A method for extracting acetabular lunar surface model based on hybrid curved surface fitting

Abstract

The present application relates to a method for extracting an acetabular facies lunata model, in particular to a method for extracting an acetabular facies lunata model based on mixed surface fitting, which solves the technical problem of low precision and robustness of the existing method for extracting an acetabular facies lunata model. The present application realizes rough extraction of the acetabular facies lunata model through spherical surface fitting, and then realizes accurate extraction of the acetabular facies lunata model in combination with ellipsoidal surface fitting, which combines the stability of spherical surface fitting and the anatomical fidelity of ellipsoidal surface fitting. In particular, in the case of complex pathologies such as hip dysplasia, the present application can still maintain high extraction accuracy, and improve the accuracy and reliability of clinical diagnosis and surgical planning. Meanwhile, in the iteration process of ellipsoidal surface fitting, an adaptive fault tolerance threshold is introduced, which effectively solves the problems of sensitivity to noise and strong parameter dependence of traditional methods, and has better robustness in processing complex acetabular morphology.

Description

Technical Field

[0001] This invention relates to a method for extracting the lunate surface model of the acetabulum, specifically a method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting. Background Technology

[0002] With the development of 3D imaging and computer-aided technology, precise quantification of the lunate surface of the acetabulum has become an important means to improve the diagnosis and treatment of hip joint diseases. The acetabulum, part of the pelvis, is a cup-shaped structure that forms the hip joint with the femoral head. The inner surface of the acetabulum is not completely covered by articular cartilage; the weight-bearing area is called the lunate surface, named for its crescent-shaped appearance. The extraction and parameter measurement of the acenate lunate surface model not only accurately reflects the morphology of the acetabulum but is also fundamental to understanding the anatomy and pathology of the hip joint. Furthermore, it provides an accurate and reliable data foundation for hip joint imaging diagnosis, preoperative planning, health assessment, and biomechanical research.

[0003] Chinese patent CN114947905A discloses a system and method for generating a lunate surface structure model of the acetabulum. This method extracts a hip bone model based on three-dimensional images (CT and MRI scan data) and extracts acetabular surface points by traversing vertex segments along the normal direction to intersect the femoral head. This process is repeated to extract multiple acetabular surface points, forming a set of acetabular surface points. Next, a set of triangular facets is generated by traversing the triangular facets in the hip bone model, each containing at least three vertices as subsets of the acetabular surface point set. Finally, isolated noise is removed based on the triangular facet set, resulting in an accurate lunate surface model of the acetabulum. However, this method fails to accurately extract the lunate surface model in cases of hip dislocation or subluxation, hip dysplasia, and especially when the acetabulum does not adequately cover the femoral head.

[0004] De Raedt et al. proposed an automatic acetabular lunate segmentation method (see De Raedt, Sepp, et al. "Lunate extract: fully automatic acetabular lunate segmentation and hipangle measurements." Acta Radiologica 66.11 (2025): 1208-1216). This method employs a graph cut technique, combining boundary and region terms based on surface curvature and hip-femoral surface compatibility. The boundary term includes curvature and surface compatibility metrics. First, the cost based on the maximum curvature is calculated by approximating the curvature of each point on the grid, while simultaneously assessing hip-femoral compatibility by comparing the dot product of surface normals. The region term defines two regions: one containing points identified as the acetabular lunate, and the other a region farther from the surface with lower curvature. Combining these two terms, the algorithm accurately segments the acetabular lunate by minimizing an energy function. Parameter optimization is performed through a grid search, aiming to minimize the squared distance between the automatic segmentation and the manual reference points to achieve optimal segmentation results. This method relies heavily on accurate curvature calculation and the quality of the original model. If the input data is of poor quality or the curvature calculation is not accurate enough, the robustness and applicability of the method will be greatly reduced, which may lead to inaccurate segmentation results, especially in complex or diseased areas.

