Mitral valve leaflet length automatic measurement method based on 3D-to-2D morphological transformation

Abstract

The invention provides a mitral valve leaflet length automatic measurement method based on 3D-to-2D morphological transformation. The method comprises the following steps: firstly, acquiring three-dimensional medical image data of the heart of a patient, and extracting six partitions of a mitral valve and related heart structures; an orthogonal plane is calculated by fitting a mitral valve structure point set, and a cKDTree algorithm is used to determine a connected area of each partition, and a corresponding projection plane is established. And then extracting the contour of the projection point set through an Alpha shape algorithm, correcting the contour into a linear three-dimensional point set, and finally calculating the length of the valve leaflet of each partition. According to the method, the accuracy, consistency and efficiency of measurement are improved through an automation technology, personal errors are reduced, the method is suitable for various three-dimensional medical image data, and a more reliable basis is provided for clinical diagnosis and treatment of mitral valve diseases.

Description

Technical Field

[0001] The present invention belongs to the field of medical image processing technology, and specifically relates to a method for measuring the six-division leaflets of the mitral valve based on three-dimensional medical image data. The present invention adopts a 3D to 2D morphological transformation technology to automatically measure the length of the six-division leaflets of the mitral valve. Background Art

[0002] The mitral valve is an important valve structure between the left ventricle and the left atrium of the heart, and its normal function is crucial to maintaining the cardiovascular system. The mitral valve consists of anterior and posterior leaflets, which are further divided into six partitions: A1, A2, A3 and P1, P2, P3 according to the Carpentier classification. Accurately measuring the leaflet lengths of these six partitions is of great significance for diagnosing mitral valve disease, assessing the severity of the disease, and formulating treatment plans.

[0003] Traditional measurement of mitral valve leaflet length mainly relies on two-dimensional echocardiography, which has the following limitations:

[0004] 1. Two-dimensional imaging cannot fully display the three-dimensional structure of the mitral valve, which may lead to measurement errors in clinical diagnosis;

[0005] 2. The repeatability and comparability of the measurement results are poor and rely on the operator’s experience;

[0006] 3. It is difficult to accurately distinguish the six partitions, especially the boundaries between adjacent partitions.

[0007] With the development of medical imaging technology, three-dimensional CT and MRI have provided new possibilities for the observation of mitral valve structure. However, how to accurately extract the leaflet length of the six-division mitral valve from these three-dimensional data remains a challenge. At present, most methods still rely on manual measurement, which is not only time-consuming and labor-intensive, but also easily affected by subjective factors, making it difficult to ensure the consistency and accuracy of the measurement. Therefore, it is of great clinical significance to develop a method that can automatically, accurately and repeatably measure the leaflet length of the six-division mitral valve. This method should be able to make full use of three-dimensional medical image data, overcome the limitations of two-dimensional measurement, and be able to reduce human errors, improve the efficiency and accuracy of measurement, provide doctors with more accurate diagnostic basis, and reduce the risk of misdiagnosis and delayed treatment.

[0008] The present invention aims to solve the above problems and proposes a method for automatically measuring the length of mitral valve leaflets based on 3D to 2D morphological transformation. The method automatically identifies the mitral valve structure by processing three-dimensional medical image data, and converts the three-dimensional leaflet structure into a two-dimensional projection representation through a series of mathematical algorithms, and finally achieves accurate measurement of the length of the six partitioned leaflets. This method can not only improve the accuracy and consistency of measurement, but also greatly improve work efficiency, providing a more reliable basis for the diagnosis and treatment of mitral valve-related diseases. Summary of the invention

[0009] The present invention aims to propose a method for automatically measuring the length of mitral valve leaflets based on 3D to 2D morphological transformation by addressing the lack of current analysis and processing of three-dimensional data of left heart mitral valve structure, comprising the following steps:

[0010] Step 1: Acquire three-dimensional medical image data of the patient's heart;

[0011] Step 2: extracting the mitral valve structure from the acquired three-dimensional medical image data, including six partitions of the mitral valve leaflet structure, namely, A1 partition, A2 partition, A3 partition, P1 partition, P2 partition, and P3 partition, and extracting related cardiac structures, including the aorta, left ventricle, left myocardium, and left atrium;

