicates that its left ventricle is relatively more spherical, so the standard template needs to be morphologically adjusted.

[0036] When adjusting, the volume of the template point cloud is kept basically constant, and a scaling factor is applied along its long axis, which is approximately , while applying a compensatory scaling factor (approximately , to approximately maintain the volume), so that the overall aspect ratio of the adjusted template approaches 1.27. This non-rigid deformation parameter (i.e., the long-axis scaling factor of 0.79375 and the short-axis scaling factor of 1.122) is the "compensation model parameter", and its setting goal is to match the geometric characteristics of the deformed template with the left ventricular aspect ratio of 1.27 of the target case.

[0037] Subsequently, the template point cloud that has undergone the above morphological adjustment is spatially aligned with the actual left ventricular cavity point cloud data extracted from the patient's CT data. This alignment uses a variant of the iterative closest point (ICP) algorithm. By iteratively calculating and applying a rigid transformation matrix containing three-dimensional translation and three-dimensional rotation, the average distance between the corresponding point pairs between the two point clouds is minimized. Initially, the centroids of the two point clouds may have a spatial offset of about 15 mm. After, for example, 20 iterations, an optimal translation vector ( mm, mm, mm) and rotation matrices (describing the rotation angles around each axis, such as around the X axis , around the Y axis , around the Z axis ), this transformation matrix is ​​applied to the patient's original CT point cloud data to correct its position and posture in three-dimensional space, effectively adjusting the spatial offset between the "multi-point cloud" of the template point cloud and the actual CT point cloud data. Ultimately, the coordinates of each point in the patient's left ventricular cavity CT point cloud are updated to a new coordinate system aligned with the morphologically adjusted template, thereby generating a registration space coordinate set.

[0038] S202: Based on the registration spatial coordinate set, a region growing algorithm is applied to identify the anterior septum region and the sidewall region, a region growing starting point is selected, and based on the density and topological structure characteristics of the spatial coordinates, the partitions to which the multiple points belong are determined. The point cloud is grouped into spatial partitions to obtain a partitioned point cloud set. According to the registration space coordinate set generated in step S201, this data set is a three-dimensional point cloud of the left ventricular cavity after precise alignment and morphological matching, containing about 60,000 points with corrected (X', Y', Z') coordinates. On this basis, the point clouds of the anterior septum region and the lateral wall region in the cardiac cavity are finely identified and segmented. This process is achieved through the region growing method. First, the starting point of the region growth is selected, that is, the seed point is selected. For the identification of the anterior septum region, the operator refers to the standard anatomical orientation and selects a point as the seed point in the middle of the surface of the ventricular septum facing the left ventricular cavity in the registration point cloud. The coordinates of this point are mm (in the registered coordinate system). For the lateral wall region, a seed point is selected at the middle of the inner surface of the opposite left ventricular free wall with coordinates of mm.

[0039] Starting from the selected seed point, based on the density of spatial coordinates and local topological structure characteristics, determine whether its neighboring points should belong to the same partition. Regarding the density of spatial coordinates, a neighborhood search radius of 3.5 mm is set. For a candidate neighboring point, if its nearest distance to an existing point in the current growth area is less than a distance threshold, it is regarded as a potential belonging point. The distance threshold is set based on the average point density of the reference point cloud. The average distance between adjacent points is calculated by local sampling of the point cloud and then multiplied by a coefficient (such as 0.9), which is set to 1.8 mm. When the distance between any point in the point cloud and its nearest neighbor is less than 0.5 mm, the point is considered to be a dense point; when the distance is between 0.5 mm and 1.8 mm, it is a normal density; when it is greater than 1.8 mm, it is a sparse point. Growth gives priority to normal and dense points. To determine the local topological structure features, the impact of the candidate point on the consistency of the local surface normal vector after addition is calculated, the normal vector of the candidate point and its neighborhood is estimated, and compared with the average normal vector of the current growth area. If the angle between the two is less than an angle threshold, the topological structure is considered compatible. This angle threshold is set to 25 degrees. This value is based on the observation of the smooth continuity of the inner wall of the normal cardiac cavity to ensure that the growth process does not mistakenly cross the boundaries or sharp edges of the anatomical structure.