[0005] The study by Upasani et al. (see Upasani, Vidyadhar V., et al. "Assessment of three-dimensional acetabular coverage angles." Journal of Hip PreservationSurgery 7.2 (2020): 305-312.) mentions using a least-squares regression method to fit a sphere to calculate the coverage angle and surface area of ​​the acetabulum, and measuring the coverage angle in different regions. However, due to the inherent errors in the fitting algorithm, the center and radius of the fitted sphere will have certain errors, making it difficult to adapt to the anatomical relationship between the acetabulum and femoral head under different conditions, resulting in incomplete or over-extraction of the lunate plane of the acetabulum. Summary of the Invention

[0006] The purpose of this invention is to solve the technical problems of low accuracy and robustness of existing methods for extracting the acetabular lunate surface model, and to provide a method for extracting the acetabular lunate surface model based on hybrid surface fitting.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] A method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting, characterized by the following steps:

[0009] Step 1: Obtain the acetabular model and the corresponding femoral model of the acetabular lunate plane model to be extracted;

[0010] Step 2: Obtain the gold standard model of the femur, map the anatomical markers of the femoral head on the gold standard model of the femur to the femoral model, and perform spherical fitting on the anatomical markers of the femoral head mapped to the femoral model to obtain the fitted sphere and its center and radius;

[0011] Step 3: Enlarge the radius of the fitted sphere to obtain the radius of the spherical bounding box. Then, using the center of the fitted sphere as the center and the radius of the spherical bounding box as the radius, extract the initial region of interest of the lunate surface of the acetabulum in the acetabular model.

[0012] Step 4: Calculate the residuals from all points in the initial region of interest of the acetabular lunate to the fitted sphere, and then extract the initial point set of the acetabular lunate from the initial region of interest of the acetabular lunate based on the residuals.

[0013] Step 5: Fit an ellipsoid to the initial point set of the acetabular lunate surface to obtain the fitted ellipsoid, its center, and semi-axis length. Then, calculate the residuals from all points in the initial region of interest of the acetabular lunate surface to the fitted ellipsoid, and extract candidate point sets from the initial region of interest of the acetabular lunate surface based on the residuals.

[0014] Step 6: Use the candidate point set as the initial point set of the acetabular lunate plane, and then return to step 5 until the center of the fitted ellipsoid and the length of the semi-axis satisfy the convergence condition, and use the candidate point set as the point set of the acetabular lunate plane.

[0015] Step 7: Map the point set of the acetabular lunate surface to the point-face relationship of the acetabular model, extract the vertices and triangular facets corresponding to the point set of the acetabular lunate surface in the acetabular model, and obtain the acetabular lunate surface model, thus completing the extraction of the acetabular lunate surface model.

[0016] Furthermore, step 4 specifically involves:

[0017] Step 4.1: Calculate the residuals from the initial region of interest in the lunate plane of the acetabulum to the fitted sphere using the following formula:

[0018]

[0019] in, Let be the residual from the i-th point in the initial region of interest of the acetabular lunate to the fitted sphere, where i is an integer and 1≤i≤I, and I is the number of points in the initial region of interest of the acetabular lunate. Let i be the i-th point in the initial region of interest of the lunate plane of the acetabulum. , Let be the center and radius of the fitted sphere, respectively. This represents the distance from the i-th point in the initial region of interest of the acetabular lunate to the center of the fitted sphere;

[0020] Step 4.2: Define the fault tolerance threshold according to the following formula:

[0021]

[0022] in, This is the fault tolerance threshold; This represents the set of residuals between the initial region of interest on the lunate surface of the acetabulum and the fitted sphere. ; , They are respectively The median and standard deviation;

[0023] Step 4.3: Select points with residuals less than the tolerance threshold in the initial region of interest of the acetabular lunate plane to obtain the region of interest point set. Then, perform connectivity filtering on the region of interest point set to obtain multiple connected regions. Select the points in the connected region with the largest area as the initial point set of the acetabular lunate plane.

[0024] Furthermore, step 5 specifically includes:

[0025] Step 5.1: Use the least squares method to fit the initial point set of the acetabular lunate surface to an ellipsoid, and obtain the fitted ellipsoid and its center and semi-axis length;

[0026] Step 5.2: Calculate the residuals from the fitted ellipsoid to all points in the initial region of interest on the lunate surface of the acetabulum using the following formula:

[0027]

[0028]

[0029] in, Let be the residual from the i-th point in the initial region of interest of the acetabular lunate surface to the fitted ellipsoid. , Let be the center of the fitted ellipsoid and the equivalent radius, respectively. This represents the distance from the i-th point in the initial region of interest of the acetabular lunate to the center of the fitted ellipsoid. , , These represent the lengths of the three semi-axes of the fitted ellipsoid;

[0030] Step 5.3: Define the adaptive fault tolerance threshold according to the following formula:

[0031]

[0032] in, An adaptive fault tolerance threshold; This represents the set of residuals between the initial region of interest on the lunate surface of the acetabulum and the fitted ellipsoid. ; , They are respectively The median and standard deviation;

[0033] Step 5.4: Select points with residuals less than the adaptive fault tolerance threshold in the initial region of interest of the acetabular lunate plane to obtain a preliminary set of points. Then, perform connectivity filtering on the preliminary set of points to obtain multiple candidate connected regions. Select the points in the candidate connected region with the largest area as the candidate set of points.

