Aortic stent plane automatic positioning and related index measuring method
By using deep learning algorithms and point cloud feature analysis, we have achieved multi-plane automatic localization and measurement of aortic stents after TAVR, which solves the problem of complex three-dimensional structure evaluation in existing technologies and provides a precise multi-dimensional evaluation method that is suitable for standardized measurement of complex morphologies such as aortic valve annulus.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANJING FIRST HOSPITAL
- Filing Date
- 2026-01-08
- Publication Date
- 2026-06-09
AI Technical Summary
Existing technologies lack systematic and automated methods for locating and measuring multiple key anatomical planes after transcatheter aortic valve replacement, resulting in complex and poorly repeatable assessment of valve structures, making it difficult to meet the needs for accurate assessment of complex three-dimensional structures.
By acquiring three-dimensional medical image data, deep learning algorithms are used to segment aortic stents and related structures. Based on the spatial distribution characteristics of point clouds, lateral and longitudinal reference planes are determined, multiple stent planes are generated, and relevant geometric indices are calculated to achieve automated and standardized multi-plane positioning and measurement.
It enables rapid and precise localization and multi-dimensional assessment of aortic stents, provides quantitative measurements of stent area, perimeter and other related indicators, supports standardized assessment of stent position and deformation after TAVR, and improves the repeatability and accuracy of assessment.
Smart Images
Figure CN122175855A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of medical image processing technology and relates to automatic positioning and measurement technology in the image evaluation after aortic valve replacement surgery. Specifically, it relates to a method for automatic positioning of aortic stent plane and measurement of related indicators. This method is applicable to the automatic positioning and analysis of multiple key planes around the implanted stent after interventional valve treatments such as transcatheter aortic valve replacement. Background Technology
[0002] Transcatheter aortic valve replacement (TAVR) has become an important interventional procedure for treating severe aortic stenosis, especially suitable for elderly patients or those at high risk of surgery. As the indications gradually expand to low- and intermediate-risk populations, the durability of the valve and its long-term functional assessment are receiving increasing attention post-procedure.
[0003] Recent follow-up studies based on cardiac computed tomography (CT) scans have revealed a radiographic manifestation known as "subclinical leaflet thrombosis" (SLT) after TAVR (Transcatheter Aortic Valve Replacement). Its main characteristics include leaflet thickening (HALT) and decreased leaflet mobility (RELM). Furthermore, it is significantly correlated with a reduction in the effective valvular opening area (EOA), suggesting that SLT may affect valvular hemodynamics and long-term durability.
[0004] In addition to HALT, post-TAVR CT can also identify sinus filling defects (SFD) and prosthesis filling defects (PFD), which are collectively referred to as prosthetic-associated subclinical thrombootic events (PASTE). SFD is significantly associated with major adverse cardiovascular events (MACCO) and negatively correlated with a decrease in left ventricular ejection fraction (LVEF), suggesting it may be an important factor to consider in postoperative anticoagulation therapy.
[0005] Although CT imaging plays a crucial role in identifying postoperative subclinical thrombotic events, current clinical practice for quantitative assessment of the valve annulus and leaflet structures after TAVR still relies primarily on manual point selection and subjective judgment, which suffers from problems such as operational complexity, poor repeatability, and low efficiency. Furthermore, existing technologies mostly focus on single-plane measurements, lacking systematic and automated methods for locating and measuring multiple key anatomical planes (such as the transverse, longitudinal, top, bottom, coronary ostia, and Warburg sinus planes), making it difficult to meet the needs for accurate assessment of complex three-dimensional structures.