[0040] During the specific execution, for each seed point, the unassigned points within its 3.5 mm neighborhood are checked. If a neighboring point satisfies all the following conditions at the same time: the distance to the nearest point in the current area is less than 1.8 mm, and its addition will not cause the local normal vector to change by more than 25 degrees, then the neighboring point is marked as the same partition as the seed point (anterior septum or lateral wall), and the newly added point is used as a new growth source to continue iterative expansion outward until no neighboring points that meet the conditions can be added in the current partition. In this way, the original left ventricular cavity point cloud is effectively divided into a point cloud subset representing the anterior septum area (containing approximately 10,500 points) and a point cloud subset representing the lateral wall area (containing approximately 11,200 points). These two subsets together constitute the spatial partition, thereby obtaining a partitioned point cloud set.

[0041] S203: Based on the partitioned point cloud set, using the Gaussian curvature calculation model, select the three-dimensional spatial neighborhood of multiple points in the partition, analyze the trend of surface normal changes between the point sets, and perform curvature gradient calculation based on the numerical distribution of adjacent normal angles between points to obtain the curvature gradient difference value; Based on the partitioned point cloud set obtained in step S202, which includes the point cloud of the front septum area (10,500 points) and the point cloud of the sidewall area (11,200 points), local curvature analysis is performed on the point cloud data in these two partitions. For each point in the partition , first select its 3D spatial neighborhood, which is determined by its nearest Neighbor points are used to define The value of is set according to the point cloud density and the desired curvature scale. Here, , that is, select the 25 nearest points around each point Constitute its local point set.

[0042] Then, for each point and its corresponding local point set, by fitting a quadratic surface to the local point set (In the local coordinate system To estimate its Gaussian curvature , the Gaussian curvature is calculated from the quadratic coefficient obtained by fitting: , unit is mm ,By this method, a Gaussian curvature value is calculated for each point in the anterior septal region and the lateral wall region.

[0043] Next, calculate the curvature gradient, which reflects the rate and direction of change of curvature in space. , whose curvature gradient is a vector whose magnitude By calculation The Gaussian curvature value of and all points in its neighborhood The Gaussian curvature value of The ratio of the weighted difference between the two values ​​to the distance is used to estimate the The calculated The value is fitted with a local linear function, and the gradient of the function is Estimate of the size in mm , calculate the arithmetic mean of the curvature gradient of all points in the anterior septum area, and get the average anterior septum curvature gradient , calculated mm , similarly, the average sidewall curvature gradient is calculated , calculated mm .

[0044] Finally, the absolute difference between the average curvature gradient values ​​of the two regions is calculated as the curvature gradient difference value: Curvature gradient difference value = Substitute the values: Curvature gradient difference = mm = mm , thereby obtaining the curvature gradient difference value.

[0045] In the Gaussian curvature calculation model, the curvature gradient is associated with the spatial distribution characteristics of the myocardial fiber orientation.

[0046] See also Figure 4 , the specific steps for obtaining S3 are: S301: obtaining raw data of curvature gradient difference values, detecting thickness measurements of the ventricular septum and lateral wall in ultrasound images, calculating ratios of the curvature gradient difference values ​​to the ventricular septum thickness and lateral wall thickness using point-by-point division, establishing a ventricular septum thickness ratio parameter and a lateral wall thickness ratio parameter, and fusing the two parameters using a weighted average algorithm to generate a ventricular wall thickness ratio parameter; Get the original data of the curvature gradient difference value calculated in step S203, which is At the same time, the end-diastolic thickness measurements of the ventricular septum and lateral wall (or left ventricular posterior wall, here referring to the lateral wall) were retrieved from the two-dimensional echocardiography examination report of the same patient. These thickness values ​​were obtained by accurately measuring the distance from the inner edge to the outer edge of the myocardium using an electronic caliper at the level of the papillary muscle in the parasternal short-axis section at the peak phase of the R wave of the electrocardiogram. The patient's ventricular septum thickness IVSth was measured to be 14.0 mm, and the lateral wall thickness PWth was measured to be 9.5 mm.

[0047] Next, the curvature gradient difference value is calculated by ratio with the ventricular septum thickness and the side wall thickness respectively. First, the ventricular septum thickness ratio parameter (Ratio_IVS) is calculated: Ratio_IVS = curvature gradient difference value / IVSth Substitute the value: Ratio_IVS = Then calculate the side wall thickness ratio parameter (Ratio_PW): Ratio_PW=curvature gradient difference value / PWth Substitute the value: Ratio_PW= , retain four significant figures for The two calculated ratio parameters, namely the ventricular septum thickness ratio parameter and sidewall thickness ratio parameters , was established.

