Wound positioning and registering method applied to orthopedic robot-assisted surgery

Abstract

The invention relates to the technical field of skeleton feature analysis, in particular to a wound positioning and registering method applied to an orthopedic robot-assisted operation. The method comprises the following steps: dividing density distribution of point cloud data in a to-be-positioned region to obtain a skeleton region cluster; screening high-density blocks through high-density distribution of the point cloud data; performing curvature analysis on the spatial local surface based on the high-density block to obtain a local surface feature degree; and combining the distribution density and the local surface feature degree of the point cloud data, determining the angular point distribution importance of different skeleton region clusters, determining the optimized angular point detection number according to the relationship between the expected angular point distribution importance and the detection number, adjusting the point cloud data, and obtaining the optimized point cloud data for registration. According to the method, the curvature and density characteristics of skeleton features in different regions are analyzed to obtain the distribution importance of angular points in the regions, so that the robustness of angular point feature position matching during registration is improved, and the accuracy of registration in a wound positioning process is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of bone feature analysis, and in particular to a trauma positioning and registration method applied to orthopedic robot-assisted surgery. Background Art

[0002] In orthopedic robot-assisted surgery, trauma positioning and registration is a very important part. It uses image recognition to assist doctors in accurately locating the site of trauma to achieve precise surgical operations. To this end, the following algorithms and methods are commonly used, involving image recognition, data registration, and accuracy assurance. Among them, the classic and effective registration method is shape-based registration. Because bones are usually rigid objects, a registration algorithm (Iterative Closest Point, ICP) can be used to accurately match preoperative models with real-time intraoperative image data. The ICP algorithm completes registration by minimizing the distance between point clouds. This algorithm is extremely helpful for 3D bone positioning in orthopedic surgery.

[0003] Registration algorithms rely heavily on the effectiveness of early feature extraction, specifically corner marking. This process can often lead to discrepancies in the number and location of corners due to inconsistent pre- and intra-operative shooting angles, lighting, and other environmental factors. This means that different locations have varying degrees of significance depending on the shooting environment. Inaccuracies in corner marking during the pre-registration phase can lead to suboptimal registration due to environmental issues. Summary of the Invention

[0004] In order to solve the technical problems in the prior art of inaccurate diagonal point marking in the pre-registration preparation stage, which may lead to poor registration results due to environmental shooting problems, the purpose of the present invention is to provide a wound positioning and registration method for orthopedic robot-assisted surgery. The technical solutions adopted are as follows: The present invention provides a wound positioning and registration method for orthopedic robot-assisted surgery, the method comprising: Acquire point cloud data of intraoperative three-dimensional images in the area to be located; cluster the point cloud data according to the distribution number of unit volume blocks in the three-dimensional image and the distance between the unit volume blocks to obtain bone region clusters; Analyze the high-density distribution of point cloud data in unit volume blocks and determine high-density blocks; obtain the local surface characteristics of each high-density block based on the curvature of the space composed of unit volume blocks in the local range; The importance of corner point allocation for each skeletal region cluster is determined based on the distribution of local surface features of high-density blocks in each skeletal region cluster and the distribution of point cloud data in unit volume blocks. The optimized number of corner point detections for each skeletal region cluster is obtained based on the correspondence between the expected importance of corner point allocation and the number of corner point detections, combined with the corner point allocation importance of the skeletal region cluster. The number of point cloud data in each skeletal region cluster is adjusted according to the number of optimized corner point detections to obtain optimized point cloud data for positioning and registration.

[0005] Furthermore, the method for obtaining the skeleton region clusters includes: Determining a density value for each unit volume block based on the amount of point cloud data in each unit volume block; The density value difference between each two unit volume blocks is used as the density deviation between each two unit volume blocks; the product of the density deviation and the distance between each two unit volume blocks is used as the clustering measure between each two unit volume blocks; The unit volume blocks are clustered based on the clustering metric between the unit volume blocks to obtain the bone region clusters.

[0006] Furthermore, the method for obtaining the high-density block includes: Normalize the number of point cloud data in each unit volume block to obtain the quantity index of each unit volume block; The unit volume blocks whose quantity index is greater than the preset high distribution threshold are regarded as high-density blocks.

[0007] Furthermore, the method for obtaining the local surface characteristic comprises: For any high-density block, perform surface reconstruction on all point cloud data within a preset local range of the high-density block to obtain the local surface of the high-density block; The curvature value of the local surface is calculated as the local surface characteristic of the high-density block.