[0006] Therefore, there is an urgent need for a multi-plane automatic positioning and measurement method based on three-dimensional medical images to achieve automated, standardized, and precise assessment of the valve annulus and leaflet structure after TAVR, providing a reliable basis for postoperative valve function monitoring, thrombosis risk identification, and anticoagulation strategy formulation. Summary of the Invention
[0007] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method for automatic aortic stent plane positioning and related index measurement, including: Step 1: Acquire the patient's postoperative three-dimensional medical imaging data and perform standardized processing on the images; Step 2: Identify the aortic stent and its related structures in the image, including the aorta, coronary arteries, left ventricle, and calcification, to obtain the three-dimensional morphology of the stent and surrounding anatomical area; Step 3: Based on the three-dimensional point set of the aortic stent, establish the lateral and longitudinal reference planes of the aortic stent using the spatial distribution characteristics of the point cloud; Step 4: Based on the horizontal reference plane, search for the lowest and highest points of the support in the support point set, and determine the bottom and top planes of the support accordingly; Step 5: By analyzing the junction area between the coronary arteries and the aorta, determine the locations of the openings of the left and right coronary arteries, and construct the left and right coronary artery planes based on the characteristics of the intersection area; Step 6: Generate multiple parallel cross-sections between the base plane of the stent and the coronary reference plane, and determine the Valsalva sinus plane based on the changing trend of the aortic cross-sectional area; Step 7: Using the longitudinal reference plane as the center, rotate at set angular intervals to generate multiple longitudinal planes for the support; Step 8: Intersect each plane with the original image, calculate the area, perimeter and other relevant geometric parameters of the support, and generate the corresponding two-dimensional cross-sectional image.
[0008] Furthermore, the three-dimensional medical image data mentioned in step 1 can be any kind of three-dimensional medical image data containing information about the aortic stent, aorta, and left and right coronary arteries, including but not limited to CT, CTA, or other medical imaging data that can reflect the stent and its surrounding anatomical structure.
[0009] Furthermore, in step 2, the image is segmented by manual annotation, or by traditional image segmentation algorithms, or by automatic segmentation by the system using deep learning algorithms, including MedNeXt, NNUNET, and UNETR++. After segmentation, structural information of the aorta, coronary arteries, left ventricle, aortic stent, and calcified regions is obtained for subsequent planar localization and geometric measurement.
[0010] Furthermore, in step 3, the specific method for calculating the transverse and longitudinal reference planes of the aortic stent is as follows: First, the position of the aortic stent in three-dimensional space is extracted, its center point is calculated, and a coordinate system is established with the center point as the reference. Then, three orthogonal reference planes are calculated based on the spatial distribution characteristics of the stent point cloud. By comparing the circularity of the aortic stent's projection on different planes, the transverse and longitudinal planes of the aortic stent are determined. The specific implementation steps are as follows.
[0011] Step 3.1: Extract the 3D point set of the aortic stent:
[0012] in, Let n be the three-dimensional coordinates of the i-th point, and n be the number of aortic stent points in the sampling points. Find the center point of the point set
[0013] in, coordinates of the center point n is the number of sample points. Decentralize the point set:
[0014] Define a decentralized data matrix Its size is , ] Calculate the covariance matrix and eigenvectors:
[0015] in, for The matrix formed by these elements is a (3, n) matrix. C is the covariance matrix. For finding eigenvalues and eigenvectors of a matrix:
[0016] The eigenvalues are: For eigenvalues, These are the eigenvectors.
[0017]
[0018] in Eigenvalues The corresponding standardized feature vector.
[0019] Step 3.2: Determine the candidate plane: based on the center point of the support. To generate three candidate planes, each passing through a point on the plane, we use three standardized eigenvectors as normal vectors. The plane equation is as follows:
[0020] Calculate the circularity of the projection of the support point set onto the candidate plane: 1. For any point on the support In the candidate plane Projection:
[0021]
[0022] in express Dot at The projection point obtained on the plane.
[0023] 2. Obtain the projection set of all points on each of the three planes. And calculate the centroid of the projection point:
[0024]
[0025]
[0026] 3. Calculate the distance from each projection point to the centroid: 4. Calculate the mean and standard deviation of the distances:
[0027]
[0028] 5. Calculate the roundness index:
[0029] Calculate the support points respectively The circularity of the projection, where, Indicates the support point is at Circularity of a planar projection.
[0030] 6. Establish the transverse and longitudinal planes of the aortic stent: The plane with the smallest circularity is selected as the transverse plane of the aortic stent: The normal vector is represented as: ( ) The rest and Orthogonal planes serve as the longitudinal plane of the aortic stent: The normal vector is represented as ( ) Furthermore, the specific method for calculating the bottom plane of the support in step 4 is as follows: First, by analyzing the spatial distribution of the aortic stent point cloud in the longitudinal direction, the lowest point of the stent was determined. and the highest point Based on this, the transverse reference plane of the aortic stent obtained in step 3 is translated along its normal direction to the lowest and highest points, respectively, to obtain the bottom plane and top plane of the aortic stent.