[0048] Finally, these two ratio parameters are fused through weighted average operation to generate a single wall thickness ratio parameter (WallThicknessRatioParam), which is calculated as follows: WallThicknessRatioParam= Weight coefficient and The setting is based on previous studies and expert consensus on the impact of different ventricular wall area lesions on the overall left ventricular function, ensuring Considering that the structural and functional characteristics of the ventricular septum are slightly more important than the free lateral wall in the assessment of ventricular stiffness, ,but These weights were determined based on a retrospective analysis of data from 300 patients with various heart diseases. An optimization algorithm was used to find the weight combination that maximized the distinction between normal and abnormal diastolic function. Ultimately, a septum weight of 0.55 was determined to be optimal, with an area under the receiver operating characteristic curve (AUC) of 0.85. Substituting these values ​​into the calculations: WallThicknessRatioParam= ; WallThicknessRatioParam= ; WallThicknessRatioParam= , retain four significant figures for , thus generating the wall thickness ratio parameter. To facilitate subsequent network input, this parameter is multiplied by Scale to obtain unitless input value .

[0049] S302: Calling the wall thickness ratio parameter as the network input layer node, constructing the radial basis function network hidden layer Gaussian kernel function, calculating the Euclidean distance between the input parameter and the hidden layer center point, activating the hidden layer nodes through the exponential function, calculating the output layer weight coefficient using the least squares method, performing a linear weighted sum operation, and generating a regional stiffness correction factor; The wall thickness ratio parameter after scale adjustment generated in step S301 is called, and its value is , this value is used as a single node input of the radial basis function (RBF) neural network. The RBF network structure is pre-set, and its hidden layer contains three neurons. Each neuron uses a Gaussian kernel function as its activation function. Gaussian kernel function The mathematical expression is ,in It is The center point of the hidden layer neurons, are its width parameters. These parameters (center point and width) are based on unsupervised learning (e.g., K-means clustering algorithm to determine the center point) of a training dataset containing 250 patients with known wall thickness ratio parameters and corresponding clinically assessed regional stiffness grades. , and set the width according to the multiple of the standard deviation of the sample distribution of each cluster ) and subsequent supervised fine-tuning to ensure the representativeness and discrimination ability of the network. The specific parameter values ​​are shown in the table below.

[0050] Table 2 Parameters of hidden layer neurons in radial basis function network

[0051] As shown in Table 2, the table lists the preset center points and width parameters of the three Gaussian kernel functions of the hidden layer of the RBF network in this embodiment.

[0052] Next, calculate the input parameters With the center point of each hidden layer neuron The square of the Euclidean distance, that is : For neurons : ; For neurons : ; For neurons : ; Then, the activation output of each hidden layer neuron is calculated by the exponential function : ; ; ; The weight coefficient of the output layer It is determined by applying the least squares method to the above 250 training data sets during the network training phase. The goal is to minimize the mean square error between the regional stiffness correction factor predicted by the network and the actual clinical assessment grade, solving the linear equation system (in is the hidden layer output matrix of the training sample, is the corresponding target stiffness correction factor vector, is the weight vector to be found), we get , the output layer weight coefficient determined by this process is , , ,These weights were cross-validated, and the Kappa value of the consistency between the prediction results and the expert evaluation was 0.78 on an independent test set of 50 cases, indicating that the model has good generalization ability.

[0053] Finally, a linear weighted sum operation is performed to calculate the final output of the network, which is the regional stiffness correction factor : ; Substituting the values: ; Keep three significant figures , thereby generating a regional stiffness correction factor.

[0054] S303: Calling DICOM format image data of CT angiography, extracting the aortic root contour through a threshold segmentation algorithm, establishing a three-dimensional rectangular coordinate system, and calculating the angle between the aortic root central axis and the spatial vector of the horizontal plane to generate the aortic root inclination angle.

[0055] The DICOM format image data of the thoracic aorta CT angiography (CTA) obtained during the same examination period were retrieved. This data was obtained through rapid scanning technology after intravenous injection of iodine contrast agent, with a layer thickness of 0.75 mm, and can clearly show the three-dimensional anatomical structure of the aorta and its root. First, in order to accurately extract the three-dimensional contour of the aortic root, a threshold segmentation method based on Hounsfield units (HU) was used. The aorta cavity shows high density due to the filling of contrast agent, and its HU value is usually between 200HU and 550HU. This HU range was set as the segmentation Threshold, the HU value of each voxel in the CTA image is compared with this threshold range. All voxels with HU values ​​within this range are preliminarily marked as belonging to the aortic vascular structure. For a voxel with a HU value of 380, it is classified as the aorta, while the voxel with a HU value of 80 is excluded. This preliminary segmentation result is further optimized by three-dimensional connected region analysis, retaining only the connected region with the largest volume and consistent with the anatomical position of the aorta. Morphological closing operation (dilation followed by erosion) is used to fill small internal holes and smooth the contour edges, thereby obtaining an accurate three-dimensional point cloud contour of the aortic root.