[0008] Furthermore, the method for obtaining the importance of corner point allocation includes: For any bone region cluster, calculate the mean value of the number of point cloud data contained in all unit volume blocks in the bone region cluster to obtain the bone distribution significance value of the bone region cluster; Normalizing the mean of the local surface features of all high-density blocks in the bone region cluster to obtain the bone structure saliency value of the bone region cluster; The product of the bone distribution significance value and the bone structure significance value of the bone region cluster is used as the corner point allocation importance of the bone region cluster.

[0009] Furthermore, the method for obtaining the optimized number of corner point detections includes: For any skeleton region cluster, the ratio of the expected corner point allocation importance to the corner point allocation importance of the skeleton region cluster is taken as the importance ratio of the skeleton region cluster; Multiplying the importance ratio of the skeleton region cluster by the product of the preset correction coefficient to obtain the corner point detection ratio of the skeleton region cluster; The ratio of the expected number of corner detections to the proportion of corner detections in the skeleton region cluster is used as the optimized number of corner detections in the skeleton region cluster.

[0010] Furthermore, the method for obtaining optimized point cloud data includes: For any skeletal region cluster, when the number of optimized corner point detections is less than the total number of point cloud data in the skeletal region cluster, the unit volume blocks in the skeletal region cluster are arranged in ascending order according to the number of point cloud data in the unit volume blocks to obtain a elimination sequence; point cloud data are eliminated in sequence according to the arrangement order of the unit volume blocks in the elimination sequence until the total number of point cloud data in the skeletal region cluster is the same as the number of optimized corner point detections, thereby obtaining optimized point cloud data; When the number of optimized corner point detections is greater than the total number of point cloud data in the skeletal region cluster, the unit volume blocks in the skeletal region cluster are arranged in descending order according to the number of point cloud data in the unit volume blocks to obtain an addition sequence; interpolation and addition of point cloud data are performed in sequence according to the arrangement order of the unit volume blocks in the addition sequence until the total number of point cloud data in the skeletal region cluster is the same as the number of optimized corner point detections, thereby obtaining optimized point cloud data; When the number of optimized corner point detections is equal to the total number of point cloud data in the skeleton region cluster, the point cloud data in the skeleton region cluster is used as optimized point cloud data.

[0011] Furthermore, the point cloud data is eliminated in sequence according to the arrangement order of the unit volume blocks in the elimination sequence until the total number of point cloud data in the skeleton region cluster is the same as the number of optimized corner point detections, thereby obtaining optimized point cloud data, including: Traverse the unit volume blocks in the elimination sequence in sequence, and calculate the mean distance between each point cloud data in the unit volume block and other point cloud data as the discreteness of each point cloud data; When the discreteness of the point cloud data is greater than the preset discrete threshold, the corresponding point cloud data is used as the data to be eliminated, and the data to be eliminated are eliminated in descending order of discreteness; When the discreteness of the non-existent point cloud data is greater than the preset discrete threshold, the point cloud data corresponding to the maximum discreteness is eliminated; the process stops when the total number of point cloud data in the elimination sequence is the same as the number of optimized corner point detections, and the optimized point cloud data is obtained.

[0012] Furthermore, determining the density value of each unit volume block based on the amount of point cloud data in each unit volume block includes: When point cloud data exists in the unit volume block, the number of point cloud data in the unit volume block is used as the density value of the unit volume block; otherwise, a preset density value is used as the density value of the unit volume block; the preset density value is a non-zero positive integer.

[0013] Furthermore, the unit volume block is a volume of 1 in the three-dimensional image. Cube blocks of different sizes.

[0014] The present invention has the following beneficial effects: The present invention performs preliminary clustering based on the density distribution of point cloud data in the area to be located, and obtains the classification results of point cloud data in characterizing different bone areas, namely bone area clusters, which facilitates the subsequent analysis of different bone structure characterization capabilities for different bone area clusters. By screening high-density blocks with high density distribution in unit volume blocks, a significant analysis of local arrangement curvature is performed. By using high-density distribution analysis to limit volume blocks that may be bone parts to perform surface curvature analysis, a local surface characteristic degree is obtained, which reflects the degree to which point cloud data in local space can reflect the curved structural characteristics of bones. Finally, in the divided bone area clusters, the distribution density and local surface characteristic degree are combined to determine the better corner point allocation importance for feature corner point matching analysis in different bone area clusters, so as to subsequently reduce the influencing corner points with weak characterization such as ambient light. Through the relationship between the expected corner point allocation importance and the number of detections, the preferred number of corner point detections for each cluster is determined, and the number of point cloud data is adjusted based on the preferred number of detections to improve the registration accuracy. The present invention analyzes the curvature and density characteristics of bone features in different areas to obtain the importance of corner point distribution in the area, thereby improving the robustness of corner point feature position matching during alignment and enhancing the accuracy of alignment during trauma localization. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] Figure 1A flowchart of a wound positioning and registration method for orthopedic robot-assisted surgery provided by one embodiment of the present invention; Figure 2 A schematic diagram of local point cloud data of a knee area provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0017] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effectiveness of a wound localization and registration method for robotic-assisted orthopedic surgery proposed in accordance with the present invention. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0019] The following describes in detail a specific solution of a wound positioning and registration method for orthopedic robot-assisted surgery provided by the present invention with reference to the accompanying drawings.