[0034] Furthermore, in step 2, a geodesic-based Bayesian coherence point drift algorithm is used to map the femoral head anatomical markers on the gold standard femoral model to the femoral model.

[0035] Furthermore, step 1 also includes smoothing the acetabular model and the femoral model respectively.

[0036] Furthermore, in step 1, a window function Singer smoothing filter is used to smooth the acetabular model and the femur model respectively.

[0037] Furthermore, in step 3, the radius of the spherical bounding box is R0+Z, where R0 is the radius of the fitted sphere, Z is the magnification value, and 3mm≤Z≤10mm.

[0038] Furthermore, in step 2, the anatomical landmarks of the femoral head mapped onto the femoral model are spherically fitted using the least squares method.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] This invention provides a method for extracting the acetabular lunate model based on hybrid surface fitting. It achieves coarse extraction of the acetabular lunate model through spherical fitting, and then combines this with ellipsoidal fitting for precise extraction. This method combines the stability of spherical fitting with the anatomical fidelity of ellipsoidal fitting, maintaining high extraction accuracy, especially in complex conditions such as hip dysplasia, thus improving the accuracy and reliability of clinical diagnosis and surgical planning. Furthermore, an adaptive fault-tolerant threshold is introduced during the iterative process of ellipsoidal fitting, effectively solving the problems of noise sensitivity and strong parameter dependence in traditional methods, making it more robust when dealing with complex acetabular morphologies. Attached Figure Description

[0041] Figure 1 This is a flowchart of a method according to an embodiment of the present invention;

[0042] Figure 2 This is a flowchart of step 4 of an embodiment of the present invention;

[0043] Figure 3 This is a flowchart of steps 5-7 of an embodiment of the present invention;

[0044] Figure 4 The following are comparison diagrams of the initial region of interest, the initial point set / candidate point set of the acetabular lunate surface, and the fitted sphere / fitted ellipsoid in the acetabular model obtained in the embodiments of the present invention. Among them, (a) is a schematic diagram of the initial region of interest, the initial point set of the acetabular lunate surface, and the fitted sphere in the acetabular model, and (b)-(f) are schematic diagrams of the initial region of interest, the candidate point set, and the fitted ellipsoid in the acetabular model for different iteration rounds, respectively.

[0045] Figure 5 The following is a comparison diagram of the initial region of interest, the initial point set / candidate point set of the acetabular lunate surface, and the fitted sphere / fitted ellipsoid in the acetabular model from another perspective obtained in the embodiments of the present invention. Among them, (a) is a schematic diagram of the initial region of interest, the initial point set of the acetabular lunate surface, and the fitted sphere in the acetabular model, and (b)-(f) are schematic diagrams of the initial region of interest, the candidate point set, and the fitted ellipsoid in the acetabular model for different iteration rounds, respectively. Detailed Implementation

[0046] The following detailed description, in conjunction with the accompanying drawings and specific embodiments, provides a method for extracting the acetabular lunate surface model based on hybrid surface fitting, as proposed in this invention. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of this invention and are not intended to limit the scope of protection of this invention.

[0047] A method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting, such as... Figure 1 As shown, it includes the following steps:

[0048] Step 1: Obtain the acetabular model and the corresponding femoral model of the acetabular lunate plane model to be extracted, and use the window function Singer smoothing filter to smooth the acetabular model and the femoral model respectively.

[0049] In step 1, a window function Singer smoothing filter is used to smooth the acetabular and femoral models, which can remove the step artifacts and high-frequency noise generated during medical image segmentation.

[0050] Step 2: Obtain the gold standard model of the femur, and then use the geodesic-aware Bayesian Coherent Point Drift (G-BCPD) algorithm to map the anatomical markers of the femoral head on the gold standard model to the femoral model. Then, use the least squares method to perform spherical fitting on the anatomical markers of the femoral head mapped to the femoral model to obtain the fitted sphere and its center and radius.