[0031] The plane equation of the bottom plane is:
[0032] The plane equation of the top plane is:
[0033] in, The base plane of the aortic stent The top plane of the aortic stent.
[0034] Furthermore, the specific method for calculating the left and right coronary artery planes in step 5 is as follows: First, based on the segmentation results from step 2, the left and right coronary artery regions are extracted. Common morphological dilation operations are then applied to both the aorta and coronary artery regions. Next, the intersection point of the left coronary artery and aorta is calculated to obtain the intersection region. The center point of this intersection region is then calculated. Its coordinates are Similarly, calculate the intersection point of the right coronary artery and the aorta to obtain the intersection area, and calculate the center point of the intersection area. Its coordinates are The center point is used to characterize the anatomical location of the coronary artery ostium in space. The normal vector of the aortic stent transverse reference plane obtained in step 3 ( As a directional consistency constraint, the left and right coronary artery planes are constructed respectively, and their equations are as follows:
[0035]
[0036] in, The level of the left coronary artery. This is the plane of the right coronary artery.
[0037] Further, calculate the midpoint coordinates of the center points of the left and right coronary arteries. Using the midpoint as the point of passage and keeping the plane normal vector perpendicular to the midpoint, Consistent, the reference plane at the midpoint of the coronary artery is obtained: .
[0038] Furthermore, the specific method for calculating the Warburg sinus (SOV) plane in step 6 is as follows: Based on the bottom plane of the bracket obtained in step 4 and the reference plane of the midpoint of the left and right coronary arteries obtained in step 5. Using the spatial region between the two as the search range, multiple parallel slicing planes are uniformly generated within this range along the normal direction of the bottom plane of the stent. The cross-sectional area of each slicing plane and the aortic segmentation model is calculated, and the plane with the largest cross-sectional area is selected as the Valsalva sinus feature plane, denoted as... .
[0039] Furthermore, the specific method for calculating the longitudinal plane of the support in step 7 is as follows: Based on the longitudinal reference plane of the aortic stent obtained in step 3 Using this plane as the initial reference plane, multiple longitudinal planes of the stent at different angles are obtained by rotating it sequentially according to a preset angle. The equation of the longitudinal reference plane of the aortic stent is:
[0040] The direction of the rotation axis is the transverse reference plane. The normal vector u ( The axis of rotation passes through the center point. .
[0041] According to Rodriguez's rotation formula, the rotated normal vector is:
[0042] N is the rotation angle. For the normal vector of the rotated plane, ) is a vector cross product representing a new vector perpendicular to u and N.
[0043] Furthermore, the specific method for calculating the relevant indicators and mapping them to the original image in step 8 is as follows: Based on the stent reference planes obtained in steps 3 to 7, calculate the perimeter and area of the corresponding cross section of each stent reference plane; and further calculate the perimeter and area of the area where the stent reference plane intersects with the aorta by calculating the spatial intersection point of the stent reference plane and the aortic structure.
[0044] Further, based on the plane equations obtained from steps 3 to 7, let the plane equation of the stent reference plane be ax + by + cz + d = 0. Then, interpolate the plane with the volumetric data matrix composed of the original medical images. The intersection of the planes yields a two-dimensional cross section. For a given (x, y) coordinate, solving the plane equation yields the corresponding z coordinate:
[0045] When z is not an integer , ,in It is the size of the original image in the z-axis. To round down the calculated z value, w is the weighting coefficient. Therefore, the voxel value of the original image corresponding to (x, y) is:
[0046] By interpolating the voxel values at each point on the entire plane using the above method and normalizing the interpolation results, a corresponding two-dimensional cross-sectional image can be obtained for subsequent structural observation and quantitative analysis.
