Method, device, terminal and medium for registration of guide plate three-dimensional model with maxillary sinus structure
By extracting the curvature distribution and texture characteristics of the guide plate and maxillary sinus structure, the composite feature vector is generated for matching, which solves the problem of insufficient registration accuracy in guide plate design, and achieves high-precision matching of guide plate and maxillary sinus structure, improving dental implant surgery safety and effectiveness.
Patent Information
- Application Number
- CN202411429592.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-14
- Publication Date
- 2024-12-27
- Estimated Expiration
- 2044-10-14
AI Technical Summary
In the prior art, in the three-dimensional model design process of guide plates, the registration accuracy of guide plates and maxillary sinus structures is insufficient, especially in the details, which affects the surgical effect.
By obtaining scanning data of the guide plate three-dimensional model and maxillary sinus structure, the curvature distribution is extracted and dense and sparse sampling is performed to obtain feature point sets; combining multi-scale analysis and local texture features, composite feature vectors are generated for feature matching, and registration accuracy is improved.
High-precision registration of the three-dimensional model of the guide plate and the maxillary sinus structure is achieved, ensuring the matching of details and improving the accuracy and safety of surgical design.
Smart Images
Figure CN118967766B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of 3D printing model design, and in particular to a method, device, terminal and medium for aligning a guide plate three-dimensional model with a maxillary sinus structure. Background Art
[0002] Insufficient bone in the maxillary posterior area, reduced alveolar bone volume, atrophy of the maxillary posterior bone, pneumatization of the maxillary sinus, and abnormal hyperplasia of the maxillary gingiva are the difficulties in dental implant technology. These conditions can be solved by maxillary sinus floor augmentation.
[0003] At present, digital 3D printing guide technology is a new method for maxillary sinus lift. It designs the guide plate three-dimensional model through 3D printing technology to achieve the purpose of avoiding important anatomical structures during surgery, so as to effectively reduce the risk of surgery. At present, the design steps of the digital model of the guide plate mainly include: obtaining the user's oral scan data and reconstructing the jaw in three dimensions; determining the boundary of the maxillary sinus lateral wall window guide plate, and determining the contour of the guide plate on the basis of avoiding specific anatomical structures (such as nerves and blood vessels, maxillary sinus floor or thick bone wall). However, this kind of guide plate obtained by the design idea of avoiding specific structures is relatively rough in some areas, and the details often need to be aligned. Therefore, in the design process of the guide plate in the prior art, there is a problem of poor alignment accuracy between the three-dimensional model of the guide plate and the maxillary sinus structure data. Summary of the invention
[0004] The present invention application provides a method, device, terminal and medium for aligning a three-dimensional guide plate model with a maxillary sinus structure, thereby improving the alignment accuracy of the three-dimensional guide plate model with the maxillary sinus structure data in details.
[0005] In order to solve the above technical problems, the present invention provides a method for aligning a guide plate three-dimensional model with a maxillary sinus structure, comprising:
[0006] Acquire a three-dimensional model of the guide plate and scan data of the maxillary sinus structure; extract the curvature distribution of the three-dimensional model of the guide plate and the scan data of the maxillary sinus structure;
[0007] According to the curvature distribution of the guide plate three-dimensional model, dense sampling is performed on the area where the curvature is greater than a first threshold, and sparse sampling is performed on the area where the curvature is less than or equal to the first threshold, so as to obtain a first feature point set; according to the curvature distribution of the maxillary sinus structure scanning data, dense sampling is performed on the area where the curvature is greater than a second threshold, and sparse sampling is performed on the area where the curvature is less than or equal to the second threshold, so as to obtain a second feature point set; wherein the first threshold and the second threshold are both preset curvature thresholds;
[0008] Extracting local geometric information of each first feature point in the first feature point set and local geometric information of each second feature point in the second feature point set through multi-scale analysis; collecting local texture features of the guide plate three-dimensional model and local texture features of the maxillary sinus structure scan data;
[0009] Generate a first composite feature vector for each first feature point based on the local geometric information of each first feature point and the local texture features of the guide plate three-dimensional model; generate a second composite feature vector for each second feature point based on the local geometric information of each second feature point and the local texture features of the maxillary sinus structure scan data;
[0010] The first composite feature vector and the second composite feature vector are used to perform feature matching to obtain a registration result.
[0011] As a preferred solution, the step of performing feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result includes:
[0012] Using a principal component analysis method, performing dimensionality reduction processing on the first composite feature vector to obtain a first reduced-dimensionality feature vector; using a principal component analysis method, performing dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensionality feature vector;
[0013] Constructing a data space, and dividing the data space into a plurality of sub-areas by a recursive method;
[0014] Corresponding nodes are constructed based on each of the sub-regions; based on the constructed nodes, a kd-tree data structure is used to perform a feature matching operation on the first reduced-dimensionality feature vector and the second reduced-dimensionality feature vector to obtain the registration result.
[0015] As a preferred solution, the principal component analysis method is used to perform dimensionality reduction processing on the first composite feature vector to obtain a first reduced dimensionality feature vector, including:
[0016] Obtain the covariance matrix and total data variance of the first composite eigenvector; obtain the weight of each first principal component in the total data variance according to the covariance matrix of the first composite eigenvector; screen out multiple first principal components that are greater than a preset weight threshold; respectively obtain the corresponding relationship between each first principal component and the first composite eigenvector to obtain a first expression; query a preset database to select the first expression with the highest matching degree with the preset database; use the selected first expression to simplify the first composite eigenvector to obtain a first reduced-dimensionality eigenvector.