[0051] Step 3: Enlarge the radius of the fitted sphere to obtain the radius of the spherical bounding box. Then, using the center of the fitted sphere as the center and the radius of the spherical bounding box as the radius, extract the initial region of interest (ROI) of the lunate surface of the acetabulum in the acetabular model. The radius of the spherical bounding box is R0 + Z, where R0 is the radius of the fitted sphere, and Z is the magnification value. The value of Z ranges from 3mm to 10mm, preferably 5mm to 6mm. In this embodiment, Z is 5mm.

[0052] In step 3, a spherical bounding box is constructed with the center of the fitted sphere as the center and R0+Z as the radius. The portion of the acetabular model located within this bounding box is extracted and set as the initial region of interest for the acetabular lunate surface. This process effectively eliminates pelvic regions unrelated to the anatomical structure of the acetabular lunate surface, significantly reducing the complexity of subsequent calculations and improving the efficiency of acetabular lunate surface model extraction.

[0053] Step 4: Calculate the residuals from all points in the initial region of interest (ROI) of the acetabular lunate to the fitted sphere, and then extract the initial point set of the acetabular lunate from the ROI based on these residuals. For example... Figure 2 As shown, step 4 specifically involves:

[0054] Step 4.1: Calculate the residuals from the initial region of interest in the lunate plane of the acetabulum to the fitted sphere using the following formula:

[0055]

[0056] in, Let be the residual from the i-th point in the initial region of interest of the acetabular lunate to the fitted sphere, where i is an integer and 1≤i≤I, and I is the number of points in the initial region of interest of the acetabular lunate. Let i be the i-th point in the initial region of interest of the lunate plane of the acetabulum. , Let be the center and radius of the fitted sphere, respectively. This represents the distance from the i-th point in the initial region of interest of the acetabular lunate to the center of the fitted sphere;

[0057] Step 4.2: Define the fault tolerance threshold according to the following formula:

[0058]

[0059] in, This is the fault tolerance threshold; This represents the set of residuals between the initial region of interest on the lunate surface of the acetabulum and the fitted sphere. ; , They are respectively The median and standard deviation;

[0060] Step 4.3: Select points with residuals less than the tolerance threshold in the initial region of interest of the acetabular lunate plane to obtain the region of interest point set. Then, perform connectivity filtering on the region of interest point set to obtain multiple connected regions. Select the points in the connected region with the largest area as the initial point set of the acetabular lunate plane.

[0061] Based on the anatomical characteristics of the acetabulum, the inner surface contour of the bony articular surface of the acetabulum more closely approximates the geometry of an ellipsoid. To further improve the anatomical consistency and boundary accuracy of the extracted acetabular lunate surface, this embodiment, after coarsely extracting the acetabular lunate surface model based on a fitted sphere, further introduces an iterative search strategy of ellipsoid fitting, residual filtering, and connectivity constraints to achieve precise extraction of the acetabular lunate surface model. The specific method is as follows: Figure 3 As shown.

[0062] Step 5: Fit an ellipsoid to the initial point set of the acetabular lunate surface to obtain the fitted ellipsoid, its center, and semi-axis length. Then, calculate the residuals from all points in the initial region of interest of the acetabular lunate surface to the fitted ellipsoid, and extract candidate point sets from the initial region of interest of the acetabular lunate surface based on the residuals. Specifically:

[0063] Step 5.1: Use the least squares method to fit the initial point set of the acetabular lunate surface to an ellipsoid, and obtain the fitted ellipsoid and its center and semi-axis length;

[0064] Step 5.2: Calculate the residuals from the fitted ellipsoid to all points in the initial region of interest on the lunate surface of the acetabulum using the following formula:

[0065]

[0066]

[0067] in, Let be the residual from the i-th point in the initial region of interest of the acetabular lunate surface to the fitted ellipsoid. , Let be the center of the fitted ellipsoid and the equivalent radius, respectively. This represents the distance from the i-th point in the initial region of interest of the acetabular lunate to the center of the fitted ellipsoid. , , These represent the lengths of the three semi-axes of the fitted ellipsoid;

[0068] Step 5.3: Define the adaptive fault tolerance threshold according to the following formula:

[0069]

[0070] in, An adaptive fault tolerance threshold; This represents the set of residuals between the initial region of interest on the lunate surface of the acetabulum and the fitted ellipsoid. ; , They are respectively The median and standard deviation;

[0071] Step 5.4: Select points with residuals less than the adaptive fault tolerance threshold in the initial region of interest of the acetabular lunate plane to obtain a preliminary set of points. Then, perform connectivity filtering on the preliminary set of points to obtain multiple candidate connected regions. Select the points in the candidate connected region with the largest area as the candidate set of points.