[0047] Furthermore, by calculating the intersection points of the aorta and aortic stents, parameters such as area and perimeter are calculated. For the area and perimeter of a plane passing through the aorta, the Graham scan algorithm (commonly used in point clouds) is employed to calculate the convex hull (i.e., the outermost circle of the intersection points with the aorta). Spatial resampling is performed on the volume data corresponding to the two-dimensional cross-sectional image to achieve image normalization. Trilinear interpolation is used to adjust the original voxel spacing (e.g., 0.68 × 0.68 × 0.5 mm). 3 Unified to 1×1×1 mm 3 The normalized 3D image dimensions are calculated as follows:
[0048]
[0049]
[0050] Where (D, H, W) is the original image size, ( ) represents the original image spacing, ( ) represents the normalized image spacing. , , The image size is the normalized value. Therefore, the original coordinates (x, y, z) can be transformed into:
[0051] y
[0052] z
[0053] but( () is in a voxel spacing of 1×1×1 mm 3 The coordinates of the point below.
[0054] The beneficial effects of this invention are: This invention utilizes innovative technologies to automate the calculation of key geometric parameters for postoperative aortic stents, achieving rapid and precise positioning of multiple aortic stent planes without manual point selection. It is applicable to complex morphologies such as the aortic valve annulus and exhibits strong robustness. It provides multi-dimensional reference planes; by precisely mapping these planes to the original three-dimensional medical image, multiple postoperative evaluation indicators can be obtained, including stent area, stent perimeter, and intersection parameters between the stent and the aortic structure. This supports the quantitative, standardized, and repeatable measurement of stent position, stent deformation, and potential complication risks after TAVR. Attached Figure Description
[0055] Figure 1 This is a schematic diagram of the overall process of the automatic plane positioning of the aortic stent and the measurement of related indicators of the present invention.
[0056] Figure 2 This is a schematic diagram of the image segmentation results of the aorta, coronary artery, left ventricle, aortic stent, and calcified region in this invention.
[0057] Figure 3 This is a schematic diagram showing the segmentation result of the aortic stent structure in the method of the present invention, as well as the horizontal and vertical initialization reference planes.
[0058] Figure 4 This is a schematic diagram of the planar spatial position of the aortic stent bottom in the method of the present invention.
[0059] Figure 5 This is a schematic diagram showing the spatial position of the top plane of the aortic stent in the method of the present invention.
[0060] Figure 6 This is a schematic diagram of the coronary artery plane where the aortic stent is located in the method of the present invention.
[0061] Figure 7This is a schematic diagram of the aortic stent located in the SOV plane of the Warburg sinus in the method of the present invention. Detailed Implementation
[0062] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that the embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. Furthermore, it should be understood that after reading the disclosure of this invention, those skilled in the art can make various modifications or alterations to the invention, and these equivalent forms also fall within the scope of protection defined by this invention.
[0063] Figure 1 A method for automatically locating the plane of an aortic stent and measuring related parameters is shown, including: Step 1: Image Data Acquisition A contrast-enhanced cardiac scan was performed on a patient after TAVR using a CT scanner to acquire three-dimensional CT image data, which was stored in DICOM format. In one embodiment, the image resolution was 512×512×313, the voxel spacing was 0.39 mm×0.39 mm, and the slice thickness was 0.5 mm.
[0064] Step 2: Structural segmentation The nnU-Net deep learning model was used and trained on a labeled dataset containing 100 post-TAVR CT images (labeled with stents, annulus, and leaflets). After training, the CT data of the test patients were automatically segmented. The segmentation accuracy was evaluated by the Dice similarity coefficient (DSC), with stent DSC ≥ 0.90, aortic stent DSC ≥ 0.90, and aortic DSC ≥ 0.85.
[0065] Step 3: Establish the transverse and longitudinal reference planes for the aortic stent The main steps include extracting the position of the aortic stent in three-dimensional space, calculating its center point, establishing a coordinate system based on the center point, and then calculating three pairwise orthogonal candidate reference planes based on the spatial distribution characteristics of the stent point cloud. By comparing the circularity of the aortic stent's projection on different candidate planes, the transverse and longitudinal reference planes of the aortic stent are determined.
[0066] Step 3.1: Extract the aortic stent point set ,in Given the three-dimensional coordinates of the i-th voxel, calculate the center point: .
[0067] Decentralize the point set: Define a decentralized data matrix ( ]), its size is Calculate the covariance matrix and eigenvectors:
[0068] in, for The matrix formed by these elements is a (3, n) matrix. C is the covariance matrix. For finding eigenvalues and eigenvectors of a matrix:
[0069] The eigenvalues are: For eigenvalues, These are the eigenvectors.