[0056] Based on the extracted aortic root contour, a local three-dimensional rectangular coordinate system is established to perform standardized angle measurement. The origin of the coordinate system is It is set at the geometric center of the aortic valve annulus. This center is obtained by fitting a plane to the aortic valve annulus contour point set (usually three aortic valve intersection points) and calculating the center of the intersection ring between the plane and the starting segment of the aortic root. The Z axis is defined as perpendicular to the aortic valve annulus plane and pointing in the direction of ascending aortic blood flow. The X and Y axes are perpendicular to each other in the annulus plane and are determined according to the standard anatomical orientation (such as the X axis points to the patient's left side and the Y axis points to the patient's front).

[0057] Then, the central axis of the aortic root was calculated. From the starting point of the central axis (the center of the valve annulus), a cross-sectional profile of the ascending aorta perpendicular to the local Z axis was obtained every 1.0 mm within a length range of 30 mm along the ascending aorta. The geometric center point of each cross-sectional profile was calculated. These consecutive center points were connected to form a three-dimensional spatial curve representing the direction of the aortic root, which is the central axis of the aortic root. The first point on this central axis close to the valve annulus was taken. mm and a point 20 mm from the annulus Millimeters (all coordinates in the local coordinate system), which defines the direction vector of the aortic root axis .

[0058] Define the horizontal plane when the patient is scanned (usually corresponding to the plane of the CT scanning bed). In the standard CT coordinate system, the normal vector of the horizontal plane is usually in the Z-axis direction, that is, (Assuming the CT scanner's Z axis is vertically upward), calculate the aortic root midline vector Normal vector to this horizontal plane The angle between , this angle is calculated by the vector dot product formula:

[0059] ; ; so but , this angle The angle between the central axis of the aortic root and the vertical direction (normal direction of the horizontal plane) is the desired aortic root inclination angle, thus generating the aortic root inclination angle as .

[0060] See also Figure 5 , the steps for obtaining the left ventricular compliance assessment value are as follows: S401: Obtain the regional stiffness correction factor and the aortic root inclination angle, input the ultrasound E / e' ratio into the multi-parameter coupling model, establish a parameter correlation matrix through matrix determinant operations, use the least squares method to eliminate collinearity interference between parameters, and generate the initial value of the filling pressure weight; Obtain the regional stiffness correction factor calculated in step S302, which is 0.608, and the aortic root inclination angle calculated in step S303, which is At the same time, the E / e' ratio was obtained from the patient's echocardiographic examination report. This ratio was calculated by dividing the peak velocity of mitral valve early diastole, E (measured E = 0.95 m / s), by the average of the peak velocity of mitral valve early diastole, e', on the septal and lateral sides (measured average e' = 0.065 m / s, i.e., 6.5 cm / s). First, the units were unified and the E wave velocity was converted to cm / s: , then the E / e' ratio is .

[0061] The three core parameters: regional stiffness correction factor (Factor = 0.608, unitless), aortic root inclination angle (Angle = , unit: degrees), and the ultrasound E / e' ratio (E_e_prime = 14.62, unitless), along with other clinical parameters that have potential influence on left ventricular filling pressures, such as patient age (Age = 65 years) and left atrial volume index (LAVI = 38 ml / m ), together as a set of input features.

[0062] Table 3 Multi-parameter coupling model input characteristic data

[0063] As shown in Table 3, the table lists the key input features and their specific values ​​used for subsequent model analysis in this embodiment.

[0064] To explore these parameters ( Representing the above five characteristics respectively) and the relationship between left ventricular filling pressure (target variable Y, measured by gold standard such as right cardiac catheterization, unit mmHg), and providing initial weights for the subsequent filling pressure equation establishment, first construct a Historical cases (e.g. )of Data Matrix ( features) and The target vector , calculate the correlation matrix between these features (yes Matrix, whose elements Characterized by and Pearson correlation coefficient between features and target variables) and the correlation coefficient vector between features and target variables If there is significant collinearity between features (for example, the absolute value of the correlation coefficient between two features is greater than 0.8, or by calculating the variance inflation factor VIF, if the VIF value is greater than 5, it is considered that there is collinearity), the Ridge Regression method is used to estimate the initial weights. Ridge Regression handles the collinearity problem by adding the L2 regularization term to the loss function of the ordinary least squares method. Its weight vector The calculation formula is ,in It is centralized and standardized feature matrix, is the centered target vector, is the identity matrix, is the ridge parameter (regularization strength), The value of is selected on the training set by 10-fold cross validation, with the goal of minimizing the mean squared prediction error (MSE). For example, the optimal , use this The initial weight coefficient vector (only for the first three core parameters) calculated by the value is: , , , these are the initial values ​​of the generated filling pressure weights.