[0017] The principal component analysis method is used to perform dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensional feature vector, including: obtaining the covariance matrix and total data variance of the second composite feature vector; obtaining the weight of each second principal component in the total data variance according to the covariance matrix of the second composite feature vector; screening out multiple second principal components greater than a preset weight threshold; respectively obtaining the corresponding relationship between each second principal component and the second composite feature vector to obtain a second expression; querying a preset database to select the second expression with the highest matching degree with the preset database; using the selected second expression to simplify the second composite feature vector to obtain a second reduced-dimensional feature vector.
[0018] As a preferred solution, the extracting of the curvature distribution of the guide plate three-dimensional model and the maxillary sinus structure scanning data includes:
[0019] The three-dimensional model of the guide plate and the scanning data of the maxillary sinus structure are converted into a guide plate curved surface model and a maxillary sinus curved surface model respectively, and then the coordinate information and surface normal vector information of each point in the guide plate curved surface model and the maxillary sinus curved surface model are obtained;
[0020] Obtaining the principal curvatures of the guide plate curved surface model and the maxillary sinus curved surface model;
[0021] According to the principal curvature of the guide plate surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the guide plate three-dimensional model is obtained; according to the principal curvature of the maxillary sinus surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the maxillary sinus structure scanning data is obtained.
[0022] As a preferred solution, before collecting the local texture features of the guide plate three-dimensional model and the local texture features of the maxillary sinus structure scanning data, the method further includes:
[0023] Calculating the anisotropy value of each scanning data point in the maxillary sinus structure scanning data;
[0024] For the area in the maxillary sinus structure scan data where the anisotropy value is greater than a preset value, a first smoothing intensity is used for smoothing along the structure direction, and a second smoothing intensity is used for smoothing along the direction perpendicular to the structure; wherein the first smoothing intensity is greater than the second smoothing intensity;
[0025] Neumann boundary conditions and finite difference method were used to construct the corresponding partial differential equation model for the smoothed maxillary sinus structure scanning data.
[0026] The diffusion tensor of each second feature point is calculated, the diffusion tensor is applied to the partial differential equation model, and nonlinear anisotropic filtering is performed through multiple iterations to obtain the maxillary sinus structure scanning data after filtering.
[0027] As a preferred embodiment, the local geometric information includes local curvature information, local normal vector and Gaussian curvature; the local curvature information is the curvature radius of a 5 mm grid point, the local normal vector is obtained by cross product, and the Gaussian curvature is obtained by local differential geometry calculation.
[0028] As a preferred solution, the local texture features include a gray level co-occurrence matrix and a Sobel operator.
[0029] Correspondingly, an embodiment of the present invention further provides a device for aligning a guide plate three-dimensional model with a maxillary sinus structure, comprising an extraction module, a sampling module, a collection module, a generation module and an alignment module; wherein:
[0030] The extraction module is used to obtain the three-dimensional model of the guide plate and the scanning data of the maxillary sinus structure; extract the curvature distribution of the three-dimensional model of the guide plate and the scanning data of the maxillary sinus structure;
[0031] The sampling module is used to perform dense sampling on areas where the curvature is greater than a first threshold value and perform sparse sampling on areas where the curvature is less than or equal to the first threshold value according to the curvature distribution of the guide plate three-dimensional model, so as to obtain a first feature point set; and perform dense sampling on areas where the curvature is greater than a second threshold value and perform sparse sampling on areas where the curvature is less than or equal to the second threshold value according to the curvature distribution of the maxillary sinus structure scanning data, so as to obtain a second feature point set; wherein the first threshold value and the second threshold value are both preset curvature threshold values;
[0032] The acquisition module is used to extract local geometric information of each first feature point in the first feature point set and local geometric information of each second feature point in the second feature point set through multi-scale analysis; and to acquire local texture features of the guide plate three-dimensional model and local texture features of the maxillary sinus structure scan data;
[0033] The generating module is used to generate a first composite feature vector of each first feature point according to the local geometric information of each first feature point and the local texture feature of the guide plate three-dimensional model; and to generate a second composite feature vector of each second feature point according to the local geometric information of each second feature point and the local texture feature of the maxillary sinus structure scanning data;
[0034] The registration module is used to perform feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result.
[0035] As a preferred solution, the registration module uses the first composite feature vector and the second composite feature vector to perform feature matching to obtain a registration result, including:
[0036] The registration module uses a principal component analysis method to perform dimensionality reduction processing on the first composite feature vector to obtain a first reduced-dimensionality feature vector; and uses a principal component analysis method to perform dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensionality feature vector;
[0037] Constructing a data space, and dividing the data space into a plurality of sub-areas by a recursive method;
[0038] Corresponding nodes are constructed based on each of the sub-regions; based on the constructed nodes, a kd-tree data structure is used to perform a feature matching operation on the first reduced-dimensionality feature vector and the second reduced-dimensionality feature vector to obtain the registration result.
[0039] As a preferred solution, the registration module uses principal component analysis to perform dimensionality reduction processing on the first composite feature vector to obtain a first reduced-dimensionality feature vector, including:
[0040] The registration module obtains the covariance matrix and total data variance of the first composite feature vector; obtains the weight of each first principal component in the total data variance according to the covariance matrix of the first composite feature vector; screens out multiple first principal components that are greater than a preset weight threshold; obtains the corresponding relationship between each first principal component and the first composite feature vector respectively to obtain a first expression; queries a preset database to select the first expression with the highest matching degree with the preset database; uses the selected first expression to simplify the first composite feature vector to obtain a first reduced-dimensional feature vector.