[0020] The main steps of the ICP registration algorithm used to match the preoperative model with the real-time intraoperative image data include: 1) Feature extraction: Key features of the bone structure are extracted by processing preoperative images (such as CT or MRI images). Common feature extraction methods include SIFT (Scale-Invariant Feature Transform) and SURF (Speeded Up Robust Features).

[0021] 2) Initial alignment (coarse registration): This step typically involves coarse registration (based on manual landmarks, rough registration methods, etc.) to roughly align the two images. This is achieved by translation and rotation between the images. For example, in knee joint registration, known bony landmarks (such as the center of the patella) are first used to roughly align the preoperative and intraoperative images.

[0022] 3) Fine registration (precise alignment): Fine alignment is performed by calculating the geometric transformation between the two sets of feature points. The method of minimizing the error function is usually used. The goal is to minimize the distance between the two sets of feature points. The most commonly used fine registration algorithm is the iterative closest point algorithm (ICP). Its process is as follows: Closest Point Matching: The ICP algorithm finds the closest point pair between two sets of data points (e.g., a preoperative bone model and a bone point cloud from a live intraoperative image).

[0023] Transformation optimization: Based on the current registration result, a translation and rotation matrix is ​​calculated to minimize the distance between the point clouds. This transformation is then used to update the registration result.

[0024] Iterative process: This process is repeated until the registration error (the distance between point clouds) is less than a set threshold. Ultimately, the preoperative 3D model and the point cloud in the intraoperative image are precisely aligned.

[0025] Acquiring a preoperative 3D model primarily involves the following: Image acquisition: CT (computed tomography) is used to capture images of the knee area. Scanning: The patient remains still and is positioned according to the imaging equipment's requirements. The scanner acquires tomographic image data of the knee area from multiple angles. Point cloud data generation: Using multi-layer tomographic image data (CT or MRI images), reconstruction algorithms, such as voxel reconstruction, are used to convert the sliced ​​tomographic image data into point cloud data within the 3D image, thus forming a 3D model.

[0026] During orthopedic robot-assisted surgery, a related registration algorithm is usually used for positioning in the registration phase. However, since this algorithm relies on the corner point positions in the pre-processing, the registration results may be erroneous when the corner point positions deviate in different environments. The embodiments of the present invention quantify the features of the bone region to improve the robustness of the corner point position analysis and acquisition process, thereby improving the accuracy of the positioning results. Figure 1 , which shows a flow chart of a wound positioning and registration method for orthopedic robot-assisted surgery provided by one embodiment of the present invention, the method comprising the following steps: S1: Acquire point cloud data of intraoperative three-dimensional images in the area to be located; cluster the data according to the number of point cloud data distribution per unit volume block in the three-dimensional image and the distance between the unit volume blocks to obtain bone region clusters.

[0027] In an embodiment of the present invention, for bone trauma sites, such as the knee area, a C-arm is usually used during surgery to perform real-time image acquisition to obtain point cloud data of a three-dimensional image of the knee area, and filters, such as a median filter, are used to remove noise from the image and improve image quality. The point cloud map of the intraoperative three-dimensional image is aligned with the preoperative three-dimensional model so that the two can correspond to each other in area.

[0028] Corner point comparison and registration can be performed between pre-processed images to achieve intraoperative identification and positioning. Taking into account the inconsistencies of environmental factors such as preoperative and intraoperative shooting angles and lighting factors, the selection of feature corner points is optimized and adjusted before corner point matching to improve the reliability of the registration results.