[0065] S402: Calling the initial value of the filling pressure weight, constructing the residual function of the left ventricular filling pressure equation, calculating the partial derivative of the weight coefficient using the gradient descent algorithm, updating the weight parameter along the negative gradient direction, terminating the iteration when the residual sum of squares is less than the convergence threshold, and obtaining the optimized value of the weight coefficient; Call the initial value of the filling pressure weight for the core parameter generated in step S401 , , , as well as initial weights for other auxiliary parameters (e.g., age, LAVI) and the intercept term (e.g., , , , all obtained by ridge regression), to construct the prediction equation for left ventricular filling pressure , its specific form is a linear combination: in It is a vector containing all weight coefficients to be optimized, and its initial value is .

[0066] Define the residual function, for the first samples (including the five input feature values ​​and the actual left ventricular filling pressure measured by the gold standard ), its residual Defined as the difference between the predicted value and the actual value: , the optimization goal is to minimize all The sum of squared errors (SumofSquaredErrors, SSE) of the training samples, that is, the loss function .

[0067] Gradient descent algorithm is used to iteratively optimize the weight vector , in each iteration, the loss function is calculated Relative to each weight coefficient The partial derivative of , for a linear model, this partial derivative is: ; in, , , , and so on. The weight update rule is: ; in is the number of iterations, Is the learning rate, set to a small positive number to ensure convergence, the learning rate The value is set to 0.0005 through line search or based on experience. This value shows good convergence speed and stability in preliminary experiments. The iterative process continues, and each iteration recalculates the predicted filling pressure, residual and total residual sum of squares of all training samples using the updated weights until the residual sum of squares is Less than a preset convergence threshold, or reaches a preset maximum number of iterations (e.g. 500), the convergence threshold is set to This threshold is based on historical model training experience. When the SSE change is less than this value, further iteration will have little effect on model performance. In this case, after 185 iterations, the SSE reached , the convergence condition is met and the iteration is terminated. The weight vector obtained at this time is the optimized weight coefficient. For example, the final optimized core parameter weight is: , , , (other weights are also updated accordingly) to obtain the optimized value of the weight coefficient.

[0068] S403: Substituting the optimized weight coefficient value into the left ventricular filling pressure equation, calculating the convolution integral of the myocardial stress tensor and the ventricular volume change rate, extracting the principal strain components through eigenvalue decomposition, and outputting the left ventricular compliance assessment value; The optimized value of the core parameter weight coefficient obtained in step S402 ( , , ) and other parameters’ optimization weights (e.g., , , ) into the established left ventricular filling pressure equation, for the input characteristics of the current case (Factor=0.608,Angle= ,E_e_prime=14.62,Age=65 years,LAVI=38 ml / m ), calculate the estimated left ventricular end-diastolic filling pressure : ; ; mmHg; About 21.5mmHg. As an important parameter for subsequent evaluation.

[0069] Next, the dynamic characteristics of the ventricular diastole were evaluated, which involved the myocardial stress tensor and the rate of change of ventricular volume. It is a term that describes how the myocardium withstands load during diastole. Symmetric tensor whose components (such as hoop stress , longitudinal stress , radial stress and shear stress) by combining the individualized left ventricular geometry (from CT 3D reconstruction) and the left ventricular intracavitary pressure calculated above. (as internal boundary conditions) and the passive material properties of the myocardium (for example, the material constants of the Mooney-Rivlin model, C1 = 1.5 kPa, C2 = 0.5 kPa, which are obtained by fitting the literature references or isolated myocardial tensile test data), the calculated average circumferential wall stress at end-diastole is 18.0 kPa and the longitudinal wall stress is 14.5 kPa. Obtain left ventricular volumes throughout the cardiac cycle using time-resolved three-dimensional ultrasound or cardiac MRI The curve is then differentiated numerically to obtain that during the rapid filling period, Up to 450 ml / s; in late diastole (atrial contraction), About 150 ml / s.

[0070] Subsequently, a measure of global ventricular-vascular coupling or energy transfer is calculated, which is expressed as myocardial stress (or a scalar representation thereof, such as mean wall stress). ) and ventricular volume change rate During the entire diastole (from mitral valve opening to end diastole, the time interval ) To achieve: At discrete time points, this integral is approximated by the sum ,in is the time step, , calculated The value is 2500kPa ml / s Second = 2500kPa This integral value reflects the energy status during the filling process.