[0072] Step 6: Use the candidate point set as the initial point set of the acetabular lunate surface, then return to step 5 until the center of the fitted ellipsoid and the length of the semi-axis satisfy the convergence condition, and use the candidate point set as the point set of the acetabular lunate surface.

[0073] In step 6, the changes in the center and semi-axis lengths of the fitted ellipsoid are compared. If both are not less than the given convergence threshold, the candidate point set is used as the initial point set for the acetabular lunate, and the process returns to step 5 to continue iterating. Otherwise, the iteration process is considered converged, the iteration ends, and the candidate point set is output as the point set for the acetabular lunate. Figure 4 , Figure 5 The figure shows a comparison of the initial region of interest, the initial point set / candidate point set of the acetabular lunate surface, and the fitted sphere / fitted ellipsoid in the acetabular model for different iteration rounds.

[0074] Step 7: Map the point set of the acetabular lunate surface to the point-face relationship of the acetabular model, extract the vertices and triangular facets corresponding to the point set of the acetabular lunate surface in the acetabular model, and obtain the acetabular lunate surface model, thus completing the extraction of the acetabular lunate surface model.

[0075] This invention provides a method for extracting the acetabular meniscus model based on hybrid surface fitting. Using the acetabular model and its corresponding femoral model as input, the method extracts the acetabular meniscus model, mainly in two stages. The first stage is a spherical fitting extraction stage: spherical fitting is performed on the femoral model of the hip joint to obtain a fitted sphere, which is used as the initial surface for extracting the acetabular meniscus model. The fitted sphere is then used to trim the acetabular model, obtaining the initial region of interest (ROI) of the acetabular meniscus. Then, residual calculation is performed by combining the fitted sphere and the initial ROI of the acetabular meniscus. Based on a tolerance threshold, a set of points that meet the tolerance threshold requirements is extracted from the acetabular model to obtain the initial point set of the acetabular meniscus, which then proceeds to the next stage. The second stage is the ellipsoid fitting and extraction stage: Ellipsoid fitting is performed using a point set that meets the fault tolerance threshold requirement to obtain the fitted ellipsoid, and the residual between the fitted ellipsoid and the initial region of interest (ROI) of the acetabular lunate is calculated. Then, based on the adaptive fault tolerance threshold, a point set that meets the adaptive fault tolerance threshold requirement is extracted from the initial ROI of the acetabular lunate to obtain a candidate point set. The ellipsoid extraction stage iterates according to the iterative decision conditions. After iteration, the finally extracted point set is used to extract the corresponding triangular facets from the original acetabular model, completing the extraction of the acetabular lunate model.

[0076] This invention combines the stability of spherical fitting with the anatomical fidelity of ellipsoidal fitting. By evaluating the residual distribution of distances between points within the acetabular model and the fitted ellipsoid, a point set satisfying the residual requirements is extracted. This is then iteratively refined, and finally, the extracted point set and the acetabular model are used to extract the lunate surface of the acetabulum and reconstruct the model. Simultaneously, an adaptive fault-tolerant threshold is introduced during the iterative refinement process, effectively addressing the problems of noise sensitivity and strong parameter dependence inherent in traditional methods, making it more robust when handling complex acetabular morphologies. This method not only demonstrates excellent accuracy but also automates the extraction process, reducing human interference. Especially in complex conditions such as hip dysplasia or dislocation, it maintains high extraction accuracy, improving the accuracy and reliability of clinical diagnosis and surgical planning.

Claims

1. A method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting, characterized in that, Includes the following steps: Step 1: Obtain the acetabular model and the corresponding femoral model of the acetabular lunate plane model to be extracted; Step 2: Obtain the gold standard model of the femur, map the anatomical markers of the femoral head on the gold standard model of the femur to the femoral model, and perform spherical fitting on the anatomical markers of the femoral head mapped to the femoral model to obtain the fitted sphere and its center and radius; Step 3: Enlarge the radius of the fitted sphere to obtain the radius of the spherical bounding box. Then, using the center of the fitted sphere as the center and the radius of the spherical bounding box as the radius, extract the initial region of interest of the lunate surface of the acetabulum in the acetabular model. Step 4: Calculate the residuals from all points in the initial region of interest of the acetabular lunate to the fitted sphere, and then extract the initial point set of the acetabular lunate from the initial region of interest of the acetabular lunate based on the residuals. Step 5: Fit an ellipsoid to the initial point set of the acetabular lunate surface to obtain the fitted ellipsoid, its center, and the length of its semi-axis. Then, calculate the residuals from all points in the initial region of interest of the acetabular lunate surface to the fitted ellipsoid, and extract a candidate point set from the initial region of interest of the acetabular lunate surface based on the residuals. Step 6: Use the candidate point set as the initial point set of the acetabular lunate plane, and then return to step 5 until the center of the fitted ellipsoid and the length of the semi-axis satisfy the convergence condition, and use the candidate point set as the point set of the acetabular lunate plane. Step 7: Map the point set of the acetabular lunate surface to the point-face relationship of the acetabular model, extract the vertices and triangular facets corresponding to the point set of the acetabular lunate surface in the acetabular model, and obtain the acetabular lunate surface model, thus completing the extraction of the acetabular lunate surface model.