[0012] Step 3: Based on the extracted mitral valve structure, the orthogonal plane of the mitral valve ring structure is calculated by fitting the structural point set;

[0013] Step 4: Based on the orthogonal planes calculated in step 3, using the mitral valve A1, P1 partition structure point set, A3, P3 partition structure point set, calculate the intersection points of the mitral valve anterior leaflet and posterior leaflet contours in sequence;

[0014] Step 5: Based on the orthogonal plane of the mitral valve annulus and any two points in space, a unique plane expression orthogonal to the plane is constructed to perform further length measurement;

[0015] Step 6: Use the cKDTree algorithm to determine the connecting areas between the six partitions of the leaflet, and define the end close to the left atrium as the vertex and the end close to the left ventricle as the bottom point respectively;

[0016] Step 7: Using the apex and bottom points of the connecting areas of the six partitions of the mitral valve determined in step 6 respectively, determine the leaflet projection planes of the six partitions in turn;

[0017] Step 8: Filter the point set of the mitral valve based on the spatial diagonal distance, and calculate the corresponding projection point set;

[0018] Step 9: Using the Alpha shapes algorithm, for the projection point sets of the six partitions, extract the contours of the projection points of each partition respectively, and further correct them into linear three-dimensional point sets;

[0019] Step 10: Calculate the spatial length of the three-dimensional point set corresponding to each partition, and finally determine the leaflet lengths of the six partitions of the mitral valve.

[0020] Furthermore, in step 1, the three-dimensional medical image data may be any medical image data containing information of the left heart region of the human body.

[0021] Furthermore, in step 2, medical annotation software is used to manually annotate the structure, or an AI deep learning algorithm is used to automatically segment the main structures of the mitral valve; the deep learning algorithms include nnUNet and SwinUNETR; the six partitions of the mitral valve after annotation or segmentation are A1 partition, A2 partition, A3 partition, P1 partition, P2 partition and P3 partition respectively.

[0022] Furthermore, in step 3, the orthogonal plane of the mitral valve ring structure Plane is determined by fitting the structure point set. m The specific method is:

[0023]

[0024] H=X T X

[0025] Hv=λv

[0026] n·(xc)=0

[0027] where p i is the i-th point in the mitral valve structure point set, n is the total number of points in the point set, x i is the centralized point, X is the matrix of all centralized points, X T is its transpose, λ is the eigenvalue, v is the corresponding eigenvector, and the parameter vector of the final plane is [n; d], the orthogonal plane Plane m The equation is expressed as:

[0028] n x x+n y y+n z zn·c=0.

[0029] Furthermore, in step 4, based on the calculated orthogonal plane, by calculating the intersection point set of the A1 and P1 partition structures and the intersection point set of the A3 and P3 partition structures, the left and right contact points of the outer contour are calculated. The specific steps are as follows:

[0030] Step 4.1: Using the A3 and P3 partition structures, calculate all adjacent point sets at the junction of the two structures, which are counted as PointsLCC; similarly, using the A1 and P1 partition structures, calculate all adjacent point sets at the junction of the two structures, which are counted as PointsRCC;

[0031] Step 4.2: Reconstrain the mitral valve annulus orthogonal plane Plane calculated in step 3 m The direction of the left atrium is defined as the positive direction;

[0032] Step 4.3: Calculate the distance between each point in the two sets of points PointsLCC and PointsRCC and Plane m Distance Dis PointsLcc 、Dis PointsRcc ;

[0033] Step 4.4: Select the points with the largest distance from the two sets of points and count them as PointLC and PointRC respectively.

[0034] PointLC=PointsLCC[Index[max(Dis PointsLcc )]]

[0035] PointRC=PointsRCC[Index[max(Dis PointsRcc )]]

[0036] Among them, Index[max(Dis)] is the index of the maximum distance point, PointLC is the intersection vertex of the mitral valve A3 and P3 partition structures, and PointRC is the intersection vertex of the mitral valve A1 and P1 partition structures.

[0037] Furthermore, in step 5, based on the orthogonal plane Plane m and two arbitrary points p in space 0 、p 1 , we get the line segment p 0 、p 1 And with Plane m The only orthogonal plane l expression.