[0071] At the same time, the end-diastolic myocardial strain tensor obtained by FEM calculation (The strain tensor is derived from the stress tensor through the myocardial constitutive relationship) to extract the principal strain components and Perform eigenvalue decomposition to solve the equation , and obtain the three principal strain values , which represent the maximum deformation of the myocardium in three mutually perpendicular principal directions. In this case, the end-diastolic principal strain (expressed as a unitless decimal to indicate the ratio of tension or compression) is: hoop strain (shortened by 18%), longitudinal strain (shortened by 15%), radial strain (35% thicker).

[0072] Finally, the left ventricular compliance index (LVCI) is a composite index that combines the above-calculated left ventricular end-diastolic filling pressure with a predefined empirical formula or machine learning model. , the convolution integral value , and the principal strain components (e.g., taking the overall volumetric strain or some combination of principal strains, such as ), output a quantitative compliance score, for example, define LVCI as: LVCI= in, is an empirical coefficient, such as , , assuming that the end-diastolic volume is 120 mL and the minimum early diastolic volume is 50 mL, then mL. LVCI= First, To convert from mmHg to kPa, use the conversion rule: 1 mmHg 0.133322kPa. kPa.

[0073] LVCI= ; LVCI= ; LVCI= mL / kPa. mL / kPa is the output left ventricular compliance assessment value. Lower values ​​indicate poorer left ventricular compliance. For example, normal left ventricular compliance values ​​are usually between 40 and 60 mL / kPa, while values ​​below 30 mL / kPa indicate a significant decrease in compliance.

[0074] The myocardial stress tensor is defined as the stress distribution matrix of myocardial tissue in three-dimensional space; The ventricular volume change rate is the rate of change of the left ventricular cavity volume per unit time.

[0075] A left ventricular compliance assessment system based on medical imaging, the left ventricular compliance assessment system based on medical imaging is used to perform the above-mentioned left ventricular compliance assessment method based on medical imaging, and the system includes: The long-axis modeling module is used to obtain the three-dimensional coordinates of the left ventricular apex and mitral valve annulus plane through CT images, call the thin plate spline interpolation algorithm to construct the left ventricular long-axis geometric model, simultaneously obtain the ultrasound end-diastolic internal diameter data, input the long-axis geometric model vertex coordinates and the ultrasound internal diameter data into the Euclidean distance calculation function to generate the left ventricular aspect ratio, and pass the left ventricular aspect ratio to the curvature registration module; The curvature registration module is used to perform adaptive volume compensation model registration on the CT left ventricular cavity point cloud data based on the left ventricular aspect ratio, segment the anterior septum and lateral wall anatomical regions using the region growing algorithm, extract the curvature distribution characteristics of the anterior septum and lateral wall using the Gaussian curvature calculation model, calculate the curvature gradient difference between the two regions, and pass the curvature gradient difference value to the stiffness correction module; A stiffness correction module is used to perform a ratio operation on the curvature gradient difference value and the ultrasonic ventricular septum thickness and side wall thickness, input the operation result into the radial basis function network to generate a regional stiffness correction factor, simultaneously obtain the aortic root tilt angle measurement value during CT angiography, and transmit the regional stiffness correction factor and the aortic root tilt angle to the compliance coupling module; The compliance coupling module is used to weight the ultrasound E / e' ratio using a regional stiffness correction factor, construct a multi-parameter coupling equation based on the aortic root inclination angle, and use a gradient descent optimization algorithm to iteratively solve the weight coefficients of early diastolic pressure and atrial systolic pressure in the left ventricular filling pressure equation to output a quantitative assessment value of left ventricular compliance. The input parameters of the multi-parameter coupling equation correspond one-to-one to the regional stiffness correction factor, the aortic root inclination angle, and the ultrasound E / e' ratio.

[0076] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A left ventricular compliance assessment method based on medical imaging, characterized in that: The following steps are involved: S1: Obtain the three-dimensional coordinates of the plane from the left ventricular apex to the mitral valve annulus through CT imaging, call the thin plate spline interpolation algorithm to establish a left ventricular long axis model, synchronously obtain the ultrasound end-diastolic internal diameter, and input the left ventricular long axis model and the ultrasound end-diastolic internal diameter into the Euclidean distance calculation function to generate the left ventricular aspect ratio; S2: Based on the left ventricular aspect ratio, an adaptive volume compensation model is used to register the CT left ventricular cavity point cloud data, the anterior septum and lateral wall regions are segmented using a region growing algorithm, and the curvature gradient difference value is extracted using a Gaussian curvature calculation model; S3: performing a ratio operation on the curvature gradient difference value and the ultrasound ventricular septum thickness and side wall thickness, inputting the calculated values ​​into a radial basis function network to generate a regional stiffness correction factor, and simultaneously obtaining the aortic root inclination angle of CT angiography; S4: Based on the regional stiffness correction factor and the aortic root inclination angle, the ultrasound E / e' ratio is input into a multi-parameter coupling model, and the weight coefficient of the left ventricular filling pressure equation is iteratively calculated through a gradient descent optimization algorithm to output a left ventricular compliance assessment value.