[0041] The registration module uses principal component analysis to perform dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensional feature vector, including: the registration module obtains the covariance matrix and total data variance of the second composite feature vector; obtains the weight of each second principal component in the total data variance according to the covariance matrix of the second composite feature vector; screens out multiple second principal components that are greater than a preset weight threshold; obtains the corresponding relationship between each second principal component and the second composite feature vector respectively to obtain a second expression; queries a preset database to select a second expression with the highest matching degree with the preset database; uses the selected second expression to simplify the second composite feature vector to obtain a second reduced-dimensional feature vector.
[0042] As a preferred solution, the extraction module extracts the curvature distribution of the guide plate three-dimensional model and the maxillary sinus structure scanning data, including:
[0043] The extraction module converts the guide plate three-dimensional model and the maxillary sinus structure scanning data into a guide plate curved surface model and a maxillary sinus curved surface model respectively, and then obtains the coordinate information and surface normal vector information of each point in the guide plate curved surface model and the maxillary sinus curved surface model;
[0044] Obtaining the principal curvatures of the guide plate curved surface model and the maxillary sinus curved surface model;
[0045] According to the principal curvature of the guide plate surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the guide plate three-dimensional model is obtained; according to the principal curvature of the maxillary sinus surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the maxillary sinus structure scanning data is obtained.
[0046] As a preferred solution, the guide plate three-dimensional model and maxillary sinus structure registration device further includes a filtering module, which is used for: before the acquisition module acquires the local texture features of the guide plate three-dimensional model and the local texture features of the maxillary sinus structure scanning data:
[0047] Calculating the anisotropy value of each scanning data point in the maxillary sinus structure scanning data;
[0048] For the area in the maxillary sinus structure scan data where the anisotropy value is greater than a preset value, a first smoothing intensity is used for smoothing along the structure direction, and a second smoothing intensity is used for smoothing along the direction perpendicular to the structure; wherein the first smoothing intensity is greater than the second smoothing intensity;
[0049] Neumann boundary conditions and finite difference method were used to construct the corresponding partial differential equation model for the smoothed maxillary sinus structure scanning data.
[0050] The diffusion tensor of each second feature point is calculated, the diffusion tensor is applied to the partial differential equation model, and nonlinear anisotropic filtering is performed through multiple iterations to obtain the maxillary sinus structure scanning data after filtering.
[0051] As a preferred embodiment, the local geometric information includes local curvature information, local normal vector and Gaussian curvature; the local curvature information is the curvature radius of a 5 mm grid point, the local normal vector is obtained by cross product, and the Gaussian curvature is obtained by local differential geometry calculation.
[0052] As a preferred solution, the local texture features include a gray level co-occurrence matrix and a Sobel operator.
[0053] Correspondingly, the present application also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the method for aligning the guide plate three-dimensional model with the maxillary sinus structure is implemented.
[0054] Correspondingly, the present invention application also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for aligning the guide plate three-dimensional model with the maxillary sinus structure.
[0055] Compared with the prior art, the present invention has the following beneficial effects:
[0056] The present invention application provides a method, device, terminal and medium for registration of a guide plate three-dimensional model and a maxillary sinus structure. The registration method comprises: acquiring a guide plate three-dimensional model and maxillary sinus structure scanning data; extracting the curvature distribution of the guide plate three-dimensional model and the maxillary sinus structure scanning data; based on the curvature distribution of the guide plate three-dimensional model, densely sampling the area where the curvature is greater than a first threshold, and sparsely sampling the area where the curvature is less than or equal to the first threshold, to obtain a first feature point set; based on the curvature distribution of the maxillary sinus structure scanning data, densely sampling the area where the curvature is greater than a second threshold, and sparsely sampling the area where the curvature is less than or equal to the second threshold, to obtain a second feature point set; wherein the first threshold and the second threshold are both A preset curvature threshold; extracting local geometric information of each first feature point in the first feature point set and local geometric information of each second feature point in the second feature point set through multi-scale analysis; collecting local texture features of the three-dimensional model of the guide plate and local texture features of the maxillary sinus structure scan data; generating a first composite feature vector of each first feature point based on the local geometric information of each first feature point and the local texture features of the three-dimensional model of the guide plate; generating a second composite feature vector of each second feature point based on the local geometric information of each second feature point and the local texture features of the maxillary sinus structure scan data; performing feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result. The present invention applies a method of sampling the three-dimensional model of the guide plate and the scan data of the maxillary sinus structure based on the curvature distribution, obtaining a feature point set, extracting local geometric information and local texture features, and then obtaining a registration result using the composite feature vector obtained. The composite feature vector contains comprehensive information from geometric morphology to texture changes. Compared with the prior art, it can ensure that detailed geometric features are fully captured, realize registration at details, and effectively improve the registration accuracy. The registration result is helpful for further optimization of the guide plate design in subsequent steps. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 : A flow chart of an embodiment of a method for aligning a guide plate three-dimensional model with a maxillary sinus structure provided in the present application.
[0058] Figure 2 : A schematic diagram of a three-dimensional model of a guide plate according to an application example of optimizing the alignment results between the three-dimensional model of the guide plate and the maxillary sinus structure provided in the present application.