2. The method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting according to claim 1, characterized in that, Step 4 specifically involves: Step 4.1: Calculate the residuals from the initial region of interest in the lunate plane of the acetabulum to the fitted sphere using the following formula: ； in, Let be the residual from the i-th point in the initial region of interest of the acetabular lunate to the fitted sphere, where i is an integer and 1≤i≤I, and I is the number of points in the initial region of interest of the acetabular lunate. Let i be the i-th point in the initial region of interest of the lunate plane of the acetabulum. , Let be the center and radius of the fitted sphere, respectively. This represents the distance from the i-th point in the initial region of interest of the acetabular lunate to the center of the fitted sphere; Step 4.2: Define the fault tolerance threshold according to the following formula: ； in, This is the fault tolerance threshold; This represents the set of residuals between the initial region of interest on the lunate surface of the acetabulum and the fitted sphere. ; , They are respectively The median and standard deviation; Step 4.3: Select points with residuals less than the tolerance threshold in the initial region of interest of the acetabular lunate plane to obtain the region of interest point set. Then, perform connectivity filtering on the region of interest point set to obtain multiple connected regions. Select the points in the connected region with the largest area as the initial point set of the acetabular lunate plane.

3. The method for extracting the acetabular lunate surface model based on hybrid surface fitting according to claim 2, characterized in that, Step 5 specifically involves: Step 5.1: Use the least squares method to fit the initial point set of the acetabular lunate surface to an ellipsoid, and obtain the fitted ellipsoid and its center and semi-axis length; Step 5.2: Calculate the residuals from the fitted ellipsoid to all points in the initial region of interest on the lunate surface of the acetabulum using the following formula: ； ； in, Let be the residual from the i-th point in the initial region of interest of the acetabular lunate surface to the fitted ellipsoid. , Let be the center of the fitted ellipsoid and the equivalent radius, respectively. This represents the distance from the i-th point in the initial region of interest of the acetabular lunate to the center of the fitted ellipsoid. , , These represent the lengths of the three semi-axes of the fitted ellipsoid; Step 5.3: Define the adaptive fault tolerance threshold according to the following formula: ； in, An adaptive fault tolerance threshold; This represents the set of residuals between the initial region of interest on the lunate surface of the acetabulum and the fitted ellipsoid. ; , They are respectively The median and standard deviation; Step 5.4: Select points with residuals less than the adaptive fault tolerance threshold in the initial region of interest of the acetabular lunate plane to obtain a preliminary set of points. Then, perform connectivity filtering on the preliminary set of points to obtain multiple candidate connected regions. Select the points in the candidate connected region with the largest area as the candidate set of points.

4. A method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting according to any one of claims 1-3, characterized in that: In step 2, the Bayesian coherence point drift algorithm based on geodesics is used to map the femoral head anatomical markers on the gold standard femoral model to the femoral model.

5. The method for extracting the acetabular lunate surface model based on hybrid surface fitting according to claim 4, characterized in that, Step 1 also includes smoothing the acetabular model and the femur model respectively.

6. The method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting according to claim 5, characterized in that: In step 1, a window function Singer smoothing filter is used to smooth the acetabular model and the femur model respectively.

7. The method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting according to claim 6, characterized in that: In step 3, the radius of the spherical bounding box is R0+Z, where R0 is the radius of the fitted sphere and Z is the magnification value, 3mm≤Z≤10mm.

8. The method for extracting the lunate surface model of the acetabulum based on hybrid surface fitting according to claim 7, characterized in that: In step 2, the anatomical landmarks of the femoral head mapped onto the femoral model are spherically fitted using the least squares method.

Images