2. The method for evaluating left ventricular compliance based on medical imaging according to claim 1, characterized in that: The left ventricular aspect ratio is specifically the geometric ratio of the long axis distance to the end-diastolic internal diameter, the curvature gradient difference value includes the anterior septal curvature distribution characteristics and the side wall curvature distribution characteristics, the regional stiffness correction factor specifically refers to the product term of the ratio of the ventricular septum to the side wall thickness and the curvature gradient, the aortic root inclination angle includes the coronal plane projection angle and the sagittal plane projection angle, and the left ventricular compliance assessment value includes the filling pressure weight coefficient and the multimodal parameter coupling value.

3. The method for evaluating left ventricular compliance based on medical imaging according to claim 2, characterized in that: The thin plate spline interpolation algorithm satisfies the continuity condition of the first-order derivative at the nodes; The Euclidean distance calculation function is defined as: ； in, The left ventricular long axis model The three-dimensional coordinates of the vertices, represents the normalized spatial coordinates corresponding to the end-diastolic diameter of ultrasound, is the total number of model vertices; During the calculation of the curvature gradient difference value, the CT point cloud spatial coordinate unit and the ultrasonic thickness measurement value are uniformly converted into millimeter dimensions; A normalization layer is set before the ratio calculation to eliminate the dimension differences of the curvature gradient difference value and the ventricular septum thickness and the side wall thickness.

4. The method for evaluating left ventricular compliance based on medical imaging according to claim 3, characterized in that: The steps for obtaining the left ventricular aspect ratio are specifically as follows: S101: Acquire CT image data, detect spatial positioning points of the left ventricular apex and the anterior and posterior mitral valves based on the left ventricular apex and mitral valve annulus planes in the CT image, organize the three-dimensional coordinates of multiple positioning points, summarize the coordinate data into a spatial sequence, and generate a spatial coordinate sequence; S102: Calling a thin plate spline interpolation algorithm to perform long axis curve fitting based on the multiple three-dimensional coordinate data of the spatial coordinate sequence to establish a three-dimensional structure of the long axis of the left ventricle, collecting the left ventricular internal diameter at end-diastole of the ultrasound, and grouping the three-dimensional structural parameters and the inner diameter values ​​into a parameter group to generate a three-dimensional structural parameter group; S103: Based on the three-dimensional structural parameter group, a Euclidean distance calculation function is called to combine the left ventricular long axis model with the ultrasound end-diastolic internal diameter value, perform Euclidean distance calculation between each parameter, and obtain the left ventricular aspect ratio.

5. The method for evaluating left ventricular compliance based on medical imaging according to claim 4, characterized in that: The steps for obtaining the curvature gradient difference value are specifically as follows: S201: Based on the left ventricular aspect ratio, register the left ventricular cavity point cloud data acquired by CT using an adaptive volume compensation model, match the compensation model parameters with the left ventricular aspect ratio value according to the three-dimensional spatial distribution of the point cloud, adjust the spatial offset between multiple point clouds, and generate a registration spatial coordinate set; S202: Based on the registration spatial coordinate set, a region growing algorithm is applied to identify the front septum region and the side wall region, a region growing starting point is selected, and based on the density and topological structure characteristics of the spatial coordinates, the partitions to which the multiple points belong are determined, the point clouds are grouped into spatial partitions, and a partitioned point cloud set is obtained; S203: Based on the partitioned point cloud set, using the Gaussian curvature calculation model, select the three-dimensional spatial neighborhood of multiple points in the partition, analyze the trend of surface normal changes between the point sets, and perform curvature gradient calculation through the numerical distribution of adjacent normal angles between points to obtain the curvature gradient difference value.

6. The method for evaluating left ventricular compliance based on medical imaging according to claim 5, characterized in that: In the Gaussian curvature calculation model, the curvature gradient is associated with the spatial distribution characteristics of the myocardial fiber orientation.