[0070]
[0071] in Eigenvalues The corresponding standardized feature vectors are decentralized to construct a data matrix, and the covariance matrix is calculated.
[0072] In this example: the aortic stent point set is extracted, consisting of 38,886 voxels; the center point is calculated; and the coordinates of the center point (ras) are determined. After decentralizing (-9.177, 152.383, 877.864), a data matrix is constructed, the covariance matrix is calculated, and the eigenvalues are obtained. The corresponding feature vectors are [0.722443, -0.071620, 0.687711], [-0.296906, 0.866119, 0.402101], and [0.624438, 0.494681, -0.604457].
[0073] Step 3.2: Determine the candidate plane: based on the center point of the support. To generate three candidate planes, each passing through a point on the plane, we use three standardized eigenvectors as normal vectors. The plane equation is as follows:
[0074] In this example, the standardized eigenvectors are [0.624438, 0.494681, -0.604457], [-0.296906, 0.866119, 0.402101], and [0.722443, -0.071620, 0.687711]. The coordinates of the center point ras are... The equations of the plane (-9.177, 152.383, 877.864) are as follows:
[0075]
[0076]
[0077] Step 3.3: Based on Step 3.2, calculate the circularity of the projection of the support point set onto the candidate plane: 1. For any point on the support In the candidate plane Projection:
[0078]
[0079] 2. Obtain the projection sets of the three planes respectively. And calculate the centroid of the projection point:
[0080] 3. Calculate the distance from each projection point to the centroid: 4. Calculate the mean and standard deviation of the distances:
[0081]
[0082] 5. Calculate the roundness index:
[0083] Calculate the support points respectively The circularity of the projection, where, Indicates the support point is at Circularity of a planar projection.
[0084] 6. Establish the transverse and longitudinal planes of the aortic stent: The plane with the smallest circularity is selected as the transverse plane of the aortic stent: The normal vector is represented as: ( ) The rest and Orthogonal planes serve as the longitudinal plane of the aortic stent: The normal vector is represented as ( ) In this embodiment: three candidate planes are generated, and the circularity of the projected profile of each plane is calculated. 0.4502, 0.4273, 0.2054, 3-plane circularity: 0.2054, select the corresponding... plane as a transverse reference plane , As a longitudinal reference plane: ,like Figure 3 As shown.
[0085] Step 4: Locating the bottom and top planes of the aortic stent. By analyzing the spatial distribution of the aortic stent point cloud in the longitudinal direction, the lowest point of the stent is determined. and the highest point Based on this, the transverse reference plane of the aortic stent obtained in step 3 is translated along its normal direction, so that the translated plane passes through the lowest point and the highest point respectively, thereby obtaining the bottom plane and top plane of the aortic stent.
[0086] The plane equation of the bottom plane is:
[0087] The plane equation of the top plane is:
[0088] in, For the aortic stent base plane, This refers to the top plane of the aortic stent. In this embodiment, the lowest point at the bottom is selected. (-39.3828125, 158.6640625, 869.0), will along By translating the direction, we obtain the bottom plane, and the equation is: The calculation plane is used to draw a 2D plane through the points of the original image, such as... Figure 4 As shown; the highest point of the top of the screening bracket (21.5546875, 160.6171875, 883.5), the top plane is obtained, and the equation is: : The calculation plane draws a two-dimensional cross-sectional image through the points in the original image, such as... Figure 5 As shown.
[0089] Step 5: Left and right coronary artery planes: Calculate the intersection point of the left coronary artery and the aorta to obtain the intersection region, and calculate the center point of the intersection region. Its coordinates Calculate the intersection point of the right coronary artery and the aorta to obtain the intersection area, and calculate the center point of the intersection area. The coordinates (4.848, 168.063, 870.695) are calculated to be:
[0090]
[0091] The coordinates are (-7.251, 152.861, 880.59).
[0092] Passing Point The plane at the midpoint of the left and right coronary arteries is as follows Figure 6 As shown.
[0093] Step 6: Wassef sinus (SOV) plane Based on the planes obtained in steps 4 and 5, the bottom plane of the bracket obtained in step 4 And the reference plane of the midpoint of the left and right coronary arteries in step 5. The model is divided into 30 planes based on the distance between them. The spatial region between the planes is used as the search range. Within this range, multiple parallel slice planes are generated uniformly along the normal direction. The cross-sectional area of each slice plane and the aortic segmentation model is calculated. The intersection point of the plane and the aorta is calculated to determine the largest plane. The plane equation is as follows:
[0094] in Plane Figure 7 As shown.