[0059] Figure 3 : A structural schematic diagram of an embodiment of a guide plate three-dimensional model and maxillary sinus structure alignment device provided in the present application. DETAILED DESCRIPTION
[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0061] Embodiment 1
[0062] Please refer to Figure 1 , Figure 1 The present invention provides a method for aligning a guide plate three-dimensional model with a maxillary sinus structure, comprising steps S1 to S5; wherein:
[0063] Step S1, obtaining a three-dimensional guide plate model and maxillary sinus structure scanning data; extracting the curvature distribution of the three-dimensional guide plate model and the maxillary sinus structure scanning data.
[0064] In this step, an intraoral 3D scanner can be used to scan the patient's dentition and gums to obtain the STL data of the 3D surface, and obtain the user's wide-field CBCT (oral and maxillofacial cone beam CT) and DICOM (Digital Imaging and Communications in Medicine) format data including the maxillary sinus, thereby obtaining the maxillary sinus structure scan data.
[0065] In some preferred embodiments, the above-mentioned oral scan data and DICOM data can be imported into the design software 3 Shape to perform three-dimensional reconstruction of the jaw, mark the three-dimensional structure of the maxillary sinus in the implant area and the three-dimensional boundary of the maxillary sinus mucosa, and mark the three-dimensional boundary of important blood vessels; determine the boundaries of the window opening guide plate on the side wall of the maxillary sinus, including the mesial, distal, and upper and lower boundaries, so as to further design a three-dimensional model of the guide plate.
[0066] The three-dimensional model of the guide plate should ensure that the window boundary avoids the thick part of the bone wall and does not injure the maxillary sinus floor and surrounding nerves and blood vessels.
[0067] In some embodiments, for the three-dimensional guide plate model and the maxillary sinus structure scan data obtained above, extracting the curvature distribution of the three-dimensional guide plate model and the maxillary sinus structure scan data includes:
[0068] The guide plate three-dimensional model and the maxillary sinus structure scanning data are respectively converted into a guide plate surface model and a maxillary sinus surface model, thereby obtaining the coordinate information and surface normal vector information of each point in the guide plate surface model and the maxillary sinus surface model; obtaining the principal curvature of the guide plate surface model and the maxillary sinus surface model; obtaining the curvature distribution of the guide plate three-dimensional model according to the principal curvature of the guide plate surface model and the coordinate information and surface normal vector information of each point; obtaining the curvature distribution of the maxillary sinus structure scanning data according to the principal curvature of the maxillary sinus surface model and the coordinate information and surface normal vector information of each point.
[0069] By implementing the implementation mode of the present application, by calculating the principal curvature of each point, the coordinate information of each point and the surface normal vector information can be combined to generate curvature distribution data with higher accuracy, thereby ensuring the position accuracy of each point and the curvature calculation accuracy. Among them, the principal curvature can be obtained by using a data set of the guide plate three-dimensional model and the maxillary sinus structure scanning data, and adopting a numerical analysis method (such as finite difference) to obtain the local surface morphology of each vertex and then analyze it. Based on information such as the principal curvature, position, and direction (such as the normal vector), the curvature radius can be determined while optimizing the vertex coordinate position information, helping to extract the bending of the model in different directions, as well as the location of the bending, and accurately extract the bending information.
[0070] Further, based on the calculated principal curvature values, Gaussian curvature can be obtained by product calculation, and Gaussian curvature reflects the intrinsic concave-convex properties of the surface at each vertex, which helps to understand the structural stability and stress distribution of the model. In addition, based on the average curvature of two or more vertices, the extrinsic curvature of the surface at each point can also be described, thereby providing information about the overall curvature of the model. The information such as Gaussian curvature and average curvature obtained in the above preferred embodiment can also be used as an aid or reference for registration, which is conducive to assisting the implementation of subsequent steps with higher precision.
[0071] Step S2, according to the curvature distribution of the guide plate three-dimensional model, densely sample the areas where the curvature is greater than a first threshold, and sparsely sample the areas where the curvature is less than or equal to the first threshold, to obtain a first feature point set; according to the curvature distribution of the maxillary sinus structure scanning data, densely sample the areas where the curvature is greater than a second threshold, and sparsely sample the areas where the curvature is less than or equal to the second threshold, to obtain a second feature point set; wherein the first threshold and the second threshold are both preset curvature thresholds.
[0072] In this embodiment, a first threshold and a second threshold may be preset before step S2 to determine intervals of different sampling methods. The first threshold and the second threshold may be the same or different, which is determined specifically according to application scenario requirements.
[0073] In step S2, points above the first threshold or the second threshold can be marked as high curvature points; conversely, points less than or equal to the first threshold or the second threshold can be marked as low curvature points. Dense sampling is performed on the marked high curvature areas, and the number of feature points in the high curvature areas can be increased, which is convenient for capturing more details and obtaining detailed geometric features; sparse sampling is used for the marked low curvature areas, which can reduce the number of feature points in these areas, thereby reducing the complexity of calculations and improving the processing speed of the system and hardware.
[0074] Before collecting the local texture features of the maxillary sinus structure scan data in step S3, the maxillary sinus structure scan data is filtered, specifically:
[0075] Calculate the anisotropy value of each scanning data point in the maxillary sinus structure scanning data; for the area in the maxillary sinus structure scanning data where the anisotropy value is greater than a preset value, use a first smoothing intensity to perform smoothing along the structural direction, and use a second smoothing intensity to perform smoothing along the direction perpendicular to the structure; wherein the first smoothing intensity is greater than the second smoothing intensity; use Neumann boundary conditions and a finite difference method to construct a corresponding partial differential equation model for the smoothed maxillary sinus structure scanning data; calculate the diffusion tensor of each second feature point, apply the diffusion tensor to the partial differential equation model, perform nonlinear anisotropic filtering through multiple iterations, and obtain the filtered maxillary sinus structure scanning data.