[0029] The point cloud data composed of several bone edges and other structures reflects the arrangement and degree of arrangement of different bones in the knee area. Figure 2 , which shows a schematic diagram of local point cloud data of a knee region provided by one embodiment of the present invention. The knee region has relatively obvious bone features, mainly consisting of the femur, tibia, patella, etc. Such regions usually have relatively obvious imaging features, specifically: The femoral condyle has a clear outline and a curved surface, especially on the articular surface, with obvious edge features; the tibia is relatively straight in shape, and the top of the bone (tibial plateau) is complex in shape, often showing a more obvious curve or stepped outline; the patella is usually hemispherical, located in front of the knee joint, with clear edges and a relatively smooth surface structure.

[0030] Furthermore, due to the different bone features in the knee area, the degree of point cloud aggregation reflected is inconsistent. The most obvious manifestation is that the density values ​​per unit volume of point cloud data in different bone feature areas are inconsistent. Therefore, different features can be divided according to the bone point cloud density, which serves the purpose of preliminarily marking the clustering of different bone areas.

[0031] For example, the more complex the shape or the clearer the bone area, the higher the corresponding point cloud density value, such as the patella and femur area. Conversely, the more blurred the shape or the smaller the volume of the bone itself, the lower the corresponding point cloud density value, such as the articular cartilage or ligament area.

[0032] Then, first cluster the point cloud results according to the density value within the unit volume to obtain the region division result. At this time, each region is roughly a division item of a skeleton type. When performing subsequent corner point optimization, different allocation importance calculations can be performed for different corner point types to improve optimization efficiency.

[0033] In the embodiment of the present invention, a unit volume block is a volume of 1 in a three-dimensional image. The larger the size of the cubic block, the higher the number of point cloud data in the unit volume block, the higher the density. The skeleton region clustering cluster can be obtained according to the density and distance between the unit volume blocks.

[0034] Preferably, in an embodiment of the present invention, the method for obtaining skeleton region clusters includes: The density value of each unit volume block is determined based on the number of point cloud data within each unit volume block. In this embodiment of the present invention, to ensure that all volume blocks are clusterable, when point cloud data exists within a unit volume block, the number of point cloud data within the unit volume block is used as the density value for the unit volume block; otherwise, a preset density value is used as the density value for the unit volume block. It will be understood that the preset density value is set to ensure that the unit block is meaningful during clustering. Therefore, the preset density value is a non-zero positive integer, and in this embodiment of the present invention, it is set to 1. The specific value can be adjusted by the implementer based on the specific implementation scenario.

[0035] The difference in density between each unit volume block is further calculated as the density deviation between the two unit volume blocks, reflecting the degree of deviation in density representation. A smaller density deviation indicates a more similar density distribution. Furthermore, combined with distance, the product of the density deviation between the two unit volume blocks and the distance between the two unit volume blocks is used as the clustering metric between the two unit volume blocks. The clustering metric serves as the dimension for clustering.

[0036] Finally, the unit volume blocks are clustered based on the clustering metric between the unit volume blocks to obtain skeletal region clusters. In an embodiment of the present invention, the clustering method adopts the K-means clustering algorithm for clustering. The smaller the clustering metric, that is, the closer the density value and the closer the distance between the unit volume blocks, the higher the possibility of being clustered into one category, because they have relatively similar or equal characteristics and the reflected skeletal region components are closer. It should be noted that the clustering process is a technical means well known to those skilled in the art and will not be elaborated or limited here.

[0037] At this point, the preliminary division of several bone regions is completed, and each bone region cluster represents a type of bone characteristics.

[0038] S2: Analyze the high-density distribution of point cloud data in the unit volume block and determine the high-density block; obtain the local surface feature of the high-density block based on the curvature of the space composed of unit volume blocks in the local range of each high-density block.

[0039] The edges of knee bones, such as the femur, tibia, and patella, typically exhibit significant contrast changes, allowing the SIFT algorithm to detect strong corners at these edges. Such corners typically appear at the corners of the bone contours, particularly at the femoral condyle, tibial plateau, and patellar edge. Furthermore, the rounded structure of the patella itself, as well as the contact area between the patella and femur, form a distinct boundary. The SIFT algorithm can extract corners at these boundaries or locations where the surface curvature suddenly changes, particularly at the upper and lower edges of the patella or the glenoid fossa.

[0040] Therefore, different skeletal regions will produce different detection results due to their structural characteristics. When using the corner detection operator to detect different skeletal regions, the higher the edge curvature value of the point cloud data within the skeletal image structure, the more likely the corresponding corner point will appear, and the more it reflects the structural characteristics of the bone, providing greater value in subsequent registration.