[0095] The specific method for calculating the longitudinal plane of the stent in step 7 is as follows: based on the longitudinal reference plane of the aortic stent obtained in step 3. Using this as the initial reference plane, the stent is rotated around an axis defined by the normal vector of the transverse reference plane, with the stent center point as the center of rotation, at preset angular intervals to obtain multiple longitudinal planes of the stent at different angles. The initial equation of the aortic stent longitudinal reference plane is:
[0096] The axis of rotation is The normal vector u ( Center of rotation .
[0097] According to Rodrigues's rotation formula, the normal vector after rotation is:
[0098] N is the rotation angle. For the normal vector of the rotated plane, ) is the vector cross product representing a new vector perpendicular to u and N. In this embodiment, the center vector of rotation, The central axis is The normal vector u has a rotation center axis of [0.722443, -0.071620, 0.68771] and a rotation center point of (-9.177, 152.383, 877.864). The resulting plane equation is:
[0099]
[0100]
[0101]
[0102]
[0103] in, The planes obtained after rotating by 30, 60, 90, 120, and 150 degrees.
[0104] Step 8: Calculation of relevant indicators: Based on steps 4, 5, 6, and 7, the plane equation is obtained. The segmentation result of the plane and the original medical image is calculated. The plane equation is the matrix formed by ax + by + cz + d = 0 and the original medical image. Given intersecting two-dimensional planes, for a given (x, y) coordinate, solving the plane equation yields the corresponding z-coordinate: When z is not an integer, , , It is the size of the z-dimensional dimension of the original image. To round down the calculated z value, w is the weighting coefficient. Therefore, the voxel value of the original image corresponding to (x, y) is:
[0105] This leads to the cross-section of the plane passing through the three-dimensional image, such as... Figure 4-7 As shown; The area and perimeter are calculated by calculating the intersection points of the plane and the aortic stent segmentation model. For the area and perimeter of a certain plane passing through the aorta, the Graham scan algorithm (commonly used in point cloud analysis) is used to calculate the convex hull (i.e. the outermost circle of the intersection point with the aorta).
[0106] The area and perimeter are calculated by determining the coordinates of points in the CT image at a standardized interval. The area S is calculated using the shoelace method, and (x, y) represents the set of n points representing the convex hull calculated using the Graham scan algorithm.
[0107] How to calculate the perimeter L:
[0108] In this embodiment, parameters such as area and perimeter are calculated.
[0109]
[0110] The embodiments of the present invention have been described above. However, the present invention is not limited to the above embodiments. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for automatic positioning of aortic stent plane and measurement of related indexes, characterized in that, include: Step 1: Acquire the patient's postoperative three-dimensional medical image data and perform standardized processing on the three-dimensional medical image data; Step 2: Identify the aortic stent and its related structures in the 3D medical imaging data. These related structures include the aorta, coronary arteries, left ventricle, and calcified areas, to obtain the 3D morphology of the stent and surrounding anatomical region. Step 3: Based on the 3D point set of the aortic stent, establish the lateral and longitudinal reference planes of the aortic stent using the spatial distribution characteristics of the point cloud. Step 4: Search for the lowest and highest points of the stent in the stent point set according to the lateral reference plane, and determine the bottom and top planes of the stent accordingly. Step 5: By analyzing the junction area between the coronary artery and the aorta, determine the opening of the left coronary artery. Step 6: The location of the stent ostium and the location of the right coronary artery ostium are determined, and the left and right coronary artery planes are constructed based on the characteristics of the intersecting region. Step 7: Multiple parallel cross-sections are generated between the stent base plane and the coronary artery reference plane, and the Wassermann sinus plane is determined according to the changing trend of the aortic cross-sectional area. Step 8: Multiple stent longitudinal planes are generated by rotating around the longitudinal reference plane at set angular intervals. Step 9: The stent reference planes determined in steps 3 to 7 are intersected with the original image, the area, perimeter and other relevant geometric parameters of the stent are calculated, and the corresponding two-dimensional cross-sectional images are generated.