[0076] In this embodiment, a relatively strong smoothing is applied along the structural direction, while a relatively weak smoothing is applied perpendicular to the structural direction, so that the retention of the structural edge can be optimized. If there is a high noise level, the noise suppression effect can be improved by adjusting the smoothing strength parameter and increasing the smoothing strength along the structural direction; the diffusion tensor is applied to the partial differential equation model, and the diffusion tensor can be used to guide the filtering direction and strength; the numerical solution method can be selected in the iterative process, and the stability and convergence of the filtering process can be guaranteed by selecting the explicit or implicit format; nonlinear anisotropic filtering is performed through multiple iterations to obtain the filtered, clear and detail-retained maxillary sinus structure scan data.
[0077] Step S3, through multi-scale analysis, extract the local geometric information of each first feature point in the first feature point set, and the local geometric information of each second feature point in the second feature point set; collect the local texture features of the guide plate three-dimensional model and the local texture features of the maxillary sinus structure scanning data.
[0078] In this step, the local geometric information includes local curvature information, local normal vector and Gaussian curvature; the local curvature information can be the curvature radius of a 5 mm grid point, the local normal vector can be obtained by cross product, and the Gaussian curvature can be obtained by local differential geometry calculation. The local texture features include gray level co-occurrence matrix and Sobel operator.
[0079] Specifically, local curvature information, local normal vectors, and Gaussian curvature, which describe local geometric information, can be extracted at different scales to enhance the robustness and adaptability of feature description.
[0080] Step S4, generating a first composite feature vector for each first feature point based on the local geometric information of each first feature point and the local texture features of the guide plate three-dimensional model; generating a second composite feature vector for each second feature point based on the local geometric information of each second feature point and the local texture features of the maxillary sinus structure scanning data.
[0081] In this step, the local texture features and local geometric information obtained in the above steps are cascaded into a high-dimensional vector, and a feature representation containing changes from geometric form to texture can be obtained, namely the first composite feature vector and the second composite feature vector of this step. These high-dimensional vectors provide rich information for subsequent registration and analysis.
[0082] Step S5: performing feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result.
[0083] In this embodiment, the use of the first composite feature vector and the second composite feature vector to perform feature matching to obtain a registration result includes:
[0084] The first composite feature vector is processed by principal component analysis to reduce the dimension, so as to obtain a first reduced dimension feature vector; the second composite feature vector is processed by principal component analysis to reduce the dimension, so as to obtain a second reduced dimension feature vector; a data space is constructed, and the data space is divided into a plurality of sub-regions by a recursive method; corresponding nodes are constructed based on each of the sub-regions; a kd tree data structure is used based on the constructed nodes to perform feature matching operations on the first reduced dimension feature vector and the second reduced dimension feature vector to obtain the registration result.
[0085] In this embodiment, the kd tree is also called kd tree or kd-tree, which is a data structure used in computer science to organize points in k-dimensional space and is a special case of a spatial binary tree. The constructed node adopts this data structure and uses the nearest neighbor search algorithm to find the leaf node where the target point is located; then, based on the distance between the current nearest point and the target point, the adjacent area where there may be closer points is explored in a backtracking manner to determine whether to update the current nearest point; by comparing feature descriptions, potential matching points are obtained in the feature matching process to achieve accurate matching.
[0086] In some preferred embodiments, the principal component analysis method is used to perform dimensionality reduction processing on the first composite feature vector to obtain a first reduced dimensionality feature vector, including:
[0087] Obtain the covariance matrix and total data variance of the first composite eigenvector; obtain the weight of each first principal component in the total data variance of the first composite eigenvector according to the covariance matrix of the first composite eigenvector; screen out multiple first principal components that are greater than a preset weight threshold; respectively obtain the corresponding relationship between each first principal component and the first composite eigenvector to obtain a first expression; query a preset database to select the first expression with the highest matching degree with the preset database; use the selected first expression to simplify the first composite eigenvector to obtain a first reduced-dimensionality eigenvector.
[0088] The principal component analysis method is used to perform dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensional feature vector, including: obtaining the covariance matrix and total data variance of the second composite feature vector; obtaining the weight of each second principal component in the total data variance of the second composite feature vector according to the covariance matrix of the second composite feature vector; screening out multiple second principal components greater than a preset weight threshold; respectively obtaining the corresponding relationship between each second principal component and the second composite feature vector to obtain a second expression; querying a preset database to select the second expression with the highest matching degree with the preset database; using the selected second expression to simplify the second composite feature vector to obtain a second reduced-dimensional feature vector.
[0089] In this preferred embodiment, the covariance matrix can reflect the correlation between each feature. The importance of the principal component in the data set (such as the degree of information retention) can be determined through the proportion or weight of the principal component in the total variance of the data. Multiple principal components greater than the preset weight threshold are screened out, and the correspondence between the principal component and the composite feature vector is used to obtain an expression. The correspondence between the principal component and the composite feature vector can be extracted, which is used for dimensionality reduction of the composite feature vector, thereby improving computing efficiency and data storage performance while retaining important information.
[0090] In addition, the present invention also provides an application example of using the registration method, such as Figure 2 This application example uses the registration result obtained by the registration method to further optimize the three-dimensional model of the guide plate.