[0041] Therefore, we analyze the bone structure information that can be represented by the unit volume block, and analyze the curvature value through local geometric morphology to characterize the local surface features. It can be understood that the bone structure mostly appears in clear bone areas, so the curvature analysis is only performed on unit blocks with high density distribution.

[0042] Therefore, firstly, the high-density volume blocks are determined by high-quantity distribution. In an embodiment of the present invention, the quantity of point cloud data in each unit volume block is normalized to obtain a quantity index for each unit volume block, which reflects the degree of density distribution. The unit volume blocks whose quantity index is greater than the preset high distribution threshold are regarded as high-density blocks. The higher the quantity index, the higher the proportion of point cloud data in the unit block. In an embodiment of the present invention, the preset high distribution threshold is set to 0.8 to screen unit volume blocks with high-density distribution. The specific numerical value can be adjusted by the implementer. It should be noted that normalization is a technical means well known to those skilled in the art. Methods such as linear normalization or standard normalization can be used and are not limited here.

[0043] The high-density block can be further analyzed for the degree of local and overall surface curvature to reflect the amount of bone structure information. Preferably, in an embodiment of the present invention, the method for obtaining the local surface characteristic of the high-density block includes: For any high-density block, surface reconstruction is performed on all point cloud data within a preset local range of the high-density block to obtain the local surface of the high-density block. Surface reconstruction is the extraction of a continuous surface mesh, such as a triangular mesh, from discrete point cloud data. In an embodiment of the present invention, the preset local range is set to a sphere with a radius of 10 unit volume blocks centered on the high-density block. The specific range size can be adjusted by the implementer. The Poisson Surface Reconstruction algorithm is used to convert discrete point cloud data into a continuous three-dimensional surface mesh as the local surface. Poisson Surface Reconstruction is a reconstruction method based on the Poisson equation. It generates a smooth, continuous three-dimensional surface mesh by modeling the normal direction and point position of the point cloud data.

[0044] Therefore, the local surface curvature value can be further calculated as the local surface characteristic of the high-density block. In an embodiment of the present invention, the principal curvature method (Curvature Computation for Triangular Meshes) can be used. First, the normal vector of each point is calculated, and then the curvature is estimated based on the change in the normal vectors of adjacent triangles. It should be noted that the application of the surface reconstruction algorithm and the calculation of the surface curvature value are both publicly available technical means well known to those skilled in the art and will not be detailed here.

[0045] At this point, further analysis of the bone structure information represented by the unit volume block is completed.

[0046] S3: Determine the importance of corner point allocation for each skeletal region cluster based on the distribution of local surface features of high-density blocks in each skeletal region cluster and the distribution of point cloud data in unit volume blocks; based on the correspondence between the expected importance of corner point allocation and the number of corner point detections, combined with the importance of corner point allocation for the skeletal region cluster, obtain the optimized number of corner point detections for each skeletal region cluster.

[0047] The curvature of each cluster area also indirectly reflects the possibility of corner points appearing, that is, the effectiveness of the corner points. The effectiveness refers to the structural characteristics that the corner points can reflect, rather than the noise corner points caused by environmental problems such as angle gloss, which affect the final registration effect.

[0048] Therefore, combining the local surface feature degree reflecting the corner point features in the skeleton region clusters and the distribution of point cloud data reflecting the more significant skeleton regions, a comprehensive analysis of the importance of corner point allocation is performed on each skeleton region cluster. Preferably, in an embodiment of the present invention, the method for obtaining the importance of corner point allocation includes: First, for any bone region cluster, the mean value of the number of point cloud data contained in all unit volume blocks in the bone region cluster is calculated to obtain the bone distribution significance value of the bone region cluster. The denser the distribution of unit point cloud data in the cluster, the higher the possibility that the bone region cluster is a bone distribution.

[0049] Furthermore, the mean of the local surface characteristic degrees of all high-density blocks in the bone region cluster is normalized to obtain the bone structure significance value of the bone region cluster. The higher the characteristic degree shown by the overall high-density block, the more significant the bone structure characteristics in the bone region cluster.

[0050] Finally, the product of the bone distribution significance value and the bone structure significance value of the skeletal region cluster is used as the corner point allocation importance of the skeletal region cluster. The higher the corner point allocation importance, the more corner points in the regional cluster need to be supported to avoid corner point registration issues caused by angle or insufficient lighting. Conversely, the lower the corner point allocation importance, the lower the corresponding bone region density, the fewer point clouds affecting the skeletal structure, and theoretically the fewer corner points in the regional cluster, avoiding false corner points caused by lighting concentration and other factors.