[0038] Furthermore, in step 6, the cKDTree algorithm is used to calculate the vertex near the left atrium. and the bottom point near the left ventricle The specific steps are as follows:

[0039] Step 6.1: Use the kd tree to extract the point sets of the connected areas of the mitral valve A1-A2 partition, A2-A3 partition, P1-P2 partition and P2-P3 partition, respectively, and count them as Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3 ;

[0040] Step 6.2: Use the left atrium and left ventricle structures obtained in step 2 to calculate the center points of the left atrium and left ventricle three-dimensional structures, respectively, and record them as Center_Point Atrium 、Center_Point Ventricle ;

[0041] Step 6.3: Calculate the center point of the left atrium respectively Atrium With Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3 The distance of the point set, select the point with the smallest distance as the vertex

[0042]

[0043] Step 6.4: Calculate the center point of the left ventricle Center_Point Ventricle With Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3 The distance of the point set, the point with the smallest distance is selected as the bottom point

[0044]

[0045] Furthermore, in step 7, based on the apex and bottom point determined in step 6, the leaflet projection plane of each partition is calculated respectively, and the steps are as follows:

[0046] Step 7.1: Use Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then obtain the projection plane A2_Proj_Plane;

[0047] Step 7.2: Use Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane lExpression, and then obtain the projection plane P2_Proj_Plane;

[0048] Step 7.3: Using PointLC, Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then obtain the projection plane A1_Proj_Plane;

[0049] Step 7.4: PointLC Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then obtain the projection plane P1_Proj_Plane;

[0050] Step 7.5: Using PointRC, Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then obtain the projection plane A3_Proj_Plane.

[0051] Step 7.6: PointRC Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then obtain the projection plane P3_Proj_Plane.

[0052] Furthermore, in step 8, the specific steps of using the mitral valve point set to obtain the projection point set are as follows:

[0053] Step 8.1: Calculate the distance D from each point in each partition of the mitral valve to the corresponding projection plane i :

[0054]

[0055] Among them, ax i +by i +cz i +d∈(A1 / A2 / A3 / P1 / P2 / P3_Proj_Plane),{(x i ,y i , z i )}∈(A1 / A2 / A3 / P1 / P2 / P3Points) is the point set of each partition;

[0056] Step 8.2: Based on the calculated distance, select points whose distance from the projection plane is less than the threshold threshold to form a new point set P′. The threshold is defined as:

[0057] Step 8.3: For each point p in the selected point set P′ i , project it onto the plane and get the projection point p i，proj :

[0058]

[0059] So far, six groups of projection point sets are obtained in sequence for the six partitions of the mitral valve.

[0060] Furthermore, in step 9, the final leaflet point set is extracted using the Alpha shapes algorithm and spline difference, and the specific steps are as follows:

[0061] Step 9.1: For the point set and parameter α, the alpha shape is described as

[0062]

[0063] Where S α Convex hull contours that make up the shape;

[0064] Step 9.2: Projection point set of each partition in turn Apply the Alpha shapes algorithm in step 9.1 to generate the corresponding contour curve

[0065] Step 9.3: Use the center point of the left atrium and left ventricle Center_Point Atrium 、Center_Point Ventricle , determine the starting and ending points of each set of points and then shape the Alpha curve of each partition Corrected to a linear point set

[0066] Step 9.4: Apply cubic spline interpolation to the corrected linear point set, for Generate a uniform three-dimensional resampling point set, denoted as

[0067]

[0068] Furthermore, in step 10, the distance between each point is calculated in sequence using the uniformly sampled point set:

[0069]

[0070] Among them, p i and p i+1 are two adjacent points in the same partition;

[0071] Calculate the leaflet length of each partition in turn. For each partition, add up the distances between all points and calculate the corresponding leaflet length, specifically:

[0072]

[0073] Among them, L Ai , L Pi They represent the leaflet lengths of the three divisions of the anterior leaflet and the three divisions of the posterior leaflet of the mitral valve respectively.