[0091] The guide shown in the figure is a short square frame-type maxillary sinus floor external lifting guide. Specifically: it is a solid short square design, and the guide frame is used as the indicator edge when bone removal and window opening. The upper boundary is horizontally parallel to the branch of the superior alveolar artery and is at a safe distance of more than 2mm, which can protect the blood vessels and prevent damage to the blood vessels and cause bleeding during the operation; and its lower boundary is flush with the maxillary sinus floor, the front boundary is flush with the anterior wall of the maxillary sinus, and the rear boundary is positioned corresponding to the most distal planned implant. The height of the window does not exceed 6mm.
[0092] The guide has finer details, can better match the maxillary sinus structure data, and has better registration accuracy.
[0093] Accordingly, refer to Figure 3 The embodiment of the present invention further provides a device for aligning a guide plate three-dimensional model with a maxillary sinus structure, comprising an extraction module 101, a sampling module 102, a collection module 103, a generation module 104 and an alignment module 105; wherein,
[0094] The extraction module 101 is used to obtain the three-dimensional model of the guide plate and the scanning data of the maxillary sinus structure; extract the curvature distribution of the three-dimensional model of the guide plate and the scanning data of the maxillary sinus structure;
[0095] The sampling module 102 is used to perform dense sampling on areas where the curvature is greater than a first threshold value according to the curvature distribution of the guide plate three-dimensional model, and perform sparse sampling on areas where the curvature is less than or equal to the first threshold value, to obtain a first feature point set; according to the curvature distribution of the maxillary sinus structure scanning data, perform dense sampling on areas where the curvature is greater than a second threshold value, and perform sparse sampling on areas where the curvature is less than or equal to the second threshold value, to obtain a second feature point set; wherein the first threshold value and the second threshold value are both preset curvature threshold values;
[0096] The acquisition module 103 is used to extract local geometric information of each first feature point in the first feature point set and local geometric information of each second feature point in the second feature point set through multi-scale analysis; and to acquire local texture features of the guide plate three-dimensional model and local texture features of the maxillary sinus structure scan data;
[0097] The generating module 104 is used to generate a first composite feature vector for each first feature point according to the local geometric information of each first feature point and the local texture feature of the guide plate three-dimensional model; and to generate a second composite feature vector for each second feature point according to the local geometric information of each second feature point and the local texture feature of the maxillary sinus structure scan data;
[0098] The registration module 105 is used to perform feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result.
[0099] As a preferred solution, the registration module 105 uses the first composite feature vector and the second composite feature vector to perform feature matching to obtain a registration result, including:
[0100] The registration module 105 uses the principal component analysis method to perform dimensionality reduction processing on the first composite feature vector to obtain a first reduced-dimensionality feature vector; uses the principal component analysis method to perform dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensionality feature vector;
[0101] Constructing a data space, and dividing the data space into a plurality of sub-areas by a recursive method;
[0102] Corresponding nodes are constructed based on each of the sub-regions; based on the constructed nodes, a kd-tree data structure is used to perform a feature matching operation on the first reduced-dimensionality feature vector and the second reduced-dimensionality feature vector to obtain the registration result.
[0103] As a preferred solution, the registration module 105 uses principal component analysis to perform dimensionality reduction processing on the first composite feature vector to obtain a first reduced dimensionality feature vector, including:
[0104] The registration module 105 obtains the covariance matrix and total data variance of the first composite feature vector; obtains the weight of each first principal component in the total data variance according to the covariance matrix of the first composite feature vector; screens out multiple first principal components that are greater than a preset weight threshold; obtains the corresponding relationship between each first principal component and the first composite feature vector respectively to obtain a first expression; queries a preset database to select the first expression with the highest matching degree with the preset database; uses the selected first expression to simplify the first composite feature vector to obtain a first reduced-dimensional feature vector.
[0105] The registration module 105 uses principal component analysis to perform dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensional feature vector, including: the registration module 105 obtains the covariance matrix and total data variance of the second composite feature vector; obtains the weight of each second principal component in the total data variance according to the covariance matrix of the second composite feature vector; screens out multiple second principal components that are greater than a preset weight threshold; obtains the corresponding relationship between each second principal component and the second composite feature vector respectively to obtain a second expression; queries a preset database to select the second expression with the highest matching degree with the preset database; uses the selected second expression to simplify the second composite feature vector to obtain a second reduced-dimensional feature vector.
[0106] As a preferred solution, the extraction module 101 extracts the curvature distribution of the guide plate three-dimensional model and the maxillary sinus structure scanning data, including:
[0107] The extraction module 101 converts the guide plate three-dimensional model and the maxillary sinus structure scanning data into a guide plate curved surface model and a maxillary sinus curved surface model respectively, and then obtains the coordinate information and surface normal vector information of each point in the guide plate curved surface model and the maxillary sinus curved surface model;
[0108] Obtaining the principal curvatures of the guide plate curved surface model and the maxillary sinus curved surface model;
[0109] According to the principal curvature of the guide plate surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the guide plate three-dimensional model is obtained; according to the principal curvature of the maxillary sinus surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the maxillary sinus structure scanning data is obtained.
[0110] As a preferred solution, the guide plate three-dimensional model and maxillary sinus structure registration device further includes a filtering module, which is used for: before the acquisition module acquires the local texture features of the guide plate three-dimensional model and the local texture features of the maxillary sinus structure scanning data:
[0111] Calculating the anisotropy value of each scanning data point in the maxillary sinus structure scanning data;
[0112] For the area in the maxillary sinus structure scan data where the anisotropy value is greater than a preset value, a first smoothing intensity is used for smoothing along the structure direction, and a second smoothing intensity is used for smoothing along the direction perpendicular to the structure; wherein the first smoothing intensity is greater than the second smoothing intensity;
[0113] Neumann boundary conditions and finite difference method were used to construct the corresponding partial differential equation model for the smoothed maxillary sinus structure scanning data.