7. The method for evaluating left ventricular compliance based on medical imaging according to claim 6, characterized in that: The specific steps for obtaining S3 are: S301: obtaining the original data of the curvature gradient difference value, detecting the thickness measurement values ​​of the ventricular septum and the lateral wall in the ultrasound image, calculating the ratio of the curvature gradient difference value to the ventricular septum thickness and the lateral wall thickness respectively using a point-by-point division operation, establishing a ventricular septum thickness ratio parameter and a lateral wall thickness ratio parameter, fusing the two parameters using a weighted average algorithm to generate a ventricular wall thickness ratio parameter; S302: calling the wall thickness ratio parameter as a network input layer node, constructing a radial basis function network hidden layer Gaussian kernel function, calculating the Euclidean distance between the input parameter and the hidden layer center point, activating the hidden layer nodes through an exponential function, calculating the output layer weight coefficient using the least squares method, performing a linear weighted sum operation, and generating a regional stiffness correction factor; S303: Calling DICOM format image data of CT angiography, extracting the aortic root contour through a threshold segmentation algorithm, establishing a three-dimensional rectangular coordinate system, and calculating the angle between the aortic root central axis and the spatial vector of the horizontal plane to generate the aortic root inclination angle.

8. The method for evaluating left ventricular compliance based on medical imaging according to claim 7, characterized in that: The steps for obtaining the left ventricular compliance evaluation value are specifically as follows: S401: Obtain the regional stiffness correction factor and the aortic root inclination angle, input the ultrasound E / e' ratio into a multi-parameter coupling model, establish a parameter correlation matrix through matrix determinant operations, use the least squares method to eliminate collinear interference between parameters, and generate an initial value of the filling pressure weight; S402: calling the initial value of the filling pressure weight, constructing the residual function of the left ventricular filling pressure equation, calculating the partial derivative of the weight coefficient using a gradient descent algorithm, updating the weight parameter along the negative gradient direction, terminating the iteration when the residual sum of squares is less than a convergence threshold, and obtaining the optimized value of the weight coefficient; S403: Substitute the optimized value of the weight coefficient into the left ventricular filling pressure equation, calculate the convolution integral of the myocardial stress tensor and the ventricular volume change rate, extract the principal strain component through eigenvalue decomposition, and output the left ventricular compliance evaluation value.

9. The method for evaluating left ventricular compliance based on medical imaging according to claim 8, characterized in that: The myocardial stress tensor is defined as the stress distribution matrix of myocardial tissue in three-dimensional space; The ventricular volume change rate is the rate of change of the left ventricular cavity volume per unit time.

10. A left ventricular compliance assessment system based on medical imaging, characterized in that: The system is used to implement the left ventricular compliance assessment method based on medical imaging according to any one of claims 1 to 9, and the system comprises: a long-axis modeling module for obtaining the three-dimensional coordinates of the left ventricular apex and the mitral valve annulus plane through CT images, constructing a left ventricular long-axis geometric model using a thin-plate spline interpolation algorithm, synchronously obtaining ultrasound end-diastolic internal diameter data, inputting the vertex coordinates of the long-axis geometric model and the ultrasound internal diameter data into a Euclidean distance calculation function to generate a left ventricular aspect ratio, and transmitting the left ventricular aspect ratio to a curvature registration module; a curvature registration module, configured to perform adaptive volume compensation model registration on the CT left ventricular cavity point cloud data using the left ventricular aspect ratio, segment the anterior septum and lateral wall anatomical regions using a region growing algorithm, extract the curvature distribution characteristics of the anterior septum and lateral wall using a Gaussian curvature calculation model, calculate a curvature gradient difference value between the two regions, and transmit the curvature gradient difference value to the stiffness correction module; a stiffness correction module, configured to perform a ratio operation on the curvature gradient difference value and the ultrasonic ventricular septum thickness and side wall thickness, input the operation result into a radial basis function network to generate a regional stiffness correction factor, simultaneously obtain the aortic root inclination angle measurement value during CT angiography, and transmit the regional stiffness correction factor and the aortic root inclination angle to the compliance coupling module; a compliance coupling module for weighting the ultrasound E / e' ratio using the regional stiffness correction factor, constructing a multi-parameter coupling equation based on the aortic root inclination angle, and using a gradient descent optimization algorithm to iteratively solve the weight coefficients of early diastolic pressure and atrial systolic pressure in the left ventricular filling pressure equation to output a quantitative assessment value of left ventricular compliance; The input parameters of the multi-parameter coupling equation correspond one-to-one to the regional stiffness correction factor, the aortic root inclination angle, and the ultrasound E / e' ratio.

Images