[0074] Technical Effects

[0075] The automatic measurement method of mitral valve leaflet length based on 3D to 2D morphological transformation proposed in the present invention has significant technical advantages. First of all, the method has a high degree of automation and can automatically process three-dimensional medical image data without manual intervention, which greatly improves the measurement efficiency. Through 3D to 2D conversion and precise mathematical algorithms, the method can accurately measure the leaflet lengths of the six partitions of the mitral valve, reducing the measurement errors caused by human operation. In addition, due to the use of an automated algorithm, the method has a high degree of consistency among different operators or multiple measurements, ensuring the repeatability and stability of the measurement results.

[0076] This method is comprehensive and can evaluate all six partitions of the mitral valve at the same time, thus providing comprehensive valve data for clinical diagnosis. Compared with traditional manual measurement, this method greatly reduces the influence of subjective factors on the measurement results and significantly improves the objectivity and reliability of the measurement. At the same time, this method is applicable to a variety of three-dimensional medical imaging data, such as CT and MRI, which enhances its adaptability and flexibility in clinical practice.

[0077] Finally, the automated measurement process not only shortens the data processing time, but also enables doctors to obtain the required measurement results more quickly. These technical effects enable the present invention to achieve significant progress in the field of mitral valve leaflet length measurement, and provide a more reliable and efficient tool for the diagnosis and treatment of related diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 It is a schematic diagram of the process of automatic measurement of mitral valve leaflet length based on 3D to 2D morphological transformation;

[0079] Figure 2 It is the six-zone diagram described in step 2;

[0080] Figure 3 is the orthogonal plane schematic diagram described in step 3;

[0081] Figure 4 PointLC, PointRL, Schematic diagram of eight key points;

[0082] Figure 5 is a schematic diagram of the projection surface of each of the six partitions of the mitral valve described in step 7;

[0083] Figure 6 is a schematic diagram of a point set of leaflet lengths of each of the six partitions of the mitral valve described in step 9;

[0084] Figure 7 This is a schematic diagram of the leaflet length results of each of the six partitions of the mitral valve described in step 10. DETAILED DESCRIPTION

[0085] The present invention will be further described below in conjunction with specific embodiments. It should be understood that the embodiments are only used to illustrate the present invention and are not intended to limit the scope of protection of the present invention. In addition, it should be understood that after reading the contents disclosed in the present invention, those skilled in the art may make various changes or modifications to the present invention, and these equivalent forms also fall within the scope of protection defined by the present invention.

[0086] The present invention proposes a method for automatically measuring the length of mitral valve leaflets based on 3D to 2D morphological transformation. Figure 1 As shown, the following steps are included:

[0087] Step 1: Acquire three-dimensional medical image data of the patient's heart. The three-dimensional medical image data can come from different imaging technologies, including computed tomography (CT), computed tomography angiography (CTA), magnetic resonance imaging (MRI) or ultrasound imaging (US) and other imaging modes. In order to ensure the accuracy of subsequent analysis, the imaging results need to cover all relevant structures of the patient's heart. Furthermore, the image data needs to be preprocessed, including denoising and standardization, to ensure the quality of the data. These processing steps can be implemented by medical image processing software.

[0088] Step 2: Extract the mitral valve structure from the acquired three-dimensional medical image data, including the six partitions of the mitral valve leaflet structure, namely, A1 partition, A2 partition, A3 partition, P1 partition, P2 partition, and P3 partition. At the same time, extract the relevant cardiac structures, including the aorta, left ventricle, left myocardium, and left atrium, such as Figure 2As shown. In this step, the image data can be segmented by manual annotation or using an AI deep learning algorithm. For manual annotation, doctors can use professional annotation auxiliary software such as 3DSlicer or MIMICS. For automatic segmentation, deep learning-based segmentation algorithms such as nnUNet or SwinUNETR are used. These algorithms need to be trained with a large amount of annotated medical image data so that they can accurately identify the six partitions of the mitral valve and other related heart structures. Through automatic segmentation, the efficiency and accuracy of segmentation can be greatly improved.