[0114] The diffusion tensor of each second feature point is calculated, the diffusion tensor is applied to the partial differential equation model, and nonlinear anisotropic filtering is performed through multiple iterations to obtain the maxillary sinus structure scanning data after filtering.
[0115] As a preferred embodiment, the local geometric information includes local curvature information, local normal vector and Gaussian curvature; the local curvature information is the curvature radius of a 5 mm grid point, the local normal vector is obtained by cross product, and the Gaussian curvature is obtained by local differential geometry calculation.
[0116] As a preferred solution, the local texture features include a gray level co-occurrence matrix and a Sobel operator.
[0117] Correspondingly, the present application also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the method for aligning the guide plate three-dimensional model with the maxillary sinus structure is implemented.
[0118] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal, and various interfaces and lines are used to connect various parts of the entire terminal.
[0119] The memory can be used to store the computer program, and the processor realizes various functions of the terminal by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.
[0120] Correspondingly, the present invention application also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for aligning the guide plate three-dimensional model with the maxillary sinus structure.
[0121] Wherein, if the module integrating the guide plate three-dimensional model and the maxillary sinus structure registration device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.
[0122] Compared with the prior art, the present invention has the following beneficial effects:
[0123] The present invention application provides a method, device, terminal and medium for registration of a guide plate three-dimensional model and a maxillary sinus structure. The registration method comprises: acquiring a guide plate three-dimensional model and maxillary sinus structure scanning data; extracting the curvature distribution of the guide plate three-dimensional model and the maxillary sinus structure scanning data; based on the curvature distribution of the guide plate three-dimensional model, densely sampling the area where the curvature is greater than a first threshold, and sparsely sampling the area where the curvature is less than or equal to the first threshold, to obtain a first feature point set; based on the curvature distribution of the maxillary sinus structure scanning data, densely sampling the area where the curvature is greater than a second threshold, and sparsely sampling the area where the curvature is less than or equal to the second threshold, to obtain a second feature point set; wherein the first threshold and the second threshold are both A preset curvature threshold; extracting local geometric information of each first feature point in the first feature point set and local geometric information of each second feature point in the second feature point set through multi-scale analysis; collecting local texture features of the three-dimensional model of the guide plate and local texture features of the maxillary sinus structure scan data; generating a first composite feature vector of each first feature point based on the local geometric information of each first feature point and the local texture features of the three-dimensional model of the guide plate; generating a second composite feature vector of each second feature point based on the local geometric information of each second feature point and the local texture features of the maxillary sinus structure scan data; performing feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result. The present invention applies a method of sampling the three-dimensional model of the guide plate and the scan data of the maxillary sinus structure based on the curvature distribution, obtaining a feature point set, extracting local geometric information and local texture features, and then obtaining a registration result using the composite feature vector obtained. The composite feature vector contains comprehensive information from geometric morphology to texture changes. Compared with the prior art, it can ensure that detailed geometric features are fully captured, realize registration at details, and effectively improve the registration accuracy. The registration result is helpful for further optimization of the guide plate design in subsequent steps.
[0124] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.
Claims
1. A method for registering a guide plate three-dimensional model with a maxillary sinus structure, characterized in that: include: Acquire a three-dimensional model of the guide plate and scan data of the maxillary sinus structure; extract the curvature distribution of the three-dimensional model of the guide plate and the scan data of the maxillary sinus structure; According to the curvature distribution of the guide plate three-dimensional model, dense sampling is performed on the area where the curvature is greater than a first threshold, and sparse sampling is performed on the area where the curvature is less than or equal to the first threshold, to obtain a first feature point set; According to the curvature distribution of the maxillary sinus structure scan data, dense sampling is performed on the area where the curvature is greater than the second threshold, and sparse sampling is performed on the area where the curvature is less than or equal to the second threshold, to obtain a second feature point set; wherein the first threshold and the second threshold are both preset curvature thresholds; Extracting local geometric information of each first feature point in the first feature point set and local geometric information of each second feature point in the second feature point set through multi-scale analysis; collecting local texture features of the guide plate three-dimensional model and local texture features of the maxillary sinus structure scan data; Generate a first composite feature vector for each first feature point based on the local geometric information of each first feature point and the local texture features of the guide plate three-dimensional model; generate a second composite feature vector for each second feature point based on the local geometric information of each second feature point and the local texture features of the maxillary sinus structure scan data; Performing feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result; The extracting of the curvature distribution of the guide plate three-dimensional model and the maxillary sinus structure scanning data includes: The three-dimensional model of the guide plate and the scanning data of the maxillary sinus structure are converted into a guide plate curved surface model and a maxillary sinus curved surface model respectively, and then the coordinate information and surface normal vector information of each point in the guide plate curved surface model and the maxillary sinus curved surface model are obtained; Obtaining the principal curvatures of the guide plate curved surface model and the maxillary sinus curved surface model; According to the principal curvature of the guide plate surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the guide plate three-dimensional model is obtained; according to the principal curvature of the maxillary sinus surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the maxillary sinus structure scanning data is obtained.