2. The method of automatic positioning of aortic stent plane and measurement of related indexes according to claim 1, characterized in that, The three-dimensional medical image data mentioned in step 1 can be any type of three-dimensional medical image data that includes information on the aortic stent, aorta, and left and right coronary arteries.
3. The method for automatic aortic stent plane positioning and related index measurement according to claim 2, characterized in that, In step 2, the three-dimensional medical image data is manually labeled and segmented, or segmented using traditional image segmentation algorithms, or automatically segmented using deep learning algorithms; the structural information of the aorta, coronary arteries, left ventricle, aortic stent, and calcified areas is obtained through the segmentation process, which is used for subsequent planar positioning and geometric measurement.
4. The method for automatic aortic stent plane positioning and related index measurement according to claim 1, characterized in that, Step 3 includes: obtaining the three-dimensional point set of the aortic stent. Calculate the center point of the three-dimensional point set. The three-dimensional point set is decentralized, and the covariance matrix is calculated based on the decentralized point set to obtain three mutually orthogonal standardized feature vectors; with the center point... To construct three candidate planes, using the three standardized feature vectors as normal vectors, and passing through the point, respectively. .
5. The method for automatic aortic stent plane positioning and related index measurement according to claim 4, characterized in that, Step 3 further includes: projecting the three-dimensional point set onto the candidate plane respectively. The projection circularity index of each candidate plane is calculated based on the projection point set, where the circularity index is the ratio of the standard deviation to the mean of the distance from the projection point to the projection centroid; the candidate plane with the smallest circularity index is determined as the transverse reference plane of the aortic stent. and will be in relation to the lateral reference plane Orthogonal candidate planes are used to determine the longitudinal reference plane for aortic stents. .
6. The method for automatic aortic stent plane positioning and related index measurement according to claim 5, characterized in that, Step 4 includes: determining the lowest point of the support based on the spatial distribution of the support point set in the longitudinal direction. and the highest point and the lateral reference plane Translate it along its normal direction to pass through the lowest and highest points respectively, in order to determine the bottom plane of the support. and the top plane of the bracket .
7. The method for automatic aortic stent plane positioning and related index measurement according to claim 6, characterized in that, Step 5 includes: extracting the left coronary artery region and the right coronary artery region based on the segmentation results of Step 2, and performing morphological dilation processing on the aortic region and the coronary artery region respectively; determining the intersection region between the left coronary artery and the aorta and the right coronary artery and the aorta respectively, and calculating the center point of the intersection region as the spatial representation of the corresponding coronary artery ostium position; using the normal vector of the transverse reference plane determined in Step 3 as the direction consistency constraint, constructing the left coronary artery plane and the right coronary artery plane passing through the center point of the left coronary artery ostium and the center point of the right coronary artery ostium respectively; and further constructing a coronary artery midpoint reference plane based on the midpoint of the two center points, and the normal vector of the coronary artery midpoint reference plane is consistent with the normal vector of the transverse reference plane.
8. The method for automatic aortic stent plane positioning and related index measurement according to claim 7, characterized in that, Step 6 includes: based on the stent bottom plane obtained in Step 4 and the coronary artery midpoint reference plane obtained in Step 5, using the space between the two as the search range, and uniformly generating multiple parallel slice planes along the normal direction of the stent bottom plane within the search range; calculating the cross-sectional area of each slice plane and the aortic segmentation model respectively, and selecting the one with the largest cross-sectional area as the Valsalva sinus plane.
9. The method for automatic plane positioning of aortic stents and measurement of related indicators according to claim 8, characterized in that, Step 7 includes: using the longitudinal reference plane determined in step 3 as the initial reference plane, and rotating the initial reference plane with the normal vector of the transverse reference plane as the rotation axis and the center point of the support point set as the passing point, at preset angle intervals to generate multiple support longitudinal planes with different angles.
10. The method for automatic aortic stent plane positioning and related index measurement according to claim 9, characterized in that, Step 8 includes: intersecting the stent reference plane obtained in steps 3 to 7 with the original medical imaging volume data to generate a corresponding two-dimensional cross-sectional image, and calculating the perimeter and area based on the two-dimensional cross-sectional image or the intersection area between the stent reference plane and the aortic structure.