[0089] Step 3: Based on the extracted mitral valve structure, calculate the orthogonal plane of the mitral valve ring structure by fitting the structure point set m ,like Figure 3 As shown, the specific method is:

[0090] First, calculate the center of the point set:

[0091]

[0092] Next, construct the matrix X and calculate its eigenvalues ​​and eigenvectors:

[0093] H=X T X

[0094] Hv=λv

[0095] Through these steps, we finally get the expression of the orthogonal plane parameters:

[0096] n·(xc)=0

[0097] where p i is the i-th point in the mitral valve structure point set, n is the total number of points in the point set, x i is the centralized point, X is the matrix of all centralized points, X T is its transpose, λ is the eigenvalue, v is the corresponding eigenvector, and the parameter vector of the final plane is [n; d], the orthogonal plane Plane m The equation is expressed as:

[0098] n x x+n y y+n z zn·c=0.

[0099] Step 4: Based on the orthogonal planes calculated in step 3, use the mitral valve A1, P1 partition structure point set, A3, P3 partition structure point set, and calculate the intersection points of the mitral valve anterior leaflet and posterior leaflet contours in sequence. The specific steps are as follows:

[0100] Step 4.1: Using the A3 and P3 partition structures, calculate all adjacent point sets at the junction of the two structures, which are counted as PointsLCC; similarly, using the A1 and P1 partition structures, calculate all adjacent point sets at the junction of the two structures, which are counted as PointsRCC;

[0101] Step 4.2: Reconstrain the mitral valve annulus orthogonal plane Plane calculated in step 3 m direction, and the positive direction is defined as the direction pointing to the left atrium

[0102] Plane m =-n x xn y yn z z+d;

[0103] Step 4.3: Calculate the distance between each point in the two sets of points PointsLCC and PointsRCC and Plane m Distance Dis PointsLcc 、Dis PointsRcc

[0104]

[0105] Step 4.4: Select the points with the largest distance from the two sets of points and count them as PointLC and PointRC respectively.

[0106] PointLC=PointsLCC[Index[max(Dis PointsLcc )]]

[0107] PointRC=PointsRCC[Index[max(Dis PointsRcc )]]

[0108] Among them, Index[max(Dis)] is the index of the maximum distance point, PointLC is the intersection vertex of the mitral valve A3 and P3 partition structures, and PointRC is the intersection vertex of the mitral valve A1 and P1 partition structures, such as Figure 4 shown.

[0109] Step 5: Orthogonal Plane Based on the Mitral Annus m and two arbitrary points p in space 0 、p 1 , we get the line segment p 0 、p 1 And with Plane m The only orthogonal plane l expression.

[0110] n=(n x , ny , n z )

[0111] v=p 1 -p 0

[0112] n new =n×v

[0113]

[0114] where p 0 、p 1 are any two points in space, n is the plane m The normal vector, n new Is orthogonal to Plane m With line segment p 0 -p 1 The vector of new is a constant term, then the plane cluster Plane l The expression is:

[0115] n new ·x+d new =0.

[0116] Step 6: Use the cKDTree algorithm to determine the connecting areas between the six partitions of the leaflet and define the vertices close to the left atrium. and the bottom point near the left ventricle The specific steps are as follows:

[0117] Step 6.1: Use the kd tree to extract the point sets of the connected areas of the mitral valve A1-A2 partition, A2-A3 partition, P1-P2 partition and P2-P3 partition, respectively, and count them as Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3 ;

[0118] Step 6.2: Use the left atrium and left ventricle structures obtained in step 2 to calculate the center points of the left atrium and left ventricle three-dimensional structures, respectively, and record them as Center_Point Atrium 、Center_Point Ventricle ;

[0119] Step 6.3: Calculate the center points of the left atrium Center_Points Atrium With Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3The distance of the point set, select the point with the smallest distance as the vertex

[0120]

[0121] Step 6.4: Calculate the center point of the left ventricle Center_Pint Ventricle With Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3 The distance of the point set, the point with the smallest distance is selected as the bottom point

[0122]

[0123] Step 7: Based on the apex and base points determined in step 6, calculate the leaflet projection plane of each partition, such as Figure 5 As shown, the steps are as follows:

[0124] Step 7.1: Use Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then obtain the projection plane A2_Proj_Plane

[0125]

[0126] Step 7.2: Use Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then obtain the projection plane P2_Proj_Plane

[0127]

[0128] Step 7.3: Using PointLC, Weighted calculation to get p 0 p 1 , substitute the plane Plan obtained in step 5 l Expression, and then get the projection plane A1_Proj_Plane

[0129]

[0130] Step 7.4: PointLC Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane lExpression, and then obtain the projection plane P1_Proj_Plane

[0131]

[0132] Step 7.5: Using PointRC, Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then get the projection plane A3_Proj_Plane

[0133]

[0134] Step 7.6: PointRC Weighted calculation to get p 0 p 1 , substitute the plane obtained in step 5 into Plane l Expression, and then obtain the projection plane P3_Proj_Plane

[0135]

[0136] Step 8: Filter the point set of the mitral valve based on the spatial diagonal distance and calculate the corresponding projection point set. The specific steps are as follows:

[0137] Step 8.1: Calculate the distance D from each point in each partition of the mitral valve to the corresponding projection plane i :

[0138]

[0139] Among them, ax i +by i +cz i +d∈(A1 / A2 / A3 / P1 / P2 / P3_Proj_Plane),{(x i ,y i , z i )}∈(A1 / A2 / A3 / P1 / P2 / P3Points) is the point set of each partition;

[0140] Step 8.2: Based on the calculated distance, select points whose distance from the projection plane is less than the threshold threshold to form a new point set P′. The threshold is defined as:

[0141] Step 8.3: For each point p in the selected point set P′ i , project it onto the plane and get the projection point p i，proj :

[0142]

[0143] So far, six groups of projection point sets are obtained in sequence for the six partitions of the mitral valve.

[0144] Step 9: Using the Alpha shapes algorithm, extract the contours of the projection points of each partition for the six partitions, and further correct them into a linear three-dimensional point set. The 2D point set of the leaflet length is equivalently calculated for the six partitions of the mitral valve as follows: Figure 6 As shown, the specific steps are as follows:

[0145] Step 9.1: For the point set and parameter α, the alpha shape is described as

[0146]

[0147] Where S α Convex hull contours that make up the shape;

[0148] Step 9.2: Projection point set of each partition in turn Apply the Alpha shapes algorithm in step 9.1 to generate the corresponding contour curve

[0149] Step 9.3: Use the center point of the left atrium and left ventricle Center_Point Atrium 、Center_Point Ventricle , determine the starting and ending points of each set of points and then shape the Alpha curve of each partition Corrected to a linear point set

[0150] Step 9.4: Apply cubic spline interpolation to the corrected linear point set, for Generate a uniform three-dimensional resampling point set, denoted as

[0151]

[0152] In this step, the Alpha shapes algorithm is used to generate the projection point outline of each partition. These point sets are further corrected using spline interpolation technology to adjust them to linear three-dimensional point sets. This helps to improve the accuracy of the calculation, especially in areas with large curvature.

[0153] Step 10: Calculate the spatial length of the three-dimensional point set corresponding to each partition, and finally determine the leaflet lengths of the six partitions of the mitral valve. The calculation results of the leaflet lengths of the six partitions of the mitral valve are as follows: Figure 7 As shown, the specific method is:

[0154] Using the uniformly sampled point set, calculate the distance between each point in turn:

[0155]

[0156] Among them, p i and p i+1 are two adjacent points in the same partition;

[0157] Calculate the leaflet length of each partition in turn. For each partition, add up the distances between all points and calculate the corresponding leaflet length, specifically:

[0158]

[0159] Among them, L A1 , L A2 , L A3 , L P1 , L P2 , L P3 The three lengths of the leaflets of the six partitions of the mitral valve are represented respectively. This method can effectively eliminate human errors, ensure the accuracy and consistency of the calculation results, and is suitable for the analysis of various medical imaging data.

Claims

1. A method for automatically measuring the length of mitral valve leaflets based on 3D to 2D morphological transformation, characterized in that: The following steps are involved: Step 1: Acquire three-dimensional medical image data of the patient's heart; Step 2: extracting the mitral valve structure from the acquired three-dimensional medical image data, including six partitions of the mitral valve leaflet structure, namely, A1 partition, A2 partition, A3 partition, P1 partition, P2 partition, and P3 partition, and extracting related cardiac structures, including the aorta, left ventricle, left myocardium, and left atrium; Step 3: Based on the extracted mitral valve structure, the orthogonal plane of the mitral valve ring structure is calculated by fitting the structural point set; Step 4: Based on the orthogonal planes calculated in step 3, using the mitral valve A1, P1 partition structure point set, A3, P3 partition structure point set, calculate the intersection points of the mitral valve anterior leaflet and posterior leaflet contours in sequence; Step 5: Based on the orthogonal plane of the mitral valve annulus and any two points in space, a unique plane expression orthogonal to the plane is constructed to perform further length measurement; Step 6: Use the cKDTree algorithm to determine the connecting areas between the six partitions of the leaflet, and define the end close to the left atrium as the vertex and the end close to the left ventricle as the bottom point respectively; Step 7: Using the apex and bottom points of the connecting areas of the six partitions of the mitral valve determined in step 6 respectively, determine the leaflet projection planes of the six partitions in turn; Step 8: Filter the point set of the mitral valve based on the spatial diagonal distance, and calculate the corresponding projection point set; Step 9: Using the Alpha shapes algorithm, for the projection point sets of the six partitions, extract the contours of the projection points of each partition respectively, and further correct them into linear three-dimensional point sets; Step 10: Calculate the spatial length of the three-dimensional point set corresponding to each partition, and finally determine the leaflet lengths of the six partitions of the mitral valve.

2. The method for automatically measuring the length of the mitral valve leaflet according to claim 1, characterized in that: In step 2, medical annotation software is used to manually annotate the structure, or an AI deep learning algorithm is used to automatically segment the main structure of the mitral valve; the deep learning algorithms include nnUNet and SwinUNETR; the six partitions of the mitral valve after annotation or segmentation are A1 partition, A2 partition, A3 partition, P1 partition, P2 partition and P3 partition.

3. The method for automatically measuring the length of the mitral valve leaflet according to claim 2, characterized in that: In step 3, the orthogonal plane of the mitral valve ring structure Plane is determined by fitting the structure point set. m The specific method is: H=X T X Hv=λv n·(xc)=0 where p i is the i-th point in the mitral valve structure point set, n is the total number of points in the point set, x i is the centralized point, X is the matrix of all centralized points, X T is its transpose, λ is the eigenvalue, v is the corresponding eigenvector, and the parameter vector of the final plane is [n; d], the orthogonal plane Plane m The equation is expressed as: n x x+n y y+n z zn·c=0。 4. The method for automatically measuring the length of the mitral valve leaflet according to claim 3, characterized in that: In step 4, based on the calculated orthogonal plane, by calculating the intersection point set of the A1 and P1 partition structures and the intersection point set of the A3 and P3 partition structures, the left and right contact points of the outer contour are calculated. The specific steps are as follows: Step 4.1: Using the A3 and P3 partition structures, calculate all adjacent point sets at the junction of the two structures, which are counted as PointsLCC; similarly, using the A1 and P1 partition structures, calculate all adjacent point sets at the junction of the two structures, which are counted as PointsRCC; Step 4.2: Reconstrain the mitral valve annulus orthogonal plane Plane calculated in step 3 m The direction of the left atrium is defined as the positive direction; Step 4.3: Calculate the distance between each point in the two sets of points PointsLCC and PointsRCC and Plane m Distance Dis PointsLcc 、Dis PointsRcc ; Step 4.4: Select the points with the largest distance from the two sets of points and count them as PointLC and PointRC respectively. PointLC=PointsLCC[Index[max(Dis PointsLcc )]] PointRC=PointsRCC[Index[max(Dis PointsRcc )]] Among them, Index[max(Dis)] is the index of the maximum distance point, PointLC is the intersection vertex of the mitral valve A3 and P3 partition structures, and PointRC is the intersection vertex of the mitral valve A1 and P1 partition structures.

5. The method for automatically measuring the length of the mitral valve leaflet according to claim 4, characterized in that: In step 5, based on the orthogonal plane Plane m and two arbitrary points p0 and p1 in space, we get the line segment p0 and p1 passing through the plane. m The only orthogonal plane l expression.

6. The method for automatically measuring the length of the mitral valve leaflet according to claim 5, characterized in that: Furthermore, in step 6, the cKOTree algorithm is used to calculate the vertex close to the left atrium. and the bottom point near the left ventricle The specific steps are as follows: Step 6.1: Use the kd tree to extract the point sets of the connected areas of the mitral valve A1-A2 partition, A2-A3 partition, P1-P2 partition and P2-P3 partition in turn, and count them as Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3 ; Step 6.2: Use the left atrium and left ventricle structures obtained in step 2 to calculate the center points of the three-dimensional structures of the left atrium and left ventricle respectively, which are recorded as Center _ Point Atrium 、Center_Point Ventricle ; Step 6.3: Calculate the center point of the left atrium _ Point Atrium With Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3 The distance of the point set, select the point with the smallest distance as the vertex Step 6.4: Calculate the center point of the left ventricle Center_Point Ventricle With Points A1-A2 、Points A2-A3 、Points P1-P2 、Points P2-P3 The distance of the point set, the point with the smallest distance is selected as the bottom point 7. The method for automatically measuring the length of mitral valve leaflets according to claim 6, characterized in that: In step 7, based on the apex and bottom point determined in step 6, the leaflet projection plane of each partition is calculated respectively, and the steps are as follows: Step 7.1: Use Weighted calculation results in p0 p1, which are then inserted into the plane obtained in step 5. l Expression, and then obtain the projection plane A2_Proj_Plane; Step 7.2: Use Weighted calculation results in p0 p1, which are then inserted into the plane obtained in step 5. l Expression, and then obtain the projection plane P2_Pr0j_Plane; Step 7.3: Using PointLC, The weighted calculation results in p0p1, which is then inserted into the plane obtained in step 5. l Expression, and then obtain the projection plane A1_Proj_Plane; Step 7.4: PointLC The weighted calculation results in p0p1, which is then inserted into the plane obtained in step 5. l Expression, and then obtain the projection plane P1 _ Proj_Plane; Step 7.5: Using PointRC, The weighted calculation results in p0p1, which is then inserted into the plane obtained in step 5. l Expression, and then obtain the projection plane A3 _ Proj_Plane. Step 7.6: PointRC The weighted calculation results in p0p1, which is then inserted into the plane obtained in step 5. l Expression, and then obtain the projection plane P3 _ Proj_Plane.

8. The method for automatically measuring the length of the mitral valve leaflet according to claim 7, characterized in that: In step 8, the specific steps of using the mitral valve point set to obtain the projection point set are as follows: Step 8.1: Calculate the distance D from each point in each partition of the mitral valve to the corresponding projection plane i : Among them, ax i +by i +cz i +d∈(A1 / A2 / A3 / P1 / P2 / P3 _ Proj_Plane), {(x i ,y i , z i )}∈(A1 / A2 / A3 / P1 / P2 / P3 Points) is the point set of each partition; Step 8.2: Based on the calculated distance, select points whose distance from the projection plane is less than the threshold threshold to form a new point set P′. The threshold is defined as: Step 8.3: For each point p in the selected point set P′ i , project it onto the plane and get the projection point p i,proj : So far, six groups of projection point sets are obtained in sequence for the six partitions of the mitral valve.

9. The method for automatically measuring the length of mitral valve leaflets according to claim 8, characterized in that: In step 9, the final leaflet point set is extracted using the Alpha shapes algorithm and spline difference, and the specific steps are as follows: Step 9.1: For the point set and parameter α, the alpha shape is described as Where S α Convex hull contours that make up the shape; Step 9.2: Projection point set of each partition in turn Apply the Alpha shapes algorithm in step 9.1 to generate the corresponding contour curve Step 9.3: Use the center point of the left atrium and left ventricle Center_Point Atrium 、Center_Point Ventricle , determine the starting and ending points of each set of points and then shape the Alpha curve of each partition Corrected to a linear point set Step 9.4: Apply cubic spline interpolation to the corrected linear point set, for Generate a uniform three-dimensional resampling point set, denoted as 10. The method for automatically measuring the length of the mitral valve leaflet according to claim 9, characterized in that: In step 10, the distance between each point is calculated in turn using the uniformly sampled point set: Among them, p i and p i+1 are two adjacent points in the same partition; Calculate the leaflet length of each partition in turn. For each partition, add up the distances between all points and calculate the corresponding leaflet length, specifically: Among them, L Ai , L Pi They represent the leaflet lengths of the three divisions of the anterior leaflet and the three divisions of the posterior leaflet of the mitral valve respectively.

Images