[0051] As an example, the expression for the importance of corner point assignment is: , where Expressed as The importance of corner point assignment of each skeleton region cluster, Expressed as The total number of unit volume blocks in the bone region clusters, Expressed as The number of point cloud data in a unit volume block, Expressed as The total number of high-density blocks in the skeleton region clusters, Expressed as The local surface characteristics of a high-density block, Expressed as a normalized processing function.

[0052] At this point, the analysis of the importance of corner point distribution in different skeletal region clusters is completed. Further, through the distribution relationship between the ideal expected number of corner points and the importance of corner points, the number of corner points required for each skeletal region cluster under the expected relationship can be analyzed, that is, the number of corner point detections can be optimized.

[0053] In the embodiment of the present invention, the relationship between the expected corner point allocation importance and the number of intersection detections is: ; In the formula, represents the number of corner point detections in the expected skeleton area, represents the importance of corner point allocation in the expected skeleton area, and represents the correction coefficient. The specific value is adjusted by the implementer according to the specific implementation scenario and is not restricted here.

[0054] The ideal number of corner points corresponding to the skeleton region cluster can be obtained by combining the importance of the corner point distribution in the skeleton region cluster. The ideal number of corner points corresponding to the skeleton region cluster can be obtained by analyzing the proportion of the relational expression. Preferably, in an embodiment of the present invention, the method for obtaining the optimized number of corner point detections includes: For any skeletal region cluster, the ratio of the expected corner point allocation importance to the corner point allocation importance of the skeletal region cluster is used as the importance ratio of the skeletal region cluster, and the corresponding ideal corner points are obtained in a year-on-year manner. The importance ratio of the skeletal region cluster is multiplied by the product of the preset correction coefficient to obtain the corner point detection ratio of the skeletal region cluster, and the ratio of the expected corner point detection number to the corner point detection ratio of the skeletal region cluster is used as the optimized corner point detection number of the skeletal region cluster. By substituting the relation into the expression, the preferred corner point detection number of the skeletal region cluster can be directly obtained. As an example, the expression for optimizing the number of corner points is: Where, Expressed as The importance of corner point assignment of each skeleton region cluster, Expressed as The optimized number of corner point detections for clustering of skeleton regions.

[0055] At this point, the analysis of the ideal number of corner points in each skeletal region cluster is completed through the analysis of the importance of the required corner points in different regions after division.

[0056] S4: Adjust the number of point cloud data in each skeleton region cluster according to the number of optimized corner point detections to obtain optimized point cloud data for positioning and registration.

[0057] By optimizing the number of corner point detections, the current point cloud data distribution can be adjusted. When the number of point cloud data in a region exceeds the optimized number of corner point detections, the optimal point cloud data can be obtained by eliminating the optimal number of corner points. Conversely, when the number of point cloud data is less than the optimized number of corner point detections, the ideal number of detections can be increased by adding corner points to obtain the optimized point cloud data.

[0058] Therefore, preferably, in an embodiment of the present invention, the method for obtaining optimized point cloud data includes: For any bone region cluster, when the number of optimized corner point detections is less than the total number of point cloud data in the bone region cluster, it means that there are too many point cloud data in the cluster and they need to be eliminated. The unit volume blocks in the bone region cluster are arranged in order from low to high according to the number of point cloud data in the unit volume blocks to obtain a elimination sequence, and they are eliminated according to the density distribution in the unit volume blocks. The lower the density in the unit volume block, the weaker the information representation ability and the greater the possibility of being eliminated.

[0059] Therefore, point cloud data is eliminated in sequence according to the order of the unit volume blocks in the elimination sequence until the total number of point cloud data in the skeleton region cluster is the same as the number of optimized corner point detections, thereby obtaining optimized point cloud data. In an embodiment of the present invention, the unit volume blocks in the elimination sequence are traversed in sequence, and the mean distance between each point cloud data in the unit volume block and other point cloud data is calculated as the discreteness of each point cloud data, reflecting the redundancy of the point cloud data. When the discreteness of the point cloud data is greater than the preset discreteness threshold, it means that the discreteness of the point cloud data is large and the degree of deletability is high. The corresponding point cloud data is used as the data to be eliminated, and the data to be eliminated is eliminated in descending order of discreteness. During this process, if the total number of point cloud data in the elimination sequence is the same as the number of optimized corner point detections, the process stops; otherwise, the unit volume blocks are continued to be traversed backward after elimination.

[0060] When the discreteness of the point cloud data does not exist and is greater than the preset discreteness threshold, it indicates that the distribution of the point cloud data is optimal. At this time, only the point cloud data corresponding to the maximum discreteness is eliminated, and the process is traversed backwards. The elimination sequence can be iterated until the total number of point cloud data in the elimination sequence is equal to the number of optimized corner point detections. The remaining point cloud data is the optimized point cloud data. In this embodiment of the present invention, the preset discreteness threshold is set to 0.8. The specific value can be adjusted by the implementer according to the specific implementation situation and is not limited here.

[0061] When the number of optimized corner point detections is greater than the total number of point cloud data in the skeletal region cluster, it indicates that the overall point cloud data in the cluster is small and needs to be added. The unit volume blocks in the skeletal region cluster are arranged in descending order according to the number of point cloud data in the unit volume blocks to obtain an addition sequence. The more distributed point cloud data in a unit volume block, the higher the possibility of priority increase.

[0062] The point cloud data is interpolated and added in sequence according to the arrangement order of the unit volume blocks in the addition sequence until the total number of point cloud data in the skeletal region cluster is the same as the number of optimized corner point detections, thereby obtaining the optimized point cloud data. In an embodiment of the present invention, one point cloud data is added in each unit volume block in sequence according to the order of the unit volume blocks in the addition sequence using the interpolation method. The addition sequence can be iteratively added, and the point cloud data is stopped only when the total number of point cloud data in the addition sequence is the same as the number of optimized corner point detections. The remaining retained point cloud data is the optimized point cloud data. It should be noted that the interpolation method is a technical means well known to those skilled in the art and will not be described in detail here.

[0063] When the number of optimized corner point detections is equal to the total number of point cloud data in the skeleton region cluster, it indicates that the data distribution in the skeleton region cluster is relatively ideal, and the point cloud data in the skeleton region cluster is used as the optimized point cloud data.

[0064] Furthermore, the optimized point cloud data can be used to participate in the ICP point cloud registration of the intraoperative data of the area to be positioned and the preoperative model, which reduces the problem of registration group error caused by environmental factors and improves the registration effect.

[0065] The present invention performs preliminary clustering based on the density distribution of point cloud data in the area to be located, and obtains the classification results of point cloud data in characterizing different bone areas, namely bone area clusters, which facilitates the subsequent analysis of different bone structure characterization capabilities for different bone area clusters. By screening high-density blocks with high density distribution in unit volume blocks, a significant analysis of local arrangement curvature is performed. By using high-density distribution analysis to limit volume blocks that may be bone parts to perform surface curvature analysis, a local surface characteristic degree is obtained, which reflects the degree to which point cloud data in local space can reflect the curved structural characteristics of bones. Finally, in the divided bone area clusters, the distribution density and local surface characteristic degree are combined to determine the better corner point allocation importance for feature corner point matching analysis in different bone area clusters, so as to subsequently reduce the influencing corner points with weak characterization such as ambient light. Through the relationship between the expected corner point allocation importance and the number of detections, the preferred number of corner point detections for each cluster is determined, and the number of point cloud data is adjusted based on the preferred number of detections to improve the registration accuracy. The present invention analyzes the curvature and density characteristics of bone features in different areas to obtain the importance of corner point distribution in the area, thereby improving the robustness of corner point feature position matching during alignment and enhancing the accuracy of alignment during trauma localization.

[0066] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0067] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A wound positioning and registration method for orthopedic robot-assisted surgery, characterized in that: The method comprises: Acquire point cloud data of intraoperative three-dimensional images in the area to be located; cluster the point cloud data according to the distribution number of unit volume blocks in the three-dimensional image and the distance between the unit volume blocks to obtain bone region clusters; Analyze the high-density distribution of point cloud data in unit volume blocks and determine high-density blocks; obtain the local surface characteristics of each high-density block based on the curvature of the space composed of unit volume blocks in the local range; The importance of corner point allocation for each skeletal region cluster is determined based on the distribution of local surface features of high-density blocks in each skeletal region cluster and the distribution of point cloud data in unit volume blocks. The optimized number of corner point detections for each skeletal region cluster is obtained based on the correspondence between the expected importance of corner point allocation and the number of corner point detections, combined with the corner point allocation importance of the skeletal region cluster. The number of point cloud data in each skeletal region cluster is adjusted according to the number of optimized corner point detections to obtain optimized point cloud data for positioning and registration.

2. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 1, characterized in that: The method for obtaining the skeleton region cluster includes: Determining a density value for each unit volume block based on the amount of point cloud data in each unit volume block; The density value difference between each two unit volume blocks is used as the density deviation between each two unit volume blocks; the product of the density deviation and the distance between each two unit volume blocks is used as the clustering measure between each two unit volume blocks; The unit volume blocks are clustered based on the clustering metric between the unit volume blocks to obtain the bone region clusters.

3. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 1, characterized in that: The method for obtaining the high-density block includes: Normalize the number of point cloud data in each unit volume block to obtain the quantity index of each unit volume block; The unit volume blocks whose quantity index is greater than the preset high distribution threshold are regarded as high-density blocks.

4. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 1, characterized in that: The method for obtaining the local surface feature degree includes: For any high-density block, perform surface reconstruction on all point cloud data within a preset local range of the high-density block to obtain the local surface of the high-density block; The curvature value of the local surface is calculated as the local surface characteristic of the high-density block.

5. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 1, characterized in that: The method for obtaining the corner point allocation importance includes: For any bone region cluster, calculate the mean value of the number of point cloud data contained in all unit volume blocks in the bone region cluster to obtain the bone distribution significance value of the bone region cluster; Normalizing the mean of the local surface features of all high-density blocks in the bone region cluster to obtain the bone structure saliency value of the bone region cluster; The product of the bone distribution significance value and the bone structure significance value of the bone region cluster is used as the corner point allocation importance of the bone region cluster.

6. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 1, characterized in that: The method for obtaining the optimized number of corner point detections includes: For any skeleton region cluster, the ratio of the expected corner point allocation importance to the corner point allocation importance of the skeleton region cluster is taken as the importance ratio of the skeleton region cluster; Multiplying the importance ratio of the skeleton region cluster by the product of the preset correction coefficient to obtain the corner point detection ratio of the skeleton region cluster; The ratio of the expected number of corner detections to the proportion of corner detections in the skeleton region cluster is used as the optimized number of corner detections in the skeleton region cluster.

7. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 1, characterized in that: The method for obtaining optimized point cloud data includes: For any skeletal region cluster, when the number of optimized corner point detections is less than the total number of point cloud data in the skeletal region cluster, the unit volume blocks in the skeletal region cluster are arranged in ascending order according to the number of point cloud data in the unit volume blocks to obtain a elimination sequence; point cloud data are eliminated in sequence according to the arrangement order of the unit volume blocks in the elimination sequence until the total number of point cloud data in the skeletal region cluster is the same as the number of optimized corner point detections, thereby obtaining optimized point cloud data; When the number of optimized corner point detections is greater than the total number of point cloud data in the skeletal region cluster, the unit volume blocks in the skeletal region cluster are arranged in descending order according to the number of point cloud data in the unit volume blocks to obtain an addition sequence; interpolation and addition of point cloud data are performed in sequence according to the arrangement order of the unit volume blocks in the addition sequence until the total number of point cloud data in the skeletal region cluster is the same as the number of optimized corner point detections, thereby obtaining optimized point cloud data; When the number of optimized corner point detections is equal to the total number of point cloud data in the skeleton region cluster, the point cloud data in the skeleton region cluster is used as optimized point cloud data.

8. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 7, characterized in that: The point cloud data is eliminated in sequence according to the arrangement order of the unit volume blocks in the elimination sequence until the total number of point cloud data in the skeleton region cluster is the same as the number of optimized corner point detections, thereby obtaining optimized point cloud data, including: Traverse the unit volume blocks in the elimination sequence in sequence, and calculate the mean distance between each point cloud data in the unit volume block and other point cloud data as the discreteness of each point cloud data; When the discreteness of the point cloud data is greater than the preset discrete threshold, the corresponding point cloud data is used as the data to be eliminated, and the data to be eliminated are eliminated in descending order of discreteness; When the discreteness of the non-existent point cloud data is greater than the preset discrete threshold, the point cloud data corresponding to the maximum discreteness is eliminated; the process stops when the total number of point cloud data in the elimination sequence is the same as the number of optimized corner point detections, and the optimized point cloud data is obtained.

9. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 2, characterized in that: The determining the density value of each unit volume block based on the amount of point cloud data in each unit volume block includes: When point cloud data exists in the unit volume block, the number of point cloud data in the unit volume block is used as the density value of the unit volume block; otherwise, a preset density value is used as the density value of the unit volume block; the preset density value is a non-zero positive integer.

10. The wound positioning and registration method for orthopedic robot-assisted surgery according to claim 1, characterized in that: The unit volume block is a volume of 1 in the three-dimensional image. Cube blocks of different sizes.

Images