2. A method for aligning a guide plate three-dimensional model with a maxillary sinus structure as claimed in claim 1, characterized in that: The using the first composite feature vector and the second composite feature vector to perform feature matching to obtain a registration result includes: Using a principal component analysis method, performing dimensionality reduction processing on the first composite feature vector to obtain a first reduced-dimensionality feature vector; using a principal component analysis method, performing dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensionality feature vector; Constructing a data space, and dividing the data space into a plurality of sub-areas by a recursive method; Corresponding nodes are constructed based on each of the sub-regions; based on the constructed nodes, a kd-tree data structure is used to perform a feature matching operation on the first reduced-dimensionality feature vector and the second reduced-dimensionality feature vector to obtain the registration result.
3. A method for aligning a guide plate three-dimensional model with a maxillary sinus structure as claimed in claim 1, characterized in that: Before acquiring the local texture features of the guide plate three-dimensional model and the local texture features of the maxillary sinus structure scanning data, the method further includes: Calculating the anisotropy value of each scanning data point in the maxillary sinus structure scanning data; For the area in the maxillary sinus structure scan data where the anisotropy value is greater than a preset value, a first smoothing intensity is used for smoothing along the structure direction, and a second smoothing intensity is used for smoothing along the direction perpendicular to the structure; wherein the first smoothing intensity is greater than the second smoothing intensity; Neumann boundary conditions and finite difference method were used to construct the corresponding partial differential equation model for the smoothed maxillary sinus structure scanning data. The diffusion tensor of each second feature point is calculated, the diffusion tensor is applied to the partial differential equation model, and nonlinear anisotropic filtering is performed through multiple iterations to obtain the maxillary sinus structure scanning data after filtering.
4. A method for aligning a guide plate three-dimensional model with a maxillary sinus structure as claimed in claim 1, characterized in that: The local geometric information includes local curvature information, local normal vector and Gaussian curvature; the local curvature information is the curvature radius of a 5 mm grid point, the local normal vector is obtained by cross product, and the Gaussian curvature is obtained by local differential geometry calculation.
5. A method for aligning a guide plate three-dimensional model with a maxillary sinus structure as claimed in claim 1, characterized in that: The local texture features include gray level co-occurrence matrix and Sobel operator.
6. A device for registering a guide plate three-dimensional model with a maxillary sinus structure, characterized in that: It includes extraction module, sampling module, acquisition module, generation module and registration module; among which, The extraction module is used to obtain the three-dimensional model of the guide plate and the scanning data of the maxillary sinus structure; extract the curvature distribution of the three-dimensional model of the guide plate and the scanning data of the maxillary sinus structure; The sampling module is used to perform dense sampling on areas where the curvature is greater than a first threshold value and perform sparse sampling on areas where the curvature is less than or equal to the first threshold value according to the curvature distribution of the guide plate three-dimensional model, so as to obtain a first feature point set; and perform dense sampling on areas where the curvature is greater than a second threshold value and perform sparse sampling on areas where the curvature is less than or equal to the second threshold value according to the curvature distribution of the maxillary sinus structure scanning data, so as to obtain a second feature point set; wherein the first threshold value and the second threshold value are both preset curvature threshold values; The acquisition module is used to extract local geometric information of each first feature point in the first feature point set and local geometric information of each second feature point in the second feature point set through multi-scale analysis; and to acquire local texture features of the guide plate three-dimensional model and local texture features of the maxillary sinus structure scan data; The generating module is used to generate a first composite feature vector of each first feature point according to the local geometric information of each first feature point and the local texture feature of the guide plate three-dimensional model; and to generate a second composite feature vector of each second feature point according to the local geometric information of each second feature point and the local texture feature of the maxillary sinus structure scanning data; The registration module is used to perform feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result; The extraction module extracts the curvature distribution of the guide plate three-dimensional model and the maxillary sinus structure scanning data, including: The extraction module converts the guide plate three-dimensional model and the maxillary sinus structure scanning data into a guide plate curved surface model and a maxillary sinus curved surface model respectively, and then obtains the coordinate information and surface normal vector information of each point in the guide plate curved surface model and the maxillary sinus curved surface model; Obtaining the principal curvatures of the guide plate curved surface model and the maxillary sinus curved surface model; According to the principal curvature of the guide plate surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the guide plate three-dimensional model is obtained; according to the principal curvature of the maxillary sinus surface model and the coordinate information and surface normal vector information of each point, the curvature distribution of the maxillary sinus structure scanning data is obtained.
7. A device for aligning a guide plate three-dimensional model with a maxillary sinus structure as claimed in claim 6, characterized in that: The registration module performs feature matching using the first composite feature vector and the second composite feature vector to obtain a registration result, including: The registration module uses a principal component analysis method to perform dimensionality reduction processing on the first composite feature vector to obtain a first reduced-dimensionality feature vector; and uses a principal component analysis method to perform dimensionality reduction processing on the second composite feature vector to obtain a second reduced-dimensionality feature vector; Constructing a data space, and dividing the data space into a plurality of sub-areas by a recursive method; Corresponding nodes are constructed based on each of the sub-regions; based on the constructed nodes, a kd-tree data structure is used to perform a feature matching operation on the first reduced-dimensionality feature vector and the second reduced-dimensionality feature vector to obtain the registration result.
8. A terminal device, characterized in that: The method comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the method for aligning the three-dimensional model of the guide plate with the maxillary sinus structure as described in any one of claims 1 to 5 when executing the computer program.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for aligning the guide plate three-dimensional model with the maxillary sinus structure as described in any one of claims 1 to 5.