Positioning method of anterior cruciate ligament dead center based on three-dimensional geometric processing

Abstract

The invention discloses an anterior cruciate ligament dead center positioning method based on three-dimensional geometric processing, and the method comprises the steps: obtaining a knee joint CT image of a knee joint of a human body and an anterior cruciate ligament related skeleton, and obtaining a three-dimensional mesh model of a femur and a tibia after preprocessing; traversing all three-dimensional vertex coordinates of the femur three-dimensional mesh model, generating contour lines from the bottom to the top, and selecting a contour line with the maximum perimeter as a feature contour line; after the feature contour line is projected, an extrusion model is obtained in a linear extrusion treatment mode, and the top face and the bottom face of the extrusion model are filled to obtain a complete closed extrusion model; obtaining a shaft alignment bounding box of the extrusion model so as to obtain a sphere model used for cutting, and obtaining a dial plate model by using the cut communication partition so as to obtain a 1-o'clock characteristic line; two feature vertexes are positioned through two times of cutting; obtaining a final dead center coordinate of the thighbone three-dimensional mesh model; the feature contour line of the tibia three-dimensional mesh model is obtained by adopting the method, the mesh is drawn according to the proportion, and the dead center coordinates of the tibia three-dimensional mesh model are positioned by penetrating out of the platform.

Description

Technical Field

[0001] The present invention relates to the field of medical image processing, and in particular to a method for locating the center of the anterior cruciate ligament insertion based on three-dimensional geometric processing. Background Art

[0002] The knee joint is one of the most complex structures in the human body, containing many ligaments and soft tissues. Among them, the anterior cruciate ligament (ACL) controls the rotation of the knee joint and plays a key role in static and dynamic stability. However, in the related field of locating the center of the ACL reconstruction surgery insertion, there is no effective and personalized computer automatic processing method. Currently, arthroscopic reconstruction of the ACL has become the mainstream treatment method at home and abroad. Doctors analyze the CT and MRI images of patients before surgery. Subsequently, doctors use an arthroscope to observe the knee joint anatomical structure of the patient, determine the position of the insertion center and the direction of the bone tunnel to drill holes for transplantation. Among them, the most critical step in the surgery is to accurately locate the insertion center. However, in the traditional surgical method, the inherent vision limitation of the arthroscope is very large, and the local light will cause a certain degree of optical distortion on the patient's leg, which may cause significant deviation in visual assessment. In addition, the position of the insertion center is usually located by the doctor's experience, and the subjective consciousness is very strong. These factors may all affect the accuracy of the insertion center and lead to surgical failure. Patients will not only affect their normal motor ability, increase the risk of re-injury, but also cause scar tissue formation and require secondary surgery. Summary of the Invention

[0003] According to the problems existing in the prior art, the present invention discloses a method for locating the center of the anterior cruciate ligament insertion based on three-dimensional geometric processing. This method focuses on the imaging morphology of the bone itself, locates the insertion center of the femoral three-dimensional mesh model through a 1 o'clock feature line and two feature vertices, and locates the insertion center of the tibial three-dimensional mesh model by meshing the tibial plateau. The specific steps are as follows:

[0004] Obtain the knee joint CT image of the bones related to the anterior cruciate ligament of the human body, preprocess the knee joint CT image, segment the bone part in the knee joint CT image by using an adaptive threshold segmentation method, and reconstruct the segmented two-dimensional image by using the surface rendering method based on VTK to obtain the femoral three-dimensional mesh model and tibial three-dimensional mesh model of the knee joint;

[0005] Traverse all the three-dimensional vertex coordinates of the femoral three-dimensional mesh model, generate contour lines from bottom to top, and select the contour line with the largest perimeter as the feature contour line;

[0006] After projecting the feature contour line, perform a linear extrusion processing method to obtain an extrusion model, and fill the top surface and bottom surface of the extrusion model to obtain a complete closed extrusion model;

[0007] Obtain the axis-aligned bounding box of the extrusion model, calculate the centroid coordinates of the axis-aligned bounding box and the lengths of each side, create a sphere model for clipping, the center coordinates of the sphere model are the centroid coordinates of the axis-aligned bounding box, the diameter of the sphere model is the length of the shortest side of the axis-aligned bounding box, use the closed extrusion model to clip the sphere model, and the remaining sphere model after clipping forms two upper and lower connected partitions. Take the centroid of the lower connected partition as the center of the dial model, obtain the 1 o'clock direction vector of the dial model, and then obtain the 1 o'clock feature line;

[0008] Coarsely cut the three-dimensional femoral mesh model according to the centroid coordinates and normal vector direction of the feature contour line to obtain a half femoral three-dimensional mesh model, and perform fine cutting based on the axis-aligned bounding box of the half femoral three-dimensional mesh model to obtain the extraction surface of the femoral three-dimensional mesh model. The point with the largest Gaussian curvature value on this extraction surface is the feature convex point, and traverse and search along the coordinate axis direction according to the weight formula to obtain the feature mutation point;

[0009] Create a plane parallel to the three-dimensional femoral mesh model, project the feature convex point and the feature mutation point onto the plane, project the points on the 1 o'clock feature line in turn, calculate the distance difference between the points on the 1 o'clock feature line after projection and the feature convex point and the feature mutation point, and obtain the insertion point center coordinates of the final femoral three-dimensional mesh model;

[0010] Use the above method to obtain the feature contour line of the tibial three-dimensional mesh model, draw the mesh according to the ratio, and penetrate the platform to locate the insertion point center coordinates of the tibial three-dimensional mesh model.

[0011] Furthermore, when segmenting the knee joint CT image: First, convert the format of the image, perform grayscale conversion and binary preprocessing operations to obtain a complete three-dimensional femoral mesh model and a three-dimensional tibial mesh model.

[0012] Furthermore, the extrusion model of the three-dimensional femoral mesh model is obtained in the following way: Traverse all the three-dimensional vertex coordinates of the three-dimensional femoral mesh model, use the height value component as the attribute value of the vertex and generate contour lines, divide them into a set of plane contour lines from bottom to top, use the contour line with the largest perimeter as the feature contour line, and the height value where the feature contour line is located as the reference surface height value for calculating the center of the dial model. Project the feature contour line to the bottom and perform linear extrusion processing, and extrude the feature contour line to the top of the three-dimensional femoral mesh model to generate an extrusion model with the same height as the original three-dimensional femoral mesh model.

[0013] Further, the 1 o'clock feature line is obtained in the following manner: Create a spherical model for cutting, use the extrusion model to perform a cutting operation on the spherical model. After cutting, the remaining spherical model forms two connected upper and lower partitions. The lower connected partition corresponds to the intercondylar fossa part of the femoral three-dimensional mesh model. Use the centroid of the lower connected partition as the center of the dial model, and emit a 1 o'clock direction vector from the bottom to the top of the femoral three-dimensional mesh model. The connecting line of the obtained intersection points is the 1 o'clock feature line.

[0014] Further, the feature convex point is obtained in the following manner: According to the above algorithm, use the centroid of the maximum perimeter feature contour line as the center of the rough cutting plane. After cutting, obtain a half femoral three-dimensional mesh model. Calculate its axis-aligned bounding box. Determine two planes for fine cutting based on the coordinate extreme values of the axis-aligned bounding box to obtain the extraction plane. At the same time, determine the detection radius of the neighbor vertices. Use the PCA algorithm to obtain the eigenvalues and eigenvectors of the vertex values for all vertices of the extraction plane, and calculate the Gaussian curvature of each vertex. The point with the maximum Gaussian curvature is the feature convex point.

[0015] Further, the feature mutation point is obtained in the following manner: Traverse and search downward along the Z axis from the feature convex point on the half femoral three-dimensional mesh model. There is a mutation in the Y-axis coordinate near the final feature mutation point on the half femoral three-dimensional mesh model, and there is also a mutation feature in the Gaussian curvature of this point. Combine the above conditions to accurately locate the feature mutation point.

[0016] Further, the insertion point center coordinates of the femoral three-dimensional mesh model are located in the following manner: Create a projection plane parallel to the femoral three-dimensional mesh model. Project the feature convex point and the feature mutation point onto the plane. Traverse the points on the 1 o'clock feature line to calculate the projected coordinates and find the distance difference from the two feature vertices. The point with the smallest difference is the insertion point center coordinates of the final femoral three-dimensional mesh model.

[0017] Further, the insertion point center coordinates of the tibial three-dimensional mesh model are located in the following manner: Traverse all three-dimensional vertex coordinates of the tibial three-dimensional mesh model. Use the height value component as the attribute value of the vertex, and divide it into a set of contour lines. Calculate the ratio based on the plane where the contour line with the largest perimeter is located. Pierce through the tibial three-dimensional mesh model along the Z axis. The point that pierces through the tibial plateau is the insertion point center coordinates of the tibial three-dimensional mesh model.

[0018] Due to the adoption of the above technical solution, a positioning method for the center of the anterior cruciate ligament insertion point based on three-dimensional geometric processing provided by the present invention only requires the user to provide the knee joint CT images taken for the case during the implementation process. Then, according to the results of segmentation and reconstruction, the required personalized three-dimensional mesh models of the femoral and tibial bones of the knee joint can be obtained, and thus the accurate position of the insertion point center can be obtained, which is convenient for doctors to place ligament grafts. In addition, this method is completely automated and does not require any technical manual operations, thereby reducing some subjective errors caused by manual experience values, improving the accuracy of the insertion point center, enhancing the surgical efficiency, and the computer program of this method occupies less memory, has a low time complexity, and a fast running time. Brief Description of the Drawings

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only some embodiments recorded in the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0020] Figure 1 Flowchart for implementing the method of the present invention

[0021] Figure 2 Effect diagram of the contour line set of the three-dimensional mesh model of the femur in the present invention

[0022] Figure 3 Effect diagram of the projection of the contour line with the largest perimeter of the three-dimensional mesh model of the femur in the present invention

[0023] Figure 4 Effect diagram of the extrusion model in the present invention

[0024] Figure 5 Effect diagram of the extrusion model cutting and trimming the sphere model in the present invention

[0025] Figure 6 Effect diagram of the remaining connected partitions of the sphere model in the present invention

[0026] Figure 7 Effect diagram of the projection of the connected partitions of the sphere model in the present invention

[0027] Figure 8 Effect diagram of the center of the dial model in the present invention

[0028] Figure 9 Effect diagram of the dial model in the present invention

[0029] Figure 10 Effect diagram of generating the dial model from the bottom to the top of the three-dimensional mesh model of the femur in the present invention

[0030] Figure 11 Effect diagram of the 1 o'clock feature line of the femur three-dimensional mesh model in the present invention

[0031] Figure 12 Effect diagram of the half femur three-dimensional mesh model after rough cutting and trimming in the present invention

[0032] Figure 13 Effect diagram of the fine cutting and double-plane trimming in the present invention

[0033] Figure 14 Effect diagram of the surface extraction after two cuts in the present invention

[0034] Figure 15 Effect diagram of the Gaussian curvature data field of the extracted surface in the present invention

[0035] Figure 16 Effect diagram of the traversal search direction of the femur three-dimensional mesh model in the present invention

[0036] Figure 17 Effect diagram of the positioning of the feature mutation point in the present invention

[0037] Figure 18 Effect diagram of the projection plane of the femur three-dimensional mesh model in the present invention

[0038] Figure 19 Effect diagram of the projection points of the 1 o'clock feature line and the feature vertex in the present invention

[0039] Figure 20 Effect diagram of the 1 o'clock feature line and the feature vertex on the projection plane in the present invention

[0040] Figure 21 Effect diagram of the insertion point center of the femur three-dimensional mesh model in the present invention

[0041] Figure 22 Effect diagram of the projection of the maximum perimeter contour line of the tibia three-dimensional mesh model in the present invention

[0042] Figure 23 Effect diagram of the meshing of the tibia three-dimensional mesh model in the present invention

[0043] Figure 24 Effect diagram of the insertion point center at the maximum perimeter contour line of the tibia three-dimensional mesh model in the present invention

[0044] Figure 25 Effect diagram of the insertion point center of the tibia three-dimensional mesh model in the present invention Detailed implementation manners

[0045] To make the technical solutions and advantages of the present invention clearer, the following clearly and completely describes the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention:

[0046] As Figure 1 shown, a method for locating the center of the insertion point of a three-dimensional mesh model of the anterior cruciate ligament on the femur and tibia based on three-dimensional geometric processing. This method is supported by relevant medical literature and clinical requirements as the theoretical basis. The center of the insertion point of the three-dimensional mesh model of the femur is located through a 1 o'clock feature line and two feature vertices; the center of the insertion point of the three-dimensional mesh model of the tibia is located by grid method. In the specific implementation process, an extrusion model with the same height as the three-dimensional mesh model of the femur is generated through the feature contour line with the largest perimeter, and the effect after optimization and trimming is as Figures 2 to 4 shown. Create a sphere with a uniform distribution of triangular patches for trimming the extrusion model, as Figure 5 shown. After trimming, the three-dimensional mesh model of the intercondylar fossa of the femur can be obtained. Taking its centroid as the center of the dial model, the coordinates of other positions are calculated, as Figures 6 - 9 shown. Generate a dial model from the bottom to the top of the three-dimensional mesh model of the femur at a step size, and emit rays according to the given direction vector to form a 1 o'clock feature line, as Figures 10 - 11 shown. Coarsely cut the three-dimensional mesh model of the femur according to the above-mentioned feature contour line, and the result is as Figure 12 shown. Further, perform fine cutting according to the axis-aligned bounding box of the half three-dimensional model of the femur, as Figure 13 shown. After obtaining the extraction surface, generate a data field according to the Gaussian curvature to locate the feature convex points, as Figure 15 shown. Traverse and compare the mutation of the Y-axis coordinate and the mutation of the Gaussian curvature along the fixed direction of the three-dimensional model of the femur to locate the feature mutation points, as Figures 16 - 17 shown. Create a projection plane, project the 1 o'clock feature line and the two feature vertices onto the plane, and the point on the 1 o'clock feature line with the smallest difference in distance to the two feature vertices is the center of the insertion point of the three-dimensional mesh model of the femur, as Figures 18 - 21 shown. For the tibia, the above method is used to locate the center of the insertion point of the three-dimensional mesh model of the tibia by grid method with the feature contour line with the largest perimeter, as Figures 22 - 25 shown.

[0047] The specific steps of the method disclosed in the present invention are as follows:

[0048] S1: Obtain the knee joint CT images of the bones related to the anterior cruciate ligament of the human body, and preprocess the above knee joint CT images. The specific method is as follows:

[0049] S11. Convert the knee joint CT image into a grayscale image: Traverse the pixel values of each pixel point of the knee joint CT image. By processing each pixel point, the knee joint CT image can be converted into a grayscale image. Then perform binary processing on the grayscale image. Using the adaptive threshold segmentation method, set the pixel points with values greater than the threshold to 255 and the pixel points with values less than the threshold to 0.

[0050] S12. Reconstruct the segmented 2D image using the VTK-based surface rendering technique, extract the part of interest in the form of an isosurface. Here, the vtkContourFilter is used to extract the isosurface, and the value of this method is set to 170 for convenient observation and analysis. The surface rendering method is fast and suitable for real-time interactive operations.

[0051] By inputting the same pixel distance and slice thickness as when taking the knee joint CT image through the SetDataSpacing function, a three-dimensional mesh model of the same size as the patient's real femur and tibia can be obtained.

[0052] S2: Traverse all the three-dimensional vertex coordinates of the femur three-dimensional mesh model to obtain the characteristic contour line. The specific method is as follows:

[0053] S21. After inputting the femur three-dimensional mesh model obtained in S1, traverse all the three-dimensional vertex coordinates of the femur three-dimensional mesh model, take the Z value (height value) component as the attribute value of the vertex, and generate contour lines according to this attribute value. As Figure 2 shown, according to the effect of multiple tests of the algorithm, the generation step size is set to 0.50 mm. The contour lines divide the femur three-dimensional mesh model from the bottom to the top into a set of planar contour lines.

[0054] S22. Traverse this set of planar contour lines, use the vtkThreshold to simplify the data operation, and only retain the scalar values of the same contour line. If there are no points on the generated surface, that is, there are no valid contour lines, then increase the step size for the next loop iteration. Calculate the perimeter of each contour line, which is obtained by calculating the sum of the lengths of all the line segments that make up the contour line, and select the contour line with the largest perimeter as the characteristic contour line, and its height value (Z value) as the reference surface height value for subsequent calculation of the center of the dial model. The characteristic contour line is as Figure 3 shown.

[0055] S3: Obtain the extrusion model with the same height as the femur three-dimensional mesh model in the following form:

[0056] S31. Calculate the axis-aligned bounding box of the femur three-dimensional mesh model, obtain the extreme values of the model in the three axial directions, and project the characteristic contour line vertically along the negative Z axis to the bottom of the axis-aligned bounding box to obtain the projected contour line.

[0057] S32. Use the vtkLinearExtrusionFilter to obtain the extrusion model by linear extrusion processing, extrude from the bottom to the top of the axis-aligned bounding box of the femur three-dimensional mesh model to obtain a columnar extrusion model. After obtaining the extrusion model, fill the holes on the top and bottom surfaces to obtain a complete and closed extrusion model. As Figure 4 shown, the complete and closed extrusion model is more conducive to subsequent cutting operations.

[0058] S4: Use the extrusion model to cut the sphere model. The lower connected partition formed by the remaining sphere model after cutting can accurately correspond to the morphological structure at the intercondylar fossa of the femoral three-dimensional mesh model, and then obtain the center of the dial model and the 1 o'clock feature line. The specific method is as follows:

[0059] S41. Obtain the axis-aligned bounding box of the extrusion model, calculate the center-of-mass coordinates of the axis-aligned bounding box and the lengths of each side. Create a sphere model for cutting. The triangular patches of the sphere are evenly distributed and sufficient in number, as Figure 5 shown. The center-of-sphere coordinates of this sphere model are the center-of-mass coordinates of the axis-aligned bounding box, and the diameter of the sphere model is the length of the shortest side of the axis-aligned bounding box.

[0060] S42. Use the closed extrusion model to cut the sphere model. Use the vtkImplicitPolyDataDistance function to convert the closed extrusion model into a corresponding implicit function. Calculate the shortest distance from any point in the closed extrusion model to the surface of the sphere model, and then convert the polygon data into a distance field, and convert the cutting into an execution based on the distance from the point to the surface of the mesh model, which is suitable for processing three-dimensional mesh models with complex geometric shapes.

[0061] S43. Obtain the remaining part of the sphere model after being cut in the above way. As Figure 6 shown, the remaining part will form two connected partitions. The lower connected partition, that is, the partition with the largest projected area, corresponds to the position of the intercondylar fossa of the femoral three-dimensional mesh model. Set the height value of the lower connected partition to the height value where the feature contour line is located, and then move the lower connected partition to the horizontal plane at the height where the feature contour line is located. Calculate the centroid of the lower connected partition, and use the centroid as the center of the dial model, as Figures 7 - 8 shown.

[0062] According to the center of the dial model, declare the unit vectors of the 12 scales of the dial and add them to the center coordinates of the dial. Therefore, the directions of each scale can be obtained, and the vector length is customized to generate the dial model. Calculate the coordinates of the 12 scale positions according to the formula. Among them, the center of the dial model is located at (x0, y0), the radius is R, and the number of hours on the dial model is h (where h = 1, 2... 12), then the coordinates of different positions can be expressed as (x h , y h ). Use the 12 scale point coordinates as control points to generate a spline function as the circular smooth boundary line of the dial model, and finally draw the complete dial model. As Figure 9 shown.

[0063]

[0064] S44. Launch a ray with the center coordinates of the dial model as the starting point, infinity as the end point, and 1 o'clock as the direction, find the intersection of the ray and the femur 3D mesh model, modify the height value of the dial model center, traverse from the bottom to the top of the axis-aligned bounding box of the femur 3D mesh model, check whether the center of the dial model is inside the femur 3D mesh model, set different step lengths, compare the results, and finally select 0.50mm as the step length to launch rays for each position as described above to find the intersection, such as Figure 10 As shown, the result is stored to obtain a set of intersection points, and the set of intersection points is connected to obtain the 1 o'clock feature line that best fits the morphological structure of the femur three-dimensional mesh model. Figure 11 shown.

[0065] S5: Based on the characteristic contour line and the axis-aligned bounding box of the half-femur 3D mesh model, the femur 3D mesh model is roughly cut and finely cut to obtain the extracted surface of the femur 3D mesh model, and Gaussian curvature is calculated to locate the characteristic convex points, and the characteristic mutation points are obtained by traversing and searching along the coordinate axis. The specific method is as follows:

[0066] S51. The method of extracting characteristic contour lines is consistent with that in S2. First, the characteristic contour line with the largest circumference is extracted. Then, the centroid coordinates of the characteristic contour line are calculated. Subsequently, the centroid coordinates of the characteristic contour line are used as the control points for segmenting the femoral 3D mesh model to generate a cutting plane. A parameter plane with a normal vector of (1,0,0) is generated at the centroid, that is, the normal vector points to the positive direction of the X-axis. The parameter plane is used to roughly cut the femoral 3D mesh model, and the half of the femoral 3D mesh model pointed to by the normal vector is retained, such as Figure 12 shown.

[0067] S52. Obtain two mutually perpendicular fine cutting planes according to the axis-aligned bounding box of the half-femur 3D mesh model, with normal vectors (0,0,1), (0,-1,0) and the detection radius of the subsequent neighbor vertex set. The radius size is set to 1 / 20 of the diagonal length of the bounding box of the half-femur 3D mesh model. After cutting, the extraction surface of the femur 3D mesh model can be obtained, such as Figure 13 As shown in Figure 2, two cuts can not only save the running time of the algorithm, but also solve the problem of under-adoption of vertices at the edge.

[0068] S53. Using the PCA statistical method, traverse each vertex of the extracted face, and select all neighboring vertices within a given radius for PCA calculation, and find the main direction of each vertex, including eigenvectors and eigenvalues. The eigenvector of each vertex is as follows: Figure 14 shown.

[0069] After PCA calculation, each vertex has three eigenvalues and three eigenvectors. The eigenvector corresponding to the largest eigenvalue λ1 is the direction of the maximum normal curvature, and the eigenvector corresponding to the smallest eigenvalue λ0 is the direction of the minimum normal curvature. Subsequently, the product K of the maximum normal curvature and the minimum normal curvature is calculated according to the formula as the Gaussian curvature of the vertex. After traversing and extracting the face vertices for calculation, a data field is generated, as Figure 15 shown. The point with the largest Gaussian curvature value is the characteristic convex point.

[0070]

[0071] S54. Traverse and search downward along the Z-axis coordinate direction from the characteristic convex point on the three-dimensional mesh model of the half femur. It can be easily seen from the coordinate axis that the characteristic mutation point is the vertex with a mutation in the Y-axis coordinate, as Figure 16 shown. Compare the Y-axis coordinates of the two pointers P and Q along the search direction. Δy(z) = ∣y p(z) -y p(z) ∣. The intersection point with the largest value is the characteristic mutation point. And there is also a certain mutation characteristic in that the Gaussian curvature data field at this point changes from red to yellow. Combining the above, the characteristic mutation point can be accurately located according to the weight formula. Take α = 0.40 and β = 0.60. The result is as Figure 17 shown.

[0072] W(z) = α·Δy(z) + βGaussianCurvatureChange(z)

[0073] S6: Locate the center coordinates of the insertion point of the final three-dimensional mesh model of the femur based on the plane parallel to the three-dimensional mesh model of the femur. The specific method is as follows:

[0074] Create a projection plane parallel to the three-dimensional model of the femur. According to the theory in medical literature, the center of the insertion point of the three-dimensional mesh model of the femur is located on the 1 o'clock feature line and at the center of the circle of the lateral condyle of the femur. The two characteristic vertices are the tangent points of the circle. Therefore, first project the two characteristic vertices onto the plane. Assume the projection plane is ax + by + cz + d = 0. The point P = (x p , y p , z p ) projected onto the plane is the point S = (x s , y s , z s ). The projection coordinate conversion formula is

[0075]

[0076] Project the feature convex points and feature mutation points onto a plane, traverse the points on the 1 o'clock feature line, project the points on the 1 o'clock feature line in turn, calculate the distance difference between the projected points on the 1 o'clock feature line and the projected coordinates of the feature convex points and feature mutation points, and the one with the smallest difference is the center of the insertion point of the final femoral three-dimensional mesh model. Extract the id value of the target point and restore it to the femoral three-dimensional model to obtain the coordinates of the center of the insertion point of the femoral three-dimensional mesh model. The results are as Figures 18 - 21 shown.

[0077] S7: The center of the insertion point of the tibia is located by drawing a grid according to a ratio, and the specific method is as follows:

[0078] S71. Fine-tune the angle of the tibial three-dimensional mesh model to make the left and right parts of the tibial plateau more horizontal. First, obtain the tibial feature contour line according to the method in S2, calculate the geometric center point and centroid of the tibial feature contour line, and find the rotation value that makes the minimum distance according to the distance between the centroid and the center of the tibial feature contour line, and align it so that the centroid of the rotated tibial three-dimensional mesh model is aligned with the geometric center of the tibial feature contour line.

[0079] S72. After rotating the tibial three-dimensional mesh model, obtain the extreme value range to grid the tibial plateau according to the axis-aligned bounding box of the tibial feature contour line. The range bounds = [x min , x max , y min , y max , z min , z max . Draw the grid according to the ratio.

[0080] As Figure 22 shown, after meshing, traverse and locate all the points inside the tibia at the height of the tibial feature contour line according to different coordinate ratios. As Figure 23 shown. According to the ratio in medical theory, the coordinates of the center of the insertion point of the internal tibial three-dimensional mesh model can be initially located according to the following formula c = (x c , y c , z c ). Among them, frontBackPercent takes 0.5 and inOutPercent takes 0.5.

[0081] x c = x max - frontBackPercent · (x max - x min )

[0082] y c = y max - inOutPercent · (y max - y min)

[0083] z c = z max

[0084] Subsequently, a vertex finder is set up to emit a ray from the inside of the three-dimensional tibial mesh model with the center of the internal insertion point as the starting point and infinity as the ending point. The ray passes through the tibial plateau, and the intersection point with the tibial plateau is the final center of the insertion point of the three-dimensional tibial mesh model, as Figures 24 - 25 shown.

[0085] The present invention has the characteristics of personalization and precision. It can directly automatically locate the center of the insertion point of the anterior cruciate ligament of the human knee joint according to the corresponding algorithm by inputting the knee joint CT image, without too many complex operations. The 1 o'clock feature line, feature vertices, and meshes involved in this method are all located based on the graphic features of the bone model itself. Each parameter of the algorithm is obtained personalized based on the three-dimensional mesh model features of the patient's femur and tibia. Compared with the traditional method of visually locating from the knee joint CT image by orthopedic doctors relying on empirical values, the accuracy is greatly improved. At the same time, this method has strong prior medical theory support and meets clinical requirements. For bones with various shapes, the algorithm can achieve precise positioning, with a short running time, low time complexity, and high reliability. It solves the problems faced by doctors.

[0086] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes an equivalent substitution or change, and should be covered within the protection scope of the present invention.

Claims

1. A method for locating the center of the anterior cruciate ligament insertion point based on three-dimensional geometric processing, the characteristics of which include: Acquire a knee joint CT image of a human knee joint and anterior cruciate ligament-related bones, preprocess the knee joint CT image, segment the bone part in the knee joint CT image using an adaptive threshold segmentation method, reconstruct the segmented two-dimensional image based on a VTK surface rendering method, and obtain a knee joint femur three-dimensional mesh model and a knee joint tibia three-dimensional mesh model; Traverse all the three-dimensional vertex coordinates of the femur three-dimensional mesh model, generate contour lines from the bottom to the top, and select the contour line with the maximum circumference as the characteristic contour line; After projecting the characteristic contour line, a linear extrusion processing method is used to obtain an extrusion model, and the top and bottom surfaces of the extrusion model are filled to obtain a complete closed extrusion model; Get the axis-aligned bounding box of the extrusion model, calculate the body-center coordinates and the lengths of each side of the axis-aligned bounding box, create a spherical model for clipping, the spherical center coordinates of the spherical model are the body-center coordinates of the axis-aligned bounding box, the diameter of the spherical model is the length of the shortest side of the axis-aligned bounding box, and the spherical model is clipped using the closed extrusion model. The remaining spherical model after clipping forms two upper and lower connected partitions, and the centroid of the connected partition at the bottom is used as the center of the dial model. The 1 o'clock direction vector of the dial model is obtained, and then the 1 o'clock feature line is obtained; The femur 3D mesh model is roughly cut according to the centroid coordinates and normal vector direction of the characteristic contour line to obtain the half-femur 3D mesh model, and the half-femur 3D mesh model is finely cut according to the axis-aligned bounding box of the half-femur 3D mesh model to obtain the extraction surface of the femur 3D mesh model, wherein the point with the largest Gaussian curvature value of the extraction surface is the characteristic convex point, and the characteristic mutation point is obtained by traversing and searching along the coordinate axis according to the weight formula; Create a plane parallel to the femoral 3D mesh model, project the characteristic convex points and characteristic mutation points onto the plane, project the points on the 1 o'clock characteristic line in turn, calculate the distance difference between the points on the 1 o'clock characteristic line after projection and the characteristic convex points and characteristic mutation points, and obtain the final stop center coordinates of the femoral 3D mesh model; The above method is used to obtain the characteristic contour line of the tibial three-dimensional mesh model, draw the mesh in proportion, and penetrate the platform to locate the center coordinate of the stop point of the tibial three-dimensional mesh model.

2. The method for locating the center of the anterior cruciate ligament insertion point based on three-dimensional geometric processing according to claim 1, characterized in that: When segmenting the knee joint CT image: first convert the image format, perform grayscale conversion and binarization preprocessing operations to obtain a complete femur 3D mesh model and a tibia 3D mesh model.

3. The method for locating the center of the anterior cruciate ligament insertion point based on three-dimensional geometric processing according to claim 2, characterized in that: The extrusion model of the femur three-dimensional mesh model is obtained in the following manner: traverse all the three-dimensional vertex coordinates of the femur three-dimensional mesh model, use the height value component as the attribute value of the vertex and generate contour lines, divide it into a set of plane contour lines from bottom to top, use the contour line with the largest circumference as the characteristic contour line, and use the height value of the characteristic contour line as the reference surface height value for calculating the center of the dial model. The characteristic contour line is projected to the bottom and linearly extruded, and the characteristic contour line is extruded to the top of the femur three-dimensional mesh model to generate an extrusion model with the same height as the original femur three-dimensional mesh model.

4. The method for locating the anterior cruciate ligament insertion center based on three-dimensional geometric processing according to claim 3, characterized in that: The 1 o'clock feature line is obtained in the following manner: a spherical model for cropping is created, and the spherical model is cropped using an extrusion model. The remaining spherical model after cropping forms two upper and lower connected partitions. The connected partition at the bottom corresponds to the intercondylar fossa of the femoral three-dimensional mesh model. The centroid of the connected partition at the bottom is used as the center of the dial model. A 1 o'clock direction vector is emitted from the bottom to the top of the femoral three-dimensional mesh model, and the resulting intersection line is the 1 o'clock feature line.

5. The method for locating the anterior cruciate ligament insertion center based on three-dimensional geometric processing according to claim 1, characterized in that: The characteristic convex points are obtained in the following manner: according to the above algorithm, the centroid of the characteristic contour line with the maximum perimeter is used as the center of the rough cutting plane, and a three-dimensional mesh model of half of the femur is obtained after cutting, and an axis-aligned bounding box is calculated for it. According to the coordinate extreme values ​​of the axis-aligned bounding box, two planes are determined for fine cutting to obtain an extracted surface, and the detection radius of neighbor vertices is determined at the same time. The PCA algorithm is used to obtain the eigenvalues ​​and eigenvectors of the vertex values ​​of all vertices of the extracted surface, and the Gaussian curvature of each vertex is calculated. The point with the largest Gaussian curvature is the characteristic convex point.

6. The method for locating the anterior cruciate ligament insertion center based on three-dimensional geometric processing according to claim 5, characterized in that: The characteristic mutation point is obtained in the following manner: a downward traversal search is performed along the Z axis from the characteristic convex point on the three-dimensional mesh model of the half femur. The three-dimensional mesh model of the half femur has a Y-axis coordinate mutation near the final characteristic mutation point, and the Gaussian curvature of the point also has a mutation feature. The characteristic mutation point is accurately located by combining the above conditions.

7. The method for locating the anterior cruciate ligament insertion center based on three-dimensional geometric processing according to claim 6, characterized in that: The center coordinates of the end point of the femoral 3D mesh model are located in the following way: create a projection plane parallel to the femoral 3D mesh model, project the feature convex points and feature mutation points onto the plane, traverse the points on the 1 o'clock feature line to calculate the projected coordinates and find the distance difference with the two feature vertices. The one with the smallest difference is the final center coordinates of the end point of the femoral 3D mesh model.

8. The method for locating the anterior cruciate ligament insertion center based on three-dimensional geometric processing according to claim 7, characterized in that: The center coordinates of the stop point of the tibia three-dimensional mesh model are located in the following way: traverse all the three-dimensional vertex coordinates of the tibia three-dimensional mesh model, use the height value component as the attribute value of the vertex, divide it into a set of contour lines, calculate the proportion based on the plane where the contour line with the largest circumference is located, and pass through the tibia three-dimensional mesh model along the Z axis. The point that passes through the tibial plateau is the center coordinates of the stop point of the tibia three-dimensional mesh